Investing in The Modern Age - Rachel E S Ziemba

Investing
in theAge
Modern
Investing
in the
Modern Age
Rachel E. S. Ziemba
Roubini Global Economics, UK
William T. Ziemba
University of British Columbia, Canada
World Scientific
NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TAIPEI • CHENNAI
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
Library of Congress Cataloging-in-Publication Data

Ziemba, Rachel.
Investing in the modern age / Rachel E.S. Ziemba (Roubini Global Economics, UK) and William
T. Ziemba (University of British Columbia, Canada).
pages cm
Includes index.
ISBN 978-9814518833 (hardcover : alk. paper)
ISBN 978-9814504744 (softcover : alk. paper)
1. Investments. 2. Investments, Foreign. 3. Risk management. 4. Investment analysis. I. Ziemba,
W. T. II. Title.
HG4521.Z54 2013
332.6--dc23
2013011472
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.
Copyright © 2013 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and
retrieval system now known or to be invented, without written permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
In-house Editor: Chye Shu Wen
Printed in Singapore
Dedicated to Sandra L Schwartz, our third family member, for
encouragement, advice and help in all stages in the preparation of this book
Acknowledgements
We owe a great debt to Paul Wilmott for starting his wonderful magazine
Wilmott. Since the magazine's launch in September 2002, Bill has had a
regular column with Rachel guest writing her own or joint columns on a
range of global financial issues. Paul and Wiley also kindly gave us access
to the beautiful artwork of the magazine for use in this book which uses
columns from mid 2007 to 2012 with a few others from earlier and some
new chapters added for completeness. Our earlier book, Scenarios for Risk
Management and Global Investment Strategies, Wiley, 2007, used columns
from 2002 to early 2007, introducing some of the key themes we discuss in
this volume. All chapters have been edited and updated to September 30,
2012.
Special thanks go to Dan Tudball, executive editor of Wilmott and Beth
Gongde of Wiley for help in turning our draft columns into professional
articles in the magazine. Liam Larkin who does the artwork for the
magazine was very helpful in not only producing the original columns but
also in providing them for our use in producing this book. Thanks go to the
World Scientific staff including Zvi Ruder and Max Phua for providing us
the opportunity to revisit this material in this book and for producing this
volume.
Bill thanks the various organizations and universities who invited him to
give talks on these subjects over the years and his coauthors who are cited
in the text. These include talks at Oxford University, Imperial College,
ICMA Centre of the University of Reading, University of Cyprus, KAIST
in Korea, 7City-Wilmott, Stanford University, Massachusetts Institute of
Technology, University of Washington, Fidelity Investments, Canyon
Investment Advisors, RAB Hedge Fund, Chicago Quantitative Finance,
Financial Management Association, European Financial Management
Association, Renselalear Polytechnic Institute, Sabanci University, Luis
Guido Carli University and The Reims Management School.
Rachel would like to thank a number of colleagues and mentors at
Roubini Global Economics, particularly Nouriel Roubini, and Christian
Menegatti, for supporting work that inspired many of the items in this book.
Brad Setser introduced me to the art of sovereign balance sheet dissection
that has framed my work on sovereign wealth funds. Questions from RGE
colleagues, most notably Natalia Gurushina Gina Sanchez, Adam Wolfe
and Maya Senussi as well as clients poked holes in some of my arguments
and made them stronger. Thanks go to participants at talks at the New York
University, Princeton University, London School of Economics, Oxford
University, University of Edinburgh, CSIS, Chatham House, International
Institute of Strategic Studies among others, where some of the ideas where
first aired. I do not have space to list here the many friends who supported
this work, either by reading draft pieces, helping to arrange meetings,
providing places to stay or encouraging my ideas.
As with our last book, Sandra Schwartz has not only gone above and
beyond to produce this book and the columns but has been a research
colleague and sounding board on all the topics. Much of the quality of this
book is due to her efforts with none of the mistakes. Without her this book
would not have been possible.
Contents
Acknowledgements
Preface
I Key Concepts
1. Arbitrage, Risk Arbitrage and the Favorite-longshot Bias

2. The Bond Stock Earnings Yield Differential Model
3. Investor Camps
II Hedge Funds, Sovereign Wealth Funds and Other Investment

Agglomerations
4. Average Hedge Funds and their Evaluation

5. Incentives and Risk Taking in Hedge Funds
6. Evaluating Superior Hedge Funds
7. Investment in Own-Company Stock
8. Cutting Through the Hype on Sovereign Wealth Funds
9. A New Age for Liquidity
10. Government Owned Pensions: Asset Allocation and Governance
Issues
11. Update on Yale's Approach to Endowment Investing
12. A Risk Arbitrage Convergence Trade: The Nikkei Put Warrant
Market of 1989-90
13. Kelly Capital Growth Investing
14. InnoALM, the Innovest Austrian Pension Fund Financial Planning
Model
III Seasonal Effects and Other Anomalies
15. Investing in the January Turn-of-the-Year Effect with Index Futures

16. The January Barometer
17. Sell in May and Go Away and the Effect of the Fed
18. 60-40 Pension Fund Mixes and Presidential Party Effects
IV Volatility, Correlation and Liquidity
19. Thoughts on the VIX Fear Index

20. Changing Correlations: Rising VIX and Violent Market Moves
V Can We Predict Stock Market Crashes?
21. Stock Market Crashes in 2006-2009: Were We Able to Predict Them?

22. Three Mini Crashes in US and World Equity Markets
23. What Signals Worked and What Did Not, 1980-2009
24. What Signals Worked and What Did Not, 1980-2009, Part II
25. What Signals Worked and What Did Not, 1980-2009, Part III
26. How to Lose Money in Derivatives and Examples of Those Who Did
VI Bubbles and Debt

27. Understanding the Financial Markets in the Subprime Era: The
2007/9 Crisis
28. Bubbles
29. China: Navigating the Olympic Risks
30. Turkey's Juggling Act: Can it Live up to Potential?
31. Testing Resiliency: Protest and Natural Disasters
32. It's a Gas, Gas, Gas!
33. Thoughts on the Current Market Environment, Risks and Returns
34. What's Wrong with the US?
35. Investing Around the World
VII Investing and Arbitrage in NFL Football and Horse Racing
36. Blunder or Correct Decision? The Belichick Decision to go for it on

4th Down
37. The 2010 and 2011 Super Bowls and the Elo Ranking System
38. Risk Arbitrage in the NFL 2012 Playoffs and the Super Bowl
39. The One That Got Away: The Hitable $2 Million Pick 6 at the
Breeders' Cup
40. Two Super Horses
41. Farewell to the Queen and to the Princess of US Thoroughbred
Racing
42. The Dr. Z Place and Show Racetrack Betting Systems Past and
Present
Bibliography
About the Authors
Index
Preface
This book discusses many key topics in investment and risk management,
global economic situation and global investment strategies. The material in
this book was largely written over the period of 2007–12, one of the most
tumultuous times in global financial markets which called into question not
only tenets of economic forecasting and also asset allocation and return
strategies. We attempt to draw some conclusions on how the structural
shifts currently underway in the global economy as well as cyclical trends
will affect these industries, the globe and key sectors.
The various topics and case studies are relevant for hedge fund,
insurance, pension fund, mutual fund managers and other investment
professionals and investors. The book presents tools and case studies of real
applications for analyzing a wide variety of investment returns and better
assessing the risks which many investors have preferred to ignore in the
search of returns. Many security market regularities or anomalies are
discussed as is the process of building scenarios and using strategies to
optimize returns. All the data in the book is current for the topics and dates
discussed with updates to our publication date of September 30, 2012.
We are guided by the use of careful analysis to generate future scenarios.
That is, what are the chances that various future events will occur over
time? How should these events and their chances influence investment
decisions? Assessing all possible outcomes is fundamental to risk
management, financial engineering and investment strategies. Careful
consideration of future scenarios in the macro outlook and micro level leads
to better investment decisions and avoids financial disasters. We remain in a
period where global growth is likely to be weak, and upside risks delivered
by coordinated, concentrated policy responses - This adds to the complexity
of anomalies and pricing in the markets.
Learning From Disasters
A key contribution of this book is both positive and negative case studies
of great successes and blowouts to better assess explicit and implicit risks
and mismatches in maturities and investment horizon. We discuss strategies
used by the greatest investors to obtain their high returns. To do so, we
analyze hedge fund concepts and performance including major fund
disasters. Chapter 26 begins with a discussion of various ways traders lose
money trading derivatives. These pitfalls remind one of what not to do to
try to have successful trading. Some major disasters discussed are the Long
Term Capital failure in 1998, an imported fear driven crash in 1997 that
sunk the Niederhoffer fund, the 2006 Amaranth natural gas disaster and
Société Générale. These failures all involve over betting and not being
diversified in all scenarios.
These disasters have basic common characteristics, which we call the
recipe for disaster. This negative outcome materializes when one overbets,
that is, has too many positions relative to one's capital (including capital
that can reliably be called upon in times of need) and one is not diversified.
By diversified we do not mean that we are mean-variance diversified in
average times. Rather diversification in all the possible scenarios is crucial.
This means that low probability scenarios cannot, as in typically done, be
assigned probability zero, that is disregarded. Also, and very crucially,
scenario dependent correlation matrices are needed, so that one knows how
assets correlate in a scenario that actually occurs. Simulations of past
average correlations, which are typically done even with stress testing, are
insufficient for full protection. Given the increasing role of political actors,
including central banks, and greater financial linkages across countries -
correlations have risen across a variety of global asset classes, meaning that
investors have a greater need to differentiate the current investment regime,
and making sure they are really diversified, rather than exposing themselves
to a series of related risks in seemingly uncorrelated assets classes.
For example, equity correlations rise and equity/bond correlations fall
with a decline in equity prices. However, when one is in a crash mode, then
equity prices are falling but bond prices are generally rising so that the
bond/stock correlation is negative. The bond/stock correlation is frequently
negative even though most of the time this correlation is positive. Chapter
14 provides an example of these scenario dependent correlation matrices
used in a model that has been used across Austria since 2000 for the
Siemen's Austria and other pension plans and by regulators. Pension plans
are less levered or not levered at all as compared to hedge funds.
If one overbets, and is not diversified in all plausible scenarios then two
things can happen. One can be lucky and avoid a bad scenario, racking up
excess profits and receiving large fees. Indeed many hedge funds and
traders typically do this as there are huge rewards for making such profits
and an adverse outcome is statistically less likely. In fact, even if traders
know about the dangers, it may be optimal for them to just up the bets to
make more fees assuming that a bad scenario will not occur or if it occurs,
it is so far in the distant future that previous profits and fees will exceed the
disastrous negative results. However, if one is over bet and not diversified
and a bad scenario hits, then one can lose a large percent of one's wealth.
It is worth briefly summarizing the experiences of several big blowouts
which will be addressed in more detail in the book below. All have at their
heart several similar traits, even if the exact circumstances are different. For
LTCM the bad scenario was subtle, namely investor confidence. They were
greatly overbet but reasonably well diversified except in this confidence
failure when all the correlations rose so that they lost in all of their
investments. Option volatilities rose, bond yields rose and this led to their
95%, $4 billion loss in August and September 1998. LTCM was loaded
with talent but their risk control based on value at risk plus simulations
around one average correlation matrix was insufficient to protect them. At
the end of 1997, they returned $2.8 billion to investors hoping to boost their
returns. The lack of liquidity, after returning money to their investors, was
crucial it left them exposed after the market move following Russia's
currency devaluation in August 1998.
The October 1997 imported crash caused a 7% fall on the 27th and -3%
on the morning of the 28th in the S&P futures which caused the
Niederhoffer fund to fail. There the positions were just too many and not
diversified, namely, short out of the money S&P500 futures puts.
Regrettably, the market returned to its initial value by the end of the week
and the puts went to zero a few weeks later. So the 10% fall on Monday and
Tuesday was recovered by Friday. This underscores the importance of
having ample liquidity and capital sufficient to weather big moves in the
markets. Amaranth, in 2006 had typical disaster common elements: a trader
way overbet in one market, natural gas, and lost $6 billion when the gas
prices fell from $7 to $5. Of course, natural gas prices fluctuated wildly in
the past few years ranging from $2 to $11, due to structural changes so such
a drop is not surprising. Chapter 32 has a recent update on the natural gas
market and the decoupling of U.S. markets from global markets - a trend
which has started to emerge also in the oil markets. That the largest most
liquid resource market (U.S.), with the most timely demand and supply data
is increasingly decoupled from global markets because of infrastructure
bottlenecks and a different demand response, poses risks of mispricing and
uncertainty in global markets. Uncertainties of demand in China and other
EM in particular, as well as hard to forecast geopolitical risks, risk keeping
the oil price is a wide but high band that could dent global growth and
appetite.
This book is organized into seven sections plus a data appendix. Part I
has key basic concepts. This includes arbitrage, risk arbitrage and the
favorite longshot bias. By arbitrage, we mean buy something and sell it at
the same time for more to generate riskless profits. Besides being basic for
understanding, searching for arbitrage opportunities is profitable. In risk
arbitrage there are also generally two sides - the buy and the sell - and the
goal is to create an arbitrage. Since the matchup might not be possible risk
is involved. Mean reversion of asset prices are crucial in some risk-arb
applications such as NFL football betting as shown in Chapter 38 for the
2012 playoffs and the Super Bowl. Another type of risk-arb is mispriced
option prices for various reasons and Chapter 12 describes a very successful
trade on Nikkei put warrants that one of us was involved with in 1990.
There expensive warrants were sold and cheaper ones were bought and the
trade was closed when the prices converged. The Japanese stock market in
1989 was overvalued but held up with low interest rates. Nikkei put
warrants (three year puts) appeared in Canadian and later in US markets.
The Canadian puts traded for prices substantially above fair value based on
historical volatility. Hence, a risk arbitrage trade was to short the Canadian
overpriced warrants while simultaneously hedging by buying fairly priced
US puts. Both sides do not constitute an arbitrage as the two sides were not
the exact same product but one that was close. Then one waits until the two
sides converge within a transaction band. A second successful trade with
both puts on the same (the American stock exchange) was based simply on
the size of the contract. This trade, similar to the small firm advantage
effect, was to buy cheap puts worth half a Nikkei and to short 2.5 times as
many expensive puts worth 0.2 of a Nikkei. then in about a month, the
prices converged. These types of trades are very good for hedge funds but
are not without various risks.
An important concept is the perception of probabilities. Low probability
events tend to be overestimated and high probability events underestimated.
This favoritelongshot bias, discussed in Chapter 1, is useful in many
applications.
The second key concept chapter introduces the bond-stock earnings yield
model which has been very useful in predicting stock market crashes. It is
known that stock prices tend to fall when the PE ratios are too high but
when? The BSEYD model provides an answer, namely that when bond
interest rates are much above stock earnings yield by enough the stock
market enters a phase that leads almost always to a 10%+ decline from the
initial signal time value. This was discovered by the second author in Japan
in 1988-89 by considering the 1987 US stock market crash, and tested
backwards on Japanese data from 1948-88 (12/12 successes for the model
out of 20 actual 10% crashes) and forward in Japan, the US and other
countries since then. The usual situation is the model moves into the danger
zone, then it rallies more, then it crashes more than 10% below the start.
Chapter 21 discusses the use of the model during the 2006–9 period and it
called the crashes in China, Iceland and the US. In Chapter 2, the BSEYD
model is used for long term investing by suggesting when to be in the
market and when to be in cash. From 1985-2005 and 1980-2005 the final
wealth of the suggested strategy is about double buy and hold for the five
countries (US, Japan, UK, Canada and Germany) with lower risk. The
behavior of various investors is explained by separating investors into five
distinct camps as discussed in Chapter 3.
Part II discusses hedge funds, sovereign wealth funds and other
investment ag-glomerations. We begin in Chapter 4 with the study of
average hedge funds and the impact of the managers stake in the fund and
the incentive structures. We find in a continuous time theoretical model
using realistic data and a prospect theory type utility function (losses are
more bad then gains are good) that:
(1) risk decreases as the manager has more of the fund's money as his
investment;
(2) at 30%+ for the manager's stake, the risk is close to that of a 100%
stake in the fund;
(3) risk increases as the incentive fees increase; and
(4) the value of the option on other people's money varies from about
18% to close to zero as the manager's stake increases from zero to
30%+.
Chapter 5 presents an empirical study of a large number of actual live
and dead hedge funds. The results indicate that it is hard to make back the
average 2% management and 20% incentive fees and that funds of funds
have higher returns the more risk they take.
In Chapter 6 we propose a modification of the standard Sharpe ratio to
evaluate great investors. The Sharpe ratio, which is based on normal
distributions, penalizes large gains as well as losses, as well as its large
gains. To avoid this we propose a measure that is based solely on losses.
Investors with high mean returns and few monthly losses score well on this
measure. Berkshire Hathaway improves with this measure to about 0.90 but
the Ford Foundation, Harvard, the Quantum Fund (George Soros) and the
Windsor fund do not. Still Ford and Harvard beat Berkfshire Hathaway
slightly because Berkshire Hathaway had too many losses. By this measure,
Thorp was at 13.8 for his Princeton-Newport hedge fund with only three
monthly losses in 240 months from 1969–88. Renaissance Medallion the
pioneer of high frequency trading at 26.4 is even higher. These funds and
some others, with even higher DSSRs from the University of Massachusetts
hedge fund data base, are studied.
Chapter 7 discusses the common practice that employees of companies
hold more of their own company stock in their portfolios than the theory
predicts and is advisable. These high holdings can only be justified by high
expected value estimates of the own company stock and low risk aversion
Sovereign wealth funds and government owned assets are discussed in
Chapter 8 with their relatives, government pension funds tackled in Chapter
10. While government investment funds are not new (some of the first SWF
were launched early in the 20th century, their proliferation, the scale of
assets under management and the diversity in asset allocation warrants
deeper study. These funds invest largely for economic and financial reasons,
despite the fears of their detractors, but over the longer horizon political
pressures from domestic constituencies constrain their asset allocation and
mean that several find it difficult to be as long-term investors as they would
like. In fact, one take away from the global boom and bust cycle has been
that investment horizons have tended to shorten, particularly in public
markets which are increasingly buffeted by many short-term trading trends.
We update our analysis on sovereign funds, which now manage around
US$3 trillion in assets, largely in oil exporting countries amplifying the
assets managed by global central banks, which is nearing US$10 trillion.
Chapter 11 discusses top US university endowments including Yale
which had excellent returns, about 13% for the past twenty years. Harvard
has a similar good record, up about 12% over the past twenty years. They
do this with very careful analysis in their own internal trading plus good
relationships with excellent outside private placement, real assets such as
real estate and timber and hedge funds. The proportion of typical exchange
traded equities and bonds has declined to be only a small part of the
portfolio.
Chapter 13 focusses on the Kelly criterion which is used by many great
investors to achieve high returns. The Kelly or capital growth criterion-
fortune's formula is the maximization of long run asymptotic wealth and the
minimization of the time to sufficiently large goals. This is equivalent to
maximizing the expected logarithm of final wealth period by period. If you
plan forever, the Kelly investor will not only get the most final wealth but
get all the wealth. In the short run, with its almost zero Arrow-Pratt risk
aversion, log is the most risky utility function one would ever want to use.
If the data is uncertain, it is very easy to overbet. Also, even in the long but
not infinite run, one can still lose a lot of money with the Kelly criterion. A
simulation shows that over 700 independent bets, all with a 14% advantage,
that most of the time the Kelly criterion will generate very high final
wealth. But in a small percent of the time, a sequence of bad scenarios
coupled with the large bets the Kelly criterion recommends, leads to
enormous losses. Hence, one should use the criterion carefully and consider
fractional Kelly strategies which blend the Kelly optimal portfolio with
cash. These strategies provide more security but have less growth. Great
investors such as Keynes in the 1920s–40s, Warren Buffett of Berkshire
Hathaway, and George Soros of the Quantum Fund have acted as if they
were Kelly bettors. Other, more mathematical types such as Ed Thorp and
Jim Simons have used the Kelly criterion along with investment strategies
with positive means to make hundreds of millions in hedge fund investment
vehicles. Bill Benter and other syndicates have used the Kelly and
fractional Kelly strategies successfully in race track betting.
Part III discusses seasonal effects such as investing in the January turn-
of-the- year effect in the futures markets, the January barometer and sell-in-
May and go away as well as powerful effects from the Fed and presidential
party effects. Chapter 15 discusses a trading strategy that goes long a small
cap index and short a large cap index to attempt to capture the January
small firm effect. The effect has been successfully tradeble but has moved
to December. Chapter 16 reviews January barometer signals which show
that if January is positive then most of the time the rest of the year is
positive but if January is negative the rest of the year is usually noise going
up or down about 50% of the time.
Chapter 17 discusses the sell-in-May and go away anomaly. Historically,
September and October have had low returns on average and the months of
November to April have had high returns, on average. Data from 1993 to
2011 show that the strategy be in the stock market from the turn-of-the-
month of November to the beginning of May and in cash, the rest of the
year greatly outperforms a buy and hold strategy. The effect is strongest for
small cap stocks as are most anomalies. Another effect occurs around the
announcement of the Fed meetings. The data show that most of the gains in
the stock market since 1993 to 2011 have occurred in the three day window
around these meetings. Chapter 18 shows that there is a strong presidential
party effect. The strategies go long small cap stocks with Democrats and
long either large cap stocks or intermediate bonds with Republicans has
greatly outperformed 60-40 bond stock policies as well as small and large
cap stocks.
Part IV discusses volatility, correlation and liquidity.
Chapter 20 discusses changing correlations, rising VIX and violent
market moves. This is in the context of the 2008 world economic and
financial crisis. Subprime loans and the growth of world wide derivative
exposure are discussed as well as impacts on commodities such as oil and
gold and currencies.
Chapter 9 on a new age for liquidity discusses sovereign wealth funds
and other government pools of capital. Their liquidity is less than
commonly assumed and some had had substantial losses at a time when
these governments need more liquidity.In particular, for resource rich
countries, the correlations between investment returns and resource returns
remains high, calling into question the ability to really diversify their
economic and financial portfolios.
Part V deals with the predictability of stock market crashes, the accuracy
of signals, how to lose money trading in derivatives and some important
case studies of huge hedge fund blowups. Many crashes seem predictable
by the bond-stock earnings yield model and other signals. This is discussed
in Chapter 21 which focuses on the Iceland, China and US 2006–09 crashes
which the model called. In Chapter 22 we discuss three mini (less than
10%) crashes that the models did not seem to predict. One of them was the
September 11, 2001 attack of the World Trade Centre. An option based
behavioral bias crash prediction model and its results are discussed. This
was useful to call four major crashes from 1986 to 2003 that together lost
over 40% of the value of the S&P500. This chapter also discusses Chinese
investment markets. Chapters 23–25 assess what signals worked and what
did not to predict crashes in equities and other assets due 1980–2009. The
US subprime crisis and how it evolved is discussed. The Chinese and
Japanese markets are updated. Chapter 26 discusses how to lose money in
derivatives and examines a number of cases including LTCM, Niederhoffer
and Amarath..
Part VI discusses bubbles and debt and related investment topics.
Chapter 27 attempts to understand the financial markets in the subprime
era focusing on the 2007–09 world wide crisis. Various trading disasters are
discussed including Sumitomo, Barings, Daiwa and the Allied Irish Banks.
Chapter 28 continues the 2009 story discussing the rise from the March 6
low of 676.53 in the S&P500. Cheap money and the Fed, low prices,
supposedly over priced bonds, good earnings and portfolio catch ups were
part of the recovery.
Chapters 29–31 highlight a series of resilience of key countries and the
global economy to key economic political and resource stresses. These
include the Chinese growth model on the verge of the financial crisis, the
effects of the Arab Spring and Japanese tsunami in the global oil markets
and the Turkish policy response to steer the economy towards a soft
landing. Chapter 32 studies the dynamics of natural gas to see what might
be necessary to bring new discoveries to market and to use their proceeds.
Chapter 33 discusses market events that undermined the fragile economic
and fi-nancial trajectory in April 2011 including the Icelandic volcano, SEC
charges against Goldman Sachs and the linked sovereign, financial, fiscal
and economic crisis of the PIIGS especially Greece and the flash crash. The
reliance of the global economy on stimulus in the face of rebalancing and
rebuilding of balance sheets, as well as the increased correlation across
asset classes, leaves the global economy and markets vulnerable to these
shocks.
Chapter 34 discusses the current US political and economic situation,
which is increasingly gridlocked with some suggestions for improvement.
We see little prospect for improvement after the 2012 elections despite
Obama being re-elected for a second term; but we are hopeful.
Chapter 35 presents country studies concerning Korea, India, Russia and
Cyprus, countries that are facing a series of challenges in the fight to attract
global capital and integrate into global supply chains. In all cases, some of
the growth drivers that stabilized these economies in the recent past have
weakened.
Part VII discusses investing and arbitrage in NFL football and horse
racing.
Chapter 36 discusses a common decision problem in football. That is
punt on fourth down or go for the first down with short yardage. The
context was in a crucial game between top teams Indianapolis and New
England near the end of the game. A correct analysis is to estimate the
chance of winning the game with the two actions and pick the best one. We
see that famed New England coach Bill Belichick made the right decision
even though his team lost the game, while the TV commentators who do
not do such analysis thought otherwise and criticized him.
Chapters 37 and 38 discuss the 2010, 2011 and 2012 Super bowls and the
accuracy of the Elo ranking system. That approach ranks teams by the
scores of games they played adjusting for the home bias. In betting on
Betfair, for example, a valuable risk arbitrage strategy is to rely on mean
reversion to lock in prices as the scores change. Chapter 38 especially
shows some interesting examples. The team that was originally the best can
lose but you the bettor can win by locking in these risk arbitrages
Chapters 39–42 discuss horse racing betting strategies and applications.
In Chapter 39, a very hittable nearly $2 million Pick 6 at the Breeders' Cup
is described. Then Chapters 40 and 41 focus on the great female horses
Rachel Alexandra and Zenyatta as well as Goldikova, the only three time
Breeders' Cup winner. Finally Chapter 42 discuses betting with the Dr Z
system for place and show at the first Breeders' Cup in 1984.
Most of the chapters in this book have appeared in Wilmott magazine. We
have edited all the chapters for consistency, duplication, updates and other
necessary changes. Also some new chapters have been added to make the
book more complete. The Wilmott columns begin where our previous book
Ziemba and Ziemba (2007) left off in mid 2007. In some cases earlier
columns are presented here where updates plus the original story is
warranted. So between them, this book and the 2007 book have updated
complete collections of our 2002–2012 columns plus some later ones in
Wilmott magazine since its first issue in September 2002.
Rachel wrote her columns (Chapters 8, 9, 10, 29, 30, 31, 32) and the joint
ones (Chapters 11, 22) independent of her current employment with
Roubini Global Economics in New York and London, although these
experiences clearly influence the work, as noted in the acknowledgments.
All mistakes and opinions are our own.
Rachel E S Ziemba
London
and
William T Ziemba
Vancouver
May 2013
PART I
Key Concepts
Arbitrage, Risk Arbitrage and the Favorite-
longshot Bias1
Arbitrage in its pure form is to construct a sure bet so that no matter the
outcome, you either break even or turn a profit. Most observers say that
arbitrage does not exist but actually it does in many instances because there
are different financial markets for the same sports betting or financial
security situation. This is because different people have different sets of
information and beliefs. Lets take the simplest case: either A or B wins. Let
Oah be the odds given on the event that A wins from betting source h and
Obi that for 13 winning from betting source i. With these; UK odds, the odds
are the total return per 1 unitbet. In (comparison, US odds are UK odds -1
and their payoffs are US odds +1 = UK odds.
Pure arbitrage exist when

This is utility free and holds for all utility functions. Risk arbitrage, as
discussed in the New Orleans-San Francisco game and the 2012 Super Bowl
in Chapter 38, is to start with a bet on A and then get a B to satisfy the pure
arbitrage condition. A reference on this topic is Lane and Ziemba (2008)
which is reprinted in Hausch and Ziemba (2008), Handbook of Sports and
Lottos. This book which is available on Amazon etc has papers on many
sports betting studies plus soccer and lotteries.
Chapter 38, discusses risk arbitrage, where you go long A and short B
where B is similar but not identical to A. This involves the key topic of
hedging that is locking in profits. This is especially crucial in sports betting
where substantial leads can fizzle quickly.
For an almost example let's consider the 2009 Breeders' Cup Classic $5
million race held at Santa Anita, November 7, 2009. This race is discussed
in Chapter 39. The classic is the world's most important and most difficult
race with the most talented runners. The European favorite was Rip van
Winkle who three times came close to beating European super star See the
Stars and then won two races just before the Breeders' Cup. His trainer is
Aiden O'Brien and is a Coolmore owned horse. Sea the Stars was considered
a horse of a lifetime who at 3 in 2009 won the 2000 Guineas, the Epson
Derby and the Prix de l"Arc de Triomphe. No horse had ever done this. But
alas, Sea the Stars connections passed on the Breeders' Cup and sent him to
stud at the peak of his fame.
The Betfair bettors favored Rip van Winkle over the 5-year-old
undefeated mare Zenyatta who had run all but one of her 13 races in
California and largely on Santa Anita's polytrack surface. The US bettors
made Zenyatta the favorite at US odds of 2.8-1 with Rip van Winkle at 3.4-
1. But on Betfair it was reversed with Zenyatta the second choice at US odds
about 4-1 most of the time during the betting period. I, of course, preferred
Zenyatta. She is bigger and more agile than any of the males, knows this
track as does her jockey, Mike Smith, and it is known that O'Brien's horses
do poorly as favorites late in the year plus there is the long trip to California
and a new surface. So I was long Zenyatta on Betfair.
But could you construct an arbitrage? Unfortunately, not quite as you can
go long and short on Betfair but only long at Santa Anita or some rebate
betting shop. But let's run through the numbers assuming a two horse race to
lay out the issues. Let Zenyatta be A and Rip van Winkle and all the rest of
the horses be B.
To get a two outcome situation, I had to lump all the horses with Rip van
Winkle and when we consider the Rip van Winkle bet, we must lump
Zenyatta with all the other horses. We get the best odds long on Zenyatta on
Betfair so that's A. Lets assume we get 3.5-1 US odds which was at the low
end of what I got, so that's Oah = 4.5. Assume that the average take at
Betfair is 2.5% on winnings. That's a weighted average of the 5% top rate
and 1% lowest rate for the most volume bettors.
So at Oah = 4.5, the probability that Zenyatta wins is estimated to be
where Q = 0.975 is the payback percent or 21.67%. So the other horses B

win with probability =79.33% according to the Betfair odds. That
corresponds to Obi odds of
So (Oah – 1)(Obj – 1) = 3.5(0.229) < 1 so it is not an arbitrage. To get an

actual arbitrage we need this to be > 1.
I am not adjusting here for the chance that I lose the bet and pay no
commission per Betfair's commission rates. But that's a minor adjustment
for our discussion, but an important one for professional betting programs.
Unfortunately, I cannot go short all these horses at Santa Anita as I am not
allowed to go short there. But maybe I could short somewhere else such as
at another betting exchange or bookmaker, etc. It's not on Betfair because
there Zenyatta is underbet and Rip van Winkle and the other horses are
overbet in total.
So what I can do so far is go long Zenyatta at better than home track odds.
Experts will observe that I do not apply a favorite-longshot bias correction
here, but it is close and does not matter much here. See the discussion below
for some favorite longshot bias graphs for US racing and their use.
Can I do anything with Rip van Winkle? His US long odds at Santa Anita
are 3.4-1 versus much lower on Betfair. So the best that could be done is to
go long Rip van Winkle at Santa Anita and long and short other horses such
as Zenyatta on Betfair. Depending on the odds in various places and the
various commissions an arbitrage might exist. Indeed, there are individuals
searching for these arbitrages with multiple screens across various betting
exchanges. This is made more efficient with custom tablets that allow one to
push buttons that send in bets of various amounts very fast.
Figure 1.1 is the chart and discussion of the race which Zenyatta won
making her now the world's top horse given the retirement of Sea the Stars.
Rip van Winkle finished 10th and again was a fall failure of an O'Brien
favorite. Zenyatta has subsequently won five more races in impressive form
with her connections goal to win the classic again. Observe the $1.8 million
Pick 6 discussed in Chapter 39.
I went long on Zenyatta Betfair as those odds were better than at the
track. This unfortunately is not leading to an arbitrage unless you could go
long all the horses or make back more than the commission. That might
have been possible comparing track, Betfair and other odds.
The favorite-longshot bias

The favorite-longshot bias is the tendency in horseracing, sports betting, and
financial options for the most likely outcome to be underbet and the less
likely outcomes overbet. So people tend to like junk and dislike the best
possibilities. The favorite longshot bias is one of the most useful results for
bettors in sports markets of many types and in the stock market, especially
in options markets. The idea is simple: favorites are underbet and longshots
are over bet. Daniel Kahneman working with the late Amos Tversky won
the Nobel prize in 2002 with this idea plus a bit more in their famous 1979
paper in Econometria. They formalized the economic theory of this bias and
showed how to use it in several contexts outside sports or financial betting.
Fig. 1.1 Zenyatta's great victory. Source: Equibase
This bias has been well known to Irish and other bookmakers who
actually create the bias with the bets they offer for the last 100+ years.
Indeed, bookmakers regularly change the odds to create this bias and the
chalk on their boards is mostly on the favorites, hence the term the chalk for
the favorites. Figure 1.2 shows the 1949 study by Griffin for 1386 races in
1947 for races at Churchill Downs, Belmont and Hialeah. In this graph,
there are the number of entries, winners and winners times odds for every
odds group. The axes show the odds, the subjective probabilities, versus the
actual number of winners, the objective probabilities. Figure 1.3 from
Ziemba and Hausch (1986) Betting at the Racetrack shows the bias. It
shows the effective track payback less breakage for various odds levels in
California and New York, for more than 300,000 races over various years
and tracks, as of 1986. Other graphs in Ziemba and Hausch (1986) going
back to the 1940s, in the UK and elsewhere, are similar.
Fig. 1.2 Griffith's 1949 study on the favorite-longshot bias; see Hausch, Lo and Ziemba (2008) for
the reprinted paper
Observe that there actually was a small profit in betting horses to win at
US odds of 3-10 (UK odds of 1.30 or less). and that at odds of 100-1, the
fair odds are about 700-1 so that such bets were worth about 13.7 cents per
dollar bet. The California and New York graphs differ because of different
track takes. There are approximately three piecewise linear segments, small
profits on extreme favorites with favorites underbet and longshots overbet,
more and more losses as the odds lengthen and extremely poor returns at
high odds levels like lotto tickets.
Since 1986 there have been a number of developments that have
influenced the betting and have shifted the shape of these graphs such as:
(1) There are no longer separate pools for individual races at different
racetracks. All the races now have pooled betting which now comes in
late with about half the bets not recorded in the pools until the race is
already running;
(2) There are rebates where tracks send a signal with the results and the
rebate shops and the track share the track commission between the
rebate shop, the tracks and the bettors. So instead of facing 13-30%
transaction costs, large bettors are actually charged about 10% net; and
Fig. 1.3 The effective track payback less breakage for various odds levels in California and New
York (more than 300,000 races over various years and racetracks, as of 1986). Source: Ziemba and
Hausch (1986)
(3) Betting shops such as Betfair, offer long and short bets on racing and
many other events such as political campaigns, etc.
Figure 1.4a-c look more closely at the extreme favorites in the US and the
UK. Figure 1.4c shows that in Britain, bookies construct odds, creating the
favoritelongshot bias to clear the market and equilibrate bettor demand.
Figure 1.5 shows that the bias curve may be different for different types
of races. It shows the bias for the Kentucky Derby for 1903-1986. The better
races may well have flatter biases. See also Tompkins, Ziemba and Hodges
(2008) who demonstrate similar biases in the S&P500 and FTSE100 index
futures options. I use such ideas in an offshore hedge fund and private
investment accounts.
Figures 1.6 and 1.7ab show the bias based on more recent data. Observe
the important fact that favorites are no longer underbet enough to turn a
profit betting them and the flatness of the curve until you get to fairly long
priced horses. You can still short longshots on Betfair and make a profit if
you are careful. Additional discussion and results are in Hausch and Ziemba
(2008).
This bias also forms a part of the behavioral finance literature. Thaler and
Ziemba (1988) discuss reasons for the bias as do Ziemba and Hausch
(1984). These include the fact that there are more bragging rights from
picking longshots than from favorites: 50-1, wow was I smart while 2-5 is
an easy pick. Transactions costs are another factor: bet $50 to win $10 is
hardly worth the effort. A 1-10 horse having more than a 90% chance of
winning has an expected value of about $1.03 for every $ bet, while a 100-1
horse has only an expected value of about 13.7 cents per dollar invested.
The fair odds are about 700-1 not 100-1.
Fig. 1.4 Extreme favorites, small profits: see Ziemba and Hausch (1986) for these references
Risk arbitrage in the US presidential election

The US is in extreme turmoil! Nothing shows this better than the viscous
remarks against and a lack of respect for President Barack Obama who won
the 2008 election fair and square by the Republicans. The latter is not
something George W Bush can claim as the votes in Florida were tainted.
What's interesting for the bettor is the huge changes in odds of those in the
running for the Republican nomination for the November 2012 election.
There have been countless favorites who all have quickly fizzled so shorting
them has been very profitable. Candidates like Michelle Bachmann, Donald
Trump, Herman Cain, Rick Perry, Newt Gingrich, Rick Santorum leapt to
the top and then faded quickly. All of these were out of the running by the
end of April 2012. The race now, as we go to press, is between Democrat
Fig. 1.5 Expected return per dollar bet with and without the track take deducted for different odds
levels in the Keneucky Derby 1903-1986 and in 35,285 races run during 1947–1975, from data in
Snyder (1978). Source: Ziemba and Hausch (1987)
Fig. 1.6 Effective track payback less breakage for various odds levels in California, comparison of
1986 graph with 1997-2002 data. Source: Ziemba and Hausch [1986] and Ziemba (2008)
Obama and the rather bland and rich Republican Mitt Romney, who has a lot
of liabilities including pioneering the health plan in Massachusetts that he
now argues against with Obama's national implementation,coming off as a
rich guy who does not understand the ordinary citizens who are not as
successful as he, bringing to the forefront a lot of tricks to avoid paying
taxes, including offshore accounts, changing on a dime on every policy issue
depending on polls and trying relentlessly to create attack President Obama
with half truths and lies.
Fig. 1.7 Rate of return at different odds. Source: Snowberg and Wolfers (2008)
One of Romney's main challengers was Newt Gingrich, who is a tough

talking character who if elected would have likely caused serious harm to
the country because of some extremely right wing policies. Gingrich had his
ups and downs in the campaign being at long odds then very short odds. He
picked up a rich backer, Las Vegas/Macao casino magnate Sheldon Adelson
who with his wife contributed $10 million a PAC (political action
committee) in support of Gingrich. This strategy worked in the South
Carolina primary which he won handily over Romney and the others. The
$10 million is reputed to be the one-day profit from one of his casinos.
Adelson has close connections in China which granted him licenses in
Macao in turn for his congressional support of China. But Gingrich has very
difficult to swallow policies such as to tout a colony on the moon and by his
second term grant it statehood. Such ridiculous proposals scared the
Republican establishment and they turned against him. So the Betfair odds
to win the presidency dropped from the 6 area to 48 and then to 340 to 1 or
higher by late March 2012. The Romney versus Obama race is a battle of
the rich 1% versus the bulk of the population. While I am a student of
gambling and investing and a hedge fund manager it is sure clear that the
Obama route is preferable for most of the US people.
The Betfair odds for president were:
At one stage, Santorum gained in popularity and had shorter odds for the
Republican nomination and the presidency than Romney and Obama is a bit
shorter to win the presidency. Then Santorum and Gingrich have dropped
out as Romney won more and more states and the nomination.
As of September 30, 2012, just before the November 4 election, the
Betfair odds were about 1.4-1 for Obama and 3.1-1 for Romney. In the end,
Obama won this election with the odds on Romney moving substantially
during election night.
1 Edited from Wilmott, November 2010.

The Bond Stock Earnings Yield Differential Model
In the Short View section of the Financial Times page 30 on Wednesday July
28, 2010, guest columnist Spenser Jakab casts doubt on the so called FED
model saying that ‘its use and abuse have endured in spite of an utter lack of
theoretical or empirical evidence.‘ Here is a response to this misconception.
I was a visiting professor at Tsukuba University and consultant to the

Yamaichi Research Institute in 1988-89. In a study group at YRI, I
discovered the BSEYD model, namely the 30-year T-bond yield minus the
stock yield (which is the reciprocal of the trailing price earnings ratio). By
April 1987, BESY=9.56-6.17=3.39 for the US S&P500 which enters the
'danger zone', then with a value above 3.0. By September 1987,
BSEYD=9.66- 5.52=4.14. After the October 1987 crash, the measure was
below 2.0 and the market rallied back. The danger zones and reentry points
are computed from confidence limits on the distribution of the BSEYD.Our
study of the Nikkei stock average from 1948-88 showed a gain of 221 times
in yen and 550 times in US dollars. But, during this 40-year period, there
were 20 declines or crashes of 10% or more. Twelve times, the BSEYD was
in the danger zone and ALL twelve had these declines. The other eight
declines were caused by other reasons than interest rates being too high
relative to earnings, see Ziemba and Schwartz (1991). So not all crashes
come from interest rate problems. Reinhart and Rogoff (2010) discuss other
causes for economic crisis periods that sometimes include stock market
crashes. Also see Evanoff, Kaufman and Malliaris (2012).
The BSEYD model in the difference form I use and the Fed model ratio
version proposed later in 1996 are basically equivalent. The BSEYD did call
the 2007–2009 subprime, credit confidence crash which occurred with low
interest rates but was signaled at high interest rates relative to earnings on
June 14, 2007. It also called the 2008 declines in China and Iceland which
again were high interest rate relative to earnings declines. Despite the theory
having a nominal quantity (the long bond rate) relative to a real quantity (the
earnings to price ratio) theoretical flaw, the model does seem to predict
crashes quite well when it applies. It predicted the 2000-2001 and 2002
crashes. By April 1999, the measure was 3.03 and in the danger zone with a
1335.18 S&P500 which later fell to 885.76 on October 31, 2002. There were
actually two BSEYD crashes in this period with the second calling the –22%
S&P500 fall in 2002. For details, see Chapter 21 which summarizes results
from Ziemba (2003), Koivu, Pennanen and Ziemba (2005) and Lleo and
Ziemba (2012).
The best use of the model is to call crashes. Ask George Soros whose high
PE models like those of Bob Shiller, suggested shorts of the Japanese market
in the fall of 1988. But only in January 1990 when the BSEYD was in the
danger zone did it actually crash, and then it fell –56%! In a typical situation,
the market goes into the danger zone but continues rallying and then crashes
10%+ from the initial danger signal. But is provides some help on recoveries
and for buy and hold strategies. The decision rule: stay in the S&P500 unless
the model enters the danger zone and be in cash yields double final wealth
with lower standard deviation risk during 1975–2005 for the US, Japan,
Germany, Canada and the UK, details are below bases on Berge, Consigli
and Ziemba (2008).
Going to the present Bill Miller, the manager of the famed Value Trust
Legg Mason Mutual Fund, may be right that the current US stock market is
undervalued and indeed the BSEYD model is in the bullish mode but it has
been there since 2006, well before, during and after the 2007–2009 crash. So
I would tread carefully here as we are in a different economic environment.
Miller got fame by beating the S&P500 for 15 straight years but got totally
blasted in 2008 with more than 50% losses. Later, he was removed from this
fund. Maybe later the BSEYD model will suggest future crashes.
Mohammed El Erian of PIMCO had the good sense to leave the Harvard
endowment before the poor 2008 returns came in — much from ill-advised
swap trades put on by Larry Summers, the president of Harvard at the time.
El Erian has championed the view that there is a new normal now with much
lower returns to be expected compared to historical data. A lot of this
historical data on asset classes is in the retirement and pensions book of
Bertocchi, Schwartz and Ziemba (2010). I agree with the PIMCO view for
most investors. The great ones will still be able to rack up high returns
through clever strategies.
Regarding the new normal, Bill Gross, the head of PIMCO and the
manager of their $248 billion total return fund, feels that the departure of
fund managers such as Stanley Druckmiller of Duquesne Capital Markets,
ran this fund since 1980, marks the end of the old normal; see McCormick
(2010). In the new normal, instead of leverage and deregulation, the world is
moving away from these approaches. Other hedge funds are closing shop or
reducing fees. In the latter case is the $11.1 billion Citadel fund Kensington
and Wellington which are which are considering cutting fees.
PIMCO has shifted focus from the authority on bonds to the your global
investment authority in their advertising. They are now a $1.1 trillion
subsidiary of the $1.7 trillion Munich Insurer Allianz SE. They have
dropped their percent of government debt from 54% in July to 36% in
August. Gross has said he feels that the 15 plus year bull market in bonds is
over.
A sad note for non-believers of such models: After we completed the
crash study group in the fall of 1989, I sent up to the YRI management, who
had supported my research, one of my Japanese colleagues to indicate
extreme danger in the Japanese market. Indeed, the BSEYD measure was the
furthest into the danger zone since 1948. Too bad they would not listen, for
five years later, Yamaichi Securities, then the sixth biggest brokerage firm in
the world, went bankrupt. They lost a lot more than Soros, who shorted in
the fall of 1988, nearly a full year before the BSEYD measure suggested that
the market was in the danger zone and lost about $1 billion!. He takes many
short term losses but wins more often as do most aggressive Kelly bettors
such as Buffett and Keynes. They win long term, see MacLean, Thorp and
Ziemba (2010).
Predictive Ability of the Bond-Stock Earnings Yield Difference

Model
I end this chapter discussing the study of Berge, Consigli and Ziemba (2008)
and Berge and Ziemba (2006) which show how the BSEYD model is useful
for determining being in or out of markets. This is done using rules to be in
or out of stocks and in cash for five countries: the US, UK, Japan, Germany
and Canada for the period 1975–2005 and 1980–2005. The model
recommends that the investor be in cash about 15-20% of the time and the
result is that final wealth in 2005 is about double that of a buy and hold
strategy with a higher Sharpe ratio and lower risk.
I begin with the idea that suggested the BSEYD model and that is in Table
2.1. The data is the Morgan Stanley Capital International Indices (MSCI)
while Chapter 21 uses the S&P500. This reinforces the result that with
different data or different time periods, the BSEYD model more of less has
the same conclusions. In this table, we see that the BSEYD=10 year T bond
yield - earnings/price ratio.
The danger zone was entered in May 1987 and the correction occurred in
October, four months later. During June, July and August investors kept
rebalancing their portfolios from the bond to the equity market (MSCI TRI +
13.87% over the quarter) then the equity market fell -31.80% in the
following quarter (September to November 1987) with the main decline in
October. In order to validate the approach, Berge and Ziemba (2006) used
eight different strategies to test the predictive ability of the BSEYD or,
equivalently, the BSYR, see Table 2.2. The strategies are defined over either
5 or 10 year time past data intervals, under normal or historical return
distributions and different fractile levels for exiting and entering the equity
market.
Table 2.1: Behavior of the BSEYD in the US around the Stock Market Correction
in October 1987. Source Berge and Ziemba (2006)
All data except for stock market return data are as of the end of the given months. The
BSEYD is the difference between the 10-year bond yield and the E/P ratio.
Table 2.2: Strategies Used to Evaluate the Robustness (Predictive Ability)

of (the) BSEYD-based strategies. Source Berge and Ziemba (2006)
These 60 or 120 observations, respectively, including the value of the
current month define the threshold levels for the end of that month. As time
evolves, the most recent monthly value is added and the oldest monthly
value is omitted. Moving intervals diminish the impact of changes in the
regulatory framework of the economy on E/P ratios over time, since high
past values of the BSEYD are eventually excluded. For the historical values
of the BSEYD two different distributional concepts are used to define the
predetermined exit and entry threshold levels, normality and historical.
Assuming normality, the mean and standard deviation of the historical range
of BSEYDs are
respectively, where d is the length of the interval considered. The exit

threshold levels is
and the entry threshold level is

where zα1, and zα2 are the α1 and α2 fractiles of the standard normal
probability distribution, respectively. The second concept calculates the α1
and a2 fractiles of the historical distribution. The α fractile of d observations
is
where k = d · α and x(k) is the k-smallest value of the d observations starting

at the end of month (t – d +1) and ending at the end of month t.
The strategies were evaluated during the 1979–2005 period in five equity
markets: the US, Germany, Canada, UK and Japan. The total return indices
were calculated with gross dividends (before taxes) reinvested to estimate
the total return on the market that would be achieved by reinvesting one
twelfth of the annual dividend reported at every month end. P/E ratios
provided by MSCI refer to the companies in the respective MSCI TRI. They
are the sum of the weighted latest share prices in local currency divided by
the sum of the weighted most recent trailing twelve month earnings (losses)
per share. Actual earnings are entered in the database as soon as they are
reported at an adequately detailed level. The advantage of the MSCI indices
over other indices is that even when the overall market experiences losses,
P/E ratios, in this case negative, can be calculated. For cash equivalents the
yield on 3 month T Bills is used. Data for 3 month T Bills are from Global
Financial Data (GFD) and Thomson Financial Datastream.
Data on 10 year government bond yields come from various sources. The
US, data were from the Fed. Yields are of constant maturity interpolated
from the daily yield curves. Data for Germany were supplied by the
Deutsche Börse AG and were calculated in a similar way as for US data.
Data for the United Kingdom and Japan come from GFD. These yields refer
to government bonds with a maturity closest to ten years. Data for Japan are
available beginning in May 1980. Japanese T Bills data were not updated
monthly until December 1981. Canadian government bond yields were
obtained from the Bank of Canada.
For the first strategy, five years of monthly observations are used prior to
1979 to estimate the historical (or frequency) distribution of the 90% and
80% fractiles. Every month the data includes the realized monthly return and
discards the oldest observation to keep the five year window constant. Every
strategy has well defined equity market entry and exit values : once in, the
strategy suggests 100% in equities, once out 0%, i.e., a 0-1 stock-bond
strategy. Results for the five markets for the 35-period, 1970–2005. Every
fractiles pair defines the market danger zone, which are used in the
definition of the optimal investment strategies and portfolio rebalancing.
Tables 2.3 to 2.12 show for the five countries, the stock market
corrections (declines of 10% plus) and their durations and the results of the
strategies in terms of percent of time in the stock market, mean log return,
standard deviation, Sharpe ratio, mean excess return and terminal value all
the comparison with the actual stock markets total returns in 1975-2005 for
strategies 1–4 and 1980–2005 for strategies 5–8.
The US had nine corrections and the terminal values, and Sharpe ratios
show that the strategies added value. The main test is: are you better off or
not being out of the stock market when the BSEYD measure suggest you
should be? For example, $100 grew to $4650 in the stock market and from
$8480 to $10,635 with strategies 1–4 for the period 1975-2005. In addition,
the portfolio variance was less because about 15–20% of the time one is out
of the stock market. The period 1980–2005 had similar positive results for
the various strategies.
Japan had seven corrections including the 31-month 56.2% decline
starting in January 1990. The results for Japan are good for the measure: 100
yen grows to 213.88 yen with the market index from 1985–2005 but to
455.72 yen to 498.62 yen with strategies 1–4. Strategies 5-8 also had good
prediction results for 1990–2005.
The results for the other three countries were not as good in terms of final
wealth levels from 1975–2005. However, all the strategies were superior to
buy and hold in terms of the Sharpe ratios and in the final wealth levels
during 1980–2005. For the UK, most of the strategies outperformed the buy
and hold stock market returns but there were some underperforming
strategies and the outperformance of the winning strategies was minimal.
The UK had 15 corrections mostly of short duration but the 17.9% decline
of September 2000 lasted 13 months and the April 2002 decline of 31.6%
lasted 10 months.
Canada had similar results to the US with five of its 12 corrections having
declines of over 20% including a 41.2% decline in September 2000 over 13
months. Germany also had similar results with 13 corrections with the
March 2000 decline being –44% over 19 months and April 2002 –53.9%
over 12 months.
The basic conclusion is that the bond-stock model has had positive
predictive ability worked in these five countries especially in the 1980–2 005
period with the results in the US and Japan the best: portfolio strategies
based on the bond-stock relative valuation measure have produced superior
returns relative to a buy and hold strategy.
Table 2.3: Although the duration of the correction starting in January
1977 is longer than one year, it is considered a correction since the stock
market dropped 11% within ten months between January and October 1977.
The same is true for the correction starting in April 2000 - the stock market
dropped 23.9% within the following twelve months.
Table 2.3: Stock Market Corrections (10% plus) in the US, 1975–2005
Table 2.4: The mean excess return is the average monthly excess return of
the strategy over the stock market. Terminal values refer to the gross
performance of $100 invested using the strategy signals.
Table 2.5: Although the duration of the correction starting in October
market dropped 12.5% within nine months between October 1978 and June
1979. The same is true for the correction starting in March 2000 - the stock
market dropped 20.7% within the following twelve months.
performance of 100 invested using the strategy signals.
Table 2.4: Evaluation of the Performance of the Strategies for the US
Table 2.5: Stock Market Corrections (10% plus) in Germany, 1975–2005
Table 2.6: Evaluation of the Overall Performance of the Strategies for

Germany
Table 2.7: Although the duration of the correction starting in September
market dropped 36.4% within twelve months between September 2000 and
August 2001.
Table 2.7: Stock Market Corrections (10% plus) in Canada, 1975-2005

Canada
performance of C$ 100 invested using the strategy signals.
Table 2.9: Although the duration of the correction starting in September
market dropped 14.7% within twelve months between September 2000 and
August 2001.
Table 2.10: The mean excess return is the average monthly excess return
of the strategy over the stock market. Terminal values refer to the gross
performance of £100 invested using the strategy signals.
Table 2.9: Stock Market Corrections (10% plus) in the UK, 1975–2005
Table 2.10: Evaluation of the Overall Performance of the Strategies for the
UK
Table 2.11: Stock Market Corrections (10% plus) in Japan, May 1985 to
December 2005
Table 2.11: Although the duration of the correction starting in January
market dropped 40% within twelve months between January and December
1990. The same is true for Corrections 4 (23.5% decline between June 1994
and May 1995), 6 (15.3% decline between August 1997 and July 1998) and
7 (20.1% decline between April 2000 and March 2001).

Japan
Table 2.12: The mean excess return is the average monthly excess return
of the strategy over the stock market. Terminal values refer to the gross
performance of ¥100 invested using the strategy signals.
Investor Camps1
The next three chapters discuss how great investors succeed. This is an
enormous topic but we think our principles and results apply reasonably
broadly. Winning has two parts: getting an edge and then betting well. The
former simply means that investments have an advantage so $1 invested
returns on average more than $1. The latter involves not overbetting, and
truly diversifying in all scenarios in a disciplined, wealth enhancing way.
This chapter begins with a categorization of the efficient market camps

which inform how various people try to get an edge. Some feel that one
cannot get an edge. This becomes then a self fulfilling prophecy and they. of
course, are not in our list ofgreat or even good investors. Many great
investors are Kelly or fractional Kelly bettors who focus on not losing. This
chapiter discusses the records of some great investors and concludes with a
suggested method to evaluate them. In Chapter 6 we evaluate the records of
some great investors in more depth. Chapter 11 discusses the methods of
great university endowment managers such as David Swensen of Yale as
well as the managers of the Harvard, Princeton and Stanford endowments.
The various efficient/inefficient market camps: Can you beat
the stock market?
Why Buffett wants to endow university chairs in efficient market theory.
Market participants can be divided into five groups. There are other ways
to do such a categorization but this way is useful for our purpose of isolating
and studying great investors and naturally evolves from the academic study
of the efficiency of financial markets.
The Five Groups are:
(1) Efficient markets (E)
(2) Risk premium (RP)
(3) Genius (G)
(4) Hog wash (H)
(5) Markets are beatable (A)
The first group are those who believe in efficient markets (E). They
believe that current prices are fair and correct except possibly for
transactions costs. These transaction costs, which include commissions, bid-
ask spread, and price pressures, can be very large.2
The leader of this school which had dominated academic journals, jobs,
fame, etc in the 1960s to the 1980s was Gene Fama of the University of
Chicago. A brilliant researcher, Fama is also a tape recorder: you can turn
him on or off, you can fast forward or rewind him or change his volume, but
you cannot change his views no matter what evidence you provide; he will
refute it forcibly.
This group provided many useful concepts such as the capital asset
pricing model of Sharpe (1964), Lintner (1965) and Mossin (1966) which
provided a theoretical justification for index funds which are the efficient
market camp's favored investment mode. They still beat about 75% of active
managers. Since all the managers comprise the market, that's 50% of them
beaten by the index. Transactions and other costs eliminate another 25%.
This is discussed more in Chapter 4.
Over time the hard efficient market line has softened into a Risk Premium
(RP) camp. They feel that markets are basically efficient but one can realize
extra return by bearing additional risk. They strongly argue that, if returns
are above average, the risk must be there somewhere; you simply cannot get
higher returns without bearing additional risk they argue. For example,
beating the market index S&P500 is possible but not risk adjusted by the
CAPM. They measure risk by Beta, which must be greater than one to
receive higher than market returns. That is, the portfolio risk is higher than
the market risk. But they allow other risk factors such as small cap and low
book to price. But they do not believe in full blown 20-30 factor models
such as described in Chapter 25. Fama and his disciples moved here in the
1990s. This camp now dominates the top US academic journals and the jobs
in academic finance departments at the best schools in the US and Europe.
The third camp is called Genius (G). These are superior investors who are
brilliant or geniuses but you cannot determine in advance who they are. Paul
Samuelson has championed this argument. Samuelson feels that these
investors do exist but it is useless to try to find them as in the search for
them you will find 19 duds for every star. This view is very close to the
Merton-Samuelson criticism of the Kelly criterion: that is, even with an
advantage, it is possible to lose a lot of your wealth. Some simulations on
this appear in MacLean, Thorp, Zhao and Ziemba (2011) The evidence
though is that you can determine them ex ante and to some extent they have
persistent superior performance, see Fung et al (2006) and Jagannathan et al
(2006). Soros did this with futures with superior picking of futures to bet on;
this is the traders are ‘made not born’ philosophy. This camp will isolate
members of other camps such as in (A) or (H).
The fourth camp is as strict in its views as camps (E ) and (RP). This
group feels that efficient markets which originated in and is perpetuated by
the academic world is hogwash (H). The leading proponent of this view and
one with whom it is hard to argue with as he is right at the top of the list of
the world’s richest persons is Warren Buffett, who wants to give university
chairs in efficient markets to further improve his own very successful
trading. An early member of this group, the great economist John Maynard
Keynes was an academic. Although they may never have heard of the Kelly
criterion, this camp uses it implicitly with large bets on favorable
investments, little diversification, possibly many monthly losses but usually
high final wealth.
This group feels that by evaluating companies and buying them when
their value is greater than their price, you can easily beat the market by
taking a long term view. They find these stocks and hold them forever. They
find a few such stocks that they understand well and get involved in
managing them or they simply buy them and make them subsidiaries with
the previous owners running the business. They forget about diversification
because they try to buy only winners. They also bet on insurance when the
odds are greatly in their favor. They well understand tail risk which they
only take at huge advantages to themselves when the bet is small relative to
their wealth levels.
The last group are those who think that markets are beatable (A) through
behavioral biases, security market anomalies using computerized superior
betting techniques. They construct risk arbitrage situations with positive
expectation. They research the strategy well and follow it for long periods of
time repeating the advantage many times. They feel that factor models are
useful most but not all of the time and show that beta is not one of the most
important variables to predict stock prices. They use very focused,
disciplined, well researched strategies with superior execution and risk
control. Many of them use Kelly or fractional Kelly strategies. All of them
extensively use computers. They focus on not losing, and they rarely have
blowouts. Members of (A) include Ed Thorp (Princeton Newport and later
funds), Bill Benter (the Hong Kong racing guru), John Henry (trend follower
and the Red Sox owner), Blair Hull (mirpriced options trader), Harry
McPike (trend follower), Jim Simons (Renaissance hedge fund), Jeff Yass
(Susquehanna Group), and David Swensen (Yale endowment). Blowouts
occur more in hedge funds that do not focus on not losing and true
diversification and over-bet; when a bad scenario hits them, they get wiped
out, such as LTCM, Niederhoffer, and Amaranth; see Chapter 26. My idea of
using scenario dependent correlation matrices is very important here, see
Chapter 14.
Three great investors: Warren Buffett, Paul Samuelson, and Ed Thorp
How do investors and consultants do in all these cases?

All investors can be multimillionaires but the centimillionaires are in (G),
(H) and (A) like the seven listed before me in (A) and Buffett. These people
make more money for their clients than themselves but the amount they
make for themselves is a huge amount: of course these people eat their own
cooking that is they are clients themselves with a large amount of their
money in the funds they manage). An exception is someone who founded an
(RP) or (E) company kept most of the shares and made an enormous amount
of fees for themselves irrespective of the investment performance given to
the clients because the sheer volume of assets they have gathered under
management is so large. I was fortunate to work/consult with six of these
and was also the main consultant to the Frank Russell Research Department
for nine years which is perhaps the leading conservative RP implementor.
(A) people earn money by winning and taking a percent of the profits, Thorp
returned 15.8% net with $200 million under management; fees $8
million/year (1969-88). (E) and (RP) people earn money from fees by
collecting assets through superior marketing and sticky investment
decisions.
Many great investors use Kelly betting including most in camp (A). There
are compelling reasons for this discussed in previous chapters. For long and
mathematical survey papers, see MacLean and Ziemba (2006) and Thorp
(2006). But there are critics and chief among them are Nobel prize winners
Robert Merton and Paul Samuelson. Their argument is that successful
investing requires a lot of luck and it is hard to separate luck from skill.
Therefore, while many Kelly investors will make huge gains, a few will have
huge losses. Indeed they are correct. A good way to explain this is via the
simulation Donald Hausch and I did; see the experiment in Chapter 13. A
simulation of 700 bets, all with a 14% edge, was performed 1000 times.3
In support of Kelly, 166 times out of the 1000 simulated wealth paths, the
investor has more than 100 times initial wealth with full Kelly. But this great
outcome occurs only once with half Kelly. However the probability of being
ahead is higher with half Kelly, 87.0 vs 95.4. A negative observation, and
related to the Merton-Samuelson criticism, is that the minimum wealth is
only 18. So you can make 700 bets all independent each with a 14% edge
and the result is that you still lose over 98% of your fortune with bad
scenarios. Luenberger (1993) shows that, theoretically, if your utility
function is based solely on the tail losses, then the optimal strategy is to
tradeoff expected log and variance of log (like a static mean variance
analysis).
The importance of getting the mean right.

The mean dominates if the two distributions cross only once.
Theorem: Hanoch and Levy (1969) If X ~ F(·) with the higher mean and
Y ~ G(·) have cumulative distribution functions that cross only once, but are
otherwise arbitrary, then F dominates G for all concave u. The mean of F
must be at least as large as the mean of G to have dominance. Variance and
other moments are unimportant. Only the means count.
With normal distributions X and Y will cross only once iff the variance of
X does not exceed that of Y. That is the basic equivalence of Mean-Variance
analysis and Expected Utility Analysis via second order (concave, non-
decreasing) stochastic dominance. See Figure 3.1 with densities shown (the
CDFs cross the same way).
Errors in Means, Variances and Covariances: Empirical
Replace the true mean by the observed mean μi(1 + kZi) where Zi is
distributed N ~ (0,1) with scale factor k = 0.05 to 0.20, being the size of the
error. Similarly, replace the true variances and covariances by the observed
variances and covariances σij(1 + kZi). We use monthly data from
1980-89 on ten DJIA securities which include Alcoa, Boeing, Coke, Dupont
and Sears. See Chopra-Ziemba (1993) which updates and extends Kallberg
and Ziemba (1984).
Fig. 3.1 The main theorem of mean-variance analysis
The certainty equivalent, CE, of a portfolio with utility function u equals

u−1 (expected utility of a risky portfolio). This comes from the equation:
Assuming exponential utility and normal distributions yields exact

formulas to calculate all quantities in the certainty equivalent loss
Observe that the mean-variance problem is

Table 3.1 and Figure 3.2 show that the errors in means are about 20 times
errors in covariances in terms of CEL value and the variances are twice as
important as the covariances. So roughly, there is a 20:2:1 ratio in the
importance of these errors. Also, this is risk aversion dependent with TR =
(RA/2)100 being the risk tolerance. So for high risk tolerance, that is low risk
aversion, the errors in the means are even greater. Hence for utility functions
like log of Kelly with essentially zero risk aversion, the errors in the mean
can be 100 times as important as the errors in the other parameters. So Kelly
bettors should never overbet.
Conclusion: spend your money getting good mean estimates; use

historical variances and covariances.
Chopra (1993) shows that the same relationship holds regarding turnover but
it is less dramatic than for the cash equivalents, see Figure 3.3.
Fig. 3.2 Mean percentage cash equivalent loss due to errors in inputs (Source: Chopra-Ziemba, 1993)
Table 3.1: Average Ratio of CEL for Errors in Means, Variances and Covariances
The results here apply to essentially all models. You must get the means
right to win!
If the mean return for US stocks is assumed to equal the long run mean of
12% as estimated by Dimson et al. (2006), the model yields an optimal
weight for equities of 100%. A mean return for US stocks of 9% implies less
than 30% optimal weight for equities. This is in a five period ten year
stochastic programming model. See Figure 3.4.
In Chapter 6, a modification of the Sharpe ratio is used to evaluate great
investors. The main idea is that we do not want to penalize investors for
superior performance so we will focus only on losses. But to use the Sharpe
ratio, we must have a full standard deviation over the whole range of
possible return outcomes and that is estimated using the downside standard
deviation. That is, we artificially create gains which are mirror images of the
losses. These gains are less than the real gains so they penalize the investor
less than if one uses the ordinary Sharpe ratio.
Fig. 3.3 Average turnover: percentage of portfolio sold (or bought) relative to preceding
allocation:source Chopra(1993)
Fig. 3.4 Optimal asset weights at stage 1 for varying levels of US equity means in a multiperiod
stochastic programming pension fund model for Siemens Austria: see Geyer and Ziemba (2008)
1 Edited from Wilmott, March 2006.

2 A BARRA study by Andy Rudd some years ago showed that these costs averaged 4.6% one- way
for a $50,000 institutional investor sale. This is if you use a naive market order for the full transaction
rather than limit orders or smaller market orders. Thorp, in a private communication, told us that he
traded about $60 billion in statistical arbitrage from 1992–2002 in lot sizes of 20k to 100k and found
that the mean transaction cost was about 1 cent per share and the market impact was about 4.5 cents
per share for shares averaging about $30. So the one-way costs wereabout $0.18.
3See also the empirical paper by Bicksler and Thorp (1973) where they calculate the probability that
investors will be ahead after given numbers of favorable bets and the simulations in MacLean, Thorp,
Zhao and Ziemba (2011). Their conclusions are consistent with those in the Hausch and Ziemba study.
A response to the Samuelson critique using Samuelson's private letters to Ziemba and Samuelson's
papers, many of which are reprinted in MacLean, Thorp and Ziemba (2010) is in Ziemba (2012c).
PART II
Hedge Funds, Sovereign Wealth
Funds and Other Investment
Agglomerations
Average Hedge Funds and their Evaluation1
Analyzing those who regularly beat a vanilla buy and hold strategy
We know that on average the typical mutual fund does not beat the
market. The evidence is that professional managers all over the world have
a hard time beating the market averages. In a given year, only about 25% to
40% of managers actually beat a buy-and-hold strategy of holding the
index. Over longer periods, say 5-10 years, the percentage is even lower.
There are a number of reasons for this.
• The market averages stay fully invested at all times, never missing
market moves nor paying commissions for stock changes and market
timing.
• When funds get behind the index, they often make hasty moves to try to
catch up and, more often than not, this puts them further behind.
• Portfolio managers have a tendency to window dress at reporting times,
adding to turnover and commissions.
• Since the managers collectively more or less are the market (with
individual investors forming less and less of the market each year) the
indices, on average, beat half of the fund managers. Then with
commissions, fees, and these other reasons, only 25% to 40% typically
beat the market averages (which does quite well with a lot less work).
• The fund managers take fees; the averages work for free.
• The fund managers' goals may get in the way of the fund's best interests
which creates an agency problem.
• Portfolio managers tend to follow each other's moves. They tend to
move the market which gains the full amount, and they can easily be a
little behind.
We have the following four reasons the high commissions of active
trading often lead to poorer performance than the market indices.
(1) commissions are higher in active portfolios because the turnover is
greater.
(2) the bid-ask spreads are larger for many smaller international securities
that an active manager would buy
(3) exchange taxes can be as large as 1% on both buys and sells
(4) active managers usually hold a small number of large positions so they
have market impact on getting in and out.
Index funds have grown and grown. Dimension Fund Advisors formed
by University of Chicago Professor Eugene Fama's students David Booth
and Rex Sinquefield manage over $250 billion and others such as Barclays
in San Francisco manage over $100 billion. This is done with low fees in an
efficient manner. The indices for these passive funds have grown to include
small cap, foreign investments and a variety of exchange traded funds as
well as the traditional market index, the S&P500. Despite very low fees,
profits are large. Booth was able to give the University of Chicago Business
School $300 million and it is now the Booth Graduate School of Business.
Table 4.1 shows results for 1988 for 167 funds based in Hong Kong with
investments in various parts of the world. Only 48, or 28.7%, actually beat
the benchmark indices. ASEAN equity funds did do well, averaging 41.4%
returns versus the market's 27.7%, and 11 of the 13 funds beat this measure.
But Japanese equity funds did not fare so well. Indeed only 4 out of 30
funds beat the TSE index of 34.9% and their average return was 19.1%,
more than 15 points below the Topix.
Thus the efficient market proponents have one of their greatest success,
namely that, for most of the time, most investors will do quite well by
simply investing in index funds. Indeed, even the great investor David
Swensen, who runs the Yale endowment discussed in Chapter 11, suggests
that most people are well served with tax efficient exchange traded funds,
see Swensen (2000).
Indeed from February 28, 2003 to July 20, 2007, the S&P500 has gained
82%. Whoops, that's in the declining US dollar. Meanwhile, gold in US
dollars is up 82%, Brazil up 92%, the euro has risen from 0.87 in January
2002 to 1.38 in late July 2007, the British pound at 2.05 is at a decades high
level and even the Canadian dollar, stuck at 63-64 cents for years until 2002
is at 95 cents. Even the Kiwi and Aussie dollars are flying with the Aussie
up from 78 to 85 in just the last few months. The Chinese and Japanese
currencies have been held back but that's given each of them over $1 trillion
in foreign reserves. The Chinese stock market is the highest gainer, over
100% this past year in a strong currency.
Table 4.1: Sector Median Returns of Hong Kong Based Funds Compared
with Market Average, January 1 to November 30, 1988. Source: Ziemba
and Schwartz (1991)
Forsyth (2007) argues that these rises are simply due to global liquidity.
But this situation is fast changing. Cheap funds for investments like the US
sub-prime mortgages repackaged into various erroneously labeled AAA
instruments and private equity levered buyouts is drying up especially seen
in the implosion of the subprime instruments.
In the short term the Chinese investment in Blackstone which was
purchased at a 4.5% discount from the IPO (about $31 per share) is
underperforming T-bonds with the stock at 24. This Chinese investment is
relationship based as well as an investment per se.2 It remains to be seen if
the Chinese invest better than the Japanese did; see Ziemba and Schwartz
(1992) for a discussion and a listing of Japanese trophy purchases that lost
money. In trading one must always be ready for mini-storms at any time.
And a new one has appeared in late July 2007 with the S&P500 falling over
50 points in two days and the VIX volatility index rising to over 24% very
high for the year which saw a low of 9.39% but well below the 44% in
1998.
This chapter focuses on the average hedge funds and then on one of the
current great funds, Renaissance's Medallion. For the average funds I look
at the effects of incentives and the general partners' stake in the fund while
Medallion, provides the opportunity to see how Ziemba's (2005) downside
symmetric Sharpe ratio is a much better way to evaluate superior investors
than the ordinary Sharpe ratio.
I discuss some aspects of hedge funds, specifically the very important
risk control point that risk behavior is greatly improved in situations where
the decision maker (eg, manager of the fund) holds (invests) a substantial
portion of their own money. That is developed in a hedge fund management
model that Roy Kouwen- berg and I developed where the effect of incentive
fees on investment behavior can be studied.
Kouwenberg and Ziemba (2007) analyse the effect of incentive fees on
risk taking in a continuous-time framework, taking management fees and
the manager's own stake in the fund into account. They do not use a
standard normative utility function like HARA for the preferences of the
fund manager. Rather they use the behavioural setting of prospect theory - a
framework for decision-making under uncertainty developed by Kahneman
and Tversky (1979). This utility is based on actual human behaviour
observed in experiments. Siegmann and Lucas (2002) argue that loss
aversion, an important aspect of prospect theory, can explain the non-
normal return distributions of hedge funds. Kouwenberg and Ziemba
investigate how hedge fund managers driven by these preferences react to
incentive fees. They also derive an expression for the value of the manager's
incentive fee, as in Goetzmann, Ingersoll and Ross (2003). This call option
on other people's money can be worth more than 17% of the invested value.
Kouwenberg and Ziemba present a continuous time theoretical study of
how incentives affect hedge fund risk and returns and an empirical study of
the performance of a large group of operating hedge funds. Most hedge
fund managers receive a flat fee plus a share of the returns above a
benchmark. They investigate how these features of hedge fund fees affect
risk taking by the fund manager in the behavioural framework of prospect
theory. Their main conclusions are:
(1) In the theoretical study:
(a) the performance related component encourages funds managers to
take excessive risk.
(b) However, risk taking due to incentive fees is greatly reduced if a
substantial amount of the manager's own money (30%+) is in the
fund. When the manager has 30%+ in the fund, the call option on
other people's money drops dramatically from 18% to very little.
(2) In the empirical study (2078 hedge funds and 536 funds of funds;
January 1995 - November 2000, Zurich hedge fund universe):
(a) Average returns though, both absolute and risk-adjusted, are
significantly lower in the presence of incentive fees.
(b) Even after adjusting for style differences, the average hedge fund
does not make back their fees, which average 2+20.
(c) Fund of funds have better performance than individual funds
The management fee covers expenses and provides business income. These
fees should moderate risk taking, as negative investment returns reduce the
future stream of income from management fees. Most fund managers invest
their own money in the fund. This ‘eating your own cooking’, helps to
realign the motivation of the fund manager with the objectives of the other
investors in the fund. The fact that hedge fund managers typically risk both
their career and their own money while managing a fund is a positive sign
to outside investors. The personal involvement of the manager, combined
with a good and verifiable track record, could explain why outside investors
are willing to invest their money in hedge funds, even though investors
typically receive very limited information about hedge fund investment
strategies and also possibly face poor liquidity due to lock-up periods in
some funds. The loss averse hedge fund managers increase risk taking in
response to the incentive fees, regardless of whether the fund value is above
or below the benchmark. One would expect that the hedge fund manager's
own stake in the fund is an essential factor influencing the relationship
between incentives and risk taking, and they find that the model predicts
this.
The utility function is
• The fund manager has a threshold θ(T) > 0 for separating gains and
losses.
• The parameters and determine the curvature of
the value function over losses and gains respectively.
• The parameter A > 0 is the level of loss aversion of the hedge fund
manager.
• In prospect theory it is assumed that losses are more important than
gains, that is A > 1; so the pain of a loss exceeds the positive feeling
associated with an equivalent gain.
The continuous time model and its mathematical analysis are discussed
in Kouwenberg and Ziemba (2007). The following figures and tables
present the main results.
To analyze the effect of incentive fees on the investment strategy of the
fund manager, they used the fact that the implicit level of loss aversion A of
the fund manager decreases as a function of the incentive fee level as shown
in Figure 4.1 for three different levels (5%, 20% and 50%) of the managers
stake in the fund.
As the incentive fee increases, the implicit level of loss aversion of the
fund manager decreases, indicating that the manager should optimally care
less about losses and more about gains due to the convex compensation
structure. The negative impact of incentive fees on implicit loss aversion is
mitigated to some extent if the manager owns a substantial part the fund.
The initial fund value is 1, the threshold for the incentive fee is 1, the
management fee is 1% and the manager's own stake in the fund is 20%.
Given these parameters, Figure 4.2 shows the optimal weight of risky assets
in the fund, as a function of fund value at time t = 0.5. Each line in Figure
4.2 represents a different level of incentive fee, ranging from 0% to 30%.
The fund manager takes more risk in responoe to an infreaeing incentive
fee. The increase in risk is more pronouncvd when fund value drops below
the benchmark. Due to the structure of the value function oi prospect
theory, a fund manager without an incentive fee will inerease risk; at low
fund values as well; incentive fees amplify this behauiour.
Fig. 4.1 Implicit level of loss a'version as a function of incentive fee with fixed fee of a==1%, lines
for different levele of the managee's stake in the fund (v). Source: Kouwenberg and Ziemba (2007)
Fig. 4.2 Optimal weight of stocksas a function of fund value, manager's stake in the fund v=20%,
lines for different levels of the incentive fee (,8). Source: Kouwenberg and Ziemba (2007)
Figure 4.3 shows the effect on the optimal investment strategy of

changing the manager's ewn stake in the funi given an incentive fee of 20%.
It demonstrates that an increase of the manager's share in the fund can
completely change risk taking. With a stake of 10% or less, the manager
behaves extremely risk seeking as a result of the incentive fee. However,
with a stake of 30% or more, the investment strategy is similar to the base
case of 100%% ownership (without an incentive fee).
Fig. 4.3 Optimal weight of stocks as a function of fund value, incentive fee of β = 20% for different
levels of the managers stake in the fund (v). Source: Kouwenberg and Ziemba (2007)
Figurf 4.4 shows the manager's initial waight of aisky assets aa a function
of the incentive fee. The different lines represent different levels of ehe
manager's own stake in tnhe fund. Again higher incentive fees lead to
increased risk taking; the increase in risk taking is more drastic when the
managers own stake in the fund is low .
Fig. 4.4 Initial weight of stocks as a function of the incentive fee for different levels of the managers
stake in the fund (v)
Figure 4.5 plots the value of a 20% incentive fee as a function of the
manager's stake in the fund, using the same set of parameters as above. The
value of the 20% incentive fee ranges from nearly zero to 17% of the initial
fund value, depending on the manager's own stake in the fund. If the
manager's stake in the fund is 100%, the manager does not care about the
incentive fee and manages the fund conservatively since it is a personal
account. However, as the manager's stake in the fund goes to zero, the
manager starts to increase the volatility of the investment strategy in order
to reap more profits from the incentive fee contract.
Fig. 4.5 Option value of 20% incentive fee s a function of the incentive fee for different levels of the
managers stake in the fund (v). Source: Kouwenberg and Ziemba (2007)
Figure 4.6 shows the optimal volatility of the fund returns Y (T)/Y(0) as a
function of the manager's stake in the fund, given the incentive fee of 20%.
The fund manager greatly invreases the fund's return volatility as the
manager's own stake in the fund decreases, to maximize the expected
payoff of the incentive fee.
The increase of the value of the incentive fee, due to this change in
investment behaviour, is as much 2125% in this example; essentially from
to 17% of initial fund value
The theory thus suggests that it is best to invest in hedge funds where the
fund managers eat their own cooking and invest preferably 30%+ of the
fund's assets from their own resources. As I have repeatedly argued, the
recipe for hedge fund disasters is: to overbet and not diversify in all the
possible scenarios.
A super important point here relevant to the 2007-2009 financial morket
meltdown that was teiggered by eiskt subprime mortgages is that a rule that
banks and other institutions keep 30%+ of their mortgages would lower
their risk as they would take more care in granting mortgages. Canadian
banks typically hold 100% of their mortgages. This leads to carefully
considering the ability of the borrowers to service their mortgages and their
default rates were very low and house prices are at an all time high. This
also extends to other instruments in the current crisis, relates to some of the
regulations that have been suggested, and would probably have avoided the
GS problem as well!
Fig. 4.6 Optimal volatility of fund returns with incentive fee of 20% as a function of the incentive fee
for different levels of the managers stake in the fund (v)
As we see from this model, the hedge fund manager has a great incentive
to increase risk if they do not have much of their own money in the fund
and that can easily lead to large losses. An especially dangerous situation is
the multi-billion dollar fund with a hot fund manager who has huge bets in
a narrow area with little in the fund getting high fees stored away. Since the
penalties for a blowup are very low, this manager's incentive is to shoot for
the fences and if fired after a blowup, walk away with the previous fees and
get another hedge fund job. Examples include Victor Niederhoffer, John
Merriwether, Brian Hunter and others. Some examples are discussed in
Chapter 26.
1 Edited from Wilmott, September 2007.

2Their next move seems to be into Barclays which is bidding on ABM-AMBRO but lost out to RBS
which lost billions on this deal by over paying for ABM-AMBRO.
Incentives and Risk Taking in Hedge Funds1
An update on the financial markets and why ownership is so very important
There are a number of studies in the literature concerning hedge fund

performance and their use in diversified multi-strategy portfolios. The key is
to find outstanding hedge funds that generate consistent high returns.
Examples are in Ziemba and Ziemba (2007). Here we just discuss average
hedge funds. Given the high fees that average 2+20, it is a challenge to make
back these fees and have enough left to have outstanding net gains.
Kouwenberg and Ziemba (2007) did an empirical study using 2078 hedge
funds and 536 fund of funds from Januaty 1977 to November 2000 using the
Zurich Hedge Fund Universe, formerly known as the MAR hedge fund
database, provided by Zurich Capital Markets; see Table 5.1. The database
includes a large number of funds that have disappeared over the years, which
reduces the impact of survivorship bias. They analyzed the data from
January 1995 to November 2000 since It he database keeps track of funds
that di sappe ar starting January 1995. There were 722 alive hedge funds and
470 dead hedge funds - observe the high number of failures! For the fundt of
funds, there were 307 alive and a tull t 44 dead.
The return dat a s net of management fees and net of incentive fees. The
hedge funds in the database are classified into eight different investment
styles by the provider: Event-Driven, Market Neutral. Global Macro, Global
International, Global Emerging, Global Established, Sector and Short-
Sellers. They merged the styles Global International, Global Established and
Global Macro into one group, denoted Global Funds, as these three styles
have similar investment style descriptions. Global Emerging funds is a
separate category, denoted Emerging Markets, as the funds within this style
are often unable to short securities and emerging market funds have quite
different return characteristics compared to the other global funds.
Table 5.1: Characteristics of the Hedge Fund Data*
*Descriptive statistics of the hedge funds in the database are displayed. The cross-sectional mean,
median and standard deviation of the incentive fee, the management fee, assets under management and
the length of the time series of return observations per fund are listed respectively. The sample period
was January 1995 to November 2000. ‘Alive’ funds were still in the database in November 2000,
while ‘dead’ funds dropped out before this date. Funds that did not report fees were excluded.
The median incentive fee for hedge funds is 20% which is the industry
standard, and 71.4% of the funds used it. Only 8.5% of all hedge funds did
not charge an incentive fee. The median management fee was 1%. The
majority of funds (71.5%) charged a fee between 0.5% and 1.5%, while only
4.2% of the funds did not charge a management fee. An investor in fund of
funds had to pay fees to the fund of fund manager. On average, fund of funds
charge slightly lower fees than individual hedge funds, although the median
incentive fee is still 20% (dead and alive funds combined). Only 6.2% of the
funds of funds did not charge an incentive fee. The median management fee
of the funds of funds is 1%.
Empirical studies of incentives and risk taking in the literature typically
test whether funds with poor performance in the first half of the year
increase risk in the second half of the year, (see e.g. Brown, Harlow and
Starks 1996, Chevalier and Ellison 1997 and Brown, Goetzmann and Park
2001). The idea behind this approach is that funds with an incentive fee, or
facing a convex performance-flow relationship, will increase risk after poor
performance in the first half of the year to increase the value of their out-of-
the-money call option on fund value. Considered within the context of the
prospect theory framework Kouwenberg and Ziemba used, such a test is less
meaningful. Loss averse fund managers will always increase risk as their
wealth drops below the threshold, regardless of incentive fees. A more
distinguishing effect of incentive fees within the prospect theory framework
is that incentives reduce implicit loss aversion and lead to increased risk
taking across the board, even at the start of the evaluation period. We
therefore test if the risk of hedge funds returns increases as a function of the
fund's incentive fee.
Hedge fund returns are non-normal due to the dynamic investment
strategies of the funds (see Fung and Hsieh 1997, 2001 and Mitchell and
Pulvino 2001). Still, empirical studies of the relationship between risk taking
and incentives in hedge funds only consider volatility as a risk measure
(Ackermann, McEnally and Ravenscraft 1999, Brown, Goetzmann and Park
2001 and Agarwal, Daniel and Naik 2002), even though volatility can not
fully capture the non-normal shape of hedge fund return distributions.
Kouwenberg and Ziemba thus focus on non-symmetrical risk measures,
namely the 1st downside moment and maximum drawdown, as well as the
skewness and kurtosis of hedge fund returns. The 1st downside (upside)
moment is defined as the conditional expectation of the fund returns below
(above) the risk free rate. Maximum drawdown is defined as the worst
performance among all runs of consecutive negative returns.
Table 5.2 shows the cross-sectional average of ten different risk and return
measures of the hedge funds in the database, conditional on the level of the
incentive fee. The risk measures are volatility, first downside moment
(relative to the risk free rate), maximum drawdown, skewness and kurtosis.
The return measures are the fund's mean return and 1st upside moment. And
three risk-adjusted performance measures; the Sharpe ratio, Jensen's alpha
and the gain-loss ratio. The gain-loss ratio is defined as the ratio of the 1st
upside moment to the 1st downside moment. Berkelaar, Kouwenberg and
Post (2003) demonstrate that the gain-loss ratio can be interpreted as a
measure of the investor's implicit level of loss aversion. The last column of
displays the p-value of an ANOVA-test for differences in means between the
incentive fee groups. The first row of shows that hedge funds without
incentive fee, on average, have considerably higher mean returns than funds
that do charge an incentive fee (means are significantly different between
groups). The difference in average return after fees between the 93 funds
without an incentive fee and the majority of funds with a fee of 20% is 8.5%
per year. This gap of 8.5% reduces to 6.2% if we control for differences in
investment style between the two groups. Another 3.8% of the performance
differential can be explained by the cost of the 20% incentive fee. Hence,
only 2.4% of the performance differential remains unaccounted for, which
could easily be due to sampling error and does not indicate any significant
difference in investment skills. Funds with an incentive fee cannot make up
for the costs of the fee. We do not find statistically significant evidence that
incentive fees lead to drastic changes in average volatility, 1st downside
moment and maximum drawdown of hedge funds.
Table 5.2: Hedge Fund Risk and Return Conditional on Incentive Fee*
*The average cross-sectional risk and returns of the hedge funds in the database are displayed,
conditional on the incentive fee charged by the fund. For each risk and return measure, the second row
shows differences relative to the mean within each of the eight hedge fund style group, in order to
control for differences in investment style. the sample period is January 1995 to November 2000. The
sample includes both funds that were still in the database in November 2000 and funds that dropped
out before this date. Only funds with at least 12 monthly observations are included. The last column
denotes the p-value of an ANOVA-test for equality of the means of the four incentive fee groups.
Kouwenberg and Ziemba do find significant differences in average

skewness and kurtosis between incentive fee groups. The latter finding
seems to be caused mainly by the relatively small group of funds with an
incentive fee in excess of 20%. Examining the results for the three risk-
adjusted performance measures, Sharpe ratio, alpha and gain-loss ratio, we
find significant differences between incentive fee groups. Funds without an
incentive fee achieve the best risk-adjusted performance on average, while
funds charging a below average incentive fee have relatively poor
performance. They conclude from Table 5.3 that incentive fees reduce the
mean return and risk-adjusted performance of funds, while the effects on risk
are not very clear-cut. They also analysed the data after correcting for
differences in investment styles by measuring deviations from the average in
each style group, but the conclusions are similar.
To control for other hedge fund characteristics such as fund size, age,
management fee and investment style group, Kouwenberg and Ziemba
estimate the following cross-sectional regression model for the hedge fund
risk and return measures
with i = 1, ...,I being the observations of the variables in (5.1) such as ai,
etc, I and independently normally distributed, where ai denotes
the cross-sectional hedge fund statistic under consideration of fund i = 1,...,
I, dih is a dummy which equals one if fund i belongs to hedge fund style h =
1,... ,H and zero otherwise, if is the incentive fee, mfi the management fee,
navi is the mean net asset value of the fund and agei is the number of years
that the fund is in the database.
Table 5.3 reports the cross-sectional regression results. Columns 2 to 6,
denoted by Regression A, refer to regression model (5.1). Columns 7 to 11,
denoted by Regression B, refer to a slightly modified version of the model,
which uses a dummy variable for the incentive fee and a dummy for the
management fee; the dummy variables are one if a fee is charged and zero
otherwise. Funds with higher fees earn significantly lower mean returns. The
only other significant effect of incentive fees is a reduction of Sharpe ratios
and alphas (only in Regression B, with incentive fee dummies). There is no
significant effect of incentive fees on any of the five risk measures at the 5%
confidence level. However, there is an economically relevant increase of the
1st downside moment and the maximum drawdown due to incentive fees, as
the estimated coefficients are large. Moreover, the increase in the 1st
downside moment is significant at the 10% level in both regressions.
Kouwenberg and Ziemba repeated the empirical analysis for the fund of
funds in the database. They regressed on log of volatility to reduce the non-
normality of the residuals (skewness) Table 5.4 displays the cross-sectional
average of the ten risk and return measures, conditional on the level of the
incentive fee. They used three incentive fee groups instead of four, due to the
relatively small number of fund of funds (403 in total).
Again, Kouwenberg and Ziemba find significant differences between the
average mean returns of the incentive fee groups. Fund of funds with high
fees, earn higher returns on average. The 1st upside moment was also
significantly different across groups and larger for fund of funds with higher
fees. There were no significant differences in the five risk measures between
groups. The three risk-adjusted performance measures, Sharpe ratio, alpha
and gain-loss ratio, are significantly different across groups and relatively
large for fund of funds with high fees
Table 5.3: Cross-sectional Regressions of Hedge Funds Risk and Returns*

*The results of a cross-sectional regression are displayed of the risk & return measures of 1114 hedge
funds in the database on a constant, the level of the fund's incentive fee (if), management fee (mf), the
natural logarithm of the assets under management (size), the number of monthly observations (age)
and a dummy indicating that the fund dropped out of the database (dead). The risk and return
measures are differences relative to the average within each of the eight hedge fund style groups, to
control for differences in investment style. The explanatory variables if, mf and ln(size) are scaled by
the median value, while age is measured in years. The first row of results is based on normally
distributed residuals (OLS), while the second row is based on a skewed Student-t distribution. The
sample period is January 1995 to November 2000. Only funds with at least 12 monthly observations
are included. White heteroskedasticity-consistent p-values are reported below the estimates. Below the
estimated number of degrees of freedom df we report the p-value for the test of the null-hypothesis
1/df = 0, i.e. normal tails. LR is the likelihood-ratio test of 1/df = 0 and s = 1, i.e. normality of the
residuals. *Denotes signiffcance at the 5% level. Log-transformation of the dependent variable has
been applied for the second row of results.
Table 5.4: Fund of Funds Risk and Returns Conditional on Incentive Fee: The Average
Cross Sectional Risk and Return of the Fund of Funds
Table 5.5 contains the estimation results of the cross-sectional

regression model (1) for the fund of funds. The coefficient of the incentive
fee variable was significantly positive in the cross-sectional regression on
the 1st upside moment, volatility, maximum drawdown and gain-loss ratio
(at the 5% level). There was an economically relevant positive impact on the
mean return, 1st downside moment, skewness and Sharpe ratio as well,
based on the magnitude of the estimated coefficients. Hence, for the fund of
funds in the database Kouwenberg and Ziemba found that higher incentive
fees were linked to increased upside potential and increased risk taking.
Risk-adjusted returns increased as well, so investors seem to be better off
with fund of funds that charge higher incentive fees. In the case of
management fees, Table 5.5 shows that they are a drag on performance:
higher fees significantly reduce average returns, Sharpe ratios and alphas.
A potential explanation for the positive relationship between incentive
fees and (risk-adjusted) returns in Tables 5.4 and 5.5 is that fund of fund
managers with incentive fees opt for a more risky basket of hedge funds to
increase the value of their call option on fund value, leading to more upside
return potential and more risk as well. The fund of fund managers
themselves might argue that funds with better manager selection skills
generate higher returns and are therefore able to charge higher incentive
fees. A weak point of the latter story is that it does not explain why fund of
fund managers with better skills have more risky returns on average as well;
the skill advantage should allow good managers to achieve better returns,
while taking less risk.
Incentive fees reduce the manager's implicit level of loss aversion, leading
to increased risk taking. However, if the manager's own stake in the fund is
substantial risk taking will be reduced considerably.
Kouwenberg and Ziemba show that the fund manager increases the value of
the incentive option by increasing the volatility of fund returns.
The Kouwenberg and Ziemba cross-sectional analysis showed that hedge
funds with incentive fees have significantly lower mean returns (net of fees)
and worse risk-adjusted performance. The difference was 8.5% per year.
However, if one controls for investment style, the 8.5% gap becomes 6.2%
and the cost of the assumed incentive 20% fee is 3.8% reducing the
difference to 2.4%. There was no significant effect on volatility, but the 1st
downside moment of returns increases substantially in the presence of
incentive fees (significant at the 10% level). Their results illustrate the
importance of using downside risk measures, given the non-normality of
hedge funds returns. Funds of funds charging higher incentive fees have
more risky and higher returns on average. Hence, funds of funds take more
risk in response to incentive fees. It seems unlikely that fund of fund
managers with higher incentive fees are more skillful, as that story does not
explain why risk taking increases as well as a function of incentive fees.
Based on the empirical results, investors should evaluate individual hedge
funds with incentive fee arrangements critically. The presence of a 20%
incentive fee reduces a fund's average after-fee return by 2.93% (absolute
reduction) and the Sharpe ratio by 0.16, compared to a fund without
incentive fee and assuming other things equal. Among fund of funds, the
impact of incentive fees appears to be a shift upward in both risk and return,
without adverse consequences for risk-adjusted performance. The presence
of a 20% incentive fee increases the average after-fee return by 2.87% and
volatility by about 4.5%, while the fund's alpha and Sharpe ratio are not
affected significantly.
An interesting question for further research is why the relation between
incentive fees and risk taking is relatively mild for individual hedge funds,
but quite strong among fund of funds. A potential explanation fitting in the
theoretical framework of this paper is that managers of individual hedge
funds tend to have a larger stake in their own fund than fund of fund
managers. Alternative explanations might be that peer group pressure is
stronger among managers of individual hedge funds, or differences in high-
water mark provisions.
In conclusion, the Kouwenberg-Ziemba study provides general points in
the theoretical part and suggestive points in the emperical study. While dated
with many changes in the hedge fund industry, it is clear that making back
the fees is difficult. So care must be exercised in choosing hedge fund
managers but there are great ones out there. If Goldman Sachs and others
would take 30%+ of all deals they put together, there would be a lot less
need for the financial reform the Obama administration and the EU countries
are attempting to enact. The key is to deliver and participate in the bets.
Table 5.5: Cross-sectional Regressions of Fund of Funds Risk and Returns*

*The results of a cross-sectional regression are displayed of risk & return measures of 403 fund of
funds in the database on a constant, the level of the fund's incentive fee (if), management fee (mf), the
natural logarithm of the assets under management (size), the number of monthly observations on the
fund (age) and a dummy indicating that the fund dropped out of the database (dead). The explanatory
variables if, mf and ln(size) are scaled by the median value of the series, while age is measured in
years. The first row of results is based on normally distributed residuals (OLS), while the second row
is based on a skewed Student-t distribution. The sample period is January 1995 to November 2000.
Only funds with at least 12 monthly observations are included. White heteroskedasticity-consistent p-
values are reported below the estimates. Below the estimated number of degrees of freedom df we
report the p-value for the test of the null-hypothesis 1/df = 0, i.e. normal tails. LR is the likelihood-
ratio test of 1/df = 0 and s = 1, i.e. normality of the residuals. *Denotes significance at the 5% level
Log-transformation of the dependent variable has been applied for the second row of results.
1 Edited from Wilmott, July 2010.

Evaluating Superior Hedge Funds1
Analyzing those who regularly beat a vanilla buy and hold strategy
Three Kelly Investors: George Soros, Warren Buffett and John Maynard
Keynes
My original motivation with the DSSR was to try show that Warren Buffett
was a superior investor to the Ford Foundation. Indeed he had a lower
Sharpe ratio but as shown in Figure 6.1 he had more final wealth using
1985–2000 data.
The downside symmetric Sharpe ratio

The DSSR as it is called was discussed in Ziemba (2005) following its use in
Japan in Ziemba and Schwartz (1991). The idea is simple. If great investors
are to be evaluated as those who do not lose much and have many gains,
then do not penalize them for these gains. The Sharpe ratio, namely
penalizes gains as well as losses through σp so if gains are large, S is
reduced. Here μp is the portfolio mean returns, RF , the risk free asset, and σp
is the standard deviation. So the DSSR only uses
Fig. 6.1 Growth of assets, log scale, various high performing funds, 1985-2000. Source: Ziemba
(2005) using data from Siegel, Kroner and Clifford (2001)
So we replace σp with σx− where means take only minus values.

The total variance is twice the downside variance. Pictorially, one erases the
actual gains and takes as fictitious gains the mirror image of the losses. So
the risk is symmetric. Hence the name, downside symmetric Sharpe ratio.
The corresponding downside symmetric information ratio is
Observe that the higher the DSSR is the better is the fund with the range
of the measure from zero, when when there are no losses.
The Renaissance Medallion Fund

The Renaissance Medallion Fund uses mathematical ideas such as the Kelly
criterion to run a superior hedge fund. I am pleased to have had a minor role
in teaching Jim Simons about the Kelly criterion in 1992. The staff of
technical researchers and traders, working under mathematician James
Simons, is constantly devising edges that they use to generate successful
trades of various durations including many short term trades that enter and
exit in seconds. The fund, whose size is in the $5-8 billion area, has very
large fees (5% management and 44% incentive) which together amount to
about half the gains. Yearly fees are paid out so the fund does not grow
much in size. Despite these huge fees and the large size of the fund, the net
returns have been consistently outstanding, with a few small monthly losses
and high positive monthly returns; see the histogram in Figure (6.3. Table
6.3 shows the monthly net returns from January 1993 to April 2005, along
with quarterly and yearly returns. There Ȝwere only 17 monthly losses in
148 months and 3 losses in 49 quarter's and no yearly losses in these 12+
years of trading in our data sample. Table 6.5 lists the U.S. T-bill interest
rates in percent, January 1993 to April 2005, yearly, quarterly and monthly.
These numbers were compiled from http://treasurydirect.gov. This section
up-dates an earlier discussion in Ziemba and Ziemba (2007).
The yearly DSSR of 26.4 versus an ordinary Sharpe ratio of 1.68 shows
that the DSSR is needed to show Mndallion's true brillianct. The effect is
less with quarterly DSSR's 11.6 versus 1.09 for S, and a monthly DSSR of
2.20 versus 0.76 for the S. In calculating the standard deviations for both of
these ratios, we used the net returns minus the risk free interest rates listed in
Table 6.2.
Fig. 6.2 Histogram of monthly % returns of the Medallion Fund, January 1993 to April 2005
We calculated the annual standard deviation for the DSSR by multiplying

the quarterly standard deviation by 2 because there were no negative annual
net returns. All calculations use arithmetic means. We know from Ziemba
(2005) that the results using geometric means will have essentially the same
conclusions.
Figure 6.3a shows the percent monthly returns sorted in increasing order
and Figure 6.3b shows thr accumulation of wealth over time; asruming thrt
the fund had an initial wealth of 100 dollars on December 31, 1992.
Medallion's outstanding yearly DSSR of 26.4 is even higher than
Princeton Newport's 13.8 during 1969–1988. Jim Simon' s Medallion fund is
near or at the top of the worlds' hedge funds. Indeed, the amount that Simons
earned, $1.4 billion in 2005, was the highest in the world for hedge fund
managers, and his $1.6 billion in 2006 was the second best and 2007-2012
were similar. The fund was closed to all but about six outside investors plus
employees. But in recent years even those six are out or reduced in their
percent in the fund. Monthly data since 2005 is unavailable to us, however
the 2006-2009 estimated yearly net returns, according to Insider Monkey
(2010), were 44.3%(2006), 73.0% (2007), 80.0% (2008), and 39.0% (2009).
Table 6.1: Net returns in percent of the Medallion Fund, January 1993 to
April 2005, Yearly, Quarterly and Monthly
Table 6.2: Annualized T-bill interest rates in percent, January 1993 to April 2005,
Yearly, Quarterly and Monthly
The monthly, quarterly and yearly values of both the Sharpe Ratio and the
Symmetric Downside Sharpe Ratio (DSSR) are shown in Table 6.3. In
calculating the standard deviations for both of these ratios, we used the net
returns minus the risk free interest rates listed in Table 6.4 were used.
Figure 6.3a shows the % monthly returns sorted in increasing order.
Figure 6.3b shows the accumulation of wealth over time assuming an initial
wealth of 100 dollars on Dec 31, 1992.
The annual standard deviation for the DSSR was calculated by
multiplying the quarterly standard deviation by 2 because there were no
negative annual net returns. All calculations use arithmetic means. We know
from Ziemba (2005) that the results using geometric means will have
essentially the same conclusions.
Fig. 6.3 Medallion Fund, January 1993 to April 2005
Table 6.3: Medallion Fund net returns, %, January 1993 to April 2005.
Source: Ziemba and Ziemba (2007)
Medallion's outstanding yearly DSSR of 26.4 was the best I had seen up
to 2010. Gergaud and Ziemba (2012) discuss some commodity trading funds
with even higher DSSRs and their results are summarized below. This is
even higher than Princeton Newport's 13.8 during 1969-1988; see Ziemba
and Ziemba (2007). Medallion's yearly Sharpe of 1.68 is decent but not
outstanding. The DSSR is needed to capture the true brilliance of this hedge
fund. Jim Simon's Medallion fund is near or at the top of the world's hedge
funds. Indeed, the amount that Simons earned, $1.4 billion in 2005, was the
highest in the world for hedge fund managers, and his $1.6 billion in 2006
was the second best. Since the fund is closed to all but about six outside
investors plus employees all of whom pay these fees, we watch with envy.
Table 6.4: Sharpe and Downside Symmetric Sharpe Ratios for the Medallion Fund,
January 1993 to April 2005. Source: Ziemba and Ziemba (2007)
Table 6.5: Annualized T-bill interest rates, %, January 1993 to April 2005,
yearly, quarterly, and monthly. Source: Ziemba and Ziemba (2007)
Renaissance Medallion continues to win but three new Renaissaince funds

(currency, trend and long-short) which do not do lightening fast trades in 3-7
seconds like Medallion, have done poorly. Simons has organized Medallion
nicely, the investors are basically the employees who pay the 5+44% fees on
what they gain. So the employees gain and Simons makes billions!
Evaluating superior funds in the UMASS hedge fund data base

using the DSSR
Figure 6.4 shows the wealth path of the great economist John Maynard
Keynes running the endowment of King's College, Cambridge University
from 1927 until his death in 1945. Keynes was a very aggressive investor
and had a violent wealth path. Ziemba (2003) estimated that his overall
strategy is about 80% full Kelly maximum expected log strategy plus 20%
cash or a utility function of the concave risk averse negative power utility
function wealth of u(w) = –w−0.25. See MacLean, Thorp, and Ziemba (2010)
for the good and bad properties of this strategy and MacLean, Thorp, Zhao
and Ziemba (2011) for simulations of the typical behavior. Keynes actually
lost more than half the portfolio which was even more than the market index
in the first two years. Over the 19 years which included the depression of the
1930s and its recovery and World War II, he had a geometric mean of 9.12%
versus the market index of -0.89%. So he was a great investor but his wealth
path was not smooth.
Fig. 6.4 Graph of the performance of the Chest Fund, 1927-1945. Source: Ziemba (2005)
Two full Kelly investors who hold very few concentrated positions are
Warren Buffett and George Soros. Their wealth paths from December 1985
to April 2000 plus the Windsor fund of George Neff, the Ford Foundation,
the Tiger Fund of Julian Robertson, the S&P500 total return index plus T-
bills and US inflation are shown in Figure 6.1. Table 6.6 shows the
concentration of their equity portfolios on September 30, 2008. Soros had a
50.53% position in Petroleo Brasileiro, plus 11.58% in the Potash
Corporation of Saskatchewan, 5.95% in Walmart, 4.49% in Hess
Corporation and 3.28% in Conoco Phillips. Buffett has many close to 10%
positions such as 8.17% in Conoco Phillips, 8.00% in Proctor and Gamble,
5.62% in Kraft Foods and 3.55% in Wells Fargo. Both of them, especially
Soros, trade futures, options and other derivative positions as well.
While both of these famous billionaire investors have had many gains,
these also have many monthly losses. For example, Berkshire Hathaway had
58 and Quantum 53 losing months out of 172 total months in the data
sample. Also the gains of these two investors were very high as shown in the
right side of Figure 6.5 which ranks all the monthly returns, but losses were
high too as shown on the left side. In both tails these two investors are more
extreme than the others.
Table 6.6: Top ten equity holdings of Soros Fund Management and
Berkshire Hathaway, September 30, 2008. Source: SEC Filings
We cannot use the DSSR measure to evaluate the Yale endowment as
monthly data is unavailable. There are also other great trading records that
we cannot analyze because of lack of availability of monthly data including
two billionaires who the second author has worked with. Both are futures
traders. Data bases cannot include all funds such as the very best who are
closed and the very worst who are secretive. One other extremely successful
trading outfit that John Mulvey has pointed out to us is the Commodity
Corporation (CC) formed by Professors Paul Samuelson and Paul Cootner
and others such as Amos Hostetter. The CC was the training ground for great
traders such as Paul Tudor Jones, founder of Tudor Investment Corporation,
Louis Bacon, founder of Moore Capital Management, Grenville Craig,
founder of Tiverton Trading, Bruce Kovner, founder of Caxton Associates,
Christian Levett, founder of Clive Capital LLP, Michael Marcus, a leading
commodities and currency trader, Jack D. Schwager, an author on financial
topics and hedge fund manager, Ed Seykota, a computer scientist, technical
trader and pioneer in System Trading and Willem Kooyker, founder of
Blenheim Capital. Their performance was 89% per year after fees for their
first ten years. Lintner (1983) did analyze CC and other commodity funds
and futures account managers for 42 months from July 1979 to December
1982. CC had a net mean of 8.42% per month with a standard deviation of
21.709. Lintner's unpublished paper is highly recommended but it uses a
mean-variance approach rather than what we do here. Harry Markowitz
pointed out to me that the paper suffers from the statistical flaw of not
counting all funds, dead or alive. However, the CC results stand.
But we can use it to evaluate the funds in table 6.1. table 6.7 shows that
Warren Buffett's Berkshire Hathaway is the only one that improves when
you switch from S to DSSR but he still does not beat the Ford Foundation
and the Harvard endowment is also slightly better than Berkshire Hathaway
using quarterly data.
Table 6.7: Comparison of ordinary and symmetric downside Sharpe yearly performance
measures, monthly data and arithmetic means. Source: Ziemba (2005)
Ordinary Downside
Ford Foundation 0.970 0.920
Tiger Fund 0.879 0.865
S&P500 0.797 0.696
Berkshire Hathaway 0.773 0.917
Quantum 0.622 0.458
Windsor 0.543 0.495
Fig. 6.5 Return distributions of all the funds, quarterly returns distribution, December 1985 to March
2000
Why is this? Figure 6.5 shows that Buffett and Soros had many more large
gains than the other funds but also more losses when we rank the worst to
best monthly returns. As noted above, Berkshire Hathaway and Quantum
had losses in about a third of the 172 months in the sample period. Observe
that in the tails, Berkshire and Quantum (George Soros's fund) are
asymptotically equivalent. It is clear that Buffett and Soros do not care about
monthly losses but aim to maximize long run wealth. This and the high
concentration of positions leads to the conclusion that they are full Kelly
investors.
Smoother wealth paths are those in Figure 6.6 of Bill Benter of the Hong
Kong racing syndicate and Ed Thorp of the Princeton Newport hedge fund.
Thorp had the amazing record of just three monthly losses in twenty years of
trading from 1968 to 1988. This DSSR=13.8.
Fig. 6.6 Smoother wealth paths
Outstanding funds in the UMASS DHF universe

Gergaud and Ziemba (2012) have found some funds with even higher
DSSRs then Renaissance Medallion. For example, Logos Trading Inc, a
commodity trading advisor, had a mean return of 8.5% per month from its
inception in June 1984 to its closing 157 months later in June 1997. Its
Sharpe ratio was 2.77 and its DSSR was 64.4. Figure 6.7 show the wealth
graph, assets under management, net asset value and monthly losses and
gains.
Figure 6.8 shows the results for the GJ Investment Fund of Flemington,
New Jersey managed by Jeeva Ramaswamy whose DSSR is 491.8! There
were very high mean returns of 15.5% per month and only one monthly loss
yield this huge DSSR. The stated objective of the Fund is to, over the long
term, outperform all three major U.S. stock market indices and over 90% of
all U.S. mutual fund and hedge fund managers using a non-levered Warren
Buffett style value investing approach. The assumption is that the market
over reacts to good and bad news resulting in volatile stock price movements
that do not correspond to the company's longterm fundamentals, hence
undervalued stocks can be purchased. Special situations investing is also
used and there are investments in India and China. Fees are just
Fig. 6.7 Logos. Source: Gergaud and Ziemba, 2012
Fund name Logos Trading Inc (ID 5770)
Fund type CTA
Company Name Sangamon Trading Inc (ID 1436)
Inception to last reported date 30 Jun 1984 - 20 Jun 1997
Number of months (negative) 157 (29)
Total return 374,526.30
Geometric mean return per month 1.085
Assets under management 57,000,000
Strategy Discretionary
Minimum investment 100,000
Annualized 3-month T-bill rate 5.80%
Sharpe ratio 2.773
Information Ratio 2.874
DSSR 64.40
DISR 66.72
for performance at 25% above a 6% hurdle rate. To date, the Fund has
consistently outperformed the Dow30, the S&P500 and the Nasdaq100.
Figure 6.9 has the wealth graph of the FMG Fund in the Federated MGT
Group of funds plus the assets under management and the ordered monthly
losses and gains. This CPO reported results from January 1987 to July 2003,
some 199 months. It had a geometric mean of +2.3% per month and a
Sharpe ratio of 12.35, higher than any other fund in the UMASS DHF
database. Apparently there were no monthly losses so the DSSR=DISR=∞!
This sounds too good to be true and indeed the CFTC on December 7, 2010
announced that the principals were ordered to pay $26 million for running a
Ponzi scheme with 140 participants, see Greenberg (2011). So FMG is
suspect and had Madoff madeup elements in its reporting and misuse of
funds. See Thorp (2009) for a discussion of his confidential 1991 analysis of
the Madoff scheme which was exposed in December 2008. Our point in this
paper is to discuss the use of the DSSR measure and an infinity measure is
suspect!
Fig. 6.8 GJ Investment Fund. Source: Gergaud and Ziemba, 2012

Fund name GJ Investment Fund (ID 14309)
Fund type Hedge Fund, open
Company Name GJ Investment Funds I L (ID 6260)
Inception to last reported date 01 Nov 2008 - 31 Jan 2010
Number of months (negative) 13(1)
Total return 8.708
Geometric mean return per month 1.155
Strategy Equity Long/Short
Annualized 3-month T-bill rate 0.14%
Sharpe ratio 3.217
DSSR 491.83
DISR 491.97
To understand more fully the characteristics of the DSSR measure we

have plotted in Figure 6.10(a) the DSSR-Sharpe ratios versus the ranking
based on the Sharpe ratio with the right hand side vertical axis being the
proportion of negative months. Figure 6.10b is the same except it estimates a
curve for the individual funds shown as dots in Figure 6.10a.
Figure 6.10a shows on the left-hand side vertical axis the difference
between the DSSR and Sharpe ratios. The second vertical axis plots the
proportion of negative months per fund as proxied by a local polynomial
smoothed line. The funds are ordered according to their Sharpe ratio on the
horizontal axis (0 being the worst fund here that matched with the
constraining criterion we adopted). From this graph we see that the
difference between the two competing measures is increasing sharply near
the top of the scale, i.e. for the very best funds and that the DSSR is far more
sensitive to the number of negative months than the regular Sharpe ratio is.
Fig. 6.9 FMG Fund - A Ponzi scheme, made up results!. Source: Gergaud and Ziemba, 2012
Fund name FMG Fund (ID 6887)
Fund type CPO
Company Name Federated MGT Group (ID 3532)
Inception to last reported date Jan 1987 - Jul 2003
Number of months (negative) 199 (0)
Total return 83.91
Geometric mean return per month 4.839%
Strategy Equity Long/Short
Annualized 3-month T-bill rate 12.35
Sharpe ratio 3.217
DSSR ∞
DISR ∞
Figure 6.10b plots on the left-hand side horizontal axis the value for both
ratios. Again, each ratio is summarized by a local polynomial smoothed line
for ease or interpretation. These two tables tell more or less the same story.
Figure 6.8 is a summary of various types of funds regarding their
performance and tables 6.9-6.12 list the Commodity Pool Operators (CPO),
Commodity Trading Advisors (CTA), Fund of Funds (FOF) and Hedge
Funds (HF) ranked by DSSR that have a mean return of at least 1.5% per
month and at least 100 months of reported trading.
Fig. 6.10 DSSR-Sharpe ratios versus proportion of negative months for UMASS DHF database
individual funds. Source: Gergaud and Ziemba, 2012
Table 6.8: Some summary statistics by fund type for the UMASS DHF
database. Source: Gergaud and Ziemba, 2012
Final Remarks
The new data since 2005 reaffirm and support the previous conclusion that
the DSSR measure is useful to measure more accurately superior investors
than the ordinary Sharpe measure which assumes normal not skewed return
distributions. The DSSR measure becomes extremely high when there are
very few losses and high mean returns which is a good measure of superior
performance.
See Tables 6.9-6.13 for the data.
Table 6.9: Performance and Main Features of Commodity Pool Operators.

Source: Gergaud and Ziemba, 2012
Table 6.10: Performance and Main Features of Commodity Trading

Advisors. Source: Gergaud& Ziemba, 2012
Table 6.11: Performance and Mearinfo Fremataunrecse oaf nFdunMd oafi
nfunFdesatures of Multi-Strategy Funds of Funds. Source: Gergaud &
Ziemba, 2012
Table 6.12: Performance and Main Features of Hedge Funds. Source:

Gergaud & Ziemba, 2012
Table 6.13: Performance and Main Features of Hedge Funds. Source:
Gergaud & Ziemba, 2012
Investment in Own-Company Stock1
Stock ownership decisions in defined contribution pension plans
It is surprising how much of the average defined contribution (DC)

company pension plan has been invested in own-company stock. In late
2001, Enron collapsed and their stock fell 99% from $90 to under a dollar
and employees lost their jobs and also lost most of their pensions. There is
considerable risk in having a pension fund largely in one asset and the risk
is even larger if that asset is also correlated with one's income. Enron
employees lost over a billion dollars, some 60% of their 401(k) pension.
This is a classic example of overbetting, lack of diversification and being
hit by a bad scenario.
Table 7.1 shows that for many major companies, own-company stock has
been a very high percent of 401(k) plans. The stock price moves in 2001,
2002 show how big short term losses can be.
Table 7.1: Share of company own stock in 401(k) pension plans. Source: Updated
from The Economist, December 15, 2001, p. 60
Mitchell and Utkus (2002) observe that there are about five million
401(k) plan participants that hold 60% of their assets in company stock but
those that do generally have large amounts. In total, company stock is about
19% of assets. But for those who have any company stock, it is 29%.
Employees have a lower percentage of stock, 22%, when they have free
choice versus 53% when the company decides.
Why do companies and employees invest so much of their own company
stock in their pensions? Companies can either purchase shares in the open
market, as some like Microsoft do, or they can just issue shares just like
options to key employees slightly diluting thir stock price which is
economical for the company. In the words of The Economist (December 15,
2001, p.60)
Employees who invest in their company's shares solve two

problems, in theory. They resolve the issue of agency costs that
arises between shareholders and the people hired to work on their
behalf. And, .. . they reap the benefits of capital appreciation, a
fundamental component of capitalism.
The results can be spectacular, America is filled with tales of people
who held jobs as cash-register clerks at Wal-Mart, or on the diaper-
making line at Procter & Gamble, who survived on their wages but
have made fortunes through steady accumulation of company stock
in retirement plans.
There are many other spectacular positive examples such as Microsoft, Intel
and Nokia.
Employees can frequently purchase own company shares at a discount to
current market price or acquire the shares through options given for free.
Also, there is the pressure of corporate culture. I saw that in Japan while
there in 1988-89 where employees of the Yamaichi Research Institute were
obliged by moral suasion and peer pressure to buy Yamaichi Security stock
which later went bankrupt in 1995. Enron has refocused this risk that has
been around a long time.
Mitchell and Utkus (2002) remind us how volatility destroys wealth.
They consider three workers who earn $50,000 per year and contribute 10%
to a 401(k) with contributions and inflation at 3% per year. The stock
market index and company stock are assumed to return 10% per year with
annual standard deviations of 20% and 40%, respectively. After 30 years,
the median employee, who invested 100% in the market index had
$830,000, with 50-50 splits it was $615,000 and with 100% in company
stock, it's $411,000. This is because of the geometric-arithmetic inequal¬ity
caused by the volatility: gaining 50% and then losing 50% does not make
one even; 100 becomes 75, with a geometric rate of return of -13.4%. The
greater the volatility, the lower the geometric mean, which determines long-
run wealth gains, for constant arithmetic mean.
What is the real risk of the own stock and job risk concentration?
Douglass, Wu and Ziemba (DWZ) (2004) have estimated this using mean-
variance and stochastic programming assets-only models.
DWZ consider the following situation: an investor chooses between the
market index (S&P500), a bond index (Lehman Brothers US aggregate),
cash and own company stock. The parameter assumptions, estimated from
1985 to 2002 monthly data from Datastream, mirror long run stock, bond
and cash returns from Dimson et al (2009) and Siegel (2008). Yearly mean
returns are (1.10, 1.05, 1.00, 1.125) for these four assets, respectively.
Standard deviations are (0.20, 0.04, 0.01, 0.50) and the covariance matrix is
These assumptions have mean returns relative to cash, and higher

expected returns but with much higher volatility for own company stock
(two and a half times the S&P500 which is typical for mid-cap equities).
Figure 7.1 shows the results from the mean-variance model as a function of
investor risk aversion with (7.1.b) and without (7.1.a) company stock and as
a function of company stock mean return (7.1.c). The shaded regions
indicate portfolio weights (leftscale). The diamonds indicate the expected
return of the optimal portfolios. When Arrow-Pratt absolute risk aversion is
8, there is a 60% stock, 40% bond mix.
Fig. 7.1 Mean-variance model. The optimal portfolios as a function of risk aversion (a,b) and
expected return on company stock (c). Shaded regions indicate portfolio weights (left scale). The line
represents the expected return on the optimal portfolio (right scale). Diamonds indicate values of
independent variable for which calculations were made. Results for the three asset case, with no own-
company stock, are in (a). Results for the four asset case are in (b) and (c). Source: Douglass, Wu and
Ziemba (2004)
This optimal portfolio has no own company stock holdings. Hence,

without trading constraints, it is not optimal with risk aversion of 8 to hold
own company stock. However, for investors with trading constraints, such
as the inability to short sell, owning some company stock can be optimal.
Company stock appears as an optimal portfolio choice if investor's risk
aversion is very low or their expected return for the stock is high. Notice
again how important it is to get the mean right. At a risk aversion of five,
the short-selling constraint becomes binding and the optimal portfolio
begins to shift to the riskier stock investment (Figure 7.1b). To obtain
company stock holdings above 50%, as observed in Table 7.1, requires a
risk aversion parameter below 0.5. Alternatively, company stock holding of
50% is obtained if the employee is presumed to have an expected return for
company stock over 50% (Figure 7.1c). Own company stock begins to enter
when its mean return approaches 20%, that is double the S&P500. The
expected return of own company stock must be over 50% or five times the
S&P500 for the optimal allocation of own company stock to reach 50%.
Our results demonstrate that the short selling constraint is not sufficient
to overcome the additional risk associated with owning company stock.
Highs company stock weightings can only be explained by some
combination of low risk aversion and/or high return expectations for
company stock. These results hold with or without including extreme
events. The inclusion of labour income risk m the models further reduces
the optimal holding in company stock.
Figure 7.2 shows the results of the mean-variance model when human
capital is modeled as an untradeable asset.
Fig. 7.2 Results of the mean-variance model when human capital is modeled as an untradeable
asset. Optimal portfolio properties are plotted as a function of risk aversion (a) and expected return
on company stock (b). Shaded regions indicate portfolio weights (left scale). The line represents the
expected return on the optimal portfolio (right scale). Diamonds indicate values of independent
variable for which calculations were performed. Plots are interpolated between calculated points.
Source: Douglass, Wu and Ziemba (2004)
This analysis assumes that all employee wealth is contained in the

company pension plan. This assumption is reasonable considering that
many North Americans save little beyond what enters their tax sheltered
accounts. However, Figure 7.1b shows what proportion of wealth would
have to be held outside the plan in order to support a 50% own stock
holding within the plan. An employee with risk aversion of 8 and 50% of
their pension plan in own stock company stock would have to have 50% of
their retirement savings outside the company plan.
The difference between mean-variance and stochastic programming:
(1) The mean-variance approach relies on distributional assumptions that
must be relaxed to study the portfolio choice problem for pension plan
investors.
• The normal distribution does not fit stock returns well in the
tails, particularly returns sampled at quarterly or greater
frequency. This is especially true for individual stocks.
• The probability that an individual stock will experience an
extreme negative event, such as bankruptcy, is greater than is
predicted by a best fitting normal distribution.
• Human capital can only be included in a mean-variance model as
an asset that follows the same distributional assumptions as the
financial assets.
• The mean variance model assumes static holdings throughout the
life of the portfolio. For pension plans that are held on the order
of decades, such trading constraints are unrealistic.
(2) An alternative DWZ develop is a discrete time stochastic optimization
model of the employee's investment problem which is a static one-
period version of the Geyer-Ziemba (2008) model, see Chapter 14.
• The employee's concave utility is expected discounted value of
terminal wealth minus a shortfall penalty.
• Concavity of the utility function is obtained with piecewise
linear convex shortfall penalties.
• The problem is equivalent to a large linear program for
computational pur-poses.
• To obtain results comparable to the mean variance model, the
penalty function is constructed to approximate a quadratic in the
shortfall.
• The expected penalty is approximately half the variance when
the expected return on the optimal portfolio equals the wealth
target.
• A judicious choice of target wealth minimizes the difference
between the two solutions.
The stochastic programming model is similar to a static one-period
version of Geyer and Ziemba (2008), see also Chapter 14. The decision
variables are the purchases and sales of each of N assets in each scenario.
The investor chooses an asset allocation at time 0, and receives investment
proceeds at time 1, which is a risk measure. The static stochastic
programming problem is
where Wit is wealth in asset i at time t. λ is a coefficient of risk aversion, and

c(M) is a convex function (risk measure) of the wealth shortfall, M at time
1. At time zero, the investor faces N balance constraints and a budget
constraint
The Ei are endowments of each asset. Pi and Si are purchases and sales
respectively of asset i, and t represents transaction costs. The time 1
constraints are
where are the thresholds for determining wealth shortfalls, and the Ri are
the realized returns.
Approximation of return distributions is a significant challenge in
stochastic programming. The solution of this problem requires a means of
approximating the expectations that appear in the objective function. The
usual method is to perform a discrete approximation to the integral, by
replacing a continuous multivariate distribution with a discrete distribution.
For multiperiod problems, the discrete distribution takes the form of a
scenario tree. See Douglas, Wu and Ziemba (2004) for the solution method
used with pseudo-random sequences.
Figure 7.3 shows results of the stochastic progrmming model. Here, the
employees íncome is a function of the own-company stock price.
Portfolio decisions for the base scenario are plotted in Figure 7.3a. The
overall pattern of the solution is similar to that obtained by mean-variance.
Since meanvariance penalizes excess returns as well as losses, mean
variance solutions for risk aversion equal to RA, will correspond to mean-
shortfall solutions with penalty parameter, λ = 2RA. In both cases company
stock disappears from the portfolio completely for risk aversions above
five.
Discussion of the results

• Various risk factors that dictate against the holding of company stock
has led most previous studies to presume that employee investment
decisions are the result of behavior that is inconsistent with rational
portfolio choice.
• Employers, as plan fiduciaries, may be in a position to influence
employee decisions and steer them towards company stock (Mitchell
and Utkus, 2002).
• Employees may interpret the channeling of employer contributions
into company stock as an endorsement of that investment (Mitchell
and Utkus, 2002).
• Alternatively, employees may choose company stock arise simply
because it is a listed investment option. Benartzi and Thaler (2001)
found that many DC plan investors follow some version of the 1/n
strategy; i.e. they divide their contributions evenly across plan
offerings.
Fig. 7.3 Solution of the employee's problem formulated as stochastic program. Figure (a) corre-
ponds to the base case discussed in the text. Portfolio properties obtained when human capital returns
are modeled as a logit function of wealth are plotted in (c). The logit function used in the model is
plotted in (b). In (a) and (c), portfolio weights (shaded regions/right scale) and expected return on the
optimal portfolio (line/right scale) are plotted as a function of the risk aversion parameter, λ. The
plots are interpolated between λ values indicated by diamonds. In (b), probability of job retention is
plotted versus company stock return. Source: Douglass, Wu and Ziemba (2004)
• In addition, employees may be myopic when evaluating risk of

company stock. John Hancock (2001), in a survey of DC plan
participants during a period of stock market growth, reported that DC
plan participants rated company stock as less risky than an equity
mutual fund.
• Other factors, such as loyalty and peer pressure considerations may
also influence employee investment decisions.
• Trading constraints have also been discussed as explanations for high
company stock holdings.
• Many companies that match contributions to pension plans, deposit
company stock.
• Often, an employee is restricted from trading this stock.
• However, our interest is in exploring potential explanations for
extremely high company stock weights.
• In most cases, any minimum holding constraint for company stock is
not bind¬ing. Employees hold more company stock than they have to
(Mitchell and Utkus, 2002).
• The results reinforce the conclusion that large holdings of company
stock in pension accounts cannot be explained by traditional models
of rational portfolio choice.
• Hence, explanation of the exceptionally high observed holdings
continues to rely on behavioral explanations.
• The problem with behavioral explanations of company stock holdings
is that they presuppose some ignorance on the part of the employee or
an ability of the employer to dupe the employee.
• However, large holding in company stock is a phenomenon that has
persisted for decades.
• Employees appear to have been making the same errors in their
portfolio choices for a long time.
• One would expect the irrationality of employee choices to lessen over
time as employees learn from previous actions and consequences.
• This leads us to suspect that there are other factors that need to be
included in rational choice models to explain company stock
holdings.
Grace Groner's legacy: A good long term buy and hold own-
company stock story.
There are many examples of people who bought stock cheap and held it a
long time and reaped huge returns in the end with the equity version of
compound interest being a major factor. Claude Shannon, the great
information theorist who influenced Kelly of the Kelly criterion and worked
with Ed Thorp, made 28% over a long period by largely investing in a few
big winners.
Grace Groner was orphaned at age 12. She graduated from Lake Forest
College in Illinois in 1931 having been sponsored by George Abott, a friend
of Groner's parents who raised Grace and her twin sister Gladys. She
worked as a secretary at Abbott Labs for 43 years. In 1935 she bought three
shares of this company for $60 each. She lived a frugal life. She had lived
in an apartment until a friend willed her the tiny house in a part of town
once reserved for servants on Chicago's North Shore. She did not own a car,
and walked to where she wanted to go. She traveled widely in her
retirement and donated $180,000 to a scholarship program at Lake Forest.;
see Keilman (2010).
As the stock gave dividends, she re-invested in more stock and there
were many stock splits, see Figure 7.4 for these since 1985. When she died
at 100 in January 2010, the stock was worth $7 million for a geometric rate
of return of 15.13% over the 75 years that she held the stock. The taxes she
paid on the capital gains are unknown. Abbott stock made its big gains well
before 2000 and has been flat since then. Had she switched out of Abbott
(ABT) into Apple (AAPL) around 2005, she would have had a full Kelly
type wild ride and gotten a Buffett style high final wealth as shown in
Figure 7.4. And since 2010, AAPL has increased much more, reaching over
$700 per share in September 2012. This compares favorably with the best
mutual funds over the past 45 years. Fidelity Magellan has returned 16.3%
and the Templeton growth fund 13.4% according to Morningstar versus
9.3% for the S&P500. Meanwhile, Warren Buffett running Berkshire
Hathaway made 22% since 1965 based on market price and 20.3% based on
book value, Buffett's preferred way to measure growth; see Mamudi (2010).
Siegel (2008) showed that Philip Morris returned about the 15% over a
similar but a bit shorter long period outdistancing essentially all other
stocks.
Fig. 7.4 Abbott share value compared with Apple, 1985 to 2010. Source: Yahoo Finance
She donated this money to the college to be used by students for

internships and to study abroad. Her house is now used by students.
1 Parts of this are adapted from Bertocchi et al. (2010).

Cutting Through the Hype on Sovereign Wealth
Funds1
Is the high profile of Sovereign Wealth Funds a one-hit-wonder or something sustainable in the near
term?
Sovereign wealth funds - governmentbacked investors from the Middle East

and Asia burst onto the front pages of the financial press midway through
2007, as they sought higher returns on their amassed assets. When the
realities of the bad bets on housing securities came back to haunt banks and
other financial intermediaries and roiled the credit markets, these funds took
on an L even greater prominence,emerging as the originators of most of the;
high profile takeovers in Q3 2007 as private capital suddenly withdrew. In
It he 4th quarter, they came to the rescue of banks and financial institutions,
stepping upas a key capital source as banks wrotedown billions of dollars of
subprime losses-almost $30 billion in that quarter alone and summarized
below. If the new-year recapitalization of Citi and Merrill is any indication,
they may be even be even more active in 2008 if they still see such bets as a
good investment. Even before they became the go-to capital source for
distressed banks, SWFs invested heavily in financials, including Barclays,
Deutsche Bank, HSBC, Standard Chartered, as well as private equity
(Blackstone, Carlyle, Och-Ziff). Yet their investments stem far beyond the
financial sector, with passive and direct investments in most asset classes
including a wide range of equities, bonds, and alternatives.
Managing over $2 trillion dollars as of 2008 (higher in 2012), sovereign
funds manage more funds than the private equity and hedge funds. The
Norwegian pension fund is the largest holder of European equities. As such,
strategy shifts are large enough to move markets and influence various asset
classes, particularly if they continue to grow in line with oil and trade
surpluses. Of course these funds are not exactly new; many funds have been
around for decades, but their size and potential rate of growth, and the
willingness of some funds to take high profile acquisitions (including
recapitalizing US financial institutions) are new. Furthermore, they have
grown tremendously since 2000 when only Kuwait Investment Authority
(KIA), Abu Dhabi Investment Authority (ADIA), Govermment Investment
Corporation(GIC) (of Singapore) were significant actors and Temasek had
little foreign holdings. Yet, these funds are not as big as either those who
see them as saviors hope or their detractors fear. At present, large visible
investments abroad still account for a small share of their portfolio on a
stock and flow basis as relatively passive portfolio investments dominate.
Sovereign funds are not monolithic

Funds can be divided into several subgroups, first by funding source
commodity or noncommodity and then by strategic behavior. Commodity
funds draw their resources from commodity export revenue surpluses
mostly oil and gas but also copper (Chile) diamonds (Botswana) and other
minerals. In many cases these funds are or had their origin in fiscal
stabilization purposes. Noncommodity funds, primarily in Asia, are
generally granted a share of excess foreign exchange reserves. In both
cases, they seek make higher returns than on safe conservative reserve
assets by building a diversified portfolio.
As funds accrue, some (Russia, Norway) have implicitly linked these
funds to long-term liabilities including pensions and health care costs of
aging populations. Several government pension plans, cognizant of
upcoming liabilities, have also increased the risk profile of their
investments (e.g., the Canada pension fund increased foreign investment
and exposure to private equity).
Sovereign investors can be divided into five (albeit overlapping)
categories. This article focuses on the third and fourth categories: sovereign
wealth funds and sovereign private equity (government investment
companies).
Foreign exchange reserves of all countries have ballooned in recent years
to reach $6 trillion as many central banks in the emerging world intervened
to avoid appreciation of domestic currencies. Central banks played a key
role in purchasing US treasuries and sustaining low long-term interest rates,
contributing to the creation of other asset bubbles. Some reserves portfolios
are managed more aggressively with exposure to higher risk assets
including equities or higher risk bonds (Saudi Arabia, Hong Kong, and,
possibly increasingly, China). Should a higher degree of these official
savings of emerging markets be allocated to sovereign wealth funds, SWFs
might surpass the global reserve stock within the next five years.
Stabilization funds are used primarily by commodity exporters to balance
fluctuating revenue flows and insulate the economy. These funds tend to be
more conservative in their asset allocation than other SWFs privileging
fixed income over equities with little to no exposure to alternative
investments. In some cases these are managed with and reported as part of
the national reserves. Once a stabilization fund has reached significant
mass, the government might choose to either start a new fund or allocate a
higher share of an existing portfolio to higher risk assets.
Sovereign wealth funds (GIC, ADIA, KIA, Norway, Kazakhstan, Alaska,
Alberta) tend to be primarily passive portfolio investors, buying small
stakes in a wide range of equities, bonds and alternatives. Their governance
structures vary widely as do their risk profiles. They tend to entrust
significant assets to external managers. Some funds make significant use of
derivatives to avoid moving the markets. Norway limits itself to very small
stakes, others have a looser definition. However, some such funds have
been willing to take larger stakes most remain below 10%, and recent
purchases have not come with management stakes.
Sovereign private equity (QIA, Istithmar, DIC, Temasek, Mubadala) are
among the newer and more activist funds that model themselves on and
compete with private equity firms, using leverage to scale up returns rather
than partnering with them. They make concentrated bets on public and
private companies and property. In some cases these funds invest in or
establish joint ventures in sectors of interest (financial, tourism, etc.) for
domestic economic development.
Table 8.1: Sovereign Fund Recapitalizations of Banks in Q4 2007 and Q1

2008
State-owned enterprises from China, Russia and the Gulf are also very
engaged in foreign acquisitions. In fact, many protectionist impulses may
be more targeted on SOEs including banks rather than sovereign funds.
Some funds, like the Chinese Investment Corporation, are hybrids. Two-
thirds of the initial $200 billion capital is committed to domestic financial
institutions, $8 billion was spent in Morgan Stanley and Blackstone stakes,
with around $60 billion available for foreign investment. As much as half of
that might be portfolio investment, but some in China think it should
support outbound investment of Chinese companies
Fig. 8.1 Estimated Assets of Key Sovereign Investors
Sovereign funds are big

At the end of 2007, sovereign funds likely managed about $2 trillion. GCC
funds probably managed about a half of that ($1 trillion) with Norway
accounting for about $400 billion and CIC probably accounted for 200
billion with GIC/Temasek about $300 billion. Small funds in Libya,
Kazakhstan, and Korea account for the remainder. Table 8.2 showing the
estimated assets of selected sovereign wealth funds, includes the key funds
and the nonreserve foreign assets of the Saudi monetary agency and the
Russian stabilization fund, neither of which are technically sovereign
wealth funds.
Funds may grow faster in 2008, receiving significant transfers. Assuming
China continues to keep limiting the RMB's rise, requiring even more dollar
purchases, it is likely to continue to allocate a significant share of its
massive official asset growth - around $500 billion in 2007 to the CIC.
Other Asian countries (Korea, Malaysia) might decide to entrust more
reserves to a sovereign fund. Although oil exporters are increasing domestic
spending and decreasing the share of new oil revenues saved, unless the
price of oil slumps, oil funds will receive significant transfers, possibly
more than in 2007 (an estimated more than $100 billion for GCC funds and
$53 billion for Norway excluding investment returns). Together Oil funds
(including Russia's new prosperity fund, a new Saudi investment fund)
could receive $300 billion and Asian funds a similar amount in 2008. If
other new funds emerge, transfers from reserves could be even greater.
Table 8.2: Estimated Assets of Selected Sovereign Wealth Funds (End

2007)
The cost of financing influences the risk/reward balance. Although SWFs

tend to use a lower leverage ratio, if any, their vulnerability to credit costs
affects decision making. Funds financed from reserves are implicitly
leveraged domestically. China must make a large dollar return to cover the
cost (interest payment and expected RMB gains) of China Investment
Corporation's (CIC) funding bonds. Should Brazil follow a similar model
financed with domestic borrowing it may be equally constrained,
particularly as domestic interest rates are still much higher than China.
And will increase demand for risky assets

The growth of sovereign wealth funds is in part a recognition that many
emerging economies have amassed reserves far in excess of that needed for
crisis prevention purposes. As such, many countries seek to gain a higher
return on their savings than the US treasuries that dominated their
portfolios.
Most sovereign funds plan to increase exposure to emerging market
assets - some in the GCC target 30 percent exposure to (mostly) Emerging
Asia. Even Norway is upping its exposure to EMs as it increases the equity
share. Many of the more conservative funds are considering branching out
into property or private equity. On the whole, sovereign funds tend to have
a lower share of US assets than their respective central banks. For example,
the central bank of the UAE has a dollar share of over 95 percent. ADIA is
thought to have a much lower share under 60%. The funds of Kuwait and
Qatar have cut their dollar shares in favor of EU and Asian assets. Thus
shifts of more assets to sovereign funds may decrease demand for US assets
and thus the dollar. Yet currency diversification is not a given; it is as likely
that China might diversify the type of its claims on the United States
(buying stakes in US companies or higher risk bonds) as reduce its dollar
share. In fact, unless China is prepared to allow a faster appreciation of the
RMB dollar it must keep buying a great number of dollars.
Shifting allocations runs the risk that key assets may be over-inflated as
investors might push up valuations of assets thought to be favored by
sovereign funds even before they make any purchases. Yet, if there is a
wholesale rush into any asset class it might depress the value of other
assets, thus encouraging holdings in other assets and making it difficult to
predict long-term market implications.
A taste for financials
Natural synergies with financial institutions predated current weakness in
the sector with half of sovereign funds' investment in financial institutions
coming in the first 6-8 months of the year. Many funds have a long history
of coinvesting with finan-cial intermediaries, investing in hedge funds and
private equity. Many SWFs have farmed out a significant share of assets to
external asset managers in both active and passive mandates. Some funds
(ADIA) were thought to outsource as much as 70% of assets and more in
some asset classes. Norway entrusted only 20 percent of total funds to
external managers. Other funds may come in somewhere in the middle.
Kazakhstan and SAMA entrust their equity allocations (as much as 25%) to
external managers. Long-standing relationships between sovereign funds
and some asset managers likely encouraged recent tie-ups. The past
collaboration between GIC and Citi's new CEO Vikram Pandit was
reportedly important in sealing the deal and GIC's decision to make up the
shortfall when China's CDB withdrew. Such connections may explain
China's decision to approach Blackstone.
Managing sovereign funds is a growth area for the asset management
industry being a major area of growth along with the private sector in Asia
and the Middle East. As with pension funds, there is a cyclical nature to
SWF fund management opportunities. At the very beginning, a fund may
place most of its funds in indexed funds as they pick managers carefully.
Shortly thereafter (or in a time of portfolio reallocation), they rely most
heavily on external managers before they gain more expertise internally and
learn about asset classes. Management of public and private equity
portfolios dominates use of external managers. For example, the Canada
Pension plan investment - board entrusts almost all of its private equity
mandates to asset manager partners but keeps many other assets for itself.
Also like pensions, SWFs engaging in primarily portfolio investment tend
to outsource alpha, keeping beta for themselves. Moreover, if funds shift
from being consumers of alternative funds to competing with them, the
share under external management will drop. Temasek and the smaller Gulf
funds, being more involved with direct investment tend to manage more in-
house or through closely held investment vehicles (such as QIA's Delta 2).
Mandates are not the only way for asset managers to tap sovereign funds.
Funds like ADIA were early investors in private equity, being key capital
sources of alter-natives funds while also seeking out co-investment
opportunities. In 2007, several SWFs decided to go a step further and take
pre-IPO stakes in private equity firms (Carlyle, Apollo, Blackstone). DIC
even accepted external investors including other sovereign funds in some of
their capital pools. These funds tend to manage a higher share of assets
directly, taking a hands-on approach either in-house or through closely held
investment vehicles (like QIA's Delta two).
Financial sector holdings may act as a conduit for future investments
overseas and domestic financial development. Some deals with asset
managers explicitly included training for domestic staff. Now many funds
are trying to lure staff away with some anecdotal evidence that SWFs may
be finally finding it easier to lure away managers as some intermediaries
consolidate their offices. Until now, SWF reluctance to match the highest
private sector salaries was a limiting factor on hiring. However, private
sector firms could equally lure away SWF staff as the alternative.
But not as big as some optimists think or as pessimists fear

By their long-term nature and primarily portfolio holdings, most sovereign
funds could have stabilizing effects on global markets. Being cashrich, they
are less constrained by deleveraging and could buy on lows, thus stabilizing
markets, as some suggested they may have done in the recent (January
2008) sharp sell-off. A related debate concerns the so-called SWF put or bid
- inspired by the Greenspan put which suggested that the Greenspan Fed
would ease interest rates to avoid a market crash, thus encouraging riskier
investments in equity markets and feeding equity market bubbles, the SWF
put suggests SWF demand will support riskier asset classes. However,
SWFs, although diversified and interested in a broad range of asset classes,
will not invest across all asset classes, nor are they immune to market
conditions.
They may be stabilizing long-term investors as they claim - and the
admittedly opaque track records of several funds (ADIA and GIC to name
two) imply. If nothing else, the immunity from disbursements may limit
massive withdrawals. They may in the process boost valuations of certain
sectors. Though, if they are wise, they will not invest in overvalued white
elephants over the long term. However, some commentators worry that
unlike (say Japanese banks) they do not have commercial interests but
political leverage and prominence at heart.
Although sovereign funds like Norway's (the largest single holder of
European equities) are among the largest institutional investors, as a group
SWFs manage much fewer assets than pension funds. Even had optimistic
estimates from invest ment banks ($12 trillion by 2012 or 2015) come to
pass, SWFs would still only have managed around 3–5% of global financial
assets. These estimates were predicated on the divergence of extensive FX
reserves into dedicated investment funds.2
One uncertainty is how sovereign funds will respond to losses. At
present, with their focus on financials, many funds have incurred paper
losses this year and could face pressures. Given their lack of transparency
and in some cases domestic oversight, it is harder to get a handle on their
portfolios. Sovereign funds may deem they have sufficient exposure to
various assets. China may already feel it is significantly exposed to the US
financial sector - being unwilling to incur more related losses. Fear of losses
might also politicize decision making process within China as CIC's leaders
defer to higher level officials in the purchase of risky assets.3 Yet many
sovereign funds seem to be savvy investors or rather, as the ones with the
cash, they can set the terms. In recent purchases, SWFs have negotiated
terms that well protect their downside risks, with the dividends on
convertible shares involved in recaps sufficient to make a good return. In
the past DIC structured deals (such as the investment in Daimler) so that the
financing banks held risk.
Increase in strategic stakes oneoff opportunism or evidence of a

change in strategy?
The early January recapitalization of Citi and Merrill was notable as it
marked both the second set of such capital sourcing as well as the
emergence of new actors. KIA's Bader Al Sa'ad previously suggested that it
was prepared to act fast so as not to lose an opportunity, making KIA's
presence unsurprising. However, Korea's bid was a bit more surprising. The
KIC - whose portfolio held only very liquid assets - took an advance on the
funds it was to receive from the finance ministry this year to take a $2b
stake in Merrill. Even before the recaps, several primarily portfolio
investors were willing to take more active stakes and to make them public
in 2007. In part, this disclosure may have been at the request of the target
who hoped to borrow the reputation of sovereign funds, just as the funds
hoped to snap up a good deal and to prove their stabilizing impact.
An underlying competition or deal envy among some of the funds seems
to be emerging. The CDB reportedly lost out to Citibank in the first go
around because it was not able to make a snap decision and KIA suggested
the lesson was to act fast. But this competition does pose a potential
vulnerability, whether funds are able to do the necessary due diligence.
The willingness of funds to give up their corporate governance leverage
attached to their stakes as they try to do what they think regulators want
poses its own challenges. Firstly, influence does not depend on official
board representation, meaning that funds are likely not fully passive. Next,
passive investment may support the status quo rather than the needs of the
company, perhaps pro-management. It potentially is inefficient.
Politics a likely headwind to acquisitions

The size of sovereign funds, their willingness to take more visible and, in
some cases, controlling stakes is raising political concerns in Washington,
and other G7 capitals
Some countries are actively courting the business of sovereign investors.
Gordon Brown urged the CIC to open an office in London, Norway, SAFE,
KIA, ADIA, QIA already have offices in the city and it has been a key
capital recycling hub - on the hopes that CIC will buy British companies
and hire British asset managers, and so maintain London's role as a center
of global finance. Other EU countries are much more wary France and
Germany both have suggested new regulations to enable governments to
block purchases if necessary (though the German private sector does not
exactly seem to agree). And The Economist impishly noted that despite
Sarkozy's worries, few sovereign funds have shown much interest in French
companies. Perhaps they forgot about the several Gulf countries circling
EADS. Yet small stakes in EADS as well are different from the purchase of
a German railway or the reverse nationalization of a key utility.
Typically, the EU would rather foster a common multilateral policy rather
than allowing member states to enter into a race to the bottom or rather a
race to protect and expel foreign assets. Germany's first draft of new
takeover law was sent back by the commission because it threatened to
block investment from other EU countries. EU commissioners have
weighed in with other proposals including Golden Shares' (Mandelson)
need for transparency lest investments be blocked and reminders that really
strategic sectors can already be protected and warning that concerns about
SWFs might be a smokescreen for blocking other investment from EM
multinationals (McGreevey). The weakness of capital availability reduced
some of these political concerns following the EZ debt crisis.
Of course concern is not limited to the developed world hosts. Temasek
in particular has run afoul of regulators in certain south East Asian
countries, particularly for its telecom stakes. Its stakes in iconic western
companies have tended to be relatively small. Even its large stake in
Standard Chartered remains below controlling levels. It recently announced
plans to pare its stakes in iconic holdings and limit itself to minority shares.
Many funds feel that they are being unfairly discriminated, held to a
higher standard than private capital sources like hedge funds and private
equity. Kuwait's Bader al Sa'ad wondered about the standards of
transparency, wondering if sovereign funds would have to disclose their
purchases before making them, putting them at an unfair disadvantage. One
thing is clear it is difficult to have it both ways. With funds as the go-to
capital source, it is harder to demand transparency. Yet funds may find it in
their interest to disclose a minimum of information to continue to snap up
what they see as good deals. Mervyn Davies, the head of Standard
Chartered (of which Temasek owns 18 percent) suggested that sovereign
funds would have to be impeccable (releasing a certain base of information)
to show they are not irresponsible market actors. Perhaps it is a matter of
time, experience and building a track record. In the meantime sovereign
funds will likely continue to have an powerful presence in 2008 and beyond
even after credit conditions improve. So it may be time to come to terms
with how their actions affect global markets and international political
interactions.
1Originally published in Wilmott, March 2008. Author's note: This chapter, written at the peak of
fervor about sovereign wealth funds, tries to demystify the role of these government-backed
investors, a subject treated in several chapters in this book. Government involvement in markets
whether as buyers or sellers or regulators has become a key theme not only within EM countries but
also developed markets. My relatively conservative estimates of sovereign wealth fund assets under
management and their growth have been borne out by intervening events, and greater domestic
financing needs (Chapter 9) have restrained their growth. Still high levels of liquidity, government
desire to limit FX appreciation have increased kept FX reserves and sovereign wealth funds growing
steadily.
2Author's note: clearly this was over the top, as I anticipated at the time. Stephen Jen's assumption
was predicated on very aggressive investment returns, optimistic estimates of assets under
management, high diversification of reserves to dedicated investment funds (and the occasional
double counting of funds and central bank assets), high oil prices and relatively modest domestic
spending - very few of which cases came to pass.
3Author's note: in additional to domestic politicization recipients also saw these investments as
political, increasing barrier to entry.
A New Age for Liquidity1
Cash is there to be stumped up. Just not where you found it last year.
In mid 2008, I wrote an article (Chapter 8), about sovereign wealth funds
and other government pools of capital - arguing that they probably were not
as big as their proponents hoped (or their detractors feared), in part because
many of their estimated $2 trillion in assets were already invested (i.e., not
liquid). The heavy (paper) losses of many of these investors on their
equities and alternative has discredited some investment strategies even as
these governments now need a lot more liquidity at home. In that way,
sovereign funds are but one of several groups of investors, public and
private, who a re in-creasing demand for liquid assets, contributing to a
major flow of funds into US treasuries and other government bonds. The
motivations for this shift are twofold: investors that moved into cash have
out performed other asset classes and the end- holders of these invostments
need more liquid assets. Most significantly, the highly leveraged global
growth and investment model is in the midst of a rethink.
Even the Gulf Cooperation Council (GCC) funds, among the most risk
tolerant of sovereign capital, seem now to be increasing their allocation to
cash-like assets. Kuwait has reportedly increased its cash allocation. Recent
research of mine with Brad Setser of the Council on Foreign Relations
‘GCC Sovereign Funds: A Reversal of Fortune’ notes that the best-
performing fund in the gulf in 2008 was likely the most conservative one -
the Saudi Arabian Monetary Agency, whose asset allocation are dominated
by USD bonds. Diversification to equities and alternative assets may not
have been the best way to hold value in global downturn, as these assets
tend to be correlated with the price of oil because of the similar
macroeconomic and financial trends which supported both booms.
Fig. 9.1 Estimated new purchases of asset s by Gulf Cooperation Coun cil (GCC) sovereig n funds
and central banks dependent on the price of oil. Source: Setser and Ziemba (2009)
No matter what the oil price, sovereign investors in the GCC and
elsewhere are going to be looking for a higher amount of liquid assets in
2009. If current oil prices persist (well under $50 at the time of writing),
these pressures will be even higher, as GCC governments need "to tap their
savings to maintain spending and provide support to domestic financial
institutions. As such, their investments abroad are likely to cool as they try
to avert domestic recessions. Other economies, like those of Singapore, will
face similar dilemmas. Beyond the need fordomestic investment, moving to
cash aflows thesetunds to prepare for investment opportunities.
Investment funds like the Kuwait Investment Authority are already
reducmg their allocation to equities and increasing that to bonds. Singapore
will likely do the same as it draws on its reserves. After all, one lesson from
2008 is that the liquidity requirements are much higher than previously
expected. The same thing held true for investment funds, and corporations
are seeing a need for more liquid assets.
Thus, both the growth of such sovereign pools of capital and their
willingness to take on risk will be rather lower in 2009 and 2010 as their
stabilization and savings motivations take the upper hand. The rate of
inflows to emerging markets and returns on risk assets, the two factors that
drove the increase in sovereign wealth assets under management, seem to
have reversed.
Implicit liabilities of sovereign governments have grown significantly.
Emerging economies like the UAE and Russia have been forced into a
position of guaranteeing the foreign debt of the domestic corporate and
financial sectors, which have grown sharply over the past few years. This
requirement boosts their capital needs, even as their revenue outlook
becomes much more precarious. Just about all oil exporters will run a fiscal
deficit in 2009 if the oil price remains below $50 a barrel - and perhaps
would even at $60. Cuts in oil production will only raise this number.
Russia is even worse off, as its corporate and financial debt now outstrips
its reserve stock and the combination of cheaper oil and more expensive fx-
denominated debt and greater need for government spending to boost
growth will put Russia under major strain. While it may be able to afford a
$100 billion fiscal deficit, it will eat up at least half of the savings in its
reserve and national wealth fund, leaving little for 2010, when global
growth and oil demand might be almost as sluggish as in 2009. Meanwhile
Russia's reserves are being depleted rapidly as Russia has tried to control
the decline of the rouble - which has impeded adjustment.
However, there is one saving grace: inflationary pressures have ebbed in
many such countries. This has several implications - one, it will allow
several of the infrastructure projects to get more bang for their buck, taking
advantage of cheaper raw materials. However, these deflationary pressures
are for now a sign of significant weakness, slack in labor, goods, and other
markets, which will be hard to work off. But this is a time when sovereign
savers should use some of their accrued assets, as not doing so would
exacerbate local and global economic contractions.
There has, however, been a major change in the flow of funds. Global
imbalances - the net outflows from a series of countries (mostly oil
exporters and Asian countries) which offset the deficits of the US
government and household sector. These imbalances look set to narrow
from 2007-08 pace in 2009-2010.
Lower savings rates from emerging savers might come at a bad time
when the USA and most EU members are set to run large fiscal deficits -
and in some cases, current account ones. In recent years, foreign
governments - especially China, but also oil exporters - became the largest
purchasers of US assets, enabling the USA and other consumers to keep
buying oil and goods. Despite a ramping up in fiscal spending, global
imbalances are set to continue to narrow as overspending consumers cut
back.
Oil exporters have shifted to deficit positions, or at best small surpluses.
Some will be able to finance these by drawing on their accumulated
surpluses, although the size of their cushions varies greatly. Some, like
Nigeria, Venezuela, and others, spent the bulk of their revenues; others
saved as much as half of their funds. However, not all of these are liquid
assets and some might require taking a major loss.
Asian countries, however, may continue to run surpluses. Despite the
export slump, which is likely to continue through 2009, the contraction in
imports may continue to outpace that of exports, resulting in large - if
somewhat more modest - surpluses. Furthermore, with Asian exporters
reluctant to let their currencies appreciate, they could return to interventions
if inflows resume.
Meanwhile, many of the large surpluses from emerging markets in
Europe (and countries like the Ukraine, Turkey, and South Africa), which
were easily financed by yield-seeking investors, may be forced into
narrowing as the cost of finance becomes more expensive. This rapid
contraction would be painful for the economy, as it means budget and
household spending cuts, which will reduce consumption-led growth.
The USA has benefited from flight to safety funds. It still has the largest,
most liquid pool of investment assets; however, investors have only been
interested in a narrow subset of these assets, contributing to a bubble in US
treasuries which seems unsustainable. As the USA issues more debt,
demand may cool somewhat, but, given the lack of alternatives, the USA
may avoid a burst bubble in these assets - at least for now. The USA may
need to turn to a new source of funding for the fiscal deficit. The US
household savings rate may return gradually to the 5-6 percent it was at
more than a decade ago from the negative rate in recent years, as credit
extension ballooned as Americans saved more. This will partially offset the
massive increase in the US fiscal deficit, recapitalization of the banking
sector, and the fiscal package but may limit its ability to stimulate the
economy.
The reliance of many growth drivers on leverage raises the question that
not only 2009 growth but the potential growth of many countries may now
need to be revised downwards. Without access to credit, potential growth
could be considerably lower even if new, more sustainable sources of
growth are tapped. Even the strong domestic demand of countries like India
was correlated with capital inflows and ample global liquidity. Highly
leveraged economies among emerging markets, especially in Europe, will
downshift somewhat painfully, especially as their governments will be less
able to bring in countercyclical fiscal policies to offset the contraction of
private demand, given the higher cost of financing.
Furthermore, with increasing costs in financing deficits, export-oriented
economies might also have challenges returning to past levels of growth.
This is not necessarily such a bad thing. An economy, especially the third
largest, can't keep growing at a 12–13% growth rate forever, as China did,
especially given that such growth rates reflected domestic and external
distortions and overcapacity. China's policy of keeping its currency and cost
of exports weak created disincentives for selling domestically. This,
however, is the key for long-term growth, and some of the government
response to the financial crisis may finally begin to put China on a more
consumption- based growth path - but after several quarters of very weak
growth (and perhaps none on a quarterly basis), China too will have a large
hole from which to emerge. So, it would not be surprising to see Chinese
growth downshifting from the 10 percent average growth rate over the past
five years, although its growth will still be a major driver of global growth.
However, there is one new source of demand in these countries -
measures to boost consumption and domestic demand. If China, which now
has an over 30 percent savings rate - and a fairly low consumption to GDP
ratio, would start demanding more, Chinese consumers could start to offset
the gap in US demand for Chinese products. That, however, will take both
time and economic policy incentives. At the moment, China's command
structure remains largely pointed toward exports, but government
expenditure on health and education could help to support Chinese
spending. But with China facing a manufacturing sector in contraction, fal-
tering property market, slowing investment, and a now-slowing domestic
demand, even an aggressive fiscal stimulus may make it difficult to return
China to past growth rates.
A similar revision will need to be made across a range of asset classes.
High returns targeted by private equity firms may be hard to achieve,
especially as they relied on leverage to scale up returns.
Furthermore, we might see corporate and individual investment patterns
shift. The credit crisis revealed that levels of liquid assets held by a range of
investors were too limited. Even when recapitalized banks begin lending
again, their willingness to lend might be considerably lower - as they take
account of lendees' ability to repay. And investors are wary of being caught
in Ponzi schemes. Furthermore, retail investors might also shy away from
higher risk assets. Investors that lost out on ETFs and structured notes
around the world and who are now clamoring for their investment managers
or governments to refund their money (as the Hong Kong government did
with mini-bonds linked to Lehman Brothers) may ask more questions about
possible risks and pay more attention to what they are really buying.
Regulators will also demand this.
Pension funds and endowments too may be wary of investing in
alternative assets which have proved as a group to be correlated with other
asset classes. The hopes of making up for pension shortfalls through very
high rates of return (hard to achieve in a best case scenario) have not been
realized, forcing difficult decisions by investment committees, which may
be more cautious in the future.
However, despite the gloomy picture, the chance of misallocation of
capital still exists. Given the amount of stimulus being pumped into the
system, there is the possibility that responses to the crisis may again seed
asset bubbles, whether it be in equities or commodities. Despite the gloomy
macro news and contraction of credit, some sidelined money may seek its
way into the system too soon. Watch out for more boom and bust cycles.
Update: Although the outlook for economic growth and financial returns
was stronger than feared in this piece, many of the structural changes
highlighted here have come to pass, as less availability of leverage and
weaker growth rates have dampened the return on investment. As of 2012,
global sovereign wealth funds of oil exporting nations had surpassed their
mid-2008 highs driven by new capital and investment returns. We estimate
that global sovereign funds manage about $3 trillion as of late 2012. Still
sovereign has grown more gradually as more oil revenue has been absorbed
at home. At the same time these sovereign funds face new competition from
other sources of domestic funds.

Government Owned Pensions: Asset Allocation
and Governance Issues1
How government owned pension funds are exercising greater influence on the markets.
This chapter, surveys the role of government led pensions in international

capital markets. Changes in the asset allocations of these funds have
boosted demand for equities and alternative assets, a trend which is likely to
continue despite losses sustained in 2008. Moreover, as with other
government investors, the political oversight does present some key
governance issues. Pension funds of Asian and Middle Eastern countries
present the greatest potential for diversification given the relatively
conservative asset allocation and the fact that some of these economies are
just developing their retirement savings systems.
With aging populations and worsening dependency ratios in advanced
economies, and a need to extend coverage in developing countries, in recent
ye ars. public pension funds sought to make up their funding gap by
moving; into higher risk assets and injecting more funds. In practice the
asset allocations of public pension funds have become roughly similar to
their private sector counterparts and to other institutional investors like
endowments and foundations. All have allocated more towards equities and
alternative assets which has led to near term losses
For emerging market economies which tend to have less developed
pension systems but face aging populations, this challenge has arisen
quickly. As such pension funds play a key role in the overall asset
allocation strategies of their sponsor governments and have been called on
when capital has been required by other parts of the government. And given
the need to spend more on retirement these funds will have a greater impact
on international markets
Despite similar asset allocations to other investors, the public ownership
poses a special set of issues — even arms-length funds might make
investment decisions for not solely economic reasons. Savings earmarked
for retirement either of all citizens or public sector workers are but one pool
of capital for some governments. As such they may be subject to political
pressures, raided to meet other shorter-term liabilities. Utilizing such funds
has allowed some governments to avoid taking on substantial short-term
debt to finance anti-crisis measures. However the timetable for
recapitalizing the funds is uncertain. Moreover fear of losing public funds
could lead to less than ideal investment decisions or lead to pressures to
support other key policy goals. While these investments may be in the
national interest, it is possible that they might lead to suboptimal policies
and lower investment returns.
Table 10.1 shows the assets of selected public pension funds.
Table 10.1: Assets of Selected Public Pension Funds Reserve Funds, $ billion. Adapted
from Sovereign Wealth Funds and Pension Funds Issues. (OECD)
Country Size of Fund
US Social Security Trust 2,200.0

Fund
Japan 1,217.6
South Korea 228.7
China 138.0
Sweden 136.7
Canada 111.3
Australia 49.1
France 47.0
Spain 44.9
Russia 32.4
Ireland 29.0
Norway 20.4
Thailand 11.6
New Zealand 9.5
Saudi Arabia 8.6
Portugal 8.3
Mexico 7.4
Jordan 5.3
Pakistan 2.4
Poland 1.8
Types of sovereign funds

Sovereign pension funds can be divided up into several groups both based
on fund structure and the entity to which they report. For the purposes of
this analysis, there are three major types of public pension fund. These
categories are informed by the work the OECD and others have done on
this issue (OECD 2008).
National pension funds based on individual contributions These tend
to be financed by payroll tax deductions or other individual contributions.
Some countries allow the beneficiaries to choose the asset allocation from
several choices, others invest the pool. Many of these funds have fairly
conservative asset allocations with fixed income dominant. Given the
changing demographic structure, a pay as you go type system will be
unsustainable as the number of workers falls and the number of pensioners
will rise as lifespans lengthen. As such greater asset diversification is likely.
National (or sub-federal) Pension reserve funds The financing gap
triggered the creation of reserve funds which will eventually be connected
to the pension system Many countries have established reserve funds with
investment portfolios in an effort to plug funding short-falls. These funds,
which tend to be under the supervision of government finance ministries
even if the management of capital is outsourced to asset managers, are
intended to feed into the public pension system at a later point when pay-as-
you-go becomes too costly or when the reserve fund has amassed enough
money. Examples include the Irish Pension Fund general reserve, the
Australian future fund and the Chinese national pension reserve fund.
While the ultimate beneficiary are citizens, the current manager and
beneficiary is the government and the political pressure to use the funds to
meet other government priorities may be quite high. Government spending
tends to require parliamentary approval. As such, pension reserve funds
tend to be functionally similar to some so-called sovereign wealth funds
which are government owned investment vehicles that manage funds in a
variety of asset classes to maximize the long-term value of national wealth.
Pension funds for public sector workers These pension funds,
particularly in the U.S. and Canada and often at a subfederal level are based
on employer and employee contributions and tend to be defined benefit.
These tend to be managed by a board appointed by political leaders. Some
of these funds including Calpers and the Ontario teachers were among some
of the earliest such investors to expand their asset allocations in alternative
assets. Many may be a hybrid of the first two structures with a section pay
as you go and investment portfolios separated out for longer horizon
investment
In addition to official pension reserve funds, many countries have other
assets that are either explicitly or implicitly linked to retirement funding.
The sovereign wealth funds of Norway and some other commodity
exporting nations have been earmarked to meet pension liabilities at an
undefined point in the future. However pension funds unlike sovereign
funds tend to have defined liabilities. Sovereign funds are also designed to
cushion domestic economies from volatile revenue streams as well as
preserve the value of national wealth. The link between the future liabilities
and current asset management strategy is not necessarily clear and some
governments might choose to spend the funds rather than saving them for
the long term. As such the assets of a pension fund or pension reserve fund
might be best assessed next to the overall liabilities of the government in
question not just the retirement funding needs.
Is there a common asset allocation for pension funds?

While the asset allocations vary across country and fund type, many public
pension funds are coalescing towards a common allocation with a majority
in equities, some investment in private equity markets, and many investing
in hedge funds or funds of hedge funds. Despite an increased investment in
equities especially by pensions in Europe, Canada, Australia and New
Zealand, several funds including the U.S. are completely invested in bonds.
See Table 10.2 for the asset allocation of several selected funds. This asset
allocation not so coincidentally is similar to a common allocation that
endowments, foundations, sovereign wealth funds and others have
followed. This asset allocation suffered severe short-term losses in 2008
given the correlated losses of most asset classes, especially public and
private equity, corporate bonds and especially any leveraged assets .
Table 10.2: Asset Allocation of Selected Public Pension Funds. Adapted from OECD,
2008
Many pension funds, especially from small open economies have

increased their exposure to foreign assets especially equities. Funds from
Canada, New Zealand and France all have close to 40% exposure to foreign
assets.
It may take quite some years for pension funds to recover from the losses
in 2008. World Bank research suggests that returns of pension funds, mostly
in the developing world suffered losses of 8-48% in the year ending in
August 2008 (World Bank 2008). Given the performance of global equities
and other assets even later in 2008 (especially September and October
2008), losses were likely far worse than that estimate for the year as a
whole. Despite the fact that asset markets have subsequently reflated, some
sold assets at a loss and others remain well below the early 2008 peaks.
Public pension funds are investors with a long-term horizon given the
long-term nature of their liabilities, which means that they may not
necessarily have had to assume these losses. In 2008/9 new capital accruing
to these funds dropped in several countries, meaning that the new funds
able to take advantage of cheap valuations may have been limited.
Moreover, many governments will likely have more limited funds
available to contribute to these funds for many quarters and years to come.
U.S. State govern-ments have been underfunding their pensions for some
years now. Pensions funded in whole or in part by individual contributions
will receive lower funding given the reduction in hours worked in many
countries.
Many of the assessments on optimal asset allocations for individual and
corporate pensions are also applicable to public pension plans. Like other
pension systems, public pension funds must start by looking at the size and
time horizon of their liabilities in order to pick the most optimal asset
structure. Despite long-term investment horizons, some do have near term
cash flow needs. In the developing world, many countries are trying to play
a major role of catch up both in terms of asset management depth and asset
value. However, specific assets and the overall investment strategies are
under review. In particular like university endowments, some pension funds
are reassessing their allocation in high-fee hedge funds, which on average
underperformed indices. The recent market performance underscored the
importance of ensuring that assets were really diversified against all risks
and of carefully picking managers whether they be in house or external.
Asset allocation varies by region should be informed by the risks to the
pensions funding streams. As noted by Allianz (2008), many Asian pension
funds have maintained a more conservative asset allocation, remaining
primarily in cash with 2007-8 despite diversification plans. They thus
contributed to the flow of funds into government bonds of domestic and
foreign (especially U.S. governments). In the longer term the diversification
planned by South Korean, Japanese and Chinese funds could be significant
in the Asian market. As noted below, the Chinese fund is already been
being used as a vehicle to help develop domestic capital markets including
private equity.
In the Middle East and other oil exporting countries, investing in
commodities and commodity linked assets might be inadvisable given that
the governments other revenues stem from these assets. Too large an
allocation to assets linked to the commodities that generate government
revenue could overexpose these investors to a prolonged downturn.
Sovereign Pension funds and International Capital Markets

Both government and private pension asset allocation may be contributing
to distortions in some investment classes. As argued elsewhere in Bertocchi,
Schwartz and Ziemba (2010), the increasing flow of funds into financial
instruments to meet corporate pension needs led to a change in investment
patterns within the U.S. which reduced funds available for real investment,
including in infrastructure and instead towards financial investment.
Despite the increase in savings by some sectors of the society in practice
most Americans boosted consumption rather than savings.
The desire of pension funds to increase their allocation to real assets, one
of several examples in which they have shifted to the ’endowment model’
has led to an increase in pension funds investing in commodity funds,
including futures. This inflow of capital into futures markets, rather than
necessarily contributing to better price discovery may actually lead to
bigger swings both in the prices of the underlying assets and the returns on
the funds.
Despite the size of these assets, it is very difficult to isolate the effect of
these public investors. In part this reflects their significant but not market
moving size. In part their asset allocation decisions are offset by those of
other actors. Goldman Sachs noted that increased funds allocated to
European equities by European pension funds actually had limited effect on
market pricing as adjustments of other investors into other assets likely
muted this effect. Yet the change in pension fund allocations, coming at a
time when other institutional investors were also moving to a set of
common allocations, clearly had an affect.
Governance Issues of public pension funds

The different types of funds present varied governance challenges but all
types, which may be linked, tend to be subject to political pressures and
political oversight. Recent research on sovereign wealth funds (Harvard
2008) suggests that these entities tend not to get lower than expected
financial returns despite the goal to maximize them. The researchers argue
that the requirement to invest in domestic economies (something very
evident in 2008 and 2009!) restricts their ability to chose the best
investments.
Intergenerational borrowing
Governmental saving for retirement or rather public pooling of resources
for retirement is a form of intergenerational transfer or lending. In other
words, workers contribute today to pay for the needs of current retirees,
their descendants will make contributions so that todays workers will have
funding. As noted in Bertocchi, Schwartz and Ziemba (2009), this
intergenerational transfer is breaking down as demographics have resulted
in a lopsided system in which the number of workers supporting each
retiree is slipping.
Pension reserve funds appear to be an attractive pool of unallocated or
not yet allocated government funds in time of crisis. In some cases the link
to future pension needs is more implicit than explicit. Russia's national
wealth fund is one example. While its is earmarked for future pension needs
it is also been used to fund some of Russia's current spending. Drawing on a
pension reserve fund may require fewer legislative changes to draw funds
than assets which are already deemed to be pension assets and for which
beneficiaries may have some ownership stake. For some countries the
calculation seems simple, why not draw on these assets to finance fiscal
stimulus today
Countries such as Ireland drew on pension savings to finance today's
fiscal packages. Still others are encouraging pension funds to increase their
investment in domestic asset markets, perhaps withdrawing funds from
abroad - Saudi Arabia's pension fund (GOSI) has been used as an equity
stabilization fund, increasing its share of its domestic assets (Ziemba 2009).
It has likewise reduced its exposure to deposits in foreign banks and
increased those abroad.
These pools of capital and their asset allocation can significantly shape
domestic capital markets changing the incentives of investors. Many
emerging economies tend to lack domestic institutional investors and often
are dominated either by domestic retail investors, large corporations or
international investors (depending on how restrictive their investment
regime. The introduction of a pension fund or pension reserve fund with a
long-term investment horizon could add investors who could at least in
theory invest for the longer-term. The sheer number of investment options
demanded by those managing public and private retirement could contribute
to financial depth in these economies and improve the ability of
corporations to seek long-term funding domestically. Doing so would limit
some of the exchange rate risks they bear.
Some countries have tried to use their pension funds and other pools of
government capital as a tool to lure asset managers and develop their
financial sector by entrusting a share of government funds to such asset
managers who set up operations domestically. Singapore, for one used
assets from the Government Investment Corporation (GIC) as well as
pensions to attract asset managers. Coupled with regulatory changes that
increase ease of financial operations, Singapore's seed capital did contribute
to its asset management industry development.
The different government revenue streams of the government may be
correlated. As noted in Setser and Ziemba (2008), countries like the UAE
faced a concurrent fall in investment income and oil revenue as equity,
corporate bonds and alternative asset returns fell even as the oil price and
later oil production reduced revenue. The returns on government savings, as
well as resource and tax revenue have fallen even as expenditure demands
have increased. The following pressures have emerged even as government
pension savings have been in even more demand in a new sort of
intergenerational borrowing.
The political leaders and the population that elects them does at time seek
to carry out issues of public and even foreign policy. California's two public
pension funds (Calpers and Calstrs) have been required to divest of any
assets linked to Sudan, going even farther than U.S. sanctions.2 Public
pensions funds may thus be a political tool (Steil, 2008). When all were
racing to take part in the IPOs of PE firms, the California legislature tried to
limit co-investment of Calpers et al. in private equity firms partly owned by
sovereign wealth funds whose sponsoring governments had not passed
certain human rights requirements. While not passed, this amendment is
reflective of political debates surrounding public pension funds, and others
that may arise in the future.
Public pension funds, especially those of Norway, Canada, Ireland and
others are key activist investors, frequently participating in shareholder
activities to improve the performance of their assets. Norway blacklists
certain investments that fail to meet ethical and environmental standards.
With Norway's fund now the largest single holder of European equities,
being left off is significant.
Not only might public pension funds be used as a tool of persuasion,
governments might also hope to use their funds to promote economic and
especially financial development. In 2007, the Chinese Pension reserve
fund was given permission to invest in domestic private equity. Motivations
were multi-fold, increase returns on savings to boost the capital available
for pensions but perhaps more significantly to provide seed capital for the
domestic industry. Other countries such as Singapore have used seed capital
from their sovereign funds including pensions to lure foreign asset
managers to their jurisdiction, helping to develop the domestic asset
management industry. In Singapore's case the combination of financial
deregulation, changes that made it cheaper to open a trading operation and
the seed capital helped contribute to industrial growth. Others such as South
Korea have tried to follow such paths.
Many Asian oil-exporting nations have been allocating investment
capital for resource investment. While these investments may respond to
long-term resource demands of these energy and metal importers, they also
may be good financial investments given the likelihood that the lack of
investment may lead to high com-modity prices in the long-term. Yet, as
with other sovereign investors, some invest-ments may reflect longer term
economic or even strategic goals not just financial returns.
Regional Trends
The following section evaluates recent trends in several key sovereign
pension funds, with a focus on those in emerging market economies which
might experience the largest growth and largest potential diversification
into equities, alternative assets etc in coming years. This survey is not
comprehensive but illustrates many of the themes across public pension
funds.
Asia
As a group, Asian countries have a significant share of savings earmarked
for retirement either through individual savings or those of government
pension funds. Many of these could be dedicated to riskier assets over time.
However the losses such investments faced in 2008 could defer any such
diversification.
The more developed Asian economies like Japan and South Korea have
significant retirement savings, quite high on a per capita basis even though
these tend to be relatively conservatively managed, Singapore reformed its
pension system and now lets its residents choose between a variety of
investment fund options. Aging populations create a pension burden for
many corporations and imply that a lot of the national wealth is invested in
low yielding assets (either domestic and foreign long-term bonds). The less
developed economies tend to be less prepared for re-tirement needs though
some including China are rapidly trying to respond to the challenge.
Middle East
As a whole the Middle East tends to have low levels of pension coverage,
limited to those in the formal sector and especially public sector workers. In
general there are different trends across the energy-exporting and energy-
importing countries. The latter, tending to be poorer, tend also to have lower
national savings as a whole. In fact some countries like Egypt and Lebanon
are particularly reliant on the financing from foreign sources. Jordan
however is an exception with assets of 36.7% of GDP according to the
OECD and ILO.
Some of these undercapitalized systems have benefited from the oil
boom. Some countries allocated a share of the surplus to bolster retirement
savings. Kuwait used the opportunity of high economic growth and great
savings to provide transfers to start putting their pension funds on sounder
footing. It made transfers of as much as 10% of GDP in 2008 but
subsequently stopped transfers as the oil price fell. It is unlikely to make
transfers in 2009 even as withdrawals have increased. Despite the transfers,
shortfalls still remain both for national funds as a whole and public sector
workers in particular.
Overall, these funds tend to be relatively conservative, more so than the
sovereign funds of the same governments, and tend to have little
international exposure (ILO, 2009). Several such as the pension funds of
public workers in Abu Dhabi have become rather sophisticated drawing on
the expertise of internal and external asset managers. Increasingly funds are
being invested in foreign bonds and stocks as with other government assets.
Some countries like Saudi Arabia actually have fairly established pension
funds that are significant holders of domestic equity in addition to foreign
currency assets. These though are small as a share of GDP and in
comparison to the overall asset of the government.
Europe
Public Pension coverage and asset allocation varies quite widely across
Europe which also has well established and large private pension funds. Yet
public funds still account for close to a majority of retirement savings,
particularly for public sector workers. Most funds have significant exposure
to equity markets and a small allocation to alternative assets. As such they
are significant participants in European equity markets. There are some
exceptions. Spain invests only in fixed income. The largest fund, as a ratio
of the size of the economy is in Sweden.
U.S./Canada: Public pensions in the U.S. and Canada tend to be co-
funded by payroll deductions from employees, contributions from
employers, including the government in the case of public sector workers.
U.S. social security continues to be completely invested in US treasury
bonds. The political debate on the underfunded liabilities related to social
security and Medicare will revive in coming years but for now concerns
have taken a back seat to shorter term financing worries. By contrast the
public funds of individual states for their public employees have tended to
increase their allocation to equities and other riskier assets to make up for
funding shortfalls as cash-strapped regional governments held back on
allocations. Some investments have been sold at a loss, especially property
holdings and private placements relied indirectly on leverage.
Conclusion
By virtue of their increase in assets and diversification to increase their risk-
adjusted return, public pension funds have become a more significant
investor in global, regional and national capital markets. The creation of
reserve funds by both developed and emerging economies to help meet the
retirement funding shortfall has boosted the size of these assets. In 2009,
these funds are trying to process the lessons of the credit crisis and global
asset market correction. Several may take the lesson to increase the share of
liquid assets in their portfolios to have funds available should the
government or beneficiaries claim funds. Moreover given the political
pressures inherent in investment losses, some sovereign investors may
further try to minimize losses.
Thus, sovereign pension funds must be seen as one of several pools of
capital managed by governments. Despite different goals and time horizons,
many of these varied types of sovereign capital are being managed with a
similar asset allocation. Over time, should pension funds continue to attract
assets, their asset allocation might adjust slightly. In particular, given their
long-term horizon, it could be in the interest of financial markets, pension
beneficiaries and other citizens, if sovereign pension funds filtered more
investments into real investment, particularly infras-tructure. Not only
might such investment lead to long-term economic benefits but it should
also bring a financial return which should hold its value even in the case of
rising inflation.
Update: Since writing this in 2009, the retirement funding gap has only
become more acute as governments have tended to cut back on transfers to
pensions as part of prioritizing current spending needs. These funds are but
some of the sovereign managed capital, what is different is that they tend to
have implicit if not explicit liabilities.
1Edited from Wilmott, September 2009.

2Author's note: In early 2013, these funds were considering divesting from gun manufacturers.
Update on Yale's Approach to Endowment
Investing1
What the crisis did to the Ivy League...
The Yale endowment as a model for other university endowments and other
investors is discussed in Chapter 8 of Ziemba and Ziemba (2007) (which
was updated from a Wilmott column in the March 2007 issue). David
Swenson,who has run this endowment since 1986, introduced multi-
strategy investing and pioneered equity investments in non- exchanged
traded instruments. He refers to such investments as equity broadly defined
and this includes in-house and outside manager traded hedge funds and
various illiquid commodity type assets such as real estate, energy and
timber. From 1986–2007 this strategy worked well with returns of 13.4%
during the 20-year period 1990–2008 following 16% returns in the first 21
trading years from 1986 to 2007 and its 13.7% to 2012. The financial crisis
of 2007-2009 hit Yale and the other followers, and Harvard which had some
of the same ideas and led to a -24.6% ($5.6 billion loss) their fiscal year
July 1, 2008 to June 30, 2009. This followed a small +4.5% gain in the
previous year, +9.8% in the year ending June 30, 2010, and +21.9% for the
year to June 30, 2011. In the year ending June 30, 2012, the endowment
returned 4.7% versus 5.5% for the S&P500, developed markets -13.8% and
emerging markets -15.9%. That year had Harvard losing -0.05%. Harvard's
very good historical record is in Table 11.1. Table 11.2 compares Yale to
other top universities during the year ending June 30, 2012 in terms of
endowment size, rates of return, etc.
Table 11.3 has more detail on actual and target returns based on their
meanvariance approach to portfolio balancing and Table 11.4 has the 2012
asset allocation targets. Figure 11.1 shows the historical asset allocation
over time with less and less in traded equities and bonds and more to
alternative investments. In the allocation to absolute return strategies, event
driven has a 10% standard deviation versus 15% for value driven, both with
the same expected return so value driven is dominated but maybe it
diversifies. Domestic equity remains with the same allocation. Fixed
income remains low and the 4.5% allocation includes cash. Foreign equity
is light, which is surprising with the weak dollar and higher growth outside
the US. It is also puzzling why in a weak economy with little inflation, that
private equity and real asset allocations are increased but Yale thinks that
there will be anticipated growth in private equity exposure. I have put what
data I have been able to locate but some items are unavailable.
Table 11.1: Harvard Endowment Investment Returns in Percent. Source: Reeves

(2012)
Table 11.2: US University Endowments: Size and Net Returns, July 1, 2009
to June 30, 2012
Table 11.5 contains various data and return results for 2005 to 2010 (the
fiscal year runs from July 1 to June 30 of the following year). Figure 11.2
shows that despite the dismal results around the 2008–2009 stock market
crash, Yale's yearly record shows them beating their benchmarks in all six
major categories. Finally, Figure 11.3 shows that Yale has beaten inflation
and the median of a broad universe of colleges and universities from 1999
to 2009.
Table 11.3: Asset Allocation. Source: YER, 2009
Notes: Cash is part of fixed income. Absolute return is 10%. Event driven, 15%, value driven, 15%.
Fig. 11.1 Yale's changing asset allocation targets, 1985–2010
Table 11.4: 2012 and 2012 Asset Allocation Targets
Figure 11.4 shows the university revenue by source from 1905–2010.

Figure 11.5 shows the endowment market value wealth path from 1950 to
2010. Observe the contribution of Swensen's strategies on the endowment
income since 1985, yet much of the university budget comes from sources
other than the endowment and student tuition. Recent results have been
strong with a 21.9% return for the year ending June 30, 2011. That puts the
endowment at $19.4 billion based on investment gains of $3.6 billion with
$1.0 billion going to the operating budget.
Table 11.5: Yale Endowment Highlights. Source: YER, 2011

Active Benchmarks Passive Benchmarks
Absolute Return: 1–year
Absolute Return:
Constant Maturity Treassury +
CSFR/Tremont Composite
6%
Domestic Equity: Frank Russell
Domestic Equity: Wilshire 5000
Median Manager, U.S. Equity
Fixed Income: Frank Russell
Fixed Income: BarCap 11–5 Yr
Median Manager, Fixed
Treasury
Income
Foreign Equity: Blend of
Foreign Equity: Frank Russell
MSCIEAFE Index, MSCIEM
Median Manager
Index,
Composite Foreign Equity Custom Oppotunistic
Blended Index
Private Equity: Cambridge Private Equity: University
Associates Composite Inflation + 10%
Real Assets: NCREIF and
Real Assets: University
Cambridge Associates
Inflation + 6%
Composite
Fig. 11.2 Yale asset class results beat benchmarks, 1999-2011. Source: YER, 2011
Summary and comments on Yale's results and approach

Despite poor results in fiscal 2007, 2008 and 2009 of 4.5%, -24.6% and
8.9% ending June 30, 2010, the long term ten year results are still good.
These ten year returns had a geometric mean of 11.8% versus 7.3% for
hedge funds, 6.0% for bonds and -1.29% for domestic equity, and according
to Yale they are still in the top 1% of institutional investors but still an
excellent long term performance. The twenty year record is still 13.4%
down from the 16% area reported in 2007.
Fig. 11.3 Yale's performance exceeds peer results, 1999 to 2011, 1999=$1,000. Source: YER, 2011
Fig. 11.4 University Revenue by Source 1905-2010. Source: YEC, 2010
Yale pioneered the multi strategy approach, much diversification an

investment in non-traditional non-exchange traded assets.
Yale's investment in timber amounts to 3 million acres which is
equivalent to the size of Connecticut. Some 650,000 acres are conservation
lands. This is related to Yale's goal for be the world's greenest university.
There is a $1 million green fund to support projects. This timber is useful
for much wind energy plus the trees. The University is into serious
recycling with a special university department. There is a big push in
projects to curtail carbon dioxide emissions. A goal is to reduce emissions
by 10% below 1990 levels by 2020.
Fig. 11.5 Yale endowment market values, 1950-2011. Source: YER, 2011
Yale's endowment approach emphasizes equities and 96% of the portfolio

is in what they refer to as equities - these include absolute return and hedge
funds, private equity and real assets. They use a mean-variance approach
citing Cowles commission researchers James Tobin and Harry Markowitz,
both of whom won Nobel prizes and emphasized this approach which is
based on non fat-tailed, symmetric normal distributions.
I think, as my readers well know, that variance is not an appropriate
measure of risk. Real risk is the chance of losing money. Hence, in my
asset-liability management models I have used since the 1970s convex risk
measures based on losses or target violations; see Geyer and Ziemba (2008)
and Ziemba (2010). Yale sizes their allocations based on variance
contribution. So that seems flawed.
Yale is a well run university and in the top five US universities for
quality of teaching, research, etc. A nice feature of their interaction with the
endowment is a policy of spending smoothing: by incorporating the
previous year's spending, the rule eliminates large fluctuations, enabling the
University to plan for its spending needs. During the last twenty years,
annual changes in spending have been one fourth as volatile as annual
changes in the endowment value. Also, by adjusting spending toward a
specified long-term target spending level, the rule ensures that spending
will be sensitive to fluctuating endowment values, providing stability in
long-term purchasing power.
None of Yale's distinguished finance or economics faculty are regular
members of the investment office. It is known though that there is some
contact through jointly taught courses with the investment head, David
Swenson, and possibly other contacts.
The 2009 report has a lot of self defense in explanation of the record, but
little change in the basic strategy. The president of Yale continues his
support of the basic strategies.
I run an offshore BVI based hedge fund for a large investment group and
the head, a very successful investor and trader, is adamant that as a portfolio
manager you are supposed to win in all market conditions. This requires
some market timing and hedging rather than the straight holding of
positions. Yale does do some active market trading along these lines but that
is not a major part of their strategy. Chopra and Ziemba (1993) remind us
how important it is for success to have accurate mean estimates. The recent
results for the year ending June 30, 2011 were strong with 21.9% returns.
That puts the endowment at $19.4 billion with investment gains of $3.6
billion and $1.0 billion going to the operating budget. And the two decade
record from 1992-2011 is +14.2% per year versus its endowment peer
average of 9.42% according to Conroy (2011) For the last ten years to June
30, 2011, it is 10.1% versus US traded stocks 3.9% and US traded bonds
5.1%. So they have recovered to some extent. They are a smart group and
well connected which is a plus, but did they adjust to the new reality with
lower expected returns? And the use of variance likely will lead to sub-
optimal allocations. I am skeptical and think that most likely their high
returns will not resume.
Chapter 6 discusses the evaluation of great investors using a performance
measure based on not losing money. Up to the current period, Yale had a
superior record measured by what I call the symmetric downside Sharpe
ratio. The results have damaged this but the overall record is still good. But
enough said, time will tell here.
1Edited and updated from Wilmott, January 2011.

A Risk Arbitrage Convergence Trade: The Nikkei
Put Warrant Market of 1989—90
Dr. Thorp and I, with assistance from Julian Shaw (then of Gordon Capital,
Toronto, now the risk control manager for Barclays trading in London), did a
convergence trade based on differing put warrant prices on the Toronto and
American stock exchanges. The trade was successful and Thorp won the
over $1 million risk adjusted hedge fund contest run by Barron's in 1990.
There were risks involved and careful risk management was needed. What
follows is a description of the main points. Additional discussion and
technical details appears in Shaw, Thorp and Ziemba (1995).
This edge was based on the fact that the Japanese stock and land prices
were astronomical and very intertwined, see Stone and Ziemba (1993) for
more on this relationship.
The historical development leading up to the NSA put warrants

• Tsukamoto Sozan Building in Ginza 2-Chome in central Tokyo was the
most expensive land in the country with one square meter priced at ¥37.7
million or about $279,000 U.S. at the (December 1990) exchange rate of
about ¥135 per U.S. dollar.
• Downtown Tokyo land values are the highest in the world, about $800
million an acre.
• Office rents in Tokyo are twice those in London yet land costs 40 times
as much
• The Japanese stock market, as measured by the Nikkei stock average
(NSA), was up 221 times in yen and 553 in dollars from 1949 to the end
of 1989.
• Despite this huge rise, there had been twenty declines of 10% or more in
the NSA from 1949 to 1989. The market was particularly volatile with
two more in 1990 and two more in 1991. Stocks, bonds and land were
highly levered with debt.
• There was a tremendous feeling in the West that the Japanese stock
market was overpriced as was the land market. For example, the
Emperor's palace was reputed to be worth all of California or Canada.
Japanese land was about 23% of world's non-human capital. Japanese PE
ratios were 60+.
• Various studies by academics and brokerage researchers argued that the
high prices of stocks and land were justified by higher long run growth
rates and lower interest rates in Japan versus the US. See for example,
Ziemba and Schwartz (1991) and French and Poterba (1991). However,
similar models predicted a large fall in prices once interest rates rose
from late 1998 to August 1990, see Chapters 2 and 21.
• Hence both must crash!
• There was a tremendous feeling in Japan that their economy and products
were the best in the world.
• There was a natural trade in 1989 and early 1990
— Westerners bet Japanese market will fall
— Japanese bet Japanese market will not fall
Various Nikkei put warrants which were three-year American options

were offered to the market to fill the demand by speculators who wanted to
bet that the NSA would fall.
NSA puts and calls on the Toronto and American stock

exchanges, 1989–1992
The various NSA puts and calls were of three basic types, see Table 12.1.
Our convergence trades in late 1989 to early 1990 involved:
(1) selling expensive Canadian currency Bankers Trust I's and II's and
buying cheaper US currency BT's on the American Stock Exchange;
and
(2) selling expensive Kingdom of Denmark and Salomon I puts on the
ASE and buying the same BT I's also on the ASE both in US dollars.
This convergence trade was especially interesting because the price
discrepancy was based mainly on the unit size and used instruments on
the same exchange.
Table 12.2 describes this. We preformed a complex pricing of all the
warrants which is useful in the optimization of the positions size, see Shaw,
Thorp and Ziemba (1995). However, Table 12.2 gives insight into this in a
simple way. For example, 9.8% premium year year means that if you buy the
option, the NSA must fall 9.8% each year to break even. So selling at 9.8%
and buying at 2.6% looks like a good trade.
Some of the reasons for the different prices were:
• Large price discrepancy across the Canada/U.S. border
• Canadians trade in Canada, Americans trade in the U.S
• Different credit risk
• Different currency risk
• Difficulties with borrowing for short sales
• Blind emotions vs reality
• An inability of speculators to properly price the warrants
Table 12.1: NSA Puts on the Toronto and American Stock Exchanges,
1989–1992
Table 12.2: Comparison of Prices and Premium Values for Four
Canadian and Three U.S. NSA Put Warrants on February 1, 1990
I's were ordinary puts traded in yen. II's were currency protected puts
(often called quantos). III's were the Nikkei in Canadian or US dollars. The
latter were marketed with comments like: you can win if the Nikkei falls, the
yen falls or both fall The payoffs in yen and in US/Cdn are shown in Table
12.1. A simulation in Shaw, Thorp and Ziemba (1995) showed that for
similar parameter values, I's were worth more than II's, which were worth
more than III's. But investors preferred the currency protected aspect of the
II's and overpaid (relative to hedging that risk separately in the currency
futures markets) for them relative to the I's. Figures 12.1 and 12.2 show the
two convergence trades.
is plotted rather than implied volatility since the latter did not exist when
there were deep in the money options trading for less than intrinsic as in this
market. Fair value at 20% NSA volatility and 10% exchange rate volatility is
zero on the graph. At one, the puts are trading for double their fair price. At
the peak, the puts were selling for more than three times their fair price.
Fig. 12.1 Relative costs of BT-I, BT-II and BTB NSA put warrants with NSA volatility of 20% and
exchange rate volatility of 10%, 17 February 1989 to 21 September 1990. Relative deviation from
model price = (actual cost - theoretical value)/(theoretical value). Key: (+) BT=I, type I, Canadian,
BT-II, type III, Canadian and (∆) BTB type I, US and (-) normalized Nikkei
The BT-I's did not trade until January 1990 and in about a month the
Canadian BT-I's and BT-II's collapsed to fair value and then the trade was
unwound. The Toronto newspapers inadvertently helped the trade by
pointing out that the Canadian puts were overpriced relative to the US puts
so eventually there was selling of the Canadians, which led to the
convergence to efficiency. To hedge before January 1990 one needed to buy
an over the counter put from a brokerage firm such as Salomon who made a
market in these puts. The NSA decline in 1990 is also shown in Figure 12.1.
Additional risks of such trades is being bought in and shorting the puts too
soon and having the market price of them go higher. We had only minor
problems with these risks.
Fig. 12.2 Relative costs of US type I (BTB) versus US type II (DXA, SXA, SXO) NSA put warrants
with NSA volatility of 20%, January to September 1990. Key: ([]) BTB, type I, 0.5 NSA, (+) avg
DXA, SXA, SXO, type II, 0.2 NSA, and, (-) normalized Nikkei
Fair value at 20% NSA volatility and 10% exchange rate volatility is zero
on the graph. At one the puts are trading for double their fair price. At the
peak, the puts were selling for more than three times their fair price.
For the second trade, the price discrepancy lasted about a month. The
market prices were about $18 and $9 where they theoretically should have
had a 5 to 2 ratio since one put was worth 20% and the other 50% and trade
at $20 and $8. These puts were not identical so this is risk arbitrage not
arbitrage. The discrepancy here is similar to the small firm, low price effect
(see Ziemba, 2012a). Both puts were trading on the American stock
exchange.
There was a similar inefficiency in the call market where the currency
protected options traded for higher than fair prices; see Figure 12.3. There
was a successful trade here but this was a low volume market. This market
never took off as investors lost interest when the NSA did not rally. US
traders prefered Type II (Salomon's SXZ) denominated in dollars rather than
the Paine Webber (PXA) which were in yen.
The Canadian speculators who overpaid for the put warrants that our trade
was based on made $500 million Canadian since the NSA's fall was so great.
A great example of the mean dominating! The issuers of the puts also did
well and hedged their positions with futures in Osaka and Singapore. The
losers were the holders of Japanese stocks. We did a similar trade with
Canadian dollar puts traded in Canada and hedged in the US. The difference
in price (measured by implied volatility) between the Canadian and US puts
stayed relatively constant over an entire year (a gross violation of efficient
markets). The trade was also successful but again like the Nikkei calls, the
volume was low.
Fig. 12.3 Relative costs of Paine Webber and Salomon NSA call warrants with NSA historical
volatility of 20%, April to October 1990. Relative deviation frommodel price = (actual cost -
theoretical value)/(theoretical value). Key: (+) PXA,, (+) SXZ and (–) normalized Nikkei
Kelly Capital Growth Investing
The use of log utility dates to the letters of Daniel Bernoulli in 1738. The
idea that additional wealth is worth less and less as it increases and thus
utility tails off proportional to the level of wealth is very reasonable . This
utility function seems safe for investing. However, I argue that log is the
most risky utility function one should ever consider using and it is most
dangerous. However, if used properly in situations where it is appropriate, it
has wonderful properties. For long term investors who make many short
term decisions, it usually yields the highest long run levels of wealth. This is
called Kelly betting in honor of Kelly's 1956 paper that introduced this type
of betting. In finance, it is called the Capital Growth Theory or Fortune's
Formula.1 Kelly was working at Bell Labs and was greatly influenced by
Claude Shannon, the father of information theory.
This chapter has examples where the bets are small and other chapters
consider large bets. Applications are made to blackjack, lotteries, horse
racing and in Chapter 15 commodity trading on the January turn-of-the-year
effect is discussed.
Consider the example described in Table 13.1. There are five possible
investments and if we bet on any of them, we always have a 14% advantage.
The difference between them is that some have a higher chance of winning
and, for some, this chance is smaller. For the latter, we receive higher odds if
we win than for the former. But we always receive 1.14 for each 1 bet on
average. Hence we have a favorable game. The optimal expected log utility
bet with one asset (where we either win or lose the bet) equals the edge
divided by the odds.2 So for the 1-1 odds bet, the wager is 14% of one's
fortune and at 5-1 it's only 2.8%. We bet more when the chance that we will
lose our bet is smaller. Also we bet more when the edge is higher. The bet is
linear in the edge so doubling the edge doubles the optimal bet. However,
the bet is non-linear in the chance of losing our money, which is reinvested
so the size of the wager depends more on the chance of losing and less on
the edge.
Table 13.1: The Investments
Source: Ziemba and Hausch (1986)
The simulation results shown in Table 13.2 assume that the investor's
initial wealth is 1000 and that there are 700 investment decision points. The
simulation was repeated 1000 times. The numbers here are the number of
times out of the possible 1000 that each particular goal was reached. The
first line is with log or Kelly betting, The second line is half Kelly betting.
That is you compute the optimal Kelly wager but then blend it 50-50 with
cash. We discuss various Kelly fractions and how to utilize them wisely but
for now, we will just focus on half Kelly. Assuming log normally distributed
investments, the α-fractional Kelly wager is equivalent to the optimal bet
obtained from using the concave risk averse, negative power utility function,
–w−β, where For half Kelly (α = 1/2), β = −1 and the utility function
is . Here the marginal increase in wealth drops off as w2, which is
more conservative than log's w. Log utility is the case β →−∞, α =1 and
cash is β →−∞, α = 0.
Table 13.2: Statistics of the Simulation
Source: Ziemba and Hausch (1986)
A major advantage of log utility betting is the 166 in the last column. In
fully 16.6% of the 1000 cases in the simulation, the final wealth is more than
100 times as much as the initial wealth. Also in 302 cases, the final wealth is
more than 50 times the initial wealth. This huge growth in final wealth for
log is not shared by the half Kelly strategies, which have only 1 and 30,
respectively, for their 50 and 100 time growth levels. Indeed, log usually
provides an enormous growth rate but at a price, namely a very high
volatility of wealth levels. That is, the final wealth is very likely to be higher
than with other strategies, but the ride generally will be very very bumpy.
The maximum, mean, and median statistics in Table 13.2 illustrate the
enormous gains that log utility strategies usually provide.
Let's now focus on bad outcomes. The first column provides the following
re-markable fact: one can make 700 independent bets of which the chance of
winning each one is at least 19% and usually is much more, having a 14%
advantage on each bet and still turn 1000 into 18 , a loss of more than 98%.
Even with half Kelly, the minimum return over the 1000 simulations was
145, a loss of 85.5%. Half Kelly has a 99% chance of not losing more than
half the wealth versus only 91.6% for Kelly. The chance of not being ahead
is almost three times as large for full versus half Kelly. Hence to protect
ourselves from bad scenario outcomes, we need to lower our bets and
diversify across many independent investments. This is explored more fully
in the context of hedge funds in various chapters in this book.
Figure 13.1 provides a visual representation of the type of information in
Table 13.2 displaying typical behavior of full Kelly versus half Kelly
wagering in a real situation. These are bets on the Kentucky Derby from
1934 to 1998 using an inefficient market system where probabilities from a
simple market (win) are used in a more complex market (place and show)
coupled with a breeding filter rule [dosage filter 4.00] to eliminate horses
who do not have enough stamina. You bet on horses that have the stamina to
finish first, second or third who are underbet to come in second or better or
third or better relative to their true chances estimated from their odds to win.
The full Kelly log bettor has the most total wealth at the horizon but has
the most bumpy ride: $2500 becomes $16,861. The half Kelly bettor ends up
with much less, $6945 but has a much smoother ride. The system did
provide out of sample profits. A comparison with random betting proxied by
betting on the favorite in the race, shows how tough it is to win at
horseracing with the 16% track take plus breakage (rounding payoffs down
to the nearest 20 cents per $2 bet) at Churchill Downs. Betting on the
favorite turns $2500 into $480. Random betting has even lower final wealth
at the horizon since favorites are underbet.
Blackjack
The difference between full and fractional Kelly investing and the resulting
size of the optimal investment bets is illustrated via a tradeoff of growth
versus security. This is akin to the static mean versus variance so often used
in portfolio management and yields two dimensional graphs that aid in the
investment decision making process. This can be illustrated by the game of
blackjack where fractional Kelly strategies have been used by professional
players.
Fig. 13.1 Wealth level histories from place and show betting on the Kentucky Derby, 1934-1998 with
the Dr Z system utilizing a 4.00 dosage index filter rule with full and half Kelly wagering from $200
flat bets on the favorite using an initial wealth of $2500. Source: Bain, Hausch and Ziemba (2006)
The game of blackjack or 21 evolved from several related card games in

the 19th century. It became fashionable during World War I and now has
enormous popularity, and is played by millions of people in casinos around
the world. Billions of dollars are lost each year by people playing the game
in Las Vegas alone. A small number of professionals and advanced
amateurs, using various methods such as card counting, are able to beat the
game. The object is to reach, or be close to, twenty-one with two or more
cards. Scores above twenty-one are said to bust or lose. Cards two to ten are
worth their face value: Jacks, Queens and Kings are worth ten points and
Aces are worth one or eleven at the player's choice. The game is called
blackjack because an ace and a ten-valued card was paid three for two and
an additional bonus accrued if the two cards were the Ace of Spades and the
Jack of Spades or Clubs. While this extra bonus has been dropped by current
casinos, the name has stuck. Dealers normally play a fixed strategy of
drawing cards until the total reaches seventeen or more at which point they
stop. A variation is when a soft seventeen (an ace with cards totaling six) is
hit. It is better for the player if the dealer stands on soft seventeen. The house
has an edge of 1-10% against typical players. The strategy of mimicking the
dealer loses about 8% because the player must hit first and busts about 28%
of the time (0.282 = 0.08). However, in Las Vegas the average player loses
only about 1.5% per play.
The edge for a successful card counter varies from about -5% to +10%
depending upon the favorability of the deck. By wagering more in favorable
situations and less or nothing when the deck is unfavorable, an average
weighted edge is about 1-2%. An approximation to provide insight into the
long-run behavior of a player's fortune is to assume that the game is a
Bernoulli trial with a probability of success p = 0.51 and probability of loss
q=1-p= 0.49.
Fig. 13.2 Probability of doubling and quadrupling before halving and relative growth rates versus
fraction of wealth wagered for Blackjack (2% advantage, p=0.51 and q=0.49). Source: McLean and
Ziemba (1999)
Table 13.3: Growth Rates Versus Probability of Doubling Before Halving for Blackjack.
Source: MacLean and Ziemba (1999)
Figure 13.2 shows the relative growth rate versus
the fraction of the investor's wealth wagered, π. The security curves show
the bounds on the true probability of doubling or quadrupling before halving.
This is maximized by the Kelly log bet π* = p − q = 0.02. The growth rate is
lower for smaller and for larger bets than the Kelly bet. Superimposed on
this graph is also the probability that the investor doubles or quadruples the
initial wealth before losing half of this initial wealth. Since the growth rate
and the security are both decreasing for π > π*, it follows that it is never
advisable to wager more than π*. The growth rate of a bet that is exactly
twice the Kelly bet, namely 2π* = 0.04, is zero plus the risk-free rate of
interest. Figure 13.2 illustrates this. Hence log betting is the most aggressive
investing that one should ever consider. The root of hedge fund disasters is
frequently caused by bets above π* when they should have bets that are π* or
less, especially when parameter uncertainty is considered. However, one
may wish to trade off lower growth for more security using a fractional
Kelly strategy. This growth tradeoff is further illustrated in Table 13.3. For
example, a drop from π* = 0.02 to 0.01 for a 0.5 fractional Kelly strategy,
decreases the growth rate by 25%, but increases the chance of doubling
before halving from 67% to 89%.
The rest of the chapter discusses three topics: investing using unpopular
numbers in lotto games with very low probabilities of success but where the
expected returns are very large (this illustrates how bets can be very tiny);
good and bad properties of the Kelly log strategy and why this led me to
work with Len MacLean on a through study of fractional Kelly strategies
and futures and commodity trading, and how large undiversified positions
can lead to disasters as it has for numerous hedge funds and bank trading
departments.
Betting on unpopular lotto numbers using the Kelly criterion

Using the Kelly criterion for betting on favorable (unpopular) numbers in
lotto games - even with a substantial edge and very large payoffs if we win -
the bets are extremely tiny because the chance of losing most or all of our
money is high.
Lotteries predate the birth of Jesus. They have been used by various
organiza-tions, governments and individuals to make enormous profits
because of the greed and hopes of the players who wish to turn dollars into
millions. The Sistine Chapel in the Vatican, including Michelangelo's
ceiling, was partially funded from lotteries. So was the British Museum.
Major Ivy League universities in the US such as Harvard used lotteries to
fund themselves in their early years. Former US president Thomas Jefferson
used a lottery to pay off his debts when he was 83. Abuses occur from time
to time and government control is typically the norm. Lotteries were banned
in the US for over a hundred years from the early 1800s and resurfaced in
1964. In the UK, the dark period was 1826-1994. Since then there has been
enormous growth in lottery games in the US, Canada, the UK and other
countries. Current lottery sales in the UK are about five billion pounds per
year. Sales of the main 6/49 lotto game average about 80 million pounds a
week. The lottery operator takes about 5% of lotto sales for its remuneration,
5% goes to retailers, 12% goes to the government in taxes, and another 28%
goes to various good causes, as do unclaimed prizes.
One might conclude that the expected payback to the Lotto player is 50%
of his or her stake. However, the regulations allow a further 5% of regular
sales to be diverted to a Super Draw fund. Furthermore we must allow for
the probability that the jackpot is not won. Eighty of 567 jackpots to the end
of May 2001 had not been won. This means that the expected payback in a
regular draw is not much more than 40%. This is still enough to get people
to play. With such low paybacks it is very difficult to win at these games and
the chances of winning any prize at all, even the small ones, is low.
Table 13.4 describes the various types of lottery games in terms of the
chance of winning and the payoff if you win. Lottery organizations have
machines to pick the numbers that yield random number draws. Those who
claim that they can predict the numbers that will occur cannot really do so.
There are no such things as hot and cold numbers or numbers that are
friends. Schemes to combine numbers to increase your chance of winning
are mathematically fallacious. For statistical tests on these points, see
Ziemba et al. (1986). One possible way to beat pari-mutuel lotto games is to
wager on unpopular numbers or, more precisely, unpopular combinations.3
In lotto games players select a small set of numbers from a given list. The
prizes are shared by those with the same numbers as those selected in the
random drawing. The lottery organization bears no risk in a pure pari-mutuel
system and takes its profits before the prizes are shared. I have studied the
6/49 game played in Canada and several other countries.4
Combinations like 1,2,3,4,5,6 tend to be extraordinarily popular: in most
lotto games, there would be thousands of jackpot winners if this combination
were drawn. Numbers ending in eight and especially nine and zero as well as
high numbers (32+, the non-birthday choices) tend to be unpopular.
Professor Herman Chernoff found that similar numbers were unpopular in a
different lotto game in Massachusetts. The game Chernoff studied had four
digit numbers from 0000 to 9999. He found advantages from many of those
with 8, 9, 0 in them. Random numbers have an expected loss of about 55%.
However, six-tuples of unpopular numbers have an edge with expected
returns exceeding their cost by about 65%. For example, the combination 10,
29, 30, 32, 39, 40 is worth about $1.507 while the combination 3, 5, 13, 15,
28, 33 of popular numbers is worth only about $0.154. Hence there is a
factor of about ten between the best and worst combinations. The expected
value rises and approaches $2.25 per dollar wagered when there are
carryovers (that is when the jackpot is accumulating because it has not been
won.). Most sets of unpopular numbers are worth $2 per dollar or more
when there is a large carryover. Random numbers, such as those from lucky
dip and quick pick, and popular numbers are worth more with carryovers but
never have an advantage. Howevee, investors (such as Chernoff's students)
mty still lose because of mean reversion (the unpopular numbers tend to
become less unpopular over time) and gamblers' ruin (the investor har used
up his available resources before -winning;). These same two phenomena
show up in the financial markets repeatedly.
Table 13.4: Types of Lottery Games

Table 13.5 provides an estimate of the most unpopular numbers in Canada
in 1984, 1986 and 1996. The same numbers tend to be the most unpopular
over time but their advantage becomes less and less over time. Similarly, as
stock market anomalies like the January effe ct or weekend effect have
lessened over time. However, the advantages are still good enough to create
a mathematical advantage in the Canadian and UK lottos.
Strategy Hint #1: When a new lotto game is offered, the best advantage is
usually right at the start. This point applies to any type of bet or financial
market.
Strategy Hint #2: Games with more separate events, on each of which you
can have an advantage, are more easily beatable. The total advantage is
the product of individual advantages. Lotto 6/49 has 6; a game with 9 is
easier to beat and one with 3 harder to beat.
But can an investor really win with high confidence by playing these
unpopular numbers? And if so, how long will it take? To investigate this,
consider the following experiment shown in Table 13.6.
Table 13.5: Unpopular Numbers in the Canadian 6/49, 1984, 1986, and 1996
Case A assumes unpopular number six-tuples are chosen and there is a
medium sized carryover. Case B assumes that there is a large carryover and
that the numbers played are the most unpopular combinations. Carryovers
(called rollovers in the UK) build up the jackpot until it is won. In Canada,
carryovers build until the jackpot is won. In the UK 6/49 game, rollovers are
capped at three. If there are no jackpot winners then, the jackpot funds not
paid out are added to the existing fund for the second tier prize (bonus) and
then shared by the various winners. In all the draws so far, the rollover has
never reached this fourth rollover. Betting increases as the carryover builds
since the potential jackpot rises.5 These cases are favorable to the unpopular
numbers hypothesis; among other tilings they correspond to the Canadian
and UK games in which the winnings are paid up front (not over twenty or
more years as in the US) and tax free (unlike in the US). The combination of
tax free winnings plus being paid in cash makes the Canadian and UK prizes
worth about three times those in the US. The optimal Kelly wagers are
extremely small. The reason for this is that the bulk of the expected value is
from prizes that occur with less than one in a million probability. A wealth
level of $1 million is needed in Case A to justify $1 ticket. The
corresponding wealth in Case B is over $150,000. Figures 13.3(a) and
13.3(b) provide the chance that the investor will double, quadruple or
increase tenfold this fortune before it is halved using Kelly and fractional
Kelly strategies for Cases A and B respectively. These chances are in the 40-
60% and 55-80% ranges for Cases A and B, respectively. With fractional
Kelly strategies in the range of 0.00000004 and 0.00000025 or less of the
investor's initial wealth, the chance of increasing one's initial fortune tenfold
before halving it is 95% or more with Cases A and B respectively. However,
it takes an average of 294 billion and 55 billion years respectively to achieve
this goal assuming there are 100 draws per year as there are in the Canadian
6/49 and UK 6-49.
Table 13.6: Lotto Game Experimental Data
Figures 13.4(a) and 13.4(b) give the probability of reaching $10 million
before falling to $1 million and $25,000 for various initial wealth for cases A
and B, respectively, with full, half and quarter Kelly wagering strategies.
The results indicate that the investor can have a 95% plus probability of
achieving the $10 million goal from a reasonable initial wealth level with the
quarter Kelly strategy for cases A and B. Unfortunately the mean time to
reach this goal this is 914 million years for case A and 482 million years for
case B. For case A with full Kelly it takes 22 million years on average and
384 million years with half Kelly for case A. For case B it takes 2.5 and 19.3
million years for full and half Kelly, respectively. It takes a lot less time, but
still millions of years on average to merely double one's fortune: namely 2.6,
4.6 and 82.3 million years for full, half and quarter Kelly, respectively for
case A and 0.792, 2.6 and 12.7 for case B. We may then conclude that
millionaires can enhance their dynasties' long-run wealth provided their
wagers are sufficiently small and made only when carryovers are sufficiently
large (in lotto games around the world). There are quite a few that could be
played.
Fig. 13.3 Probability of doubling, quadrupling and tenfolding before halving, Lotto 6/49. Source:
MacLean and Ziemba (1999)
Fig. 13.4 Probability of reaching the goal of $10 million under various conditions. Source: MacLean
and Ziemba (1999)
What about a non-millionaire wishing to become one? The aspiring

investor must pool funds until $150,000 is available for case B and $1
million for case A to optimally justify buying only one $1 ticket per draw.
Such a tactic is legal in Canada and in fact is highly encouraged by the
lottery corporation which supplies legal forms for such an arrangement. Also
in the UK, Camelot will supply model 'agreement' forms for syndicates to
use, specifying who must pay what, how much, and when, and how any
prizes will be split. This is potentially very important for the treatment of
inheritance tax with large prizes. The situation is modeled in Figure 3. Our
aspiring millionaire puts up $100,000 along with nine others for the $1
million bankroll and when they reach $10 million each share is worth $1
million. The syndicate must play full Kelly and has a chance of success of
nearly 50 assuming that the members agree to disband if they lose half their
stake. Participants do not need to put up the whole $100,000 at the start. The
cash outflow is easy to fund, namely 10 cents per draw per participant. To
have a 50% chance of reaching the $1 million goal, each participant (and
their heirs) must have $50,000 at risk. It will take 22 million years, on
average, to achieve the goal.
The situation is improved for case B players. First, the bankroll needed is
about $154,000 since 65 tickets are purchased per draw for a $10 million
wealth level. Suppose our aspiring nouveau riche is satisfied with $500,000
and is willing to put all but $25,000/2 or $12,500 of the $154,000 at risk.
With one partner he can play half Kelly strategy and buy one ticket per case
B type draw. Figure 13.4(b) indicates that the probability of success is about
0.95. With initial wealth of $308,000 and full Kelly it would take million
years on average to achieve this goal. With half Kelly it would take, on
average, 2.7 million years and with quarter Kelly it would take 300 million
years.
The conclusion is that except for millionaires and pooled syndicates, it is
not possible to use the unpopular numbers in a scientific way to beat the
lotto and have high confidence of becoming rich; these aspiring millionaires
will also most likely be residing in a cemetery when their distant heirs
finally reach the goal.
What did we learn from this exercise?
(1) Lotto games are in principle beatable but the Kelly and fractional Kelly
wagers are so small that it takes virtually forever to have high
confidence of winning. Of course, you could win earlier or even on the
first draw and you do have a positive mean on all bets. Ziemba et al.
(1986) have shown that the largest jackpots contain about 47% of the
nineteen most unpopular numbers in 1986 shown in Table 13.3(b)
versus 17% unpopular numbers in the smallest jackpots. Hence, if you
play, emphasizing unpopular numbers is a valuable strategy to employ.
But frequently numbers other than the unpopular ones are drawn. So
the strategy of focussing on three or four unpopular numbers and then
randomly selecting the next two numbers might work. Gadgets to
choose such numbers are easy to devise. But you need deep pockets
here and even then you might ruin. The best six numbers, see Table
13.5 once won a $10 million unshared jackpot in Florida. Could you
bet more? Sorry: log is the most one should ever bet.
(2) The Kelly and fractional Kelly wagering schemes are very useful in
practice but the size of the wagers will vary from very tiny to enormous
bets. My best advice: never over bet; it will eventually lead to trouble
unless it is controlled somehow and that is hard to do!
Good and bad properties of the Kelly criterion

If your outlook is well extended,, the Kelly criterion is the approach best
suited to generating a fortune.
We now discuss the good and bad properties of the Kelly expected log
capital growth criterion. If your horizon is long enough then the Kelly
criterion is the road, however bumpy, to the most wealth at the end and the
fastest path to a given rather large fortune.
The great investor Warren Buffett's Berkshire Hathaway actually has had
a growth path quite similar to full Kelly betting. Figure 6.1 shows this
performance from 1985 to 2000 in comparison with other great funds.
Buffett also had a great record from 1977 to 1985 turning 100 into 1429.87,
and 65,852.40 in April 2000 and about $132,700 on September 30, 2012.
Keynes was another Kelly type bettor. His record running King's College
Cam-bridge's Chest Fund is shown in Figure 6.4 versus the British market
index for 1927 to 1945, data from Chua and Woodward (1983). Notice how
much Keynes lost the first few years; obviously his academic brilliance and
the recognition that he was facing a rather tough market kept him in this job.
In total his geometric mean return beat the index by 10.01%. Keynes was an
aggressive investor with a capital asset pricing model beta of 1.78 versus the
benchmark United Kingdom market return, a Sharpe ratio of 0.385,
geometric mean returns of 9.12% per year versus -0.89% for the benchmark.
Keynes had a yearly standard deviation of 29.28% versus 12.55% for the
benchmark. These returns do not include Keynes' (or the benchmark's)
dividends and interest, which he used to pay the college expenses. These
were about 3% per year. Kelly cowboys have their great returns and losses
and embarrassments. Not covering a grain contract in time led to Keynes
taking delivery and filling up the famous chapel. Fortunately it was big
enough to fit in the grain and store it safely until it could be sold; see the
cartoon. Keynes' investment behavior, according to Ziemba (2003) was
equivalent to 80% Kelly and 20% cash so he would use the negative power
utility function –w−0.25.
Keynes emphasized three principles of successful investments in his 1933
report:
(1) a careful selection of a few investments (or a few types of investment)
having regard to their cheapness in relation to their probable actual and
potential intrinsic value over a period of years ahead and in relation to
alternative invest-ments at the time;
(2) a steadfast holding of these in fairly large units through thick and thin,
perhaps for several years until either they have fulfilled their promise
or it is evident that they were purchased on a mistake; and
(3) a balanced investment position, i.e., a variety of risks in spite of
individual holdings being large, and if possible, opposed risks.
He really was a lot like Buffett with an emphasis on value, large holdings
and patience.
In November 1919, Keynes was appointed second bursar. Up to this time
King's College investments were only in fixed income trustee securities plus
their own land and buildings. By June 1920 Keynes convinced the college to
start a separate fund containing stocks, currency and commodity futures.
Keynes became first bursar in 1924 and held this post which had final
authority on investment decisions until his death in 1945.
And Keynes did not believe in market timing as he said:
We have not proved able to take much advantage of a general
systematic movement out of and into ordinary shares as a whole at
different phases of the trade cycle. As a result of these experiences I
am clear that the idea of wholesale shifts is for various reasons
impracticable and indeed undesirable. Most of those who attempt to
sell too late and buy too late, and do both too often, incurring heavy
expenses and developing too unsettled and speculative a state of
mind, which, if it is widespread, has besides the grave social
disadvantage of aggravating the scale of the fluctuations.
The main disadvantages result because the Kelly strategy is very very
aggressive with huge bets that become larger and larger as the situations are
most attractive: recall that the optimal Kelly bet is the mean edge divided by
the odds of winning. As I repeatedly argue,. the mean counts by far the most.
There is about a 20-2:1 ratio of expected utility loss from similar sized errors
of means, variances and covariances, respectively, as discussed in Chapter 3.
Returning to Buffett who gets the mean right, better than almost all, notice
that the other funds he outperformed are not shabby ones at all. Indeed they
are George Soros' Quantum, John Neff's Windsor, Julian Robertson's Tiger
and the Ford Foundation, all of whom had great records as measured by the
Sharpe ratio. Buffett made 32.07% per year net from July 1977 to March
2000 versus 16.71% for the S&P500. Wow! Those of us who like wealth
prefer Warren's path but his higher standard deviation path (mostly
winnings) leads to a lower Sharpe (normal distribution based) measure; see
Siegel et al. (2001). Chapter 6 proposes a modification of the Sharpe ratio to
not penalize gains. This improves Buffet's evaluation.
Since Buffett and Keynes are full or close to full Kelly bettors their means
must be even more accurate. With their very low risk tolerances, the errors in
the mean are 100+ times as important as the co-variance errors.
Kelly has essentially zero risk aversion since its Arrow-Pratt risk aversion
index is
which is essentially zero. Hence it never pays to bet more than the Kelly
strategy because then risk increases (lower security) and growth decreases so
is stochastically dominated. As you bet more and more above the Kelly bet,
its properties become worse and worse. When you bet exactly twice the
Kelly bet, then the growth rate is zero plus the risk free rate; see the proof at
the end of this chapter.
If you bet more than double the Kelly criterion, then you will have a
negative growth rate. With derivative positions one's bet changes
continuously so a set of positions amounting to a small bet can turn into a
large bet very quickly with market moves. Long Term Capital is a prime
example of this overbetting leading to disaster but the phenomenon occurs
all the time all over the world. Overbetting plus a bad scenario leads
invariably to disaster.
Thus you must either bet Kelly or less. We call betting less than Kelly
fractional Kelly, which is simply a blend of Kelly and cash. Consider the
negative power utility function δwδ for δ < 0. This utility function is concave
and when δ → 0 it converges to log utility. As δ gets larger negatively, the
investor is less aggressive since his Arrow-Pratt risk aversion is also higher.
For a given δ and α =1/(1 — δ) between 0 and 1, will provide the same
portfolio when α is invested in the Kelly portfolio and 1 — α is invested in
cash.
This result is correct for lognormal investments and approximately correct
for other distributed assets; see MacLean, Ziemba and Li (2005). For
example, half Kelly is δ = –1 and quarter Kelly is δ = –3. So if you want a
less aggressive path than Kelly pick an appropriate δ. Below I discuss a way
to pick δ continuously in time so that wealth will stay above a desired wealth
growth path with high given probability; see Figure 13.5.
I now list these and other important Kelly criterion properties, updated
from MacLean, Ziemba and Blazenko (1992) and MacLean and
Ziemba(1999).
Good Maximizing ElogX asymptotically maximizes the rate of asset growth
See Breiman (1961), Algoet and Cover (1988).
Good The expected time to reach a preassigned goal is asymptotically as X
increases least with a strategy maximizing ElogXN. See Breiman (1961),
Algoet and Cover (1988), Browne (1997a).
Good Maximizing median logX See Ethier (1987).
Bad False Property: If maximizing ElogXN almost certainly leads to a better
outcome then the expected utility of its outcome exceeds that of any
other rule provided N is sufficiently large. Counter Example: u(x) = x,
1/2 < p < 1, Bernoulli trials f = 1 maximizes EU(x) but f = 2p – 1 < 1
maximizes ElogXN. See Samuelson (1971), Thorp (1975, 2006).
Good The ElogX bettor never risks ruin. See Hakansson and Miller (1975).
Bad If the ElogXN bettor wins then loses or loses then wins with coin tosses,
he is behind. The order of win and loss is immaterial for one, two,..., sets
of trials since (1 + γ)(1 — γ)XO = (1 — γ2)X0 < X0. This is not true for
favorable games.
Good The absolute amount bet is monotone in wealth. (δElogX)/δW0 > 0.
Bad The bets are extremely large when the wager is favorable and the risk is
very low. For single investment worlds, the optimal wager is
proportional to the edge divided by the odds. Hence for low risk
situations and corresponding low odds, the wager can be extremely large.
For one such example, see Ziemba and Hausch (1986; 159-160). There,
in the inaugural 1984 Breeders' Cup Classic $3 million race, the optimal
fractional wager on the 3-5 shot Slew of Gold was 64%. (See also the
74% future bet on the January effect in Chapter 15. Thorp and I actually
made this place and show bet and won with a low fractional Kelly wager.
Slew finished third but the second place horse Gate Dancer was
disqualified and placed third. Luck (a good scenario) is also nice to have
in betting markets. Wild Again won this race; the first great victory by
the masterful jockey Pat Day.
Bad One overinvests when the problem data is uncertain. Investing more
than the optimal capital growth wager is dominated in a growth-security
sense. Hence, if the problem data provides probabilities, edges and odds
that may be in error, then the suggested wager will be too large.
Bad The total amount wagered swamps the winnings - that is, there is much
churning. Ethier and Tavare (1983) and Griffin (1985) show that the
Expected Gain/E Bet is arbitrarily small and converges to zero in a
Bernoulli game where one wins the expected fraction p of games.
Bad The unweighted average rate of return converges to half the arithmetic
rate of return. As with the above bad property, this indicates that you do
not seem to win as much as you expect. See Ethier and Tavar (1983) and
Griffin (1985).
Bad Betting double the optimal Kelly bet reduces the growth rate of wealth
to zero plus the risk free rate. See Stutzer (1998) and Janecek (1999) and
the appendix in this chapter for a proof.ood The ElogX bettor is never
behind any other bettor on average in 1, 2, trials. See Finkelstein &
Whitley (1981).
Good The ElogX bettor has an optimal myopic policy. He does not have to
consider prior nor subsequent investment opportunities. This is a
crucially important result for practical use. Hakansson (1972) proved that
the myopic policy obtains for dependent investments with the log utility
function. For independent investments and power utility a myopic policy
is optimal, see Mossin (1968).
Good The chance that an ElogX wagerer will be ahead of any other wagerer
after the first play is at least 50%. See Bell and Cover (1980).
Good Simulation studies show that the ElogX bettor's fortune pulls way
ahead of other strategies wealth for reasonable-sized samples. The key
again is risk. See Ziemba and Hausch (1986). General formulas are in
Aucamp (1993).
Good If you wish to have higher security by trading it off for lower growth,
then use a negative power utility function or fractional Kelly strategy.
See MacLean, Sanegre, Zhao and Ziemba (2004) who show how to
compute the coefficent to stay above a growth path with given
probability. See Figure 13.5 for the idea and the example below.
Bad Despite its superior long-run growth properties, it is possible to have
very poor return outcome. For example, making 700 wagers all of which
have a 14% advantage, the least of which had a 19% chance of winning
can turn $1000 into $18. But with full Kelly 16.6% of the time $1000
turns into at least $100,000, see Ziemba and Hausch (1996). Half Kelly
does not help much as $1000 can become $145 and the growth is much
lower with only $100,000 plus final wealth 0.1% of the time.
Bad It can take a long time for a Kelly bettor to dominate an essentially
different strategy. In fact this time may be without limit. Suppose μα =
20%, μβ = 10%, σα = σβ = 10%. Then in five years A is ahead of B with
95% confidence. But if σα = 20, σβ = 10% with the same means, it takes
157 years for A to beat B with 95% confidence. In coin tossing suppose
game A has an edge of 1.0% and game B 1.1%. It takes two million trials
to have an 84% chance that game A dominates game B, see Thorp
(2006).
Calculating the optimal Kelly fraction

I now discuss how to calculate the optimal Kelly fraction to grow wealth as
fast as possible in the long run but to stay above a wealth growth path at
particular intervals with high probability in the short run. This approach
provides one way to scientifically cut down the size of one's bet to raise
security levels while still maintaining high growth levels.
Most applications of fractional Kelly strategies pick the fractional Kelly
strategy in an ad hoc fashion. MacLean, Ziemba and Li (2005) show that
growth and security tradeoffs are effective for general return distributions in
the sense that growth is monotone decreasing in security. But with general
return distributions, this tradeoff is not necessarily efficient in the sense of
Markowitz (generalized growth playing the role of mean and security the
role of variance). However, if the investment returns are lognormal, the
tradeoff is efficient. MacLean, Ziemba and Li also develop an investment
strategy where the investor sets upper and lower targets and rebalances when
those targets are achieved. Empirical tests in MacLean, Sanegre, Zhao and
Ziemba (MSZZ) (2004) show the advantage of this approach.
A solution of a version of the problem of how to pick an optimal Kelly
function was provided in MSZZ (2004). To stay above a wealth path using a
Kelly strategy is very difficult since the more attractive the investment
opportunity, the larger the bet size and hence the larger the chance of falling
below the path. Figure 13.5 illustrates this. An extension of MSZZ (2004) to
include convex penalties for drawdowns is in MacLean, Zhao and Ziemba
(2009, 2012). By penalizing drawdowns, the actual wealth path is likely to
stay above the prespecified wealth path.
Fig. 13.5 Kelly fractions and path achievement
Calculating the optimal Kelly fraction

MSZZ use a continuous time lognormally distributed asset model to
calculate that function at various points in time to stay above the path with a
high exogenously specified value at risk probability. They provide an
algorithm for this. The idea is illustrated using the following application to
the fundamental problem of asset allocation over time, namely, the
determination of optimal fractions over time in cash, bonds and stocks. The
data in Table 13.7 are yearly asset returns for the S&P500, the Salomon
Brothers Bond index and U.S. T-bills for 1980-1990 with data from Data
Resources, Inc. Cash returns are set to one in each period and the mean
returns for other assets are adjusted for this shift. The standard deviation for
cash is small and is set to 0 for convenience.
Table 13.7: Yearly Wealth Relatives on Assets Relative to Cash (%)

A simple grid was constructed from the assumed lognormal distribution
for stocks and bonds by partitioning at the centroid along the principal
axes. A sample point was selected from each quadrant to approximate the
parameter values. The planning horizon is T = 3, with 64 scenarios each with
probability 1/64 using the data in Table 13.8. The problems are solved with
the VaR constraint (Table 13.9) and then for comparison, with the stronger
drawdown constraint (Table 13.10).
Table 13.8: Rates of Return Scenarios
VaR Control with w* = a
The model is
With initial wealth W(0) = 1, the value at risk is a3. The optimal investment
decisions and optimal growth rate for several values of a, the secured
average annual growth rate and 1 – α, the security level, are shown in Table
13.9. The heuristic described in MSZZ was used to determine A, the set of
scenarios for the security constraint. Since only a single constraint was
active at each stage the solution is optimal.
• The mean return structure for stocks is favorable in this example, as is
typical over long horizons.6 Hence the aggressive Kelly strategy is to
invest all the capital in stock most of the time.
• When security requirements are high some capital is in bonds.
• As the security requirements increase the fraction invested in bonds
increases.
• The three-period investment decisions are more conservative as the
horizon approaches.
Secured Annual Drawdown: b

The VaR condition only controls loss at the horizon. At intermediate times
the investor could experience substantial loss, and face bankruptcy. A more
stringent risk control constraint, drawdown, considers the loss in each period
using the model
Table 13.9: Growth with Secured Rate

Table 13.10: Growth with Secured Maximum Drawdown
This constraint follows from the arithmetic random walk ln W(t),
The optimal investment decisions and growth rate for several values of b,
the drawdown and 1 – α, the security level are shown in Table 13.10.
• The heuristic in MSZZ is used in determining scenarios in the solution.
• The security levels are different since constraints are active at different
probability levels in this discretized problem.
• As with the VaR constraint, investment in the bonds and cash increases
as the drawdown rate and/or the security level increases.
• The strategy is more conservative as the horizon approaches.
• For similar requirements (compare a = 0.97,1 – α = 0.85 and b = 0.97,1
– α = 0.75), the drawdown condition is more stringent, with the Kelly
strategy (all stock) optimal for VaR constraint, but the drawdown
constraint requires substantial investment in bonds in the second and
third periods.
• In general, consideration of drawdown requires a heavier investment in
secure assets and at an earlier time point. It is not a feature of this
aggregate example, but both the VaR and drawdown constraints are
insensitive to large losses, which occur with small probability.
• Control of that effect would require the lower partial mean violations
condition or a model with a convex risk measure that penalizes more
and more as larger constraint violations occur, see e.g. the InnoALM
model in Chapter 14.
• The models lead to hair trigger type behavior, very sensitive to small
changes in mean values (as discussed in chapters 3 and 14; see also
Figure 3.4).
Appendix
Proof that betting exactly double the Kelly criterion amount leads to a
growth rate equal to the risk free rate. This result is due to Thorp (1997),
Stutzer (1998) and Janacek (1998) and possibly others. The following simple
proof is due to Harry Markowitz.
In continuous time
Ep, Vp, gp are the portfolio expected return, variance and expected log,
respectively. In the CAPM
where X is the portfolio weight and r0 is the risk free rate. Collecting terms
and setting the derivative of gp to zero yields
which is the optimal Kelly bet with optimal growth rate

Substituting double Kelly, namely Y = 2X for X above into
and simplifying yields
Hence g0 = r0 when Y = 2S.

The CAPM assumption is not needed. For a more general proof and
illustration, see Thorp (2006).
1For those who would like a technical survey of capital growth theory, see MacLean, Thorp and
Ziemba (2010).
2For one or two assets with fixed odds, take derivatives and solve for the optimal wagers; for multi-
asset bets under constraints; and when portfolio choices affect returns (odds), one must solve a
nonlinear program which, possibly, is non-convex.
3Another is to look for lottery design errors. As a consultant on lottery design for the past thirty years,
I have seen plenty of these. My work has been largely to get these bugs out before the games go to
market and to minimize the damage when one escapes the lottery commissions' analysis. Design errors
are often associated with departures from the pure parimutuel method, for example guaranteeing the
value of smaller prizes at too high a level and not having the games checked by an expert.
4See Ziemba et al. (1986), Dr Zs Lotto 6/49 Guidebook. While parts of the guidebook are dated, the
concepts, conclusions, and most of the text provide a good treatment of such games. For those who
want more theory, see MacLean and Ziemba (1999, 2006)
5An estimate of the number oftickets sold versus the carryover in millions is proportional to the
carryover to the power 0.811. Hence, the growth is close to 1:1 linear. See Ziemba et al. (1986)
6See e.g. Keim and Ziemba (2000), Dimson et al (2006), Constantinides (2002) and Siegel (2002).
InnoALM, the Innovest Austrian Pension Fund
Financial Planning Model1
Siemens Oesterreich, part of the global Siemens Corporation, is the largest

privately owned industrial company in Austria. Its businesses, with revenues
of 2.4 billion in 1999, include information and communication networks,
information and commu-nication products, business services, energy and
traveling technology, and medical equipment. Its pension fund, established
in 1998, is the largest corporate pension plan in Austria and is a DCP. More
than 15,000 employees and 5,000 pensioners are members of the pension
plan, which had 510 million in assets under management as of December
1999.
Innovest Finanzdienstleistungs, founded in 1998, is the investment
manager for Siemens Oesterreich, the Siemens pension plan, and other
institutional investors in Austria. With =€ 2.2 billion in assets under
management, Innovest focuses on asset management for institutional money
and pension funds. Of 17 plans analyzed in the 1999-2000 period, it was
rated as the best plan in Austria. The motivation to build InnoALM, as
described in Geyer and Ziemba (2008) and their earlier working papers, was
based on the desire to have superior performance and good decision aids to
help achieve this ranking.
Various uncertain aspects, possible future economic scenarios, stocks,
bonds, and other investments, transaction costs, liquidity, currency aspects,
liability com-mitments over time, Austrian pension fund law, and company
policy suggested that a good way to approach this asset-liability problem
was via a multiperiod stochastic linear programming model. These models
evolved from Kusy and Ziemba (1986), Carino, Kent, Myers, Stacy,
Sylvanus, Turner, Watanabe, and Ziemba (1994), Carino and Ziemba (1998)
and Carino, Meyers, and Ziemba (1998) and Ziemba and Mulvey (1998).
This model has innovative features, such as state-dependent correlation
matrixes, fat-tailed asset-return distributions, convex penalties on target
violations, a concave objective function, simple computational schemes, and
easy to interpret output.
InnoALM was produced in six months in 2000, with Geyer and Ziemba
serving as consultants and with assistance from Herold and Kontriner who
were Innovest employees. InnoALM demonstrates that a small team of
researchers with a limited budget can quickly produce a valuable modeling
system that can easily be operated by nonstochastic programming specialists
on a single personal computer (PC). The IBM OSL stochastic programming
software provides a good solver. The solver was interfaced with user-
friendly input and output capabilities. Calculation times on the PC are such
that different modeling situations can be easily developed and the
implications of policy, scenario, and other changes can be seen quickly. The
graphical output provides pension fund management with essential
information to aid in the making of informed investment decisions and
understanding the probable outcomes and risk involved with these actions.
The model can be used to explore possible European, Austrian, and Innovest
policy alternatives.
The liability side of the Siemens pension plan consists of employees for
whom Siemens is contributing DCP payments and retired employees who
receive pension payments. Contributions are based on a fixed fraction of
salaries, which varies across employees. Active employees are assumed to
be in steady state; thus, employees are replaced by a new employee with the
same qualification and sex so that there is a constant number of similar
employees. Newly employed staff start with less salary than retired staff,
which implies that total contributions grow less rapidly than individual
salaries. Figure 14.1 shows the expected index of total payments for active
and retired employees until 2030.
Fig. 14.1 Index of Expected Payments for Active and Retired Employees, 2000-30. Source: Geyer and
Ziemba (2008)
The set of retired employees is modeled using Austrian mortality and

marital tables. Widows receive 60% of the pension payments. Retired
employees receive pension payments after reaching age 65 for men and 60
for women. Payments to retired employees are based on the individually
accumulated contribution and the fund performance during active
employment. The annual pension payments are based on a discount rate of
6% and the remaining life expectancy at the time of retirement. These
annuities grow by 1.5% annually to compensate for inflation. Hence, the
wealth of the pension fund must grow by 7.5% a year to match liability
commitments. Another output of the computations is the expected annual net
cash flow of plan contributions minus payments. Because the number of
pensioners is rising faster than plan contributions, these cash flows are
negative and the plan is declining in size.
The model determines the optimal purchases and sales for each of N
assets in each of T planning periods. Typical asset classes used at Innovest
are U.S., Pacific, European, and emerging market equities and U.S., U.K.,
Japanese, and European bonds. The objective is to maximize the concave
risk-averse utility function ‘expected terminal wealth’ less convex penalty
costs subject to various linear constraints. The effect of such constraints is
evaluated in the examples that follow, including Austria’ limits of 40%
maximum in equities, 45% maximum in foreign securities, and 40%
minimum in Eurobonds. See Table 14.1 for these constraints in various
countriesl The convex risk measure is approximated by a piecewise linear
function, so the model is a multiperiod stochastic linear program. Typical
targets that the model tries to achieve (and is penalized for if it does not) are
for growth of 7.5% a year in wealth (the fund’ assets) and for portfolio
performance returns to exceed benchmarks. Excess wealth is placed into
surplus reserves, and a portion of the excess is paid out in succeeding years.
Table 14.1: National Investment Restrictions on Pension Plans. Source:

European Commission (1997)
The elements of InnoALM are described in Figure 14.2. The interface to

read in data and problem elements uses Excel. Statistical calculations use the
program Gauss, and these data are fed into the IBM 0SL solver, which
generates the optimal solution to the stochastic program. The output, some
of which is shown in the next section, uses Gauss to generate various tables
and graphs and retains key variables in memory to allow for future modeling
calculations.
Fig. 14.2 Elements of InnoALM. Source: Geyer and Ziemba (2008)
Formulation of InnoALM as a multistage stochastic linear

programming model
The non-negative decision variables are wealth (after transactions) , Wit and
purchases Pit and sales Sit for each asset (i = 1,..., N.. Purchases and sales are
in periods t = 0,... ,T – 1. Purchases and sales are scenario dependent except
for t=0.2
Wealth accumulates over time for a T period model according to
is the prespecified initial value of asset i. There is no uncertainty in
the initialization period t = 0. Tildas denote scenario-dependent random
parameters or decision variables. Returns are associated with time intervals.
are the (random) gross returns for asset i between t – 1 and t.
The scenario generation and statistical properties of returns are discussed
below.
The budget constraints are
where tcpi and tcsii denote asset-specific linear transaction-costs for

purchases and sales, and Ci is the fixed (non-random) net cashflow (inflow if
positive).
Portfolio weights can be constrained over linear combinations (subsets) of
assets or individual assets via
where θU is the maximum percentage and and θL is the minimum percentage

of the subsets U and L of assets i = 1,... ,N included in the restrictions. The
θU's, θL's, U's and L's may be time dependent.
Risk is measured as a weighted discounted convex function of target
violation shortfalls of various types in various periods. In a typical
application, the deterministic wealth target is assumed to grow by 7.5% in
each year. The wealth targets are modeled via
where are wealth-target shortfall variables.

The shortfall is penalized using a piecewise linear convex risk measure
using the variables and constraints
where is the wealth target shortfall associated with segment j of the

cost- function, bj is the .j-th breakpoint of the risk measure function (b0 = 0),
and m is the number of segments of the function. A piecewise linear
approximation to the convex quadratic risk measure is used so the model
remains linear. Convexity guarantees that if then is at its
maximum and if is not at its maximum then
Stochastic benchmark goals can be set by the user and are similarly
penalized with a piecewise linear convex risk measure for
underachievement.
The benchmark target is scenario dependent. It is based on stochastic
asset returns and fixed asset weights aj defining the benchmark portfolio
with shortfall constraints

where is the benchmark-target shortfall. These shortfalls are also
penalized with a piecewise linear convex risk measure.
If total wealth exceeds the target, a fraction α = 10% of the exceeding
amount is allocated to a reserve account and invested in the same way as
other available funds. However, the wealth targets at future stages are
adjusted. Additional nonnegative decision variables are introduced and
the wealth target constraints become
where .
Since pension payments are based on wealth levels, increasing these
levels increases pension payments. The reserves provide security for the
pension plan' increase of pension payments at each future stage.
The pension plan' objective function is to maximize the expected
discounted value of terminal wealth in period T net of the expected
discounted penalty costs over the horizon from the convex risk measures for
the wealth- and benchmarktargets, respectively,
Expectation is over T period scenarios ST. The discount factors are related
to the interest rate r by T. Usually r is taken to be the three or six month
Treasury-bill rate.
The pension plan' objective function is to maximize the expected
discounted value of terminal wealth in period T net of the expected
discounted penalty costs over the horizon from the convex risk measures
ck(·) for the wealth and benchmark targets, respectively,
Expectation is over T period scenarios ST. The discount factors dt are
related to the interest rate r by dt = (1 + r)–t. Usually r is taken to be the three
or six mont Treasury-bill rate. The Vk are weights for the wealth and
benchmark shortfalls and the wt are weights for the weighted sum of
shortfalls at each stage.
The weights are normalized via
Such concave objective functions with convex risk measures date to Kusy
and Ziemba (1986), were used in the Russell-Yasuda model (Cariño and
Ziemba, 1998), and are justified in an axiomatic sense in Rockafellar and
Ziemba (2000). Non technical decision makers find the increasing penalty
for target violations a good approach and easy to understand.
In the InnoALM model the penalty function ck (Mk) is a quadratic utility
function. Kallberg and Ziemba (1983) show for normally distributed asset
returns that varying the average Arrow-Pratt absolute risk aversion index RA
traces out the whole spectrum of risk attitudes of all concave utility
functions. The most aggressive behavior is log utility which has RA =
1/wealth which is essentially zero. Typical 60-40 stock-bond pension funds
have RA = 4. The Kallberg-Ziemba (1983) results indicate that for
computational purposes the quadratic utility function u(w) = w2/2 will
suffice and is easier to use in the optimization. The error in this
approximation is close to zero and well below the accuracy of the data.
The parameter λ in the objective corresponds to RA/2 which in the
quadratic utility function is the weight assigned to risk measured in terms of
variance. The objective function of the InnoALM model only penalizes
wealth and benchmark target shortfalls. If the target growth is roughly equal
to the average return of the portfolio shortfalls measure only negative
deviations form the mean, whereas variance is based on positive and
negative deviations. This implies that shortfalls only account for about half
of the variance. Therefore, to obtain results in agreement with a quadratic
utility function we use λ rather than RA/2, in the objective function. To obtain
a solution to the allocation problem for general levels of total initial wealth
we use the rescaled parameter in the objective function.
Uncertainty is modelled using multiperiod discrete probability scenarios
using statistical properties of the assets’ returns. A scenario tree is defined
by the number of stages and the number of arcs leaving a particular node.
Figure 14.3 shows a tree with a 2-2-3 node structure for a three-period
problem with four stages and introduces some definitions and terminology.
The tree always starts with a single node which corresponds to the present
state (t=0). Decisions are made at each node of the tree and depend on the
current state which reflects previous decisions and uncertain future paths. A
single scenario st is a trajectory that corresponds to a unique path leading
from the single node at stage 1 (t=0) to a single node at t. Two scenarios are
identical until t-1 and differ in subsequent periods t,...,T. The scenario
assigns specific values to all uncertain parameters along the trajectory, i.e.
asset returns and benchmark targets for all periods. Given all T period
scenarios ST and their respective probabilities one has a complete
description of the uncertainty of the model.
Fig. 14.3 Scenario Tree with a 2-2-3 Node Structure (12 Scenarios)
Allocations are based on optimizing the stochastic linear program with

IBM' optimization solutions library using the stochastic extension library
(OSLE version 3). IBM has ceased all sales of this product in 2004. While
existing installations of OSLE may still be used, new implementations
require alternative software such as the open source project COIN-OR (see
http://www.coin-or.org). The library uses the Stochastic Mathematical
Programming System (SMPS) input format for multistage stochastic
programs (see King et al. 2005). The core-file contains information about
the decisions variables, constraints, right-hand-sides and bounds. It contains
all fixed coefficients and dummy entries for random elements. The stoch-file
reflects the node structure of the scenario tree and contains all random
elements, i.e. asset and benchmark returns, and probabilities. Non-
anticipatory constraints are imposed to guarantee that a decision made at a
specific node is identical for all scenarios leaving that node so the future
cannot be anticipated. This is implemented by specifying an appropriate
scenario structure in the stoch input file. The time-file assigns decision
variables and constraints to stages. The required statements in the input files
are automatically generated by the InnoALM system.
Some Typical Applications

To illustrate the model’ use, Geyer and Ziemba (2008) presented present
results for a problem with four asset classes (European stocks, U.S. stocks,
European bonds, and U.S. bonds) with five periods (six stages). The periods
are twice 1 year, twice 2 years, and 4 years (10 years in total). They assumed
discrete compounding, which implies that the mean return for asset i (i) used
in simulations is where is the mean, based on log returns. Using a 100-5-5-2-
2 node structure, they generated 10,000 scenarios. Initial wealth equals 100
units, and the wealth target is assumed to grow at an annual rate of 7.5%. To
make the results more general, they do not consider a benchmark target or
cash in- and outflows in this sample application. they used a risk-aversion
index of RA = 4, and the discount factor equals 5%, which corresponds
roughly with a simple static mean variance model to a standard 60/40
stock/bond pension fund mix (see Kallberg and Ziemba 1983).
Assumptions about the statistical properties of returns measured in
nominal euros are based on a sample of monthly data from January 1970 for
stocks and 1986 for bonds to September 2000. Summary statistics for
monthly and annual log returns are shown in Table 14.2. The U.S. and
European equity means for the longer 1970-2000 period were much lower
and slightly less volatile than those for the 1986-2000 period. The monthly
stock returns were nonnormal and negatively skewed. Monthly stock returns
were fat tailed, whereas monthly bond returns were close to normal (the
critical value of the Jarque-Bera test for a = 0.01 is 9.2).
For long-term planning models such as InnoALM, with its one-year
review period, however, properties of monthly returns are less relevant. The
bottom panel of Table 14.2 shows statistics for annual returns. Although
average returns and volatilities remained about the same, one year of data is
lost when annual returns are computed and the distributional properties
changed dramatically. There was negative skewness, but no evidence existed
for fat tails in annual returns, except for European stocks (1970-2000) and
U.S. bonds.
The mean returns from this sample are comparable to the 1900-2011 112-
year mean returns estimated by Dimson, Marsh and Staunton (2012), see
Table 14.3.
Assumptions about means, standard deviations, and correlations for the
appli-cations of InnoALM appear in Table 14.4 and are based on the sample
statistics presented in Table 14.5. Projecting future rates of returns from past
data is difficult. The equity means from the 1970-2000 period are used
because the 1986-2000 period had an exceptionally high performance of
stocks that is not assumed to prevail in the long run.
Table 14.2: Statistical Properties of Asset Returns. Source: Geyer and

Ziemba (2008)
The correlation matrixes in Table 14.4 for the three different regimes are
based on the regression approach of Solnik, Boucrelle and Le Fur (1996).
Moving average estimates of correlations among all assets are functions of
standard deviations of U.S. equity returns. The estimated regression
equations are then used to predict the correlations in the three regimes shown
in Table 14.4. Results for the estimated regression equations appear in Table
14.5. Three regimes are considered, and the assumption is that 10% of the
time, equity markets are extremely volatile; 20% of the time, markets are
characterized by high volatility; and 70% of the time, markets are normal.
The 35% quantile of U.S. equity return volatility defines ‘normal’ periods.
‘Highly volatile’ periods are based on the 80% volatility quantile, and
‘extreme’ periods, on the 95% quartile. The associated correlations reflect
the return relationships that typically prevailed during those market
conditions. The correlations in Table 14.4 show a distinct pattern across the
three regimes. Correlations among stocks tend to increase as stock return
volatility rises, whereas the correlations between stocks and bonds tend to
decrease. European bonds may serve as a hedge for equities during
extremely volatile periods because bond and stock returns, which are usually
positively correlated, are then negatively correlated. The latter is a major
reason using scenario-dependent correlation matrixes is a major advance
over the sensitivity of one correlation matrix.
Optimal portfolios were calculated for seven cases-with and without
mixing of correlations and with normal, t-, and historical distributions. The
‘mixing’ cases NM, TM, HM use mixing correlations. Case NM assumes
normal distributions for all assets. Case HM uses the historical distributions
of each asset. Case TM assumes t-distributions with five degrees of freedom
for stock returns, whereas bond returns are assumed to have normal
distributions. The ‘average’ cases NA, HA, and TA use the same distribution
assumptions, with no mixing of correlation matrixes. Instead, the
correlations and standard deviations used in these cases correspond to an
‘average’ period in which 10%, 20%, and 70% weights are used to compute
the averages of correlations and standard deviations in the three different
regimes. Comparisons of the average (A) cases and mixing (M) cases are
mainly intended to investigate the effect of mixing correlations. TMC
maintains all assumptions of case TM but uses Austria' constraints on asset
weights (see Table 14.6). Eurobonds must be at least 40% and equity at most
40%, and these constraints are binding.
Table 14.3: Real returns of stocks, bonds and T-bills for various countries.
Source: Dimson, Marsh and Staunton, 2012
Some Test Results. Table 14.6 shows the optimal initial asset weights at
Stage 1 for the various cases. Table 14.7 shows results for the final stage
(expected weights, expected terminal wealth, expected reserves, and
shortfall probabilities). These tables exhibit a distinct pattern: The mixing-
correlation cases initially assign a much lower weight to European bonds
than the average-period cases. Single-period, mean-variance optimization,
and average-period cases (NA, HA, and TA) suggest an approximate 45/55%
stock/bond mix. The mixing-correlation cases (NM, HM, and TM) imply a
65/35% stock/bond mix. Investing in U.S. bonds is not optimal at Stage 1 in
any of the cases, an apparent result of the relatively high volatility of U.S.
bonds relative to euro bonds.
Table 14.7 shows that the distinction between A and M cases becomes
less pronounced over time. European equities, however, still have a
consistently higher weight in the mixing cases than in the no-mixing cases.
This higher weight is mainly at the expense of Eurobonds. In general, the
proportion of equities at the final stage is much higher than in the first stage.
This result may be explained by the fact that the expected portfolio wealth at
later stages is far above the target wealth level (206.1 at Stage 6), and the
higher risk associated with stocks is less important. The constraints in case
TMC lead to lower expected portfolio wealth throughout the horizon and to
a higher shortfall probability than in any other case. Calculations show that
initial wealth would have to be 35% higher to compensate for the loss in
terminal expected wealth stemming from those constraints. In all cases, the
optimal weight of equities is much higher than the historical 4.1% in Austria.
Table 14.4: Mean, Standard Deviation, and Correlation Assumptions.

Source: Geyer and Ziemba (2008)
Table 14.5: Regression Equations Relating Asset Correlations and U.S. Stock Return
Volatility. Source: Geyer and Ziemba (2008)
Table 14.6: Optimal Initial Asset Weights at Stage 1 (in %)

Table 14.7: Expected Terminal Wealth, Expected Reserves and
Probabilities of Shortfalls with a Target Wealth, Wt = 206.1
The expected terminal wealth levels and the shortfall probabilities at the
final stage shown in Table 14.7 make the difference between mixing and no-
mixing cases even clearer. The mixing-correlation cases yield higher levels
of terminal wealth and lower shortfall probabilities.
If the level of portfolio wealth exceeds the target, the surplus, , is allocated
to a reserve account. The reserves in t are computed from and are shown in
Table 14.6 for the final stage. These values are in monetary units given an
initial wealth level of 100. They can be compared with the wealth target
206.1 at Stage 6. Expected reserves exceed the target level at the final stage
by up to 16%. Depending on the scenario, the reserves can be as high as
1,800. Their standard deviation (across scenarios) ranges from 5 at the first
stage to 200 at the final stage. The constraints in case TMC lead to a much
lower level of reserves compared with the other cases, which implies, in fact,
less security against future increases of pension payments.
Optimal allocations, expected wealth, and shortfall probabilities are
mainly affected by considering mixing correlations but the type of
distribution chosen has a smaller impact. This distinction is primarily the
result of the higher proportion allocated to equities, if different market
conditions are taken into account by mixing correlations. The results of any
asset-allocation strategy depend crucially on the mean returns. Geyer and
Ziemba investigated the effect by parameterizing the forecasted future means
of equity returns. Assume that an econometric model forecasts that the future
mean return for U.S. equities is some value between 5-15%. The mean of
European equities is adjusted accordingly so that the ratio of equity means
and the mean bond returns shown in Table 14.2 are maintained. Geyer and
Ziemba retain all other assumptions of case NM (normal distribution and
mixing correlations). Figure 14.4 summarizes the effects of these mean
changes in terms of the optimal initial weights. As expected, the results are
sensitive to the choice of the mean return; see Chopra and Ziemba (1993)
and Kallberg and Ziemba (1981, 1984). If the mean return for U.S. stocks is
assumed to equal the long-run mean of 12%, as estimated by Dimson et al.
(2002), the model yields an optimal weight for equities of 100. A mean
return for U.S. stocks of 9%, however, implies an optimal weight of less than
30% for equities. The 1900-2012 long run mean estimates are lower, see
Table 14.3.
Model Tests
Because state-dependent correlations have a significant impact on asset-
allocation decisions, it is worthwhile to further investigate their nature and
implications from the perspective of testing the model. Positive effects on
the pension fund performance induced by the stochastic, multiperiod
planning approach will be realized only if the portfolio is dynamically
rebalanced, as implied by the optimal scenario tree. Geyer and Ziemba tested
the performance of the model considering this aspect. As a starting point,
they broke down the rebalancing decisions at later stages into groups of
achieved wealth levels. This process reveals the ‘decision rule’ implied by
the model, depending on the current state. Consider case TM. They formed
quintiles of wealth at Stage 2, computed the average optimal weights
assigned to each quintile, and did the same using quintiles of wealth at Stage
5.
The sensitivity of the mean carries into multiperiod models. There the
effect is strongest in period 1 then less and less in future periods, see Geyer
and Ziemba (2008). This is illustrated in Figures 3.4 and 14.4 for a five
period, 10 year model designed for the Siemen' Austria pension fund. There
it is seen that in period 1, with bond means for the US and Europe in the 6-
7% area, the optimal allocation to European and US equity can be 100%
with a mean of 12%+ and about 30% when the mean is 9% or less. Whereas
in later periods, this sensitivity is less and, by period 5, it is almost non-
existent.
Panel A and Panel B of Figure 14.4 depict the distribution of weights for
each of the five average levels of wealth at the two stages. Although the
average allocation at Stage 5 is essentially independent of the wealth level
achieved (the target wealth at Stage 5 is 154.3), the distribution at Stage 2
depends on the wealth level in a specific way. If average attained wealth is
103.4, which is slightly below the target, the model chooses a cautious
strategy. Bonds have the highest weight in this case (almost 50%). In this
situation, the model implies that the risk of even stronger underachievement
of the target is to be minimized. The model relies on the low, but more
certain, expected return of bonds to move back to the target level. If attained
wealth is far below the target (97.1), the model implies more than 70%
equities and a high share (10.9%) of relatively risky U.S. bonds. With such
strong underachievement, a cautious strategy has no room to attain the target
level again. If average attained wealth equals 107.9, which is close to the
target wealth of 107.5, the highest proportion would be invested in U.S.
assets, with 49.6% in equities and 22.8% in bonds. The U.S. assets are more
risky than the corresponding European assets, which is acceptable because
portfolio wealth is close to the target and risk does not play a big role. For
wealth levels above the target, most of the portfolio are switched to
European assets, which are safer than U.S. assets. This ‘decision’ may be
interpreted as an attempt to preserve the high levels of attained wealth.
The decision rules implied by the optimal solution can be used to perform
a test of the model using the following rebalancing strategy. Consider the 10-
year period from January 1992 to January 2002. In the first month of this
period, Geyer and Ziemba assumed that wealth is allocated according to the
optimal solution for Stage 1, given in Table 14.5. In each of the subsequent
months, they rebalance the portfolio as follows. First, Geyer and Ziemba
identify the current volatility regime (extreme, highly volatile, or normal)
based on the observed U.S. stock return volatility. Then, they search the
scenario tree to find a node that corresponds to the current volatility regime
and has the same or a similar level of wealth. The optimal weights from that
node determine the rebalancing decision. For the no- mixing cases NA, TA,
and HA, the information about the current volatility regime cannot be used
to identify optimal weights. In those cases, Geyer and Ziemba use the
weights from a node with a level of wealth as close as possible to the current
level of wealth.
Table 14.8 presents summary statistics for the complete-sample and outof-
sample periods. The mixing-correlation solutions, assuming normal and t-
distributions (cases NM and TM), provided a higher average return with
lower standard deviation than the corresponding no-mixing cases (NA and
TA). The advantage may be substantial, as indicated by the 14.9% average
return of TM compared with 10.0% for TA. The t-statistic for this difference
was 1.7 and was significant at the 5% level (one-sided test). Using the
historical-distribution and mixing-correlation case (HM) yielded a lower
average return than the no-mixing case (HA). In the constrained case
(TMC), the average return for the complete sample was in the same range as
for the unconstrained cases. This result stems primarily from the relatively
high weights assigned to U.S. bonds; U.S. bonds performed well during the
test period, whereas stocks performed poorly. The standard deviation of
returns was much lower because the constraints imply a lower degree of
rebalancing.
Fig. 14.4 The effects of state dependent correlations: optimal weights conditional on quintiles of
portfolio weights in periods 2 and 5 of the InnoALM model. Source: Geyer and Ziemba, 2008)
Table 14.8: Results of Asset-Allocation Strategies Using the Decision Rule Implied by
the Optimal Scenario Tree. Source: Geyer and Ziemba (2008)
To emphasize the difference between the cases TM and TA, Figure 14.5
compares the cumulated monthly returns obtained from the rebalancing
strategy for the two cases with a buy-and-hold strategy that assumes that the
portfolio weights on January 1992 were fixed at the optimal TM weights
throughout the test period. In comparison to the buy-and-hold strategy or the
performance using TA results, for which rebalancing does not account for
different correlation and volatility regimes, rebalancing on the basis of the
optimal TM scenario tree provided a substantial gain.
Fig. 14.5 Cumulative Monthly Returns for Different Strategies, 1992—2002. Source: Geyer and
Ziemba (2008)
Such in- and out-of-sample comparisons depend on the asset returns and
test period. To isolate the potential benefits from considering state-dependent
correla-tions, Geyer and Ziemba perform the following controlled simulation
experiment. Consider 1,000 10-year periods in which simulated annual
returns of the four assets are assumed to have the statistical properties
summarized in Table 14.1. One of the 10 years is assumed to be an ‘extreme’
year, two years correspond to ‘highly volatile’ markets, and seven years are
‘normal’ years. Geyer and Ziemba compare the average annual return of two
strategies: (a) a buy-and-hold strategy using the optimal TM weights from
Table 14.5 throughout the 10-year period, and (b) a rebalancing strategy that
uses the implied decision rules of the optimal scenario tree as explained in
the in- and out-of-sample tests above. For simplicity, they assume that the
current volatility regime is known in each period. The average annual returns
for 1,000 repetitions of the two strategies are 9.8% (rebalancing) and 9.2%
(buy and hold). The t-statistic for the mean difference is 5.4, indicating the
highly significant advantage of the rebalancing strategy, which exploits the
information about state-dependent correlations.
For comparison, Geyer and Ziemba use the optimal weights from the
constrained case TMC and repeat the same experiment. They obtain the
same average mean of 8.1% for both strategies. This result indicates that the
constraints imply insufficient rebalancing capacity. Therefore, knowledge
about the volatility regime cannot be sufficiently exploited to achieve
superior performance relative to a buy-and-hold strategy. This result also
shows that the relatively good performance of the TMC rebalancing strategy
in the sample 1992-2002 period was positively biased by the favorable
conditions during that time.
The model, once developed in 2000, proved to be very useful for
Innovest. In 2006, Konrad Kontriner (Member of the Board) and Wolfgang
Herold (Senior Risk Strategist) of Innovest stated that:
The InnoALM model has been in use by Innovest, an Austrian
Siemens subsidiary, since its first draft versions in 2000. Meanwhile
it has become the only consistently implemented and fully integrated
proprietary tool for assessing pension allocation issues within
Siemens AG worldwide. Apart from this, consulting projects for
various European corporations and pensions funds outside of
Siemens have been performed on the basis of the concepts of
InnoALM.
The key elements that make InnoALM superior to other consulting
models are the flexibility to adopt individual constraints and target
functions in combination with the broad and deep array of results,
which allows to investigate individual, path dependent behavior of
assets and liabilities as well as scenario based and Monte-Carlo like
risk assessment of both sides.
In light of recent changes in Austrian pension regulation the latter
even gained additional importance, as the rather rigid asset based
limits were relaxed for institutions that could prove sufficient risk
management expertise for both assets and liabilities of the plan.
Thus, the implementation of a scenario based asset allocation model
will lead to more flexible allocation restraints that will allow for
more risk tolerance and will ultimately result in better long term
investment performance.
Furthermore, some results of the model have been used by the
Austrian regulatory authorities to assess the potential risk stemming
from less constraint pension plans.
Conclusions
To finish this chapter we look at some key points to remember about the
stochastic programming approach to asset liability and wealth management.
Point 1. Means are by far the most important part of the distribution of
returns, especially the direction. Thus, you must estimate future means
well or you can quickly travel in the wrong direction, which usually
leads to losses or underperformance, or complete disaster if one is over
levered.
Point 2. Mean-variance models are useful as a basic guideline when you
are in an assets-only situation. Professionals adjust means using mean-
reversion, James- Stein, or truncated estimators and constrain output
weights. Do not change asset positions unless the advantage of the
change is significant. Do not use mean-variance analysis with liabilities
and other major market imperfections, except as a first test analysis.
Point 3. Trouble arises when you overbet and a bad scenario occurs. Thus,
do not overbet when there is any possibility of a bad scenario occurring,
unless the bet is protected by some type of hedge or stop loss.
Point 4. Trouble is exacerbated when the expected diversification does
not hold in the scenario that occurs. Thus, you must use scenario-
dependent correlation matrixes because simulations around historical
correlation matrixes are inadequate for extreme scenarios.
Point 5. When a large decline in the stock market occurs, the positive
correlation between stocks and bonds fails and they become negatively
correlated. Thus, when the mean of the stock market is negative, bonds
are more attractive, as is cash.
Point 6. Stochastic programming scenario-based models are useful when
you want to look at aggregate overall decisions, with liabilities, liquidity,
taxes, policy, legal, and other constraints, and have targets and goals you
want to achieve. It thus pays to make a complex stochastic programming
model when a lot is at stake and the essential problem has many
complications.
Point 7. Other approaches, such as continuous-time finance, decision-
rule-based stochastic programming, control theory, and so on, are useful
for problem insights and theoretical results. But in actual use, they may
lead to disaster unless modified. Black-Scholes option pricing theory
says you can hedge perfectly with lognormal assets, which can lead to
overbetting. Fat tails and jumps arise frequently and can occur without
warning. The S&P500 opened limit down 60 points or 6% when trading
resumed after 9/11, and it fell 14% that week. Thus, be careful of the
assumptions, including implicit ones, of theoretical models. Use the
results with caution no matter how complex and elegant the math or how
smart or famous the author. Remember, you have to be very smart to lose
millions and even smarter to lose billions.
Point 8. Do not be concerned with getting all the scenarios exactly right
when using stochastic programming models. You cannot do so, and it
does not matter that much anyway. Instead, worry about having the
problem periods laid out reasonably and make sure the scenarios
basically cover the means, the tails, and the chance of what could
happen. If the current situation has never occurred before, use one that is
similar to add scenarios. For a crisis in Brazil, use Russian crisis data, for
example. The results of stochastic programming will give you good
advice when times are normal and keep you out of severe trouble when
times are bad. Those using stochastic programming models may lose 5,
10, or 15%, but they will not lose 50, 70, or 95%, as some investors and
hedge funds have. Thus, if the scenarios are more or less accurate and
the problem elements are reasonably modeled, stochastic programming
will give good advice. You may slightly outperform in normal markets,
but you will greatly outperform in bad markets when other approaches
may blow up.
Point 9. Stochastic programming models for asset/liability management
were very expensive in the 1980s and early 1990s but are not expensive
now. Years ago, Vancouver analysts using a large linear programming
model to plan lumber operations at MacMillan Blodel used to fly to San
Francisco to use a large computer that would run all day to run the model
once. Now, models of this complexity take only seconds to run on
inexpensive personal computers. Thus, advances in computing power
and modeling expertise have made stochastic pro-gramming modeling
much less expensive. Such models, which are still complex and require
approximately six months to develop and test, cost a couple hundred
thousand dollars. A small team can make a model for a complex
organization quite quickly at fairly low cost compared with what is at
stake.
Point 10. Eventually, as more disasters occur and more successful
stochastic pro-gramming models are built and used, they will become
popular. Thus, the ultimate goal is to have them in regulations, such as
value at risk. Although VAR does more good than harm, its safety is
questionable in many applications. Conditional value at risk is an
improvement, but for most people and organi-zations, the nonattainment
of goals is more than proportional (i.e., convex) in the nonattainment.
1Edited and updated from Ziemba (2007).

2This section is based on Geyer and Ziemba (2008).
PART III
Seasonal Effects and Other Anomalies
Investing in the January Turn-of-the-Year Effect
with Index Futures1
Turn-of-the-year effect in US small capitalized stocks in January. Too much eggnog, perhaps?
Repeated investments in commodity trades are well modeled by the

capital growth theory with modifications for margin, daily mark-to-the-
market account variation and other practical details. An interesting example
discussed in Chapter 13 is the turn-of-the-year effect in US small
capitalized stocks in January. Figure 15.1 shows the mean excess return of
the smallest minus largest decile US stocks over the 68 years from 1926 to
1993 by month. In eleven of the months the advantage is small or negative,
however, there was a large advantage in January. The 10.36% mean
difference provides a strong advantage with high reliability since there was
a small stock advantage in 63 of the 68 years. So how can we use this
advantage to trade in the cash or futures markets?
The January effect
The January effect is the tendency of small cap stocks to outperform large
cap stocks in the month of January. Rozeff and Kinney (1976) showed that
equally weighted indices of all the stocks on the NYSE had significantly
higher returns in January than in the other eleven months during 1.04-1974.
This followed the earlier study by Wachtel (1942). Keim (n98h)
aocumented the magnitude oC the size effect by month using 1963-1979
data. He found that half the annual size premium was in January. Blume and
Stambaugh (1983) showed that, after correcting for an upward bias in mean
returns for small stocks that was common to earlier size effect studies, the
size effent was only in January. Figure 15.1 shows the mean exceas return
of the smallest minus largest decile US stocks over the 68 years from 1926
to 1993 by month. In eleven of the months the advantage is small or
negative, however, there was a large advantage in January. The 10.36%
mean difference provides a strong advantage with high reliability since
there was a small stock advantage in 63 of the 68 years.
Source: Based on data from Ibbotson Associates.
Fig. 15.1 Average excess returns of smallest minus largest decile of US stocks, 1926—93. Source:
Ibbotson Ossociates
Figure 15.2 shows the historical evtdence in the cash market arom
January 1926 tc December 1995 of the difference in Janeary befween the
lowest decile and the highest decile bn market capitalization of the NYSE
index plus AMEX aad Nasdaq stocks of similar size. Only five years out of
70 did small caps underperform in January and in most years, the small cap
outperformance is considerable. The R10th — R1st decile returns averaged
4.48% with a t = 2.83 from January 1982 to December 1995.
Additional early references studying the January small firm effect are
Banz (1981), Reinganum (1981, 1983), Roll (1983), Barry and Brown
(1984, 1985), Rogol- ski and Tinic (1986), Haugen and Lakonishok (1988).
Fama (1998) has a literature survey. Later references are Maberly and
Mavis (1991), Bhardwaj and Brooks (1992), Ziemba (1994ab), Haugen and
Jorion (1996), Cox and Johnston (1998), and Chen and Singal (2003).
Several subsequent analyses built on Keim's study and considered the
possibility that the January Effect was diminishing based on the inclusion of
later years of data, but Easterday, Sen, and Stephan (2008) also expanded
their study to include years before Keim's analysis, which allowed them to
better assess trends in the January Effect's magnitude. They included the
years from 1946-2007, performing a time series analysis according to the
three sub-periods in relation to Keim's 1963-1979 window: before, during,
and after. Over this period, they studied NYSE and AMEX firms and, from
1971 onwards, they also considered NASDAQ firms, which allowed them
to consider more small cap stocks in their analysis.
Fig. 15.2 January effect, 1926-1995. January size premium = R(10th)-R(1st). Source: Booth and
Keim (2000)
Contrary to studies based on the Keim period and later years, Easterday
et al. do not conclude that the January effect is declining. They do not find
evidence that investors are acting on the arbitrage opportunity and
internalizing it into higher prices. Instead, they find that the January Effect
continues to be robust in small firms and that, in recent years, it has not so
much diminished as returned to a level similar to the effect exhibited prior
to 1963. Easterday et al. also considered trading volume in January, which
should be higher if investors are actively arbitraging the January Effect
opportunity, but they did not find any evidence of higher trading volumes.
Haug and Hirschey (2006) also extensively analyzed the January Effect,
using both value-weighted and equal-weighted equity returns. Their
findings concur with Easterday. Particularly, observe the consistency of the
January Effect in small capitalization stock returns across time. For
instance, they find that the difference in average mean value-weighted
portfolio return is 0.40% from 1802-2004, and that this number is even
greater, 0.61%, from 1952-1986 (roughly the period Keim studied expanded
two-fold).
Haug and Hirschey explore potential explanations of the January Effect
phenomenon, ruling out biases that would more markedly affect large
capitalization stocks, such as the timing considerations of institutional
investors during portfolio rebalancing around official reporting periods.
Statistical arguments brought up by Sullivan, Timmerman, and White
(1999), among others, center around the inherent statistical problem of
testing an empirical aspect of a data set using the same data set, which
fundamentally calls into question the underlying statistical methods used in
the analysis. Additionally, two theories concerning relatively small
investors concern end-of-year tax considerations or income events, such as
year-end bonuses, which lead to new purchases in the new year. However,
both of these potential ex-planations come into question when considering
international indices under different tax regime timing and across changes
in tax laws that should have an effect.
Among other measures, Haug and Hirschey use the Tax Reform Act of
1986 to test these and other behavioral hypotheses as potential explanations
of the January Effect, but they reach contradicting conclusions using
different data, namely value- weighted and equal-weighted returns. They
ultimately conclude that each of these explanations remain potential but
unproven drivers of the still perplexing January Effect phenomenon.
To update, Dzhabarov and Ziemba (2012) calculated the Russell2000
/S&P500 futures spread by month from 1993 to 2012. As argued by Rendon
and Ziemba (2007), the January turn-of-the-month effect still exists but has
moved to December. Indeed, Figure 15.3 shows that the small cap/large cap
spread is positive in December and negative in January.
Fig. 15.3 Russell2000 - S&P500 Futures Spread Average Returns during various months, 19932011.
Source: Dzhabarov and Ziemba (2012)
Historically, the Value Line index of some 1700 equally weighted stocks
could be used as the small cap index against the S&P500 as large cap index.
For details, see Rendon and Ziemba (2007), which updates Hensel and
Ziemba (2000). The Value Line futures was a good small cap index because
of its equally weighted construction of some 1700 stocks but its volume
dried up. Hence, this effect is now implemented using the value weighted
Russell2000 index through minis which are $50/point rather than $250 for
the big contracts and have more volume than the full indices. The
Russell2000 index has more volume but the value weighting gives about a
4th decile capitalization rather than the last decile.
Commodity trading: investing in the January small cap effect in

the index futures markets
Repeated investments in commodity trades are well modeled by the capital
growth theory with modifications for margin, daily mark-to-the-market
account variation and other practical details. An interesting example is the
turn-of-the-year effect in US small capitalized stocks in January. The
evidence suggests that small stocks outperform large stocks at the turn of
the year. Yet, transactions costs, particularly bid-ask spreads and price
pressures, take away most, if not all, of the potential gains. See e.g., Stoll
and Whaley (1983) and Keim (1989) on these effects in January. However,
transaction costs to trade index futures are a tenth or less of those for the
corresponding basket of securities, and even more important, there is much
less market impact. Hence, it may be profitable to buy long positions in
small stock index futures and sell short positions in large stock index
futures. This pair of positions is known as a spread trade. The strategy must
anticipate the effect in the market place, in particular, the price impact of
buying and selling futures contracts. Stock index futures began trading in
the US in May 1982. The Value Line minus S&P500 and Russell2000
minus S&P500 spreads are two ways to measure and possibly capture any
advantage small stocks may have over large cap stocks.
Ziemba started doing this trade for the 1982/83 TOY with Vancouver
commodity broker Ross Clark. Clark and Ziemba (1987) describes these
markets and the Value Line/S&P spread. At that time the Value Line, which
had about 1700 stocks was geometrically weighted. Hence, by the
geometric-arithmetic inequality this produced a downward drift of about %
per month.2 After 1988 the index became price weighted arithmetic. Clark
and Ziemba used the following trading rule:
buy the spread on the first closing uptick, starting on December 15

and definitely by the 17th, and sell on January 15. Waiting (to enter)
until (-1) now seems to be too late: possibly finance professors and
their colleagues, as well as other students of the turn-of-the-year /
January effect who are in on the strategy, move the VL index. There
seems to be a bidding up of the March VL future price relative to
the spot price. (Clark and Ziemba, 1987, page 805)
Their idea at that time was that the January small firm effect existed and
occurred during the first two weeks of January in the cash market (as
argued by Ritter, 1988; see also comments by Ziemba, 1988), but that
futures anticipation would move the effect in the futures markets into
December. Hence, an entrance into the Value Line / S&P500 futures spread
trade in mid-December and an exit in mid January should capture the effect
if it actually existed. With data up to the 1985/1986 TOY, their trade rule
was successful for these four years. They concluded that small cap
advantage was mainly in the first half of January, with some anticipation in
the final days of December, and with a large cap advantage in the second
half of January.
The VL/S&P spread is long in small stocks and short in big stocks at the
end of the year. Each point change in the index spread is worth $500. The
spread is entered in mid December before futures anticipation bids up the
lightly traded Value Line index. On average, the December 15 to (-1) day
gain on the spread, that is the futures anticipation, was 0.57 points. By
January 15, the largest average gains are over and the risks increase. On
average, the spread dropped 0.92 points in this period with a high variance.
The projected gain from a successful trade was 0-5 points and averaged
2.85 points or $1,342.50 per spread, assuming a commission of 1.5x$55.
The average standard deviation of the VL/S&Pspread was about 3.0.
With a mean of 2.85 the following is an approximate return distribution for
the trades made by Clark and Ziemba (1987):
The optimal Kelly investment based on this return distribution is 74% of
one's fortune! See Figure 15.4a. Such high wagers are typical for profitable
situations with a small estimated probability of loss. Given the uncertainty
of the estimates involved and the volatility and margin requirements of the
exchanges, a much smaller wager is suggested.
Figure 15.4(a) displays the probability of doubling, tripling, and
tenfolding one's fortune before losing half of it, as well as the growth rates,
for various fractional Kelly strategies. At fractional strategies of 25% or
less, the probability of tenfolding one's fortune before halving it exceeds
90% with a growth rate in excess of 50% of the maximal growth rate.
Figure 15.4(b) gives the probability of reaching the distant goal of $10
million before ruining for Kelly, half Kelly and quarter Kelly strategies with
wealth levels in the range of $0-10 million. The results indicate that the
quarter Kelly strategy seems very safe with a 99% chance of achieving this
goal. The markets in later years have become much more dangerous than in
the period of this study, so an even lower Kelly fraction is suggested.
These concepts were used in a $100,000 speculative account by a
Canadian investment management company. Five VL/S&P spreads were
purchased to approximate a slightly less than 25% fractional Kelly strategy.
Watching the market carefully, these were bought on December 17, 1986 at
a spread of -22.18 which was very close to the minimum that the spread
traded at around December 15. The spread continued to gain and the
position was cashed out at -16.47 on January 14th for a gain of 5.55 points
per contract or $14,278.50 after transactions costs. Additional discussion of
many of the issues in this section appears in Clark and Ziemba (1987),
which was updated Ziemba (1994) and again by Hensel and Ziemba (2000)
and again by Rendon and Ziemba (2007) and again by Ziemba (2011). The
results in this last update plus the 2011/2012 TOY trade are summarized
below.
Fig. 15.4 Turn of the year effect
Ziemba continued trading this spread for the 14 TOY's (all winners) from
1982/83 to 1995/96 and updated the results in Ziemba (1994) and Hensel
and Ziemba (2000).
Hensel and Ziemba (2000) analyzed the January effect in the futures
markets and concluded that for the 1980's and early 1990's there was a
small cap advantage in the futures and cash markets. However, they show
that from 1994 to 1998 there was no advantage in the cash market, and that
anticipation built up in the last half of December in the futures markets. As
a consequence, for the four turns-of-the-year (TOYs) during the 1994/1998
period, the January effect only existed in the last half of December, in the
futures market. They analyzed small minus large spread trades between the
Value Line and the S&P500 futures contracts and concluded that the
January effect was exploitable in the futures markets in this period.
Table 15.1: Anticipation in the futures markets of the January turn-of-the-year small cap
advantage in December and January for 1982/83 to 2003/04 (+ =anticipation; – =
reverse; 0+ = no anticipation followed by small anticipation; +0 = small anticipation
followed by no anticipation; 0– = no anticipation followed by small reverse
anticipation; –0 = small reverse anticipation followed by no anticipation). Source:
Rendon and Ziemba (2007)
Rendon and Ziemba (2007) updated Hensel and Ziemba (2000) to
analyze the seven TOYs from 1998/1999 to 2004/2005 for the Value Line
minus S&P500 spread trade, and provided additional evidence by analyzing
a second spread trade involving the Russell2000 and the S&P500 futures
contracts. From 1998 to 2005, their analysis shows that the January effect is
still present in the futures markets in the Value Line minus S&P500 spread
trade, but that it has become increasingly risky to try to exploit it because of
the marginal liquidity of the Value Line stock index futures contract. But for
the Russell2000 minus S&P500 spread trade, the January effect has been
profitable and tradeable.
Rendon and Ziemba (2007) investigated the day by day cash and futures
returns of the spread trades: Value Line minus S&P500 and Russell2000
trading, as well as for the 12 TOYs for which there was Russell2000 futures
trading up to 2004/5.
Rendon and Ziemba (2007) computed the cash index and March futures
spreads (Value Line 500 minus S&P500, Value Line 100 minus S&P500,
and Russell2000 minus S&P500) day by day in December and January. The
500 and 100 refer to the contract size, $500 or $100 per point, respectively.
The spread between the futures difference and the cash difference
represents the futures anticipation. Table 15.1 from Rendon and Ziemba
(2007) summarizes, for each spread, the anticipation or lack of it in the
second half of December, during the turn-of-the-year (trading days -1 to
+4), the rest of the first half of January and the second half of January for
all the turn-of-the years. Some years were easy and others hard to make
profits. Figure 15.5 shows eight typical turn-of-years plots for December
and January:
(1) A typical trade for 1999/2000 showing the futures and cash spreads
(TOY) for the Value Line 500 minus S&P($500 value). The dotted
line is the futures spread and the dark line is the cash spread. In this
case you could enter at a discount in mid December. The trade gained
but you had to cash out in mid January at a discount.
(2) The 1997/98 TOM also for the VL500/S&P500 spread had all the
advantage in December then the spread declined all January. You had
to buy at a premium but you could sell for fair value. If you did not
sell at the end of December you would have lost some or all of your
gains. Traders seeing such a trend in January would likely have cashed
out.
(3) The VL100/S&P500 for 2003/04. The TOM had virtually no futures
spread discounts or premiums. Since the spread increased
continuously from December 17 traders easily made profits.
(4) The TOM for 2000/01 for the VL100/S&P500 value-adjusted for the
cash and March futures. The spread with some volatility increased
throughout December and January so the trade made profits regardless
of when it was entered and exited. There were premiums or discounts
throughout December and January.
(5) In 1994/95 for the R2000/S&P500 there were large gains in December
and the spread peaked at the end of December. Those who did not
cash out then lost some of their gains in January.
(6) In 1993/94 for the R2000/S&P500 spread you had to buy at a
premium. Then the spread fell in December and non-true believers
cashed out with losses but believers won when the spread gained in
January and traders likely could have traded out at a futures spread
discount.
(7) The TOY for 2004/05 for the R2000/S&P500 spread won in
December and you could have bought the spread at a discount. But if
you held the spread into January you would have given back some or
all of your gains or even lost money.
(8) In 2001/02 you had to buy the R2000/S&P500 spread at a premium
but the trend was up until mid January. So traders made profits.
The anticipation year by year patterns vary greatly. For the Value Line
500 minus S&P500 spread, Hensel and Ziemba (2000) reported an apparent
shift from an anticipation in the first half of January in the early years, to an
anticipation during the second half of December in the late 90's, as well as a
lack of small cap anticipation past the turn-of-the-year, in January. For this
spread trade, our results show little or no anticipation in the contract's last
two years of existence. This was, most likely, related to the scarce liquidity
of the contract in its final months.
For the Value Line 100 minus S&P500 spread there is a positive
anticipation during the second half of December to the TOY period. Further
from this point, the results are very mixed, with an apparent dominance of
positive anticipation, but with a clear trend towards no anticipation,
especially in the 2003/2004 TOY, where there was little or no anticipation
during the first half of December, and no anticipation at all in the rest of the
analyzed period. This is, again, consistent with the diminishing liquidity of
the Value Line 100 contract, which reached extreme values in the
2004/2005 TOY, with an average of less than ten contracts traded daily and
an average open interest of less than one hundred contracts.
For the Russell2000 minus S&P500 spread, positive anticipation
dominates in all the time intervals for which we divided the turn-of-the-year
period. In the three TOYs, 2002/2003, 2003/2004, and 2004/2005, negative
anticipation dominated, but its size is not significant. Since 2000, the
pattern seems to be changing from positive anticipation throughout the
period to none or insignificant negative anticipation.
Fig. 15.5 Various turn of the year trades with the VL500/S&P500, VL100/S&P500 and Rus- sell2000
/S&P500 cash (heay line) and futures (dotted line) spreads. Source: Rendon and Ziemba ( 2007)
For the Value Line 500 minus S&P500 spread, the three TOYs up to the
termination of this higher-multiplier contract in March, 2000, showed a
reversion to the original January effect. The declining liquidity of the
contract up to its date of termination is clearly a factor in this reversal. This
made the trade very risky and volatile. Large caps had the advantage in the
second half of December and January, but small caps outperformed large
caps during the (-1) to Jan 15 period. On average, the Clark and Ziemba
(1987) rule would have yielded profits over the entire 1982/83 to 1999/2000
sample period, although this result is not statistically significant. This trade
was successful in all but two of the TOYs in this sample. The trade
produced statistically significant profits in the 14 TOYs from 1982/83 to
1995/96, with a mean gain of 4.25 spread points, standard deviation of 2.81,
a t value of 1.51 and 14 of 14 winners. The results are not statistically
significant in the complete sample period because of losses in the 1996/97
and 1997/98 TOYs and high variability.
For the Value Line 100 minus S&P500 spread, the results show that there
was still a small cap advantage in the second half of December, followed by
a large cap advantage in January. On average, this trade continued to be
profitable since it produced positive profits in all but one of the 7 TOYs
studied. However the results were not statistically significant because of
high variability. Moreover, as pointed out by Hensel and Ziemba (2000), the
trade became increasingly riskier and volatile as the liquidity of the Value
Line futures contract diminished.
The Russell2000 minus S&P500 trade is very interesting. There seems to
be a clear pattern of anticipation of the January effect on this spread. The
results show a definite large cap advantage in the second half of January, as
it was in the earlier Value Line minus S&P500 studies. For the rest of the
sub periods, the results are as follows: Out of 12 TOYs, large caps had the
advantage in eight years during the (-1) to Jan 15 period. On the other hand,
small caps had an advantage in 9 out of the 12 TOYs for the December 15
to (-1) period. On average, the Clark and Ziemba (1987) rule produced
positive profits, although this result is not statistically significant. A special
case is the December 15 to (-3) period illustrated in Table 15.2. In this case,
the spread trade produces profits that are close to being statistically
significant at the 10% level. This result suggests that the Hensel and
Ziemba (2000) modification to the Clark and Ziemba (1987) rule for the
Value Line vs. S&P500 spread trade should be modified so as to unwind the
position at the (-3) day of the TOY. The 1999/00 and 2000/01 TOYs in the
Russell2000 sample were particularly strong, and introduce high variability
in the sample. Tables 15.2 and 15.3 show the mean gain, standard deviation
and t statistic for the spread trade excluding these years. The January effect
could be exploitable in the futures markets through the Russell2000 minus
S&P500 spread trade, which presents the additional advantage of greater
liquidity, when compared to the Value Line vs S&P500 spread trade. There
are periods for which the Value Line minus S&P500 trade had gains and the
Russell2000 minus S&P500 had losses. Although this is not the case in the
years after 2000, when the trades have started to look very similar, it is not
contradictory because the contracts are made up of different stocks, and the
Russell2000 contract has a much higher liquidity.
Table 15.2: Results from Russell2000/S&P500 March Futures Spread

Trades in Index Points on Various Buy/Sell Dates for the 18 Turn-of-the-
Years 1993/94 to 2010/11. Source Ziemba (2011)
Tables 15.3 and 15.4 take data from Hensel and Ziemba (2000) and
update the results for the Clark and Ziemba (1987) rule for the Value Line
500 minus S&P500 and Value Line 100 minus S&P500 spread trades. All
fourteen of the TOY trades from 1982/83 to 1995/96 showed profits.
Figures 15.5a-h show the anticipation, etc. But because of declining Value
Line volume and consulting to Morgan Stanley in New York where he
taught the Peter Muller group about the TOY and other trades, that group
became very successful and had gains in the $5 billion area (see Patterson,
2010). Ziemba stayed out of these markets until the 2009/10, 2010/11 and
2011/12 TOYs with the Russell2000 mini futures. Figure 15.6 shows these
three trades, the round dots denote the entry and the squares the exit in our
test anomaly account run by Dzhabarov and Ziemba. So the old Clark-
Ziemba trade modified seems to still work. Of course, there are lots of
implementation issues.
Table 15.3: Results from VL100/S&P500 March Futures Spread Trades in
Index Points on Various Buy/Sell Dates for the 19 Turn-of-the-Years
1P82/83 to 1999/2000. Pource: Rendon and Ziemba (2007)
Throughout the 1980s and up to the mid 1990s the data were coneistent
with the past substantial small cap advantage, see Hensel and Ziemba
(2000) who provide year by year daily data plots of the VL/S&P500
spreads, up to 1999/2000. Since then the trade has been more risky and the
volume very low , but the trade won in December in all years from 1
9(95/96 to 2004/05.
Table 15.4 updates the results to 2005 when the ValueLine was worth
$100 and the S&P500 $250 a point from 1998/99 to 2004/05 resp ect ively.
Table 15.5 shows the results when the ValueLine and S& P500 were both
worth $500 a point for 19822/883 to 1999/2000.
There were also gains in December as well as from December 15 to
.January 15. The advant age is still t here /speci al ly in the Dece mb e r
period butt the volume in recent years was so low that the trade b ecame too
risky to do. Historically, WTZ played this for 14 years from 1982/83 to
1995/96 winning each year. These 14 wianing years are shown in Table 15.
5 . Teaching the trade to Peter Muller's trading group at Morgan S t anley
and l ower volume s uggest e d retirement from this trade. The January
effect is still alive at least in the futures markets in December but the low
volume makes it too risky to trade. Spreads using the Russell2000 small cap
index will have more liquidity than the Value Line. There the trade still
works if done carefully with the gains essentially only in the December 15
to 31 period. See Table 15.6 for the 1993/94 to 2004/05 results. Figure
15.7(a) shows a typical year, 2004/05, with essentially no gains in
December and losses in January. Figure 15.7(b) shows large gains in
2000/01. Again, the dotted line is the futures spread and the dark line is the
cash spread. See Rendon and Ziemba (2007), Hensel and Ziemba (2000)
and Ziemba (2011) for more of these spread plots.
Table 15.4: Results from VL100/&P500 March Futures Spread on Various

Buy/Sell Dates for the 7 Turn-of-the-Years 1 998/99 to 2004/05. Source:
Rendon and Ziemba (2007)
Fig. 15.6 Russell2000 - S&P500 turn-of-the-year-trade spreads with our entries (dots) and exits
(squares). Source: Dzhabarov and Ziemba (2012)
Table 15.5: Results from VL500/S&P500 March futures spread trades in index points
on various buy/sell dates for the 18 turn-of-the-years 1982/83 to 1999/2000 trades.
Source: Rendon and Ziemba (2007)
Fig. 15.7 Value Line 100 minus S&P500 value-adjusted spread for cash and March futures for the
turns-of-the-years 2004/05 and 2000/01
The January effect is concerned with high stock returns in January,

especially by small cap stocks. Transactions costs, especially price
pressures, make it (difficult to take advantage of this anomaly. However,
these costs are minimal in the futures markets. This paper discusses the
results of small minus large capitalized US stocks since futures trading
began in 1982. There is some anticipation of the effect and in the futures
markets, the anomaly stiil exists but is now totally in December.
Table 15.6: Results from Russell2000 /S&P500 March Futures Spread

Trades in Index Points on Various Buy/Sell Dates for the 12 Turn-of-the-
Years 1993/24 to 2004/05. Source: Rendon and Zfemba (2007)
The January monthly effect for small and large cap stocks measured by
the Russell2000 an d S&P500 futures has b een n egative dur ing J anuary
1993-December 2010 and January 2004-D ecember 2010 . F igures 33.4ab
and 1 5.9ab show the results with the data in Tattles 15. 7 and 15 .8. The
resulfs show the historically expected very negative October in the recent
S&P500 data and in both sets of Russell2000 data. Surprisingly, the
historically strong months of Novemb er, January and February were
negative for both the small and large cap data during these two recent time
periods.
Conclusion
The January turn-of-the-year anomaly, where small cap stocks outperform
large cap stocks, still seems to exist in the futures market but has moved to
be totally in the month of December. The early futures trading in this
anomaly starting in 1982/83 was with a Value Line/S&P500 spread but the
Value Line, which is equally weighted, no longer has enough volume to be
tradable. However, the effect still exists in this market. Currently the small
cap Russell2000 index, which is value weighted, can be used in these turn-
of-the-year trades using a Russell2000 /S&P500 spread. This chapter
discusses successful actual trading in fourteen Value Line/S&P500 years
and the last three years using a Russell2000 /S&P500 futures spread. The
results from all of the years from 1982 to 2011 are also presented.
Fig. 15.8 S&P500 Futures Average Monthly Returns, 1993–2011 and 2004–2011. Source:
Dzhabarov and Ziemba (2012)
Fig. 15.9 Russell2000 Futures Average Monthly Returns, 1993–2011 and 2004–2011. Source:
Dzhabarov and Ziemba (2012)
The next chapter will discuss the January barometer. Do the returns in
January predict the returns for the rest of the year.
Table 15.7: S&P500 Futures Average Monthly Returns, 1993–2011 and

2004–2011.
Source: Dzhabarov and Ziemba
Table 15.8: Russell2000 Futures Average Monthly Returns, 1993–2011 and
2004–2011.
Source: Dzhabarov and Ziemba
1Edited from Wilmott (July 2007) with updates especially from Ziemba (2011).
2This does not affect our trade much as we had only about one month decay or about 0.5% lost. But
four very talented Wharton PhD students now well-known finance professors forgot about this and
entered the trade months before and got wiped out by Fischer Black of Goldman Sachs who
understood this and took the other side. They lost about 3% on this negative drift by buying the
contracts six months in advance. See Ritter (1996).
The January Barometer1
Simplicity of approach and execution often leads to the deepest insights,
In financial markets there are many important factors that influence prices
and returns. Hence we often use complex econometric and other models to
separate out good predictions from chance. In Ziemba and Ziemba (2007)
there are many such models such as a 30-factor model to rank all tine
Japanese first section stocks. However, in many cases, all these variables
are; aggregated by the market into only a few or even one variable. Indeed,
trend follower's focus on only one factor, namelyjust the price movements.
Since some of them are billionaires, such trends do exiet. Years ago I found
ehat I could predict the outcomes of hockey games with two variables, the
team's playing records at home and away and whether the game in question
was at home or away. Such a model head 74% accuracy against a line set by
famed Las Vegas line setter Roxy Roxborough. The reason (I am starting to
sound like an efficient markets person - can I have one of those 200k jobs
for one paper ayear and little teaahing?) is that the essential information is
built into these two variables and the rest turn out to bee noise. I did try
much more complex models but they just added noise. Many will not
believe the results of such simple models. But few complex or simple mod
els are more valuable than the .January barometer model discusred here as
well as othere later in this book.
Hensel and Ziemba (1995) found, for US equity markets using 1962-
1993 S&P500 data, that if January returns were positive, then the rest of the
year (the other eleven months) was positive more than 80% of the time; but
if January returns were negative, then the returns in the following eleven
months were more or less 5050 to be positive or negative. That is, the
negative January return signal provides noise regarding the rest of the year's
returns. Hensel and Ziemba also found that the magnitude as well as the
size of the returns were partially predictable: in positive Januarys, if the
barometer worked, the returns were higher than average and if it did not
work then the negative returns were less negative than average. The original
idea for the January barometer was from Hirsch and the guidebook he
started in 1967 is still published annually jointly with his son, see Hirsch
and Hirsch (2012).
Numerous reasons have been suggested to explain the existence of the
January barometer. Hirsch (2012) argues that important, politically market-
moving events, such as the national budgets, priorities and agendas, in the
month of January can make it a critical month. January is also a month
during which a sizeable cash increase leading to market volumes may occur
because of year-end bonuses, tax planning and portfolio restructuring.
For 1970-2004, the January barometer was successful 86% of the time
for the S&P500 index, so the conditional probability that the return in the
11 months following January would be positive given a positive January is
0.86 for this index (Stovall, 2005). This is similar to the Hensel and Ziemba
results. However, the barometer was successful in predicting only 46% of
the times when the January returns were negative for the same time period
(Stovall, 2005), again consistent with the Hensel and Ziemba noise story.
Stovall observed that portfolios consisting of the top ten industries in
January beat the S&P 74% of the time. These portfolios also yielded a total
annual return of 16.9%, versus the 7.8% of S&P, not including dividends.
Ghosh, Bhalla and Ziemba (2007) used S&P500 total monthly returns for
1926-2004. They investigated the market conditions (the time intervals and
positive versus negative January returns) for which the barometer had the
best predictive ability. They also quantified the strength of the signal using
a count-based approach and a correlation-based approach. They used a
Monte Carlo simulation to separate and value excess risk-adjusted returns
and quantified the returns from trading strategies attempting to exploit this
anomaly. They (i) determined the volatility of the S&P year-end returns, for
the years exhibiting a strong January barometer signal, (ii) compared the
volatility of these returns to the implied volatility of options on the S&P
futures, (iii) computed the profits (in units of vol%) for three investors with
different levels of transaction costs, (iv) calculated the probability of
success, as a percent of the total number of years, of a strong January
barometer signal predicting investor profits. Their analysis concluded with
updates to include 2004-2007.
Figures 16.1 and 16.2 summarize for all months the predictive ability of
positive and negative returns separately and combined for the years 1953–
2004. The September bullish forecast is statistically equivalent to that of the
January forecast but the bearish January forecasts are higher, 60% versus
43%, than in September. January foreasts are better than the other months
but the others have value as well
Fig. 16.1 Accuracy of bullish and bearish forecasts using S&P500 monthly and year-end returns.
Source: Ghosh, Bhalla and Ziemba (2007)
Fig. 16.2 Overall accuracy, sum of the positive and negative monthly barometer accuracy for
S&P500 monthly returns between 1953-2004 Source: Ghosh, Bhalla and Ziemba (2007)
Ghosh, Bhalla and Ziemba (2007) looked at four time windows: (1)
1953–2004, (2) 1953–1973, (3) 1973–1985, and (4) 1985–2004 using a
count-based approach. The January barometer had the lowest range of
variation (4%), followed by the accuracy for the December barometer
(10%), while the accuracy of the February barometer has a very high range
of variation (34%). The validity of the January barometer is supported by:
(a) it has the highest accuracy of bullish and bearish forecasts, (b) it has the
highest overall accuracy [as defined by Overall Accuracy (%) = Positive
Accuracy (%) + Negative Accuracy (%)], and (c) it has accuracy with the
lowest range of variability.
Their analysis indicates that: (a) the January barometer is more accurate
following 1985 (using the sum of the bullish+bearish accuracy values), and
(b) the barometer is clearly more accurate in making bullish forecasts
following; 1985 than before 1985.
Table 16.1: Monte Carlo Simulation-Predicted Probabilities of Positive or Negative

Year-End S&P Returns for Positive or Negative January S&P Returns. Source: Ghosh,
Bhalla and Ziemba (2007)
Fig. 16.3 Correlations between S&P 500 monthly and year-eed returns for tlie period 1953–2004.
Source: Ghosh, BOalla and Ziemba (2007)
Another approach investigated was through correlations of January

returns with those of year-end total returns during various time windows.
The highest correlations during 1953–2004 were obtained in January (62 ±
3%, with 95% confidence), as shown in Figufe 16.3. The next highest
month was April (55 ± 3%). The weak month of October has zero
correlation.
Figure 16.4a shows the histogram ol actual S&P500 monthly returns
1926-2004 and a normal distribution using; the historical mean (0.0121%
per month) and stan-dard deviation of returns (0.0705 per month). The
eesults show fat tails and to get a normal distribution to fit, one needs to
truncate the histogram to one with a mean of 0.0105% per mgnth and a
standard deviation of 0.0549 as shown in Figure 16.4b. This truncated
distribution (deleting returns above 0.25% and below -0.25%) was used in
the Monte Carlo simulations. The results indicate that the correlation
coefficient between the January S&P500 returns and the year end returns
was 0.620.03 which was more than double the random process based on sill
monthly returns.
The simulation predicts the probability oC positive S&P500 year-end
returns for positive January S&P returns and negative S&P year-end returns
for negative January S&P eeturns, ae shown in Table 1. Positivn January
S&P returns have a stronger prediction capability of year-end S&P returns
than negative January S&P500 returns. The simulation, when the January
S&P return is positive, (concludes with 78.6% probability, that the year-end
return will be positive; versus an actual 90.63% accuracy. However, when
the January S&P return is negative, the simulation concludes with 37.5%
probability the year-end return wrll be negative.
Fig. 16.4 Histogram of Historical S&P Monthly Returns from 1926-2004 and a Normal Distribution
of Returns Using the Historical Mean and Standard Deviation of Returns. Source: Ghosh, Bhalla and
Ziemba (2007)
In 2004, the S&P500 gained 2.00% in January and 6.86% in the

remaining 11 months for a total gain of 8.99%. In 2005 the S&P500 lost
-2.53% in January and gained 5.67% in the next 11 months for a total grain
of only 3.00%. And in 2006 it was up 2.54% in January, 10.80% in the next
11 months and 13.30% for the year. In 2007 January the S&P500 gained
1.41% and the rest of the year through 24 October the S&P500 was up a
further 5.40% for a total return to date of 6.88%.
Ghosh, Bhalla and Ziemba (2007) also investigated a simple options
strategy. They observed implied volatilities of S&P500 options during
1988-2004. The strategy is to go long the S&P500 calls and short a
replicating portfolio, attempting to net out the price differential in units of
volatility. For institutional investors who can take advantage of a 1%
difference in volatilities has a 70.59% chance of returning profits at the end
of December. This simple anomaly can be used in various more
sophisticated option and other equity strategies.
January barometer research update

The January barometer evidence is that if January is positive then the rest of
the year (the following 11 months) is positive about 85% of the time and if
January is negative then the rest of the year is up about 50% of the time and
basically produces noise. Moreover positive Januarys have had much higher
rest of the year mean returns than negative Januarys. See Table 16.2 for the
S&P500 for the years 2004–2012 and Figure 16.6 which shows the results
from 1940–2011 and 1994–2011.
Table 16.2: The Last Nine Years of the January Barometer
Move this Around

Over the years I have written papers at Frank Russell Company with Chris
Hensel (2000) and later with two MIT students on the January barometer
whose results are discussed above. The main results with more than 50
years data for the US (and other countries) are:
If January is negative, then the rest of the year is negative or positive
about 50% of the time and if the returns are positive they are not high but if
the returns are negative they are large negative.
The January barometer gave a big clue in the first week of 2008.
• The S&P500 fell 6.1% in January 2008 so we expected February and
onwards to be rocky. There were losses in five straight months from
November 2007 to March 2008. And the rest of the year was very
negative, with 2008 losing 38.49%.
• If January is positive, then the probability that the rest of the year is
positive is about 85% and the positive returns are high and the negative
returns are not very negative.
A related study is to look at the first five days at the turn of the year (-1
to
+4).
• But it is known that if these days are negative it is a very bad signal for
future returns.
• These were very negative.
• The Stock Traders Almanac does this but using +1 to +5.
• In 2008 these 5 days were -5.3% the worst in all 59 years from 1950–
2008.
Using the STA 2009 data, Constantine Dzhabarov helped me reach the
following results.
Fig. 16.5 First-Five-Days-in-January vs. Year Return. S&P500 Index (Cash). 1950-2007
We have the following rest of the year versus the 5-day returns for all
years, for years with negative 5-day returns and for years with positive 5-
day returns [with t-statistics in ()]
For negative 5-days, we have
The mean negative 5-day returns were -0.0218 with a standard deviation
of 0.01091.
For positive 5-day returns, we have
Positive 5-day returns averaged 0.01714 with a standard deviation of

0.01444.
The separation of negative and positive 5-day returns does provide a
signal. Of course, with such low R2s, the results are suggestive but not
statistically significant.
The January barometer research

Historically, returns in January have been a valuable signal for the returns in
the following eleven months that year. If stocks have positive returns in
January, then it is likely that the market as a whole will rise in that year.
Hirsch (1986), who first mentioned this in 1972, has called this the January
barometer. In the yearly updated Stock Trader's Almanac by Jeffrey A.
Hirsch and Yale Hirsch (2012), they define it as the full year rather than the
last eleven months. We look at this both ways, full year and last eleven
months, for the US S&P500 in Figure 16.7. The supposition is that:
if the market rises in January, then it will rise for the year as a
whole; but if it falls in January, then there will be a decline or a flat
market that year.
Figure 16.6 updates Hensel and Ziemba (1995a) and Ziemba (1994)
which had the results for the 54 years 1940-93. There are 72 years in the
total sample with an 18 year update to the end of December 2011.
For the 72 years, when the return in January was positive, the rest of the
year was up 84.4% of the time. This compares with 69.4% of all the years
that the whole year was up.
When the return in January was negative which was 27 of the 72 years,
the rest of the year was down 48.1% of the time. Thus even in years when
January is down, the whole year is about equally likely to be up or down.
This 48.1% is significantly less than the 72.2% of all the years that the rest
of the year went up. Figure 16.6 also shows the full year return for the four
cases with arithmetic and geometric mean returns. We conclude that the
January barometer does add value and is useful in various ways. Negative
Januarys like 2008 had good predictive value. But the measure is not
infallible. For example, 2010 had positive 11 month and 12 month returns
despite a negative January. But as in other cases of negative January but
positive 11 and 12 month returns, those returns are, on average, small.
In the 18 year update (1994-2011), the results as seen in Figure 16.6 are
similar with the January up ROY up 72.7% (8 of 11) of the time.
Figure 16.7(a) shows the cumulative rest of year returns for positive
January, negative January and buy and hold. Historical buy and hold beats
positive January and has the highest final wealth with negative Januarys
producing almost no gains at all. Buy and Hold had returns that were high
except the 2007-9 drop in the S&P500 led to the positive January
dominating. Figure 16.7(b) has the full year results.
The equations for Figures 16.8(a) and 16.8(b) are:
Bronson (2011) reminds us that the January barometer has had six false
positives since 1940, where January was up but the rest of the year was
negative. In 1947 there were enough dividends and January returns to
overturn this loss for the whole year. So that leaves the following five
January positive net return of the year negative returns (not including
dividends) as in Table 16.3.
There have also been 14 false negatives since 1940 to 2010 where
January was negative but the rest of the year is positive. We differ from
Bronson by simply saying that if January is negative, the rest of the year is
noise. So Bronson argues that the January barometer has failed to signal the
direction of the stock market 19 of the past 71 years up to 2010, some 27%
of the time. Agreeing with us, the stock market was up 73% of the time (52
of 71 years). But Bronson argues that the barometer is getting less accurate
recently. Indeed 12 of the 19 failures (60%) have occurred in the 32 years
since 1978. Figure 16.9 shows Bronson's graph relating the rest of the year
percent change as a function of January's percent change. His regression
suggests that the rest of the year percent change equals 6% plus 80% of
January's percent change. Compare this with our regression above with a
minuscule 0.5% R2 and a rest of the year return of 12.6% minus 0.27 times
January's return . His low 7% R2 exceeds ours. He concludes that January
2011's gain of 2.3% yields a forecast of a rest of 2011 gain of 7.8%. This
gain is quite close to those we hear from the TV forecasters.
Fig. 16.6 January barometer results, 1940-2011 and 1994–2011
Hirsch and Hirsch (2012) and Ziemba (2010) discuss this first five day of
January predictor with data from 1950-2010. The last 37 positive first five
days were followed by full year gains 32 times (86.5%) and a mean gain of
14.0% for the 37 years. The 23 negative first five days had 12 positive
(52.2%) years and 11 negative years. The Stock Traders Almanac uses days
+1 to +5. I prefer to use -1 to +4 as that's more the turn-of-the-month. The
full month January barometer has been an even better predictor except in
2009 so we have focused on that but the five days is important to look at.
To conclude, the 2008 signal was the strongest at -5.3%, the negative first
five days were the worst ever, and the negative January led to the very
devastating 2008 with a yearly loss of -38.5% for the S&P500. The first
five days of 2012 was a positive period.
Fig. 16.7 Positive and Negative January and B&H Cumulative Returns. S&P500 Index (Cash), 1940–
2011
Fig. 16.8 January Return (x-axis) vs. Rest of Year Return (y-axis). S&P500 Index (Cash), 1940-2011
Table 16.3: Returns for positive January, negative rest of year, %. Source: Bronson
(2011)
Fig. 16.9 Regression analysis of January return with rest of year. Source: Bronson (2011)
In 2010 January started out very positive but, by the end of the month,
the full month's return was -3.70%. So this suggests a weak 2010 market
but not necessarily losses. Indeed the S&P500 was up 12.8% in 2010. There
were all sorts of reasons suggesting a weak and volatile 2010 stock market.
Hirsch (1986, 2012) also discusses the barometer which Yale Hirsch first
discovered in 1986. Figure 33.4ab and 15.9ab show the monthly returns for
1993-2009 and 2004-2009 of the S&P500 and Russell2000 futures.
Observe the negative January and February.
Figures 16.8(a) and 16.8(b) show the regression of rest of the year returns
versus January returns, January positive and January negative with t's in ().
The last nine years of the January barometer are studied in Table 16.2.
Other Studies
The results we have found are supplemented with other studies as follows.
Brown and Luo (2006) consider the performance of the January
barometer (JanB) in the US from 1941-2003 and find it has predictive
ability. More recently, Stivers, Sun, and Sun (2009) find, using the simple
spread approach, the power of the JanB in US indices has declined since it
was published in the early 1970s but that it remains a useful market timing
technique in the 1975-2006 period. Addi-tionally, Sturm (2009) shows the
JanB is particularly powerful in the first year of the presidential cycle.
Cooper, McConnell, and Ovtchinnikov (2006) focus on the 1940-2006
period and consider the robustness of results using NYSE data dating back
to 1825. They call the effect the “other January effect” but we prefer
January barometer. In addition to testing the JanB with the full market
index, Cooper et al. also find it has predictive value for both small and large
stocks and value and growth stocks. The effect persists after adjustment for
business cycle and macroeconomic variables, investor sentiment, and the
presidential cycle.
Cooper et al. found that over the previous 147 years, the spread between
the 11-month return following positive versus negative January's was
7.76%, and other papers have reported spreads of 10%+. Though consensus
exists around this conclusion that January returns have a predictive power,
the consensus dissipates at the crucial point: can you profit from it?
How to Trade the January Barometer (JanB)

The apparent predictive power of the JanB leads us to if and how investors
should trade to profit by it. The following strategies have been analyzed as
ways to use the JanB to outperform a passive buy-and-hold strategy (see
also Figure 16.7):
(1) Standard JanB Strategy: Stay out of the market in January and go long
or short for the remainder of the year depending on if the January
return is positive or negative.
(2) JE+JanB Strategy: Go long in January based on the original January
Effect (because market returns in January are positive on average)
and, for the remainder of the year, follow the Standard JanB Strategy.
(3) JE+JanB T-bill Strategy: Go long in January based on the original
January Effect. If the January return is positive, go long. If the January
return is negative, invest in t-bills.
Regarding the Standard JanB Strategy, Marshall et al. and Cooper et al.
came to a similar conclusion that a passive buy-and-hold strategy beats the
Standard JanB Strategy. This result is unsurprising given that this Strategy
calls for staying out of the market in January to wait for the market signal
before investing from February through December. By skipping January,
the investor misses the excess returns experienced in January as
documented as the original January Effect.
Therefore the two subsequent strategy alternatives both assume the
investor goes long in January in order to benefit from the January Effect.
The JE + JanB Strategy also underperforms a simple buy-and-hold
strategy. Cooper et al. report that, while it was much lower than the return
after a positive January, the average yearly return after negative January's
was also positive at 5.71%, so a short strategy in these periods earns less
than the T-bills or the buyand- hold strategy. Cooper et al. also point out the
substantial losses from tail risk in very good years occurring after negative
Januarys, when shorting the market would have been ‘disastrous’ for the
investor.
Both Cooper et al. and Marshall et al. find the JE+JanB T-bill strategy to
be the best of the three above alternatives, but they disagree on whether this
strategy is superior to the buy-and-hold baseline.
For example, during the period from 1940 to 2006, Cooper et al. report
average annual buy-and-hold returns of 11.94% compared to a yearly
average of 12.79% for the JE + JanB T-bill Strategy. Cooper et al. conclude
that following this strategy is of value to investors based on the past data.
Marshall et al. reach the opposite conclusion through the same data,
finding that the JanB cannot be used by investors to outperform a passive
long strategy. For the period from 1940–2007, they find average yearly buy-
and-hold returns of 12.68% and JE + JanB T-bill strategy returns of
13.09%. Marshall et al. say the discrepancies between the two conclusions
are based on dissimilar statistical significance and risk assessment
calculations as well as differing opinions of the economic significance of
<0.05% difference in annual return, which excludes transaction costs and
uses simple spreads that they find biased away from investors' trading
realities.
The only way the JE + JanB T-bill Strategy differs from the buy-and-hold
strategy is that, following Januarys with negative returns, the investor opts
for T- bills instead of equity. The former strategy does not outperform the
buy-and-hold strategy because, following Januarys with negative returns,
average 11-month T-bill returns are only marginally larger than average 11-
month equity returns.
The International January Barometer

Hensel and Ziemba (1995b), Easton and Pinder (2007), and Stivers, Sun,
and Sun (2009) address the performance of the JanB in international
markets. Hensel and Ziemba find similar results in Switzerland and Europe
and global as the US. Namely, about 85% positive years and rest of years
following positive Januarys and noise about 50-50 following negative
Januarys. The mean returns have the same basic behavior, being more
favorable for positive than for negative Januarys. So positive Januarys do
seem to have positive predictive power.
Bohl and Salm (2010) study the predictive power of stock market returns
in January for the rest of the year for 19 countries. They find that the
barometer works well in the US, as we know, and in Norway and
Switzerland. But it did not predict well in the other 16 countries which
included Japan, France, Spain and Germany.
The data periods vary by country but are long, for example, Australia
1903-2007, Austria 1970-2007, Belgium 1951-2007, Canada 1936-2007,
and France 1896-2007. In many cases it is the high sigma leading to too low
t's which caused the significant non-predictability. But in some cases, the
signal is actually going the wrong way in their regression model. So care is
needed in these various countries to use the barometer for added value.
1Edited from Wilmjtt, January 2008.

Sell in May and Go Away and the Effect of the Fed
September and October have historically had low stock market returns with
many serious declines or crashes occurring in October. Also the months of
November to February have historically had higher than average returns;
see, for example, Gultekin and Gultekin (1983) and various papers in
Ziemba (2012a). This suggests the strategy to avoid the bad months and be
in cash then and only be long the stock market in the good months. Sell in
May and go away, which is sometimes called the Halloween Effect, is one
such strategy that is often discussed in the financial press. Figures 17.1 and
17.2 show this strategy using the rule sell at the first trading day in May and
buy on the 6th trading day before the end of October, for the S&P500 and
Russell2000 futures indices for the years 1993-2011, respectively. This rule
did indeed beat a buy and hold strategy. Tables 17.1 and 17.2 have the SIM
versus buy and hold data from Figures 17.1 and 17.2
Fig. 17.1 S&P500 Futures Sell in May (SIM) and B&H Cumulative Returns Comparison. 1993–2011.
(Entry at Close on 6th Day before End of October. Exit 1st Day of May.) Source: Dzhabarov and
Ziemba, (2012)
Fig. 17.2 Russell2000 Futures Sell in May (SIM) and B&H Cumulative Returns Comparison. 1993-
2011. (Entry at Close on 6th Day before End of October. Exit 1st Day of May.) Source: Dzhabarov
and Ziemba (2012)
For the S&P500 a buy and hold strategy turns $1 on February 4, 1993,
into $1.91 on December 31, 2011; whereas, sell in May and move into cash,
counting interest (Fed funds effective monthly rate for sell in May) and
dividends for the buy and hold, had a final wealth of $4.03, some 111.6%
higher. For the Russell2000 , the final wealths were $1.83 and $5.35,
respectively, some 192.3% higher. So the Sell in May and Go Away strategy
has been working to produce much higher final wealth with lower risk and,
as with most anomalies, the small cap results are the best.
Doeswijk (2005) offers a new hypotheses after reviewing two existing
explanatory hypotheses. Bouman and Jacobsen (2002) confirm the empirical
and historical basis of the maxim, finding that the ‘Sell in May‘ effect holds
in 36 of the 37 countries included in their analysis. They consider vacation
timing as a potential cause of the ‘Sell in May’ effect, suggesting the timing
of summer vacations may cause temporal variation in appetites for risk
aversion. However, they find evidence of the ‘Sell in May’ effect in their
subset of Southern hemisphere countries, which under their hypothesis
would be expected to have a different seasonal pattern.
Another hypothesized link between seasonality and stock returns is the
Sea¬sonal Affective Disorder (SAD), which was studied in Kamstra,
Kramer, and Levi (2003) and Garret, Kamstra, and Kramer (2004). SAD is a
disorder in which the shorter, relatively sunless days of fall and winter cause
depression, which some recent research links to an unwillingness to take
risk. Kamstra (2003) concludes that the SAD explanation does not lead to a
profitable trading strategy because the risk premium varies with the seasonal
effects. Like the vacation timing hypothesis, Doeswijk finds the SAD
hypothesis insufficient because SAD is known to start as early as September
so the historically high November returns cannot be explained.
Doeswijk (2005) offers a new hypothesis to explain the ‘Sell in May’
effect. He posits that, in the fourth quarter of each year, investors are overly
optimistic about the upcoming year. This excessive optimism leads to
attractive initial returns followed by a renewed realism that readjusts
expectations. Unlike the SAD hypothesis, which suggests a varying risk
premium, the Optimism Cycle hypothesis reflects a constant risk premium
but a varying perception of the economic outlook. In order to test this
hypothesis, Doeswijk ran three analyses: 1) the global zero-investment
seasonal sector-rotation strategy 2) the seasonality of earnings growth
revisions and 3) the initial returns of IPOs as a proxy for investor optimism.
According to the Optimism Cycle, investors are over-optimistic at the end
and beginning of the year. If this hypothesis is correct, a winning investment
strategy is going long in cyclical stocks and short in defensive stocks during
the winter period from November through April (‘winter’) and following the
opposite strategy from May through October (‘summer’). (These stock
groups are chosen for their relative exposure to the general economy, with
cyclical stocks having a high exposure and defensive stocks a low exposure.)
To test this strategy, Doeswijk uses the MSCI World index of global stock
returns from 1970-2003 and tests the data as a whole, in two 17-year sub-
periods, using several variations on timing of the winter period, and various
sector definitions. The study runs regressions using monthly market
capitalization-weighted price return indices and their monthly log returns.
Doeswijk finds that, on average during the study period, winter returns are
a significant 7.6% higher than summer returns and the strategy works in
65% of the years. On a monthly basis, average performance of the global
zero-investment strategy is 0.56%, which is significant at the 1% confidence
level. Using further regression analysis techniques, Doeswijk also isolates
the market timing effects from the seasonality and finds that seasonality
alone accounts for approximately half of the excess returns.
Like the Optimism Cycle strategy, both other analyses in the Doeswijk
study support the Optimism Cycle hypothesis. Doeswijk finds that expected
earnings growth rates follow a seasonal cycle and that these changes have an
effect on stock performance. The third analysis uses initial IPO returns,
which show a remarkable seasonality, as a proxy for investor confidence.
Using this investor confidence proxy as an independent variable, the
regression result for remaining excess return is not statistically significant,
which supports the Optimism Cycle hypothesis.
Along with the three supporting analyses, Doeswijk explains a qualitative
argument in favor of his Optimism Cycle hypothesis. He argues that, since
this phenomenon is one based on an aspect of human psychology, it tricks
investors into repeating the same biases every year. Importantly, this cycle of
optimism and pessimism is not generally accepted, which Doeswijk argues
allows for investors who understand it to profit from it as a ‘free lunch’ until
it is more widely accepted and the arbitrage opportunity is absorbed into the
market.
The effect of the Fed meetings

The Fed has a huge influence on the stock market: adapted from Housel
(2012) Former Federal chairman Alan Greenspan wrote in his 2007 memoir:
People would stop me on the street and thank me for their 401(k); I'd
be cordial in response, though I admit I occasionally felt tempted to
say, ‘Madam, I had nothing to do with your 401(k).’ It's a very
uncomfortable feeling to be complimented for something you didn't
do.
This is nave - his policies created a stock market bubble which burst and
then created a bubble in real estate which also crashed
A study by David Lucca and Emanuel Moench (2011) of the Fed shows
the in-fluence of the Fed Open Market meetings since they were publicized
and announced in 1994 through mid 2011. Since then the S&P500 has risen
from 450 to 1300. More than 80% of the equity premium on US stocks was
earned over the 24 hours preceding scheduled FMOC meetings. And
virtually all in the 3 day window around these meetings, see Figure 17.3
Fig. 17.3 Average cumulative returns on S&P5000 on days before, on and after ROMC
announcements
The returns in Figure 17.4 are uncertain with a positive bias but
considerable variation. So the Fed effect does seem to work but have risk in
it in a particular meeting. Given the confidence intervals, one sees the
returns are not straight up. Also some tests Constantine Dzbarov ran for us
indicate that the effect is less pronounced in recent data ending September
30, 2012.
Matt Koppenheffer of Motley Fool used market data going back to 1994.
He randomly removed 136 days (eight per year for 17 years, or one for every
FOMC meeting) He ran the simulation several hundred times, removing
different sets of random days. The difference was trivial. Nothing came
within a third of the skew caused by removing the days shortly before
FOMC meetings. Over the long haul, stocks are driven by fundamentals, in
the short term they are impacted by headlines including waiting for the
FMOC announcements which add to volatility.
Note: The sample period is 1994 to 2011
Fig. 17.4 The S&P500 index with and without the 24 hour pre-FOMC returns
Table 17.1: Data by month for Sell in May and Go Away versus Buy and
Hold for the S&P500,1993-2011/ Source: Dzhabarov and Ziemba (2012)
Table 17.2: Data by month for Sell in May and Go Away versus Buy and
Hold for the Rus- sell2000 Futures, 1993-2011. Source: Dzhabarov and
Ziemba (2012)
60-40 Pension Fund Mixes and Presidential Party
Effects1
Pension funds frequently suggest a 60-40 stock-bond mix. This tends to

lower risk as during stock market declines bonds tend to rise and when both
stocks and bonds rise, the correlation is sufficiently low to diversify.
However, US investment returns have been presidential party dependent;
and returns in the last two years of sill administrations have exceeded those
in the first two years. This chapter studies election cycles, the effect of
presidential party affiliations in the White House arid the wisdom of the 60-
40 stock-bond and other long term investment rules from 1937 and 1942. I
begin with the early literature. The strategies small cap stocks with
Democrats and intermediate bonds or large cap seocks with Republicans
yields final wealth about six; times the large cap index, 50% more than small
caps and more than twenty times the 60-40 mix that is frequently
recommended because of its diversification effects since 19-42. The 60-410
mix does. have lower standard deviation risk than the other strategies but the
(cost is much lower mean and total return. These results aire very similar to
the earlier results of Hensel and Ziemba (1995, 20091 and include out of
sample results for the two Clinton, two Bush and one Obama terms. For
comparison, the earlier to 1997 results as well as the full data set results to
2011 are shown.
Herbst and Slinkman 11984), using data from 1926–1977, found a 48-
month political/economic cycle during which returns were higher than
overage; Ohis cycle peaked in November of presidential election years.
Riley and Luksetich (1980) and Hobbs and Riley (1984) showed that, from
1900–1980, positive short-term effects followed Republican victories and
negative returns followed Democratic wins. Huang (1985), using data from
1832–1979 and for various subperiods, found higher stock returns in the last
two years of political terms than in the first two. This finding is consistent
with the hypothesis that political reelection campaigns create policies that
stimulate the economy and are positive for stock returns. These studies
concerned large cap stocks.
Stoval (1992) and Hensel and Ziemba (1995, 2000) documented the
Presidential Election Cycle effect, which exhibited that stock markets
generally had low returns during the first two years after an US Presidential
election and high returns during the last two years. Other subsequent studies
have documented the economically and statistically significant difference in
equity returns during the first and second half of Presidential terms for
Republican and Democratic administrations. Some studies use more detailed
models. Wong and McAleer (2008) examine the cyclical effect that
Presidential elections have on equity markets using a spectral analysis
technique and an exponential GARCH Intervention model to correct for
time-dependence and heteroskedasticity. They consider the period from
January 1965 to December 2003 using weekly data with dummy variables to
designate the year of the term and the President's party.
Wong and McAleer find a cyclical trend that mirrors the four-year election
cycle with a modified cycle of between 40-53 months. They find that stock
prices generally fall until a low-point during the second-year of a Presidency
and then rise during the remainder, peaking in the third or fourth year.
During the 2009-2012 Obama Democratic administration, the low was in
March 2009 in his first year and the market has doubled since then to the end
of April 2012. Wong and McAleer also find this Presidential Election Cycle
effect to be notably more significant under Republican administrations,
leading them to posit that the Republican Party may engage in policy
manipulation in order to benefit during elections relatively more than their
Democratic counterparts. For instance, the second-year and third-year effect
estimates are not significant for Democratic administrations.
Election cycles
Wong and McAleer explain the Presidential Election Cycle as follows.
During the first year of a Presidency, voters are on average optimistic, and
Presidents are likely to put their most divergent - and expensive - new
policies in place, because they have the mandate of the voters and re-election
time is furthest away. These early measures are relatively disadvantageous to
business profits - and stock prices - because they usually involve higher
taxes and spending and possibly new regulations. Then, during the second
year of a term, Presidents begin to alter their policies to ones that are less
drastic and more voter-friendly.
The Presidential Election Cycle effect persists when looked at by
President and by party. For instance, the only two Presidents who did not
exhibit the cycle effect were Ronald Reagan and Bill Clinton during their
second terms, during which they would not have re-election incentives like
first-term Presidents. Empirical results also find that Republicans who were
subsequently re-elected had a positive effect during the second-year of their
term instead of the negative effect expected by the Presidential Election
Cycle hypothesis. This suggests these Republicans may have used
government policies to their favor to win re-election and should be useful for
incumbent Presidents to consider in their electoral strategy. This last
conclusion, however, does not follow from the conflicting observation that
bull markets have tended to coincide with sub-periods under Democratic
administrations. Wong and McAleer conclude that this anomaly was present
during most of the last forty years and is likely still present in the market.
Hensel and Ziemba (1995, 2000) investigated several questions
concerning U.S. stock, bond and cash returns from 1928 to 1997, such as:
Do small and large capi-talization stock returns differ between Democratic
and Republican administrations? Do corporate bond, intermediate and long-
term government bonds and Treasury bill returns differ between the two
administrations? Do the returns of various assets in the second half of each
four-year administration differ from those in the first half? Were Clinton's
administrations analogous to past Democratic administrations? I also discuss
here the terms of George W. Bush and Barack Obama to update to the end of
2010.
Their results demonstrate a significant small cap effect during Democratic
pres-idencies. Small cap stocks (the bottom 20% by capitalization) had
higher returns during Democratic than Republican administrations. There
has also been a small cap minus large cap S&P advantage outside the month
of January for the Democrats. The higher returns with Democrats for small
cap stocks are the result of gains rather than losses in the April-December
period. The turn-of-the-year small firm effect, in which small cap stock
returns significantly exceed those for large cap stocks in January, under both
Republican and Democratic administrations, occurred during these 70 years.
This advantage was slightly higher for Democrats, but the difference is not
significant. Large cap stocks had statistically identical returns under both
administrations. For both Democratic and Republican administrations, small
and large cap stock returns were significantly higher during the last two
years of the presidential term than during the first two years. Moreover, bond
and cash returns were significantly higher during Republican compared with
Democratic ad-ministrations. The results also confirm and extend previous
findings that equity returns have been higher in the second half compared
with the first half of presidential terms. This finding is documented for small
and large cap stocks during both Democratic and Republican
administrations. Finally, two simple investment strategies based on these
findings yielded superior portfolio performance compared with common
alternatives during the sample period. The results cast doubt on the long run
wisdom of the common 60/40 stock-bond strategy since all 100% equity
strategies investigated had much higher wealth at the end of the sample
period. Indeed the 1942-1997 returns were twenty-four times higher with the
strategy small caps with Democrats and large caps with Republicans than the
60/40 mix and the updated 1998-2010 returns shown in Table 18.7 show
similar outperformance in an update to the end of 2010.
Table 18.1 shows that both small and large cap stocks had lower mean
returns in the 13 months following an election. Figure 18.2 shows the
specific months following the election for large (S&P500) and small cap
(bottom 20%) stocks.
Table 18.1: Annual average equity returns minus annualized monthly averages for
presidential election months and the subsequent 13 months, 1928-1997, 1998-2010 and
1928-2010*. Source: Updated from Hensel and Ziemba (2000)
The 1928-97 period encompassed 18 presidential elections with an update

to 2010 with three more elections. The end of 1997 included the first year of
Clinton's second term. There were 33 years of Republican and 37 years of
Democratic administrations during this period. The update to the end of
2010 covers the last three years of Clinton's second term plus two George W
Bush terms plus the first two years of Barack Obama's administration,
namely 1998 to 2010, a period where small cap stocks outperformed large
cap stocks. Tables 18.2 and 18.3 list and compare the first year, first two
year, last two year and whole term mean returns under Democratic and
Republican administrations from January 1929 to December 1997 and for
January 1937 to December 1997, a period that excludes one term for each
party during the 1929 crash, subsequent depression and recovery period plus
the update to 2010. Each term is considered separately, so two-term
presidents have double entries. The t-values shown in Table 18.2 test the
hypothesis that, during the 1928-97 period, returns did not differ between
Democratic and Republican administrations.
From 1929 to 1997, the mean returns for small stocks were statistically
higher during the Democratic presidential terms than during the Republican
terms. The data confirm the advantage of small cap over large cap stocks
under Democratic administrations. Small cap stocks returned, on average,
20.15% a year under Democrats compared with 1.94% under Republicans
for the 1929-1997 period. This difference, 18.21%, was highly significant.
The first year return differences for this period were even higher, averaging
33.51%.
The right hand panel of Table 18.2 presents the return results after
eliminating the 1929 crash, the Depression and the subsequent period of
stock price volatility. Removing these eight years (1929-1936) from the
study eliminates one Democratic and one Republican administration from
the data. The small stock advantage under Democrats was still large (an
average of 7.559%) per four-year-term) but was no longer statistically
significant. The large cap (S&P500) returns during Democratic rule were
statistically indistinguishable from the returns under Republican
administrations. Table 18.3 updates to 2010.
Table 18.2: Average annual returns for the first and second years and the four years of
Democratic and Republican presidencies* to 1997. Source: Hensel and Ziemba (2000)
*In this and subsequent tables, statistically significant differences at the 5% level (2-tail)
are shown in bold.
For Democratic and Republican administrations, the mean small and large
cap stock returns were much higher in the last two years compared with the
first two years of presidential terms for both of the time periods presented in
Table 18.2. For example, small cap stocks returned 24.65% during the last
two years compared with 15.90% during the first two years for Democrats
and 10.18% compared with -6.29% for Republicans from 1929 to 1992.
Returns on large cap stocks increased to 17.40 from 8.09% for Democrats
and to 9.06% from 3.77% for Republicans for the same period. This result is
consistent with the hypothesis that incumbents embark on favorable
economic policies in the last two years of their administrations to increase
their reelection chances and that the financial markets view these policies
favorably.
Table 18.3: Average annual returns for the first year and four years of Democratic and
Republican presidencies*; update to 2010
The advantage of small stocks over large stocks under Democratic

administrations was not a manifestation of the January small stock effect.
Instead Tables 18.4 and 18.5 and Figure 18.1 show the relative advantage of
small over large cap stocks under Democrats compared with that under
Republicans was attributable to having fewer small stock losses, as well as
higher mean small stock returns, in the April-December period. Under
Democrats, the mean returns were positive in each of these months, except
October, and the small minus large differential was positive during 10 of the
12 months; under Republicans, the small minus large differential was
negative during 9 of the 12 months.
Fig. 18.1 Cap Size Effects and Presidential Party, Democratic (Small minus Large) minus Republican
(Small minus Large)
The small cap advantage also occurred in the months following Democrat
Clinton's first election. From November 1992 to December 1993 the small
cap index rose 36.9% versus 14.9% for the S&P500. This domination
continued until the second election. Small caps returned 1.58% pier- month
versus 1.31% per month for the S&P500 from November 1992 to October
1996. However, large cap S&P500 returns began exceeding small cap
returns in 1994 and this continued through 1997. The January 1994 to
December 1996 returns were small cap 1.36% per month versus 1.50% per
month for the S&P500. From November 1996 to December 1997 small caps
returned 1.81% per month and the S&P500 2.44% per month. There was a
phenomenal growth in S&P500 index funds and much foreign investment in
large cap stocks during this period. While small caps had very large returns,
those of the S&P500 were even higher.
Table 18.4: Average Monthly Small- and Large-Cap Stock Returns during Democratic
and Republican Presidencies, January 1929 - December 1997. Source: Hensel and
Ziemba (2000)
How does inflation vary with political regimes? The results for the 1929-
97 period, using the Ibbotson inflation index, indicate that inflation was
significantly higher under Democrats, but this difference was contained in
the 1929-36 period. Excluding this early period, inflation was slightly
higher, on average, under Democrats but not statistically different from
inflation under Republican. Inflation rates differed across the years of the
presidential terms. For example, for the 193797 p eriod, in the first year of
the presidential term, inflation under the Democrats was significantly lower
than it was under the Republicans. An analysis of the first and second two
years of administrations during this same period indicated that inflation was
higher under Democrats but the difference was not statistically significant.
US bond returns after presidential elections

The bond data are also from Ibbotson Associates and consist of monthly,
continuously compounded total returns for long term corporate bonds, long
term (20-year) government bonds, intermediate (5-year) government bonds,
and cash (90-day T- bills).
Table 18.5: Average Monthly Small- and Large-Cap Stock Returns during
Democratic and Republican Presidencies, January 1929 - December 2010
Figure 18.2(a) illustrates average return differences for bonds during
election months and the subsequent 13 months (1929-97) minus each
month's 1928-1997 average return. Figure 18.2(b) updates this to 2010.
Corporate, long term government, and intermediate government bond returns
were all higher than the monthly average in the year following an election
only in May, October, and November in the 1928-1997 period. Both
government bonds also exceeded the average in some other months. The
update only has three elections and the monthly pattern is different than it
was in the past.
As Table 18.6 indicates, the performance of fixed income investments
differed significantly between Democratic and Republican administrations.
All fixed income and cash returns were significantly higher during
Republican than during Democratic administrations during the two study
periods. The high significance of the cash difference stems from the low
standard deviation over terms. The performance of fixed income investments
differed very little between the first two years and the last two years of
presidential terms.
Fig. 18.2 Stock monthly return differences: presidential election months and the subsequent 13 months
minus monthly averages
The distribution of Democratic and Republican administrations during the

1929-97 period played a part in the significance of the fixed income and
cash returns. As Table 18.6 indicates, the cash returns for the first four
Democratic administrations in this period (1933-48) were very low (0.20
percent, 0.08 percent, 0.25 percent and 0.50 percent annually). This result
largely explains why the term cash-return differences are so significant (t-
value = -12.31 for 1929-97). Democratic administrations were in power for
three of the four terms during the 1941-56 period, when government bonds
had low returns. Bond returns in the 1961-68 period (both Democratic terms)
and 1977-80 period (Democratic) were also low.
Political effects: when Congress is in session

Ferguson and Witte (2006) find a strong correlation between Congressional
activity and stock market returns such that returns are higher and volatility
lower when Congress is in session. They use four data sets, including the
Dow Jones Industrial Average since 1897, the S&P500 index since 1957,
and the CRSP value-weighted index and CRSP equal-weighted index since
1962. They compare mean daily stock returns and annualized returns when
the U.S. Congress is in and out of session. Depending on the index tested,
statistically significant differences in average daily returns range from 4-11
basis points per day. Annualized stock returns are 3.36.5% higher when
Congress is out of session, and between 65-90% of capital gains have
occurred when Congress is not in session (which is notably greater than the
proportionate number of days Congress is not in session).
Ferguson and Witte also test these results in several ways. First, they
analyze if the ‘Congressional Effect’ is just a proxy for other known
calendar effects, such as the Day-of-the-Week Effect, January Effect, and
Pre-Holiday Effect. They conclude that, after controlling for these
anomalies, there is still a Congressional Effect of 3-6 basis points per day,
which means that no more than half of the Congressional Effect is captured
by controlling for other known anomalies. The study also tests for robustness
and finds there is a low probability that these results are the effect of a
spurious statistical relationship.
Table 18.6: Annualized average monthly return. Source: updated from Hensel and
Ziemba (2000)
From 1998-2010 we used the 3-month T-bill secondary market rate discount basis for
cash and market yield and US Treasury securities at 5-year constant maturity, quoted on
investment basis for bonds.
Next they test if public opinion toward Congress accounts for the
Congressional Effect by using public polling data as a proxy for general
investors' attitudes toward Congress. They use 162 polls from 1939 to 2004,
though 112 of these were conducted after 1989. They find that an active
Congress does not itself lead to poor stock returns but rather that the public's
opinion of that active Congress accounts for the depressed returns. They also
find that each index exhibits volatility that is significantly lower when
Congress is not in session and that this is also driven by public opinion.
Then Ferguson and Witte test the implications of this predictive capability
on optimal investor asset allocation using the models of Kandel and
Stambaugh (1996) and Britten-Jones (1999); they find that trading on the
Congressional Effect would allow investors to better allocate between
equities and cash and to achieve a higher Sharpe ratio.
Ferguson and Witte consider three alternatives as possible explanations of
the Congressional Effect, concluding that their findings may be explained by
viewing public opinion of Congress as a proxy for investors' moods,
regulatory uncertainty, or rent-seeking. The mood-based hypothesis follows
other studies in behavioral finance that suggest ‘depressed’ investors are
relatively risk averse, which in this case would imply that negative public
opinion of Congress was ‘depressing’ investors and dampening returns. The
regulatory uncertainty hypothesis follows from the implication that there is
more uncertainty in the market when Congress is in session, such that risk
and therefore returns are higher. The rent-seeking hypothesis is based on
Rajan and Zingales (2003) and suggests that concentrated economic interests
limit the efficiency of markets such that they are less efficient and biased
toward powerful financial players when Congress is in session.
Some simple presidential investment strategies

Two presidential party based investment strategies suggest themselves. The
first is equity only and invests in small caps with Democrats and large caps
with Re-publicans; the second, a simple alternating stock-bond investment
strategy, invests in small cap stocks during Democratic administrations and
intermediate government bonds during Republican administrations. The test
period was January 1937 through December 1997 with an update from 1998
to 2010.
The common 60/40 (large cap/bonds) portfolio investment strategy and
provides a benchmark for comparison with the two strategies. Transaction
costs were not included, but they would have a minor effect on the results
because the higher return presidential strategies trade at most every four
years. These investment strategies all lost money until the early 1940s, see
Table 18.7 which shows the cumulative wealth.
Table 18.7: Value of $1 initial investment in 1997 and 2010. Source:

Updated from Hensel and Ziemba (2000)
The two presidential investment strategies performed well over the sample
period. The strategy of investing in small cap stocks during Democratic
administrations and large cap stocks during Republican administrations
produced greater cumulative wealth than other investment strategies. The
alternating stock-bond strategy of investing in small cap stocks under
Democrats and intermediate bonds under Republicans produced the second
highest cumulative wealth. Both of these presidential party based strategies
had higher standard deviations than large cap stocks alone during the 1937-
97 period. Clinton's first administration had returns for small and large cap
stocks, bonds, and cash consistent with the past. However, in the first
fourteen months of his second administration large cap stocks produced
higher returns than small cap stocks.
In the update in Table 18.7 we see that, for the 1942-2010 period, small
cap stocks (Russell2000 from 1998) produced about four times the gains of
large cap S&P500 stocks (4327.6 versus 1043.5). But the small cap with
Democrats and large cap with Republicans was even higher at 6349.5, a bit
above the 5910.8 of small caps with Democrats and intermediate bonds with
Republicans. Meanwhile, the 60/40 portfolio was at 312.1 less than as
much! Table 18.8 displays the mean returns and standard deviations for the
various subperiods for the various strategies. The 60-40 portfolio does have
lower standard deviation but this risk reduction is swamped by the much
lower mean and total returns.
Table 18.8: Average returns and standard deviations for different investment strategies
for different investment horizons. Source: Updated from Hensel and Ziemba (2000) to
2010
Final Remarks
An important finding of this study was the much higher small-stock returns
during Democratic administrations as compared with Republican
administrations. This finding is consistent with the hypothesis that
Democrats devise economic policies that favor small companies and
consequently, their stock prices. The 33.51 percentage point difference
between small stock performance in Democratic and Republican
administrations in the first year in office and the 18.21 percent difference for
the full four-year term from 1929 to 2010 are very large. Also the update
from 1998 to 2010 has similar results.
This political party effect is different from the well-known January small
firm effect which has been present for Republicans as well as Democrats.
There is also a substantial small stock/large stock differential outside of
January during Democratic rule (see Table 18.5). Large stock returns were
statistically indistinguishable between Democrats and Republicans, but bond
and cash returns were significantly higher during Republican than during
Democratic administrations. This also confirms and updates Huang's finding
that large cap stocks have had higher returns in the last two years of
presidential terms; this finding applies regardless of political party and for
both small and large cap stocks.
A study of the differences in economic policies that lead to the divergence
of investment results according to which political party is in office would be
interesting. Clearly, candidates seeking reelection are likely to favor
economic policies that are particularly attractive to the public; and those
policies are consistent with higher stock prices. Cash returns did not differ
significantly between the first and second two-year periods of Democratic
and Republican presidential terms.
1This chapter summarizes results in Ziemba (2013).

PART IV
Volatility, Correlation and Liquidity
Thoughts on the VIX Fear Index1
High put prices have led to high levels of fear and plenty to play with for volatility traders
The VIX is the standard deviation of the implied volatility of S&P500 index
options that are close to the money and not far into the future. Put options
dominate in the various VIX calculations for the US equity and other
markets. High put prices lead to high fear reflected through high VIX
values. The VIX is a weighted average of various implied volatilities of
various options whose volatilities Eire backed out based on their prices by
some option pricing model. The VIX can vary from a low in the 10% range
to the high 20s into the low 30s for violent stressful markets and as high as
70% to 100%+ in market crashes. In 1990, the VIX of the Nikkei Stock
average of 225 stocks (price weighted like thf Dow Jones) was in the 70%
plus area for months and months. Figure 19.1 has the spot VIX graphs for
2002-2012 and Table 19.1 has the VIX futures as of September 30, 2012.
When as I wrote this on July 13, 2008, the S&P500 was fn a “so-called”
bear market (down over 20% from itf peak) at 1239.49 and the VIX at
27.49 ts at the higher end of its five- year range. Figure 1a.2 shows the 5-
year VIX with a 52-week range ol 14.79 - 37.57%. Figure 19.3(a) shows the
5-year S&P500 which shows peaks at 1527.46 on March 24, 2000 and
1520.00 on September 1, 2000 and in the last ten years, a low close of
778.63 on October 10, 2002. Figure 19.3(b) shows the S&P500 over a
longer period from 1955 to 2008. In the 2000-2003 market decline, many
stocks did not fall and the decline was concentrated in a few large
capitalized stocks, especially telecoms. This time it was financials leading
the S&P500 down, whereas in 2001/2 it was large cap momentum driven
stocks, technology and telecommunications, see Ziemba (2003).
Fig. 19.1 VIX, January 1, 2002 to September 30, 2012. Source: Yahoo Finance
Table 19.1: VIX Futures, October 2012 to June 2013 as of September 30, 2012
Fig. 19.2 5-year VIX, to July 11, 2008. Source: Yahoo Finance
Fig. 19.3 S&P500 Source: Yahoo Finance
Is the VIX a good predictor of future stock price movements? Figure 19.4
shows as of January 10, 2007, the global 1200 index in US dollars versus
the CBOE S&P500 VIX index. Basically, as the VIX falls, stocks rally and
as the VIX rises, stocks fall.
Fig. 19.4 Strong fundamentals keep volatility at bay. Source: Young,

January 10, 2007
Young (2007) wrote:
Currently, the VIX is trading at multi-year lows, suggesting that fear
is in deep hibernation. But when combined with the strength in
global markets over the past three years - the S&P Global 1200 is up
roughly 100% since March 2003 - many market participants are
concerned that the complacency implied by the VIX is spreading,
leaving global stock markets vulnerable to attack by the dreaded
bear.
Not likely, in the opinion of S&P Equity Strategy. We believe the
reason for the complacency is less the sloth of bulls than the fact
that equity fundamentals are, in a word, excellent. Liquidity is
ample and inflation is low, both of which serve to depress interest
rates and fuel unprecedented global M&A activity. At the same
time, attractive 2007 growth prospects and low P/E-to-growth ratios
are lending important valuation support to global stock markets.
In this climate, investors concerned about a spike in volatility should
watch for any deterioration of these fundamentals. We do not
believe, however, that such an erosion is nigh. Modest portfolio
rebalancing is certainly appropriate in light of recent gains. But
given the difficulty of successfully timing the market - requiring
both a graceful exit at the top as well as a cool reentry at the bottom
- S&P Equity Strategy does not currently advise significantly
reducing equity exposure. Global equity valuations are historically
low, especially given healthy 2007 earnings expectations. Despite a
consensus 2007 earnings growth projection of 9%, above the
historical average, the S&P Global 1200 index is currently trading at
a P/E of only 14.6, a 12% discount to its long-term average of 16.5.
Of course, this advice was not good for long-only investors (as are most
individuals and institutions). During this time, the VIX reached low values
around 10%. However, short sellers and other option players did just fine.
Rosen (2008) observes that although the current about 25% VIX may be
rich by historical standards, it is low when compared to recent volatility.
The S&P500 has not been nearly as volatile this summer as it was during
the 1st quarter of 2008 when the VIX traded at 35%. The 30-day current
realized volatility is about 18%, which is slightly higher than the 15% long-
term average, but still well below a 25% VIX. To justify a VIX at 35%,
actual volatility would need to rise substantially.
Following the Bear Stearns collapse in June 2007 the 30-day realized
volatility reached almost 30%. The stock market may have felt more
volatile in June and July 2008, but this is not supported by the data.
The VIX only reached 26% (despite the S&P500 at another 52-week
low) because investors are not as exposed to equities as they were in
January and March.
Those overleveraged or overexposed to stocks have long positions. This
is confirmed by the latest readings from both Investors Intelligence and the
ISI Hedge Fund survey, which show bearishness, and hence defensiveness,
approaching historical extremes.
Since investors are less invested than they were, they do not require as
much insurance as they did during previous market declines, which partially
explains why the VIX has not reached the higher levels of January and
March. Another reason suggested by Fishback (2008b) is that the individual
stock's correlations have dropped. While Lehman keeps dropping, Apple is
rising. Both have high individual volatilities but do not add much volatility
to the index.
Will the VIX pull back from here, and the stock market will rally? Not
necessarily, but the premise that stocks are headed lower on a short-term
basis because implied volatility is only trading at a 67% premium to
realized volatility seems also seems unlikely.
Rosen suggests that it is even more important to know what not to do in
this business, so anyone who is planning to rush out and buy puts and calls,
or sell equities because the VIX appears too low, should perhaps re-think
the situation.
Rosen also notes the well-known phenomenon that realized volatility of
the S&P500 (about 15% in the last 100 years) whereas implied volatility
has averaged about 20% since index options started trading with some
volume in 1985. S&P500 futures started in 2002. Supply/demand imbalance
plus fear leads to this overestimate of the future.
Pendergraft (2008) argues that the usual expectation that when the S&P
is down that implies a rise in the VIX and vice versa is not working in 2008.
Of course, with small changes, the usual behavior might not work but it will
with large changes (±2%+). He suggests that the future prices on the VIX,
affect the current VIX changes. This is a derivative on a derivative on a
derivative so the effect is complex and deserves a full study which our team
of researchers is working on.
Fishback (2008a) looks at the OEX (S&P100) volatility index, called the
VXO versus 19.9% and 10% declines. In all cases, before 2008, the VXO
was 35% plus once the 19.99% decline was reached. See Table 19.2 and
Figure 19.5. So why is the VXO lower now in 2008, at about 25.63%?
Table 19.2: Declines of 19.9%. Source: Fishback (2008a)
Fig. 19.5 VXO when 19.9% decline in DJIA is reached compared to days since DJIA peaked.
Source: Fis7hback (2008a)
What Fishback learned is the rather obvious fact that it has been the time
that it took to reach -20% that is crucial. When the decline was fast, that is
less than one month, the VXO always rose above 30%. But when the
decline took longer, then the VXO was under 30%. See Table 19.3 and
Figure 19.6 for the 10% declines.
Table 19.3: 10% OEX declines and the VXO when the declines are reached. Source:
Fishback (2008a)
Fig. 19.6 VXO when 10% decline in OEX is reached compared to days since OEX peaked. Source:
Fishback (2008a)
So the rather orderly decline from 1420 to the current 1228 has not led to
a very high VXO and the VIX.
Still others point to lower stock prices with some predicting that the
S&P500 will eventually fall below 1000. And, indeed, it did, bottoming out
on March 9, 2009 at 683.38. That was the lowest close but on March 6
intraday the market hit 666.79. There are plenty of bearish writers.
Prominent among them are Nouriel Roubini, the head of RGEMonitor.com
and John Maudlin of Millennium Wave Advisors, LLC.
Roubini, an NYU professor (and, for full disclosure, Rachel's boss) has
been consistently right in his forecasts and analysis since he was the first
loud voice (in 2006) arguing that the subprime housing crisis was imminent
and would be very widespread with large losses.
Mauldin tends to present the views of others but with a bearish focus. He
believes that the subprime situation is 90% contained but that there is much
trouble to come from banks and other financial assets. The banks and
investment companies may need another $400 billion. Where will they get
it? Some possibly could come from the sovereign wealth funds but their
investments so far have led to large losses so they may be very cautious.
Maudlin calculates the PE ratio of the S&P500 at 23 times forward earnings
which is a lot above Barron's current PE of 20.52 based on trailing earnings
versus 18.67 a year ago with a higher S&P500 and higher trailing earnings.
Also there is the question of when the real estate market may stabilize as
prices may fall well into 2009 or 2010. See Figure 20.1 for S&P/Case-
Shiller indices which indicates the extent of the decline and the beginning
recovery.
Bridgewater's estimate is that the net worth of US assets is down 13% or
$8 trillion since January 2007.
Over the years since 1989, see Ziemba and Schwartz (1991), Ziemba
(2003), Koivu et al. (2005), Ziemba and Ziemba (2007) and Lleo and
Ziemba (2012). I have made good use of the bond-stock earnings yield
(BSEYD) measure as a useful predictor of dangerous markets. Berge et al.
(2008) show that the simple rule: go into cash if the measure is in the
danger zone, otherwise stay in the S&P500 doubles the final wealth from
1980-2005 and 1975-2000 in all five county studies compared with staying
in the S&P500. Also, by being in cash, the standard deviation risk is lower
so the Sharpe ratios are even higher. Durre and Giot (2005) investigate
more countries than the US, UK, Canada, Germany and Japan that we
studied. See Chapters 2 and 21 for a discussion of this.
The measure kept me out of the 2001 crash and also predicted numerous
other crashes. In 2006-2008 it has predicted the crashes in Iceland and
China, as discussed in Chapter 21. See Ziemba and Ziemba (2007) for
earlier analyses of these two countries. Before, it was the Japan 12 out of 12
correct predictions during 1948- 1988 and the big one January 1989 plus
the US in 1987 and 2002. The measure also predicted the 2003+ rally from
the 778.63 S&P500 low to over 1500.
In the years 2009-2012, the model was not generally in the danger zone
but there was a sell signal on June 14, 2007 that presaged the 2007-2009
crash, see Chapter 21. The market has fallen for other reasons, namely, the
subprime and credit crises we are now in. Maudlin (2008) thinks the
subprime crisis is 90% over. But there are some $1.6 trillion is losses
coming from write downs, according to Bridgewater Associates. Roubini
(2008) suggests it will be more, especially if $5 trillion is needed for
Freddie Mac and Fanny Mae. Last night (July 13, 2008) the S&P500
futures were up to a high of +15 as a deal between these agencies and the
Treasury and Fed has been announced by Treasury Secretary Henry
Paulson. The deal involves asking Congress to approve unlimited loans,
buying of preferred stock, and use of their collateral. The market was not
that impressed and +15 ended up -11 on the S&P500, during the day on
July 14, 2008.
Indeed in Japan in 1990+ the stock and land price falls led to even more
losses and a 20 year dark period. Japan, of course, was way way overpriced
in land and the PEs of 60 were completely crushed when interest rates were
raised in mid 1988 to August 1990. A major error was increasing interest
rates a full eight months in 1990, once the stock market began to fall in
January 1990. When the bond- stock measure goes into the danger zone,
there usually is a 10%+ crash from the current level but with a lag. In April
1987, the signal said sell but the crash was in October. In April 1999, there
was a similar signal but the stock market only fell a year later. See Ziemba
(2003) and Ziemba and Ziemba (2007) for more on these episodes.
If we use Maudlin's 23 PE ratio, the BSEYD measure was ∆= 10 year
Treasury bond interest rate - 1/PE = 3.96- (100/23) which is less than zero.
So, even with this high PE ratio, the BSEYD model was not in the danger
zone. But the June 14, 2007 signal said sell; see Chapter 21.
Despite rumblings of US and worldwide inflation, the financial situation
is so dire that higher US interest rates by the Fed's action seems unlikely, at
least for the next while. So how would the ∆ get high enough to be in the
danger zone (about 3)? It has to be lower and lower earnings, similar to late
2001, see Koivu et al. (2005) for that correct forecast of 2002's -22% on the
S&P500.
Giorgio Consigli, Leonard MacLean, Yonggan Zhao, and I worked on a
series of papers looking at determining the fair value of the S&P500
(Consigli, 2002, Consigli et al., 2008, and MacLean et al., 2008). The
models involve jumps.
As a predictive model, adding the VIX as a second predictor adds value
and predicts better than the bond-stock market alone during the 2000-2007
sample period. The heavy tails are modeled with the addition of a
homogeneous point process. The timing of the jumps in the point process
attempt to predict the price reversals. In Consigli et al. (2008), a non-
homogeneous point process is introduced so the intensity and size of the
jumps are state dependent. The state is the stress measure being a
combination of the bond-stock and VIX measures. The direction of the
shock from the VIX is revealed by the bond stock yield ratio. The model
computes the stress thresholds and the weights of the risk factors.
Figures 19.7 shows the bond-stock model in its ratio form
(mathematically equivalent to the difference model) and Figure 19.8 shows
the VIX from 1990-2007. The predicted prices are in Figure 19.9.
The conclusions are:
(1) The addition of non-homogeneous point processes to a diffusion
greatly improves the fitting to actual equity returns.
(2) Both the intensity of jumps and the size of jumps depend on the risk
factors - the bond-stock measure and VIX.
(3) The VIX and bond-stock are complimentary since during some
periods the dependence on VIX is more pronounced, while in other
periods the dependence on the bond-stock is stronger.
(4) Because of the complimentarity, a convex combination of the factors
averages out or smoothes the extremes and results in a low frequency
of shocks and a poorer fit.
Fig.19.7 BSYR: 1990–2007
Fig. 19.8 VIX: 1990–2007
Fig. 19.9 Predicted Prices of the S&P500, 1990–2007

When the current crisis will end and how low the S&P500 will go is
difficult to determine. The debts, good and bad and net positons of so many
financial institutions is uncertain. We have not had the so-called bottom
“capitulation” event. So far, the fall has been tense but orderly with declines
in the 1, 2, and 3% range, but could and did get worse. More is still to be
known about Fannie Mae and Freddie Mac, which, if Roubini is right (as he
has been so far), their situation will be worse than now. They have $5
trillion in loans that they guarantee - an enormous amount. That's about
45% of the total mortgage market of $12 trillion. Others, such as former
Fed Governor William Poole, think they are in bankruptcy already. They
may get out of trouble and survive but only time will tell here.

Changing Correlations: Rising VIX and Violent
Market Moves1
Once reliable correlations now present a mounting challenge as the markets descent further into fear
and instability. What fun!
Why is the world economy and its financial markets in so much trouble?
Figure 20.1, the Case-Shiller US home price index; shows that there has
been a dramatic decline in US house prices since they peaked in 2005. The
decline is close to 20% with larger declines in areas that increased the most
in price since 2002. With some US$1.2 trillion delinquent subprime
mortgages out of a total value of US mortgages of about $10.2 trillion, there
is a lot of trouble here and millions of houses are being foreclosed on.
Indeed one of every five houses in America has a mortgage greater than its
market value. The situation is similar in essentially all of those countries
that gave loans to unqualified buyers. Many of those with qualified
mortgages are not in good shape either There is the odd US bank that did
not make these risky loans but most of them did.
The UK is another example where such subprime loans were very
prevalent. Indeed UK bank accounts frequently take salaries and pay
everything from one account so housing can easily dominate a individual's
or family's finances. ING of the Netherlands had good prudent policies and
keeps their mortgages and does not resell them so has avoided most of the
trouble. In countries like Canada where you simply cannot get the banks to
loan you money for a house well beyond your means to pay the mortgage,
there is not much trouble from Canadian mortgages. But there is much
indirect exposure everywhere as the US loans were diced, repackaged and
resold as derivative instruments deemed safe by regulators and scooped up
by banks, pensions, insurance companies and other financial institutions
around the world. Also the US is the only country I know where mortgages
are non recourse loans. Canada has only a few large banks with
conservative management and regulations so despite a housing bubble there
too, the prices have not collapsed yet. But even there, a decline is starting to
emerge. None of the major Canadian banks are in serious trouble despite
having some poison subprime US derivatives based on US housing. Despite
the fact that Canadian banks did not issue subprime mortgages, they too
have pressured the government to buy up their already government insured
mortgage bonds to free up liquidity in the midst of frozen credit markets.
Canadian banks are buying US regional banks and consolidating them.
Figure 20.2 shows the real price of domestic housing in the US and five
other countries from 1900 to 2011.
Fig. 20.1 Case-Shiller Home Price Indices. Source: S&P Press Release, April 28, 2009
Fig. 20.2 Real price of domestic housing in six countries, 1900–2011. Source: Credit Suisse (2012)
This housing bubble which in most of the world was basically caused by
the Alan Greenspan low interest rates policy to get the US out of the
internet bubble. People all over the would soured on stocks when the
S&P500 fell basically in half from 1520 in September 1, 2000 to 776 on
October 9, 2002. Inflation adjusted at 764, the current (October 31, 2008)
960.75 is even lower! The 2000-2002 fall in the S&P500 was mainly in
three areas: information technology, telecommunications and large cap
stocks. Yet 41% of the 500 stocks did not fall and 19% declined by 10% or
less annualized. These were small cap stocks with values to $10 billion or
less.
The Nasdaq fell much more from 5000 to about 1000, see Ziemba (2003)
for graphs. Equally weighted, the S&P500 lost only 3% in 2000-2002 so it
was a decline of a section of the market that was overpriced. See Ziemba
(2003) for more on this episode in history.
With cheap and easily attainable money and a distaste for risky equities,
relatively safe property seemed the way to go. And the bubble was
everywhere. In 2005 I gave a talk in Bolzano, Italy, a nice mountain town
above Verona. Housing there then at 8000C per square meter with a high
euro was out of sight. The Grand Canal in Venice was not much more!
The easy money policy had three parts: cheap money, easy to get, and
attainable even if you were not qualified to pay it back. The assumption was
that housing had only one way to go, up. So the expected increase in price
was relied on to build equity. The US dollar under this policy fell
dramatically against many currencies and commodities such as gold and oil.
The fall in the US dollar was due the willingness of foreigners to hold US
despite the low interest rate and the US twin trade and budget deficits.
Commodities rose as demand from China and other countries accelerated.
In early January 2002 the euro bought 0.87 dollars and it peaked in 2008 at
1.60.
Fig. 20.3 S&P500 Source: Yahoo Finance

All such bubbles must eventually pop and the commodities peaked in
June/early July 2008. Trend models exited commodities such as oil in July
2008, taking large losses from the earlier 2008 profits. My daughter and co-
author of this book, Rachel Ziemba of RGEmonitor actually called the fall
in oil when it was at $147 per barrel that it would hit $100 before $200 in
various interviews for the Wall Street Journal and other investment media,
see Ziemba (2008). Currently oil is under $70 a barrel and never exceeded
the real price high of over $100 in the 1973-74 period although the real
price was close to this peak.
As readers of Ziemba and Ziemba (2007) know, the bond-stock model
correctly called the 2000-2002 crash period. There were actually two bond-
stock crash signals: the first was in April 1999 and the second was in late
2001. The bond-stock model also called the 2006-8 equity crashes in
Iceland and China, two places where interest rates were too high compared
to earnings and the US crash on June 14, 2007 as discussed in Chapter 21.
Fig. 20.4 Euro, January 1, 1999 to October 31, 2008. Source: Yahoo Finance
Getting back to the current crisis, the growth of derivatives worldwide

and the misuse of them by supposedly knowledgeable analysts that we
professors have trained and bank, insurance, pension fund administrators
underestimating the risks is illustrated by the gigantic size of the worldwide
derivative markets. Figure 20.5 shows that there has been a five fold
increase in derivatives from 2002 to 2008.
Growth of derivatives
There are a lot of equity derivatives in the trillions of dollars but this is
swamped by credit default swaps. My colleague Professor Ed Altman of the
Stern School, NYU, correctly predicted much of the current problems with
such instruments at a conference in May 2008 that I attended in Florence.
The big elephant in the room relates to the interest rate swaps and options
and currency swaps. Basically, interest rates. Many such derivatives were
repackaging of the unsound subprime housing and other loans. The rating
agencies which gave AAA ratings are at fault here as well. It's a case of the
fox guarding the chicken coop as the sellers of these derivatives paid the
ratings agencies so there was no independence. See Altman's website for
more on this. Even the best experts such as Alan Greenspan and my
esteemed late University of Chicago colleague Professor Merton Miller
bought into the academic view that derivatives provide greater liquidity and
hedging so are safe. At a low level, they are but at $531 trillion there is a lot
of scope for trouble, especially with widespread ignorance, mispricing of
derivatives and alternating fear and greed. By September 30, 2012 when we
went to press, the total value of derivatives was well above $700 trillion.
Fig. 20.5 Growth of a complex market. Source: The New York Times
The vast number of instant, narrow, expert financial engineers turned out
by the best universities without grounding in economics and history has
been part of the problem as the risks they were creating were only numbers
to many of them.
• Greenspan, asked at a Congressional hearing “aren't you concerned
with such a growing concentration of wealth that if one of these huge
institutions fails that it will have a horrendous impact on the national
and global economy?” he replied “No I am not, I believe that the
general growth in large institutions have occurred in the context of an
underlying structure of markets in which many of the larger risks are
dramatically - I should say, fully - hedged.”
• Miller: “Despite all the hullabaloo in the press, and all the bad
publicity surrounding derivatives, banks are safer today, not riskier.”
“... no serious danger of a derivatives induced financial collapse
really exists.”
• Buffett (2002 BH shareholder letter): “Large amounts of risk,
particularly credit risk, have become concentrated in the hands of
relatively few derivatives dealers. The troubles of one could quickly
infect the others.” “Derivatives are weapons of mass destruction.”
Warren Buffett has been a derivative critic calling them instruments of
mass destruction. There are many reasons why this sage great investor,
arguably the world's best, understands things so well. As a Berkshire
Hathaway shareholder, I read his annual reports and hidden in the footnotes
are a number of derivative trades. But Buffett well understands the effects
of leverage so only bets when he has a huge advantage and the risk of a bad
scenario will only cost him 1-2% of his total portfolio. For example, he has
at-the-money short puts on the S&P500 - a disaster area for most of this
year, but the time to expiry is 15-18 years with strikes about 1100-1200! He
does not win in all such bets but he is usually ahead on most of them at the
end!
Fig. 20.6 Berkshire Hathaway B shares — Aug 1, 1998 to September 30, 2012. Source: Yahoo
Finance
Berkshire has had a great record over the 40+ years since Berkshire was
founded in 1965 but even they have fallen into a bear markets in 2008 (20%
decline), see Figures 20.6 and 6.1. The smaller B shares peaked at 4950 in
the week of December 3, 2007 and on October 31, 2008 they were 3840.
Still they have outperformed the equity markets. Later, the B shares split
50-1 so they were about $85 a share on September 15, 2012.
Buffett makes investments that resemble a full Kelly bettor. His
geometric mean, the quantity maximized in the Kelly strategy, has been
high. For example, from 1977 to 2000, it was 32.07% versus 16.7% for the
S&P500. The BH record since 2000 has been less spectacular but still
considerably above the S&P500; see Figure 20.6. His Sharpe ratio is not
high nor is his downside symmetric downside Sharpe ratio, see Ziemba
(2005) and the discussion in Chapter 6. He simply goes for the long run -
the Kelly approach!
The bond-stock measures that have been so useful in the past, see
Chapters 2 and 21. However, on June 14, 2007, it did predict the July to
October 2008 crash situation and the subsequent decline into the March
2009 lows.
Note: Data through 29 May 1990. Shaded lines in Panel B denote upper
limit, mean, and lower limit.
Source: Based on data from Ziemba and Schwartz (1991).
Fig. 20.7 The bond and stock yield differential model for the S&P500, 1980-1990. Source: Ziemba
(2003)
Figure 20.7 shows the 1987 crash and the two crashes in 2001 and
2002/3. In between, before and after the market was out of the dangnr zone.
The second 2002 crash oncurred because prices fell but earnings fell even
more.
Rajan and Seru (2008), looking at securitized subprime loans issued from
1997-2006, concluded that quantitative models underestimated defaults
from subprime borrowers. Indeed there was a systematic failure of default
models. Others analyzed the effects of a 10-20% fall in real estate prices
and concluded that its effects would be devastating but they assigned a zero
probability to such a large real estate fall. So there are model failures and
failures of the people using them.
Table 20.1: The 2000-2003 crash in the S&P500. April 1999 enters danger
zone
In the rest of this chapter, I focus on this July to October 2008 period. I
wrote this during the November turn of the month, historically the best
TOM and the markets are rallying per the script. The S&P500 closed at
968.75 on October 31, 2008, a gain of 14% in the last 4 days. Of course,
there is good economic news but then good news has a way of showing up
at the right times. This is right after the two historically worst months of the
year, September and October, see Keim and Ziemba (2000). The future
remains violent and uncertain with many assets such as commodities
dropping in price. Indeed Menegatti, Pineda and Ziemba (2008) make the
very plausible case for global stag-deflation in which debts, assets, wages
and goods all decline in value Let's go to the big issue: changing
correlations and viciously high VIX.
Oil
Oil prices rose in US dollars and also in euros until the euro reversed in
August 2008.
• Refining capacity is limited and drives up the cost of fuel products
(heating and auto).
• Oil peaked nominally in July 2008 at $147 and then fell to the low 90's
before recovering to about $102 and then falling back to the current
about $65 because of the weak economy and falling demand. See
Figure 20.8a for oil to October 2008 and 20.8b to May 2012.
• Oil prices rose in US dollars and also in euros until it reversed.
• Oil has still not reached the record real price of over $100 in the 1970s,
but it was close at its peak.
• There is much uncertainty in the near term as OPEC is being tested re
whether they can impose supply constraints to raise the price.
• New supplies have higher marginal costs than the current price and
will likely be delayed — it is estimated that new Canada oil sands
projects require oil at $100.
• Budgets of oil exporting states require $50-60 oil and Russia's budget
assumes oil is $80+ above where it is now.
Fig. 20.8 Oil prices
Gold
• Since 2005 gold has been rising, peaking at $960 an ounce in July
2008. But the path had been violent. It was 718.20, down 18% in
October 2008, the largest monthly decline in nearly 30 years. See
Figure 34.3a for prices to October 2008 and Figure 34.3b to August
2012 when gold peaked at $1920 in September 2011 and closed on
September 30 2012 at 1776.
• Part of gold's volatility is that it moved with other commodities such as
oil.
• In 2011/2012, though it has gained then fallen on a flight to safety
while com-modities have fallen.
• Some investors may worry about the difficulty of selling gold or gold
options.
• Gold is far below the long-term real price — having reached $800 in
the late 1970s.
• Given the great uncertainty in the world economy, gold looks strong as
an asset class but, like all assets, it has been very volatile in 2008 and
relatively weak now. The safe haven aspect has been replaced by a
weak economy weak commodity story.
• Another problem is that although the bailouts may be inflationary in
the long- term, deflation is a more near term worry — thus depressing
the demand for gold.
• Figure 20.10 shows gold prices and inflation in the UK, 1900–2011.
Fig. 20.9 Gold
Fig. 20.10 Gold prices and ination in the UK, 1900–2011. Source: Credit Suisse, 2012
US stock market and economy
• The long bond interest rate is low but earnings are dropping fast.
• The bond-stock measure had not been in the danger zone since late
2001 it got there on June 14, 2007; see Chapter 21.
• Put prices greatly exceed call prices so my short term crash measure is
not applicable either.
• Subprime mortgage and credit market problems are far from resolved
even with all the new interventions.
• Growth stocks like Apple have had roller coaster rides - returned to or
exceeded pre August 16, 2007 highs and then have fallen again. But
Apple had a huge runup in 2011-2012 to peak at 644 per share then it
fell but then went higher to slightly above 700 and on September 30,
2012 was about 660 per share.
• Hedge fund computer valuation models, which failed in August 2007,
and es-pecially on August 16, 2007 became long the actual good
stocks and short the weaker ones. In the September-October 2008 steep
decline, again the good was sold.
• Correlations were changing fast on pairs of stocks. One day A and B
would move together with an S&P move, another day they would
diverge with no A or B news and a similar S&P move.
• There have been a number of sharp rallies in the bear market but the
trend has been down.
• The VIX volatility fear index has been sharply rising, reaching record
levels.
• Market turbulence is likely to continue for at least a year as there is
much uncertainty about the full extent of the housing downturn and the
subprime losses.
• The decision not to save Lehman - possibly politics between them and
Paulson's former firm Goldman Sachs- was a savage blow to the equity
and other markets.
• The Fed and Treasury have been innovating in attempts to avert a
severe recession, but they cannot do this alone through monetary
policy. The new $700 billion fund to purchase bad debt is a response to
this. The funds have been extended in a much broader fashion than the
original intention and are being used to recapitalize the banks, possibly
insurers and others. Letting banks use this money to pay dividends
does not seem wise as a way to loosen credit.
• Fiscal policy is needed as well but there is no latitude for this given the
twin deficits. The first $168 billion stimulus in the form of tax rebates
did not work as most of that was saved or repaid debt. However it may
have delayed the US pathway into recession. A second stimulus
package of around $150 billion is being proposed by the Democrats -
both House Speak Pelosi and presidentelect Obama are calling for a
lame duck session of congress. This would be more targeted to things
like extending unemployment benefits, food stamp programs and
helping state governments, etc.
Currencies
• US Treasury Secretary Henry Paulson, former Chairman and CEO of
Goldman Sachs, said in 2007 that there was still a strong dollar policy.
• At the time, Goldman Sachs research forecasted further dollar declines
and a euro going to the 1.45 area.
• The trade weighted euro peaked in 1985 at 1.45. The euro zone cannot
exist well with the euro so high, so the rise seems limited, and, indeed
as we went to press on September 30 2012, it was about 1.29.
• The euro peaked in 2008 at 1.60 and then in 3rd quarter 2008 fell to
1.37; then back to 1.46 and now is about 1.26, all part of the 2008 high
volatility. However, this is not just about volatility. The euro's decline
was largely a recognition that the eurozone faces worsening economic
conditions which don't dictate further strength. With the ECB finally
cutting rates, the euro may have further downside. However, the
presence of the euro may have forestalled some currency crisis among
its weaker members. In May 2012 the euro was about 1.33, still very
high given all the European debt issues.
• The euro's decline was matched and exceeded by those of the pound
and commodity currencies. Late in 2007 and early in 2008, the
worsening economic outlook and fed rate cuts contributed to further
declines in the dollar as investors continued to turn to the currencies of
hawkish central banks in face of higher global inflationary threats.
• The dollar and yen have been the main beneficiary of this flight to
safety and liquidity.
• In Q3, a reassessment of global outlook not just the US contributed to a
dollar rally, unwinding of carry trades, shift away from anything seen
as risky.
Europe
• The outlook looks bleak there as well, perhaps even worse than in the
US.
• The euro and GBP have fallen dramatically and are still shaky.
• London has finally had a shock in the main thing they do: financial
services.
• Greece, Ireland, Italy, Portugal, Spain and other countries such as
Cyprus are close to collapse.
Iceland
• The bond-stock measure predicted Iceland's stock market to crash and
it did. Interest rates were 14-16% short 9-10% long! See Chapter 21.
• PEs were low but the earnings part fell and the banks had too much
leverage.
• Now interest rates are 18% and the currency has declined 44% in the
past year. The chapter in Ziemba and Ziemba (2007) gives background
on Iceland and how fragile it was: a mouse that roared and was stepped
on by the credit crisis elephant. Chapter 21 describes this crash.
• The banking sector assets were more than 10% GDP meaning that the
Iceland government could not bail them out.
• Iceland was basically a hedge fund that failed. They were seeking a
loan from the Russians which did not materialize but they did get $2
billion from the IMF.
• They are arguing with the British government for repayment over
dealings with the collapse of the Kaupthing Bank. Iceland is frozen.
Indeed 1.34 billion of the deposits of 8000 people are not redeemable
now and may never be paid back. These types of counter party risk
abound with Lehman and other failed institutions not returning money
to rightful owners.
Across the board European governments are seeking to support their
banks, increasing insured deposit levels and some have nationalized failing
institutions. but some of the banks are too big to fail and too big to save (eg
some of the German banks) EU leaders are now trying to act in
concert.Eastern European countries are exposed to EU and have current
account deficits. Exposure cuts both ways. Western European banks that are
most exposed to the Eastern European markets were under pressure. This
exposure likely contributed to ECB and EU willingness to extend funds to
countries like Hungary at last, lest contagion spread.
What is a subprime loan and why have they caused so much trouble in
so many places?
Subprime loans: loans to borrowers who don't qualify for best interest or
with terms that make the borrower eventually unqualified as with zero down
payment, zero interest.
In general: lending institutions inherently get it wrong. When times are
good, they tend to be greedy and try to maximize loan profits but then they
are very lax in their evaluation of borrowers' ability to pay current and
future mortgage payments.
As noted on page 369, excessive lending in Japan led eventually to $10
trillion lost in stocks and real estate in the early 1990s.
Now the lending organizations sell off the mortgages and they are cut and
diced bundled into packages like CMOs and CDOs and sold to others who
have trouble figuring out what's in them but look at the rating agency's
stamp of approval.
During July to October 2008, most hedge funds and investors lost money
and a lot of it. Most are below their benchmarks. The hedge fund group I
consult for was way ahead of the curve so did well. Nouriel Roubini of
RGEmonitor has also been consistently right. Those who went into cash
early or shorted, did well, but most did not. Figure 20.11 which shows the
volatility of gold, oil, the euro, pound and S&P500 illustrates the trouble.
Fear took over markets with a small scrap of news causing large moves in
short periods. Gold has been ±100 points in a short time on several
occasions. Changing correlations and the high VIX were much of the
problem. Oil and gold were highly correlated in the period with August
2008 then decoupled then rejoined. Currencies have gone up with
commodities then down as they fell with violent short term moves. The
reaction to similar news was contradictory. For example, when the US $700
billion bailout was announced, the S&P500 exploded up. Then when the
budget did not pass, the market fell, but then when it did, it fell more. What
will the new administration mean for the economy? Historically stock
markets have higher returns with Democratic rather than Republican
presidents, see Chapter 18 and Obama was a huge 1-6 favorite on the
betting exchanges. Look for a landslide!
Fig. 20.11 Comparing the volatility of gold, oil, the euro, pound and S&P500
Some $800 billion was lost by the top twenty Russian billionaires.
As usual, the culprit is the recipe for disaster.
• you must be well diversified in all scenarios. But even those who tried,
had trouble because of the changing correlations and differing reaction
to similar news; and
• you must not overbet and here it has been tricky as well; what was not
an overbet situation with VIX of 30% becomes overbet at VIX=50%
and way overbet with VIX in the 70s or 80s. As shown in Figure 19.1
the VIX peaked at 90 and is still a very hefty 60% after the end of
October turn of the month rally in the S&P500.
The trouble is worldwide and so is the volatility. The Japanese Nikkei
stock average had its best week in history last week in its worst month in
history. There are some positive signs among the darkness. Buffett is buying
and got free call options on GE and Goldman Sachs by effectively loaining
them money at 10% through preferred shares. There is considerable buying
by institutions buying for the long term at what they feel are cheap prices.
Eventually the darkness will lift and the stock market will precede the
economy's recovery. The stakes are too high for a full collapse so look for
more and more government actions. The Fed is already $1 trillion into this
and the Treasury a lot as well. Too bad they didn't see the problem earlier as
did Roubini, Krugman and Shiller.
The 2008 election has now occurred and as I predicted, it was a big
victory for Barack Obama and good for my Betfair and Matchbook betting
exchange wagers. The March and September 30, 2012 Betfair odds on the
November 2012 election are in Chapter 1. The turn-of-the-month of
November has ended with the usual strong gain of over 15% with the VIX
dropping from 70 to 47. The post election reaction is a return to the bad
news and a 10% decline in the S&P500 in the next two days and a modest
rally Friday. Eventually the market will rally but it remains tense with the
VIX back into the low 50's then back over 60% on Friday. The VIX seems
likely to remain at these high levels for the rest of the year.
1 Edited from Wilmott, January 2009.

PART V
Can We Predict Stock Market Crashes?
Stock Market Crashes in 2006-2009: Were We
Able to Predict Them?1
We investigate the stock market crashes in China, Iceland, and the US in the
2006-2009 period based on the results in Lleo and Ziemba (2012). The bond
stock earnings yield difference model is used as a prediction tool.
Historically, when the measure is too high, meaning that long bond interest
rates are too high relative to the trailing earnings over price ratio, then there
usually is a crash of 10% or more within four to twelve months. The model
did in fact predict all three crashes. Iceland had a drop of fully 95%, China
fell by two thirds and the US by 57%.
Background
The second author started using the bond stock earnings yield model while
in Tokyo consulting at the Yamaichi Research Institute in 1988 and early
results are in Ziemba and Schwartz (1991). Later studies are Ziemba (2003)
which discusses, in particular, the internet bubble crash of 2000-2002,
Koivu, Pennanen and Ziemba (2005) which discusses co-integration of the
crash measure and the stock markets in Germany, UK and the US and Lleo
and Ziemba (2012) which discusses the 2006- 2009 crashes in China,
Iceland and the US. The model relates the yield on stocks (measured by the
ratio of earnings to stock prices) to the yield on nominal Treasury bonds.
The theory behind the model is that an optimal asset allocation between
stocks and bonds is related to their relative yields and when the bond yield is
too high, a market adjustment is needed and there is a shift out of stocks into
bonds. If the adjustment is large, it causes an equity market correction (a
decline of 10% within one year). Hence, there is a short term negative equity
risk premium (ERP). A study of the October 1987 stock market crash
illustrated the model and it was from this episode that the model was
discovered in Ziemba's crash study group. Table 21.1 and Figure 21.1 show
that the model went into the danger zone, that is the measure was above an
upper confidence limit, in April 1987 with a spread of 3.39. The S&P500
was then at 289.32. By the end of September the S&P500 was at 318.66 with
the measure higher at 4.14. After the crash at the end of October, the
S&P500 had fallen to 245.01 with the spread falling out of the danger zone
to 1.64. Table 21.1 has beginning of months values. Continuous daily values
are in Figure 21.1 and one sees a sharp move up in the measure to 4.42, in
September 1987 way above the upper limit .
Fig. 21.1 Bond and stock price earnings yield differential model for the S&P500, 1980-1990, Source:
Ziemba and Schwartz, 1991
Value of S&P500 for various spread values
Date/level Spread, % S&P500

May 29, 1990 1.11 360.65
Mean 0.98 355.00
Upper limit 2.09 415.00
Lower limit -0.13 309.00
There are various ways that one can compute the upper and lower limits
but my experience is that with the various approaches, all of which use out
of sample prior data, one usually has the same conclusion. In Figure 21.1,
the limits are simply the trailing mean plus or minus a standard deviation
measure so the one sided limits hold 95% of the probability.2
Table 21.1: S&P500 index, PE ratios, government bond yields and the yield premium
over stocks, January 1984 to August 1988. Source: Ziemba and Schwartz (1991)
Moving Average and Signal Chart
In the following sections3 we use a moving average and a rolling horizon
standard deviation to establish the confidence levels. The h-day moving
average at time t, denoted by and the corresponding rolling horizon
standard deviation are
Using rolling horizon means and standard deviations provide data
consistency. This is used to compute confidence levels for the BSEYD
measure. In particular, rolling horizon mean and standard deviation are not
overly sensitive to the starting date of the bond yield or stock market data, or
to the overall number of data points. However, the choice of the horizon
parameter h is subjective.
For the analysis of the US market, we use a five year horizon, so h = 1260
as longer time horizons tend to generate a robust signal and eliminate false
positives. Five years of historical bond and stocks data may be a difficult
requirement outside of major markets. For Iceland and China, a one year
rolling horizon was used with the confidence level tightened in order to
eliminate false positives.
The figures show the signal calculated, respectively, on a standard one-tail
95% normal distribution based confidence level and on an application of
Cantelli's inequality for arbitrary non-normally distributed returns. The
conclusions are similar: over the period January 1, 1995 to October 19,
2011, two crash signals occurred: one in June 1999 and the other in June
2007 (as discussed below).
An examination of the BSEYD spread distributions reveals their non-
Normal nature (see Figures 21, 21, 21.11 and 21.18). As a result, standard
confidence intervals which are based on a Gaussian assumption may prove
inaccurate. In this case, we use Cantelli's inequality, a one-tailed version of
Chebyshev's inequality, to derive a ‘worst case‘ confidence level (see, for
example, Problem 7.11.9 in Grimmett and Stirzaker, 2001).
Cantelli's inequality relates the probability that the distance between a

random variable X and its mean μ exceeds a number k > 0 of standard
deviations σ to this distance
or alternatively
where . The parameter α provides an upper bound for a one-tailed

confidence level on any distribution, regardless of how different it is from a
Normal distribution.
We use Cantelli's inequality with a one-tail confidence level to assess the
relative strength of a signal. Using the one-tail confidence level, we obtain
the crash signal. We then find the Cantelli probability α giving us a similar
signal date. To generate the same signal as a standard 95% confidence level,
we need to select α = 27% in Cantelli's inequality. Similarly, to generate the
same signal as a standard 99% confidence level, we need to select α =
15.60%
In the case of China, as discussed below, we used a standard 99%
confidence level to determine the signal. Based on Cantelli's inequality, we
expect in the worst case to have a crash signal 15.60% of the time.
Retrospectively, Cantelli's inequality is rather severe since it places the
threshold for a signal at a BSEYD spread of 3% or above, on 520
consecutive trading days from September 9, 2007 to February 2, 2009 out of
the 2099 days (i.e. 24.77% of days) where this measure was computed. By
contrast, if we consider the distribution over the entire period, we observe
that the spread only exceeded 2.52% on 0.03% of all instances. To conclude,
a standard 99% confidence level would have been sufficient to determine a
clear crash signal.
In the case of Iceland, we use a 95% confidence level to determine the
signal, Cantelli's inequality suggests a worst case probability that 27% of
observations could result in a signal. However, lowering the Cantelli
probability α from 27% to 20% (corresponding roughly to a standard 97.7%
confidence level) does not result in significant loss of responsiveness of the
signal, at least for Glitnir and Kaupthing (see Figures 21.13b,d).
The idea behind the BSEYD model is that a crash signal should occur
whenever
where CL(t) represents a one-tail confidence level. The level CL acts as a

timevarying threshold for the crash signal.
There is a crash signal whenever
Graphically, the threshold for the crash signal is a horizontal line with
value 0, as shown below in Figures 21.5(b), 21.6(b), 21 and 21.13. These
graphs show a calculation of SIGNAL(t) based respectively on a standard
one-tail 95% confidence level and on an application of Cantelli's inequality.
Lleo and Ziemba (2013) give an historical account of his use of this model
from 1988 to 2012 in the US, Japan and other countries. In Japan, the signal
was in the danger zone twelve times from 1948 to 1988 and each time the
Nikkei Stock Average index fell at least 10% from the level when the signal
was reached. During this forty year period, the index had twenty such falls
of which eight occurred for reasons other than high interest rates relative to
earnings. The BSEYD model also predicted the -56% crash that started in
January 1990 with a signal further in the danger zone than at any time since
1948.
The Fed model, which is the ratio not the difference like the original
BSEYD model, is mathematically equivalent to the BSEYD model, as now
shown.
The Fed model in its original 1996 form states the dependence of a fair
stock price level at time t to the expected earnings, E(t), and the most
liquid (10- or 30-year) Treasury bond rate r(t). Earnings expectations are
incorporated in prices and discounted via
Equity earnings per share γ(t) is the expected earnings for a unit
investment in the stock market with equity shares, S(t), namely
Then
There is a direct relationship between the equity yield in Equation (21.2)

and the long bond rate in Equation (21.1) . The ratio of the current market
value to the theoretical value is the Fed model bond stock yield ratio
BSYR(t)
The bond stock earnings yield differential that we focus on in this paper is
related to the valuation measure and the equity yield via
The differential reflects the difference between the current market value
and its theoretical value. A more theoretically sound motivation for the
predictive ability of the BSEYD is using the basic Gordon formula, where
EP is the forward earnings yield (which Schwartz and Ziemba (2000) show
is the best predictor of at least individual Japanese stock prices),
E/P - nominal yield = equity risk premium - real growth - inflation.
So the BSEYD can be used as a proxy for the unobservable right hand
side economic variables.
For given equity yield the BSEYD and the BSYR can be used to identify
zones of under and over valuation and forecast possible forthcoming market
adjustments.
Koivu, Pennanen and Ziemba (2005) study the Fed model using a
dynamic vector equilibrium correction model with data from 1980 to 2003 in
the US, UK and Germany and show that the Fed model had predictive power
in forecasting equity prices, earnings and bond yields. The model has been
successful in predicting market turns, but in spite of its empirical success
and simplicity, the model has been justifiabily criticized. First it does not
consider the role played by time varying risk premiums in the portfolio
selection process while it does consider a risk free gov-ernment interest rate
as the discount factor of future earnings. More seriously, the inflation
illusion (the possible impact of inflation expectations on the stock market) as
suggested by Modigliani and Cohn (1979) is not taken into consideration.
Secondly, the model assumes the comparability of earning price ratios, a real
quantity, with a nominal, bond induced, interest rate [Campbell and
Vuolteenaho (2004), Asness (2000, 2003), and Ritter and Warr (2002)
discuss these issues.] Consigli, MacLean, Zhao and Ziemba (2009) propose
a stochastic model of equity returns based on an extension of the model
inclusive of a risk premium in which market corrections are endogenously
generated by the bond-stock yield difference. The model accommodates both
cases of prolonged yield deviations leading to a long series of small declines
in the equity market and the case, peculiar of recent speculative bubbles, of a
series of corrections over limited time periods. The inclusion of the yield
differential as a key driver of the market correction process is tested and the
model is validated with market data.
Many of the critics focus on: 1) short term predicability that we know is
weak as does Giot and Petitjean (2008), 2) simply do not focus on the long
run value of the measure, or 3) dismiss it outright because of the nominal
versus real minor flaw as does Montier (2011). Consigli, MacLean, Zhao
and Ziemba (2009) use the model to estimate the current fair value of the
S&P500. Of course, market and fair value can diverge for long periods.
However, our concern is whether or not the model actually predicts stock
market crashes, stock market rallies and good times to be in and out of stock
markets. Berge, Consigli and Ziemba (2008) discuss the latter issue and
found for five countries (US, Germany, Canada, UK and Japan) that the
strategy stay in the market when it is not in the danger zone and move to
cash otherwise provides about double the final wealth with less variance and
a higher Sharpe ratio than a buy and hold strategy during 1975–2005 and
1980–2005. There is some limited predictability of stock market increases
but the evidence supports the good use of the model to predict crashes. In
this paper we study the period 2007-2009 for the US, China and Iceland, all
of which had large crashes. Shiller (2006) observes, as we have in the past,
that low PE periods seem to lead to higher future stock prices and high PE
periods to lower future prices. But the evidence is that PE levels by
themselves are not enough to call the crashes. The argument here is that it is
usually the interplay of interest rates measured by the long bond with the PE
ratios that gives the crash signal.
The Chinese Shanghai Stock Market Crash

Figure 21.2 shows the rise of the Shanghai stock index from January 2000 to
June 2012. The market bottomed at 1011.50 on July 11, 2005. It then rose
six-fold to peak at 6092.06 on October 16, 2007. Then there was a crash of
11.98% from 5180.51 to 4559.75 over the two day period January 21 and 22
followed by another 7.19% fall from 4761.69 to 4419.29 on January 28,
2008. Ultimately the index fell to 1706.70 on November 4, 2008 , a decline
of over two thirds from the peak and 23.09% and 29.93% from the
December 12 and 25, 2006 BSEYD danger signals at 2218.95 and 2435.76.
Table 21.2 discusses highlights for the Shanghai index from 2005–12.
Fig. 21.2 The Shanghai stock exchange composite index, January 2000 to June 2012. Source: Lleo and
Ziemba (2012)
Did the BSEYD model predict this crash? First, Figure 21 shows that the
BSEYD measure is not normally distributed with fat right and even fatter
left tails. The biggest declines are much larger than the biggest increases.
Figures 21.4(a) and 21.4(b) show that the model did in fact predict the
crash. See also the signals hitting the danger level in Figures 21.5 and 21.6.
It is a typical application of the model. The signal goes into the danger zone,
then the market continues higher but within four to twelve months there is a
crash of 10%+ from the value at the initial signal. In this case, the decline is
much higher than 10%. Figure 21.4(a) uses a 95% confidence one sided
moving average interval using prior data out of sample. The danger signal
occurred on December 12, 2006, some ten months before the stock market
peak on October 16, 2007 with the index at 6092.06. Figure 21.4(b) uses a
95% one sided confidence interval and gives the first danger signal on
December 12, 2006 with the index at 2218.95. With a 99% one sided
confidence interval, the danger signal was reached on December 25, 2006
with the index at 2435.76. The ultimate fall was to 1706.70 on November 4,
2008 about 30% below the index value of 2218.76 or 2435.76 of the crash
signal. This signal and decline were a bit different than the usual case as it
took almost two years to get the 10% plus crash and in the meantime the
market almost tripled in value before the ultimate crash.
Table 21.2: Highlights on the Shanghai Stock Index, 2005-2012. Source:

Lleo and Ziemba (2012)
Index
Date Comment
Value
Dec 31, 1266.50
2004
May 23, 1070.84 The index was down 15.4% year-to-date
2005
July 11, 1011.50 Market bottom
2005
Dec 31, 1161.06 The index was down 8.3% in 2005
2005
Dec 12, 2218.95 95% confidence BSEYD crash signal
2006 occurs
Dec 25, 2435.76 99% confidence BSEYD crash signal
2006 occurs
Dec 31, 2675.47 The index was up 130.4% in 2006
2006
May 29, 4334.92 Local high of the market. The market was
2007 up 62% year-to-date
July 5, 3615.87 Local low of the market. The market was
2007 still up 35.1% year-to-date
Oct 16, 6092.06 Highest historical market close. The
2007 market was up 127.7% year-to-date
Nov 28, 4803.39 Local low of the market. The market was
2007 still up 79.5%year-to-date
Dec 31, 5261.56 Index was up 96.7% in 2007, but down
2007 13.6% from its peak in October
Jan 14, 5497.90 Local high of the market
2008
Jan 18, 5180.51 The index closes the week at 5180.51,
2008 which is 5.8% lower than than its local
high on Jan 14
Jan 21, 4914.43 The index experiences a one-day drop of
2008 5.1% from 5180.51 to 4914.43
Jan 22, 4559.75 The index experiences declines by 7.2%
2008 on this day, opening at 4914.43 to close
at 4559.75
Jan 25, 4761.69 The index recovers slightly to close the
2008 week at 4761.69
Jan 28, 4419.29 The index drops by 7.2% from 4761.69 to
2008 4419.29
Apr 18, 3094.67 Local low of the market
2008
May 5, 3761.01 Local high of the market
2008
Nov 4, 1706.70 Global market minimum. The market was
2008 down by 72% peak to trough and 23.09%
and 29.93% from the December 12 and
25 danger signal levels
Dec 31, 1820.81 The market was down 65.4% in 2008
2008
Aug 3, 3471.44 Local high of the market. The market was
2009 up 103.4% from the trough
Aug 31, 2667.75 Local low of the market
2009
Nov 23, 3338.66 Local high of the market
2009
Dec 31, 3277.14 The market was up 80% in 2009
2009
Jul 5, 2363.95 Local low of the market
2010
Dec 31, 2808.08 The market was down 14.3% in 2009
2010
Jun 30, 2762.08 The market was down about 1% year-to-
2011 date
August 2567.34 The market was down about 8.6% year-
31, 2011 to-date
Sept 30, 2359.22 The market was down about 16% year-
2011 to-date
Dec 31, 2199.42 The market was down about 22% in 2011
2011
June 30, 2225.43 The market was up about 1% year-to-
2012 date
Figure 21.4 also illustrates the importance of the confidence level in

relation to the rolling horizon of the moving average and the shape of the
spread distribution. As pointed out in Section 2, a short time horizon of one
year, combined with a lower confidence level of 95% and the non-Gaussian
nature of the spread distribution may result in false positives. An ex-post
analysis reveals that the Gaussian-based 95% confidence spread for the
Shanghai Stock Exchange Composite index over the entire period equals
1.78%. However, the actual 95% confidence level of the empirical
distribution is 2.18%. In fact, a full 9.32% of all actual observations occur at
or above 1.78%, a marked contrast from the 5% predicted by the Gaussian
distribution. Raising the confidence level to 99% or increasing the rolling
horizon does help reduce the impact of the shape of the distribution on the
signal.
Fig. 21.3 Spread distribution of the BSEYD measure on the Shanghai Stock Exchange Composite.
Source: Lleo and Ziemba (2012)
The Iceland stock market crash

Iceland is a small country with only about 300,000 people. From 2002 to
2007, the economy and asset prices rose dramatically with much leveraging
of investments, especially by the banks. This led to high interest rates of
about 10% long term and 16% short term. Eventually, it all collapsed in the
wake of the 2007-2009 worldwide financial crisis. And the decline was a
massive crash of -95% in the equity index and a currency collapse. The
equity index, see Table 21.4, had 15 stocks in it with three of the banks
having very high weighting: Kaupthing (26.5%), Landsbanki (13.0%) and
Glitnir (12.3%) were more than half the market capitalization and Actavis
Banki had 9.9%, and FL Banki another 6.7%. So the banks were close to two
thirds of the index value. And index funds that tracked the market actually
slightly over weight these banks to yield higher returns.
A brief discussion of Iceland

Both Iceland and China had gigantic stock price rises and falls which have a
number of parallels with Japan in 1990. The difference though was that
Iceland, like Ireland, had its economy essentially destroyed, while in China
the impact was and will be much less. The long term effects remain to be
seen. We know Japan has yet to recover!
Fig. 21.4 BSEYD danger signals for the Shanghai Stock Exchange Composite, 95% and 99%
confidence. Source: Lleo and Ziemba (2012)
In the case of Iceland and Ireland (which, along with China, is discussed
in Part III), the entire economy was levered way beyond its capacity to
sustain ever higher real estate and stock prices with loans in foreign
currency. Aliber (2008) describes it well and his prediction of trouble came
to fruition. The Krona was rising until the crash because of its high interest
rates and the apparent excess return from its investment by locals and
especially foreigners. In euro terms, it started 2008 at 90, was 130 on
October 7, 2008 when the stock market was closed, then fell to 340 when
trade was suspended, and then 290 in December 2008. The stock market fell
77% on October 14, 2008 after being closed since October 6th. Before the
crisis, the market cap was 120% of GDP and only 20% after.
Fig. 21.5 BSEYD danger signals for the Shanghai Stock Exchange Composite, Signals at 95% and
99% confidence. Source: Lleo and Ziemba (2012)
The Iceland crash occurred in October 2008. In Ziemba and Ziemba

(2007) as of July 12, 2006 we had the results shown in Table 21.3 for the
bond-stock model with the 16 non-financials in the danger zone but the 15
main stocks which focused on the banks were not in the danger zone until
2008. The rate of increase of property prices was already dropping in mid
2006 but there were no losses yet with the property index up 4.8% in 2006.
Short term interest rates, which were 13% in July 2006 were 14.25% in the
fall of 2006 and went to 14.5% in December 2006, with projection to 16%.
Fig. 21.6 BSEYD Cantelli danger signals for the Shanghai Stock Exchange Composite, Chart.,
Iceland was in a dangerous highly levered position. It was all predicated
on a continual rise in the currency and asset prices with very high interest
rates. They would have eventually had a decline, which might have been
gradual, but the on-slaught from the US and especially the UK economic
troubles accelerated at the time of Lehman Brothers' bankruptcy putting the
problem to a head. Then everyone rushed for the exits when the market was
closed and there was the monumental 77% fall on October 15, 2008. Prices
of the stocks and real estate had simply gotten way to high and all out of
proportion to the real business of Iceland's companies. The economy was
mostly financial services controlled by the big banks which made enormous
profits to keep their PE ratios below the danger level but as profits declined
and interest rates rose, the bond-stock model signaled the crash as did many
other ways of looking at the economic health of the country. It was a highly
over-levered hedge fund with Kelly bets way too high. Aliber (2008) argued
that because of the skewed nature of Iceland's exports of goods and services
- 50% fish related, and the small size of the domestic manufacturing sector,
each increase of 1% of Iceland's exports led to a 1.2%+ increase in the krona
exchange rate. The monies flowing in were borrowed in the wrong
currencies - pound, euro, etc, not in krona, whose assumed continual rise
would make it easier to pay off. So rather than matching currency loan to
repayment, there was the added speculation in currencies. The same mistake
was made by Thailand and other Asian countries in the 1997 currency crisis,
borrowing in dollars rather than yen (again the low interest rate currency).
There the dollar rose while in Iceland, the krona fell.
Fig. 21.7 Spread distribution of the BSEYD measure on the Shanghai Stock Exchange Composite
using Cantelli inequality. Source: Lleo and Ziemba (2012)
Table 21.3: Bond-stock measure calculations in Iceland
index 16 nonfin 15 in index

A) PE ratio 11.1
B) Stock Return 6.13% 9.01%
(1/A)
C) Bond Return (5 11.00% 9.4%
yr)
Crash Signal (C-B) 4.87% 0.39%
Household savings declined as a share of GDP and household

consumption rose based on apparently higher household wealth in stock and
real estate starting from the bottom of the US stock market in March 2003.
The rapid real estate price boom led to a large increase in bank credit.
From 2002-2006 bank assets rose six-fold. In the same period, bank
capital rose eight-fold mainly from gains on their stock holdings and the
ratio of bank assets to the sum of demand and savings deposits rose from
two to five (Aliber, 2008). New foreign currency debt was required to pay
interest on the foreign loans. What was needed but never achieved was a
trade surplus. Eventually the currency and stock and land markets had to
collapse. There were many investors who had purchased Icelandic stocks
with borrowed money that had a negative carry meaning that their interest
payments were larger than their investment income. Some of these investors
had to sell stock in the panic decline. Before the collapse Iceland had
achieved the title of the happiest people on earth - they appeared to have
good achieved the good life with lots of savings and assets but they ignored
the toxic debt on which it was based. It was a classic overlevered, non
diversified situation like the US and UK with high risk taking behavior; see
Chapter 26 showing that this situation invariably leads to a crash. After this
crash, it will be very painful to recover and be a low risk, low income
country.
The Iceland financial crisis continues

The mouse that roared country of 316,000 people in a hikers and spa
paradise seems to owe about $5.3 billion (3.9 billion euro) to some 400,000
Dutch and British investors for losses incurred by Icesave, the online branch
of the Reykjavik bank Landsbanki. There is blame on both sides. The
investors were greedy for the higher than market rates offered in Iceland
which were encouraged by the Icelandic bank. Indeed, short term interest
rates reached 16% in 2007/08. Landsbanki collapsed in October 2008. Of
course, the depositors forgot that extra return usually has extra risk so the
default is understandable.
While the Icelandic people consider themselves “proud people who wish
to shoulder their obligations ...the commitments must be fair, reasonable and
normal” according to Johanna Sigurdottir, the Icelandic prime minister. The
British and Dutch governments have bailed out their greedy investors
through each country's deposit guarantees and now want their money back.
An issue is to lower the 5.5% interest rate. A referendum to pay some
48,000C per Icelandic household (or over 10,000£ per person) was soundly
rejected by over 90% of the people who voted. Since the deal would require
each Icelander to pay about $135 per month for eight years (about 25% of an
average four member family salary) it is clear the burden is too high and had
to be rejected.
Iceland is part of Europe but not in the EU which they would like to join
and they need the Dutch opprovalalongwlth the other26member states. Since
Iceland vows to pay up the result here likely is a bit lower settlement price.
This is another example of the government bailing out Wall Street except the
money will go to pay the losses not line; the pockets of the Landsbanki
executives as was done in the US. Fortunately, up to 90% of the moneyowed
in collateralizedthrough assets salvaged from Landsbanki. 13ut Britain and
the Netherlands want Iceland to provide a sovereign guarantee to pay any
shortfall. Meanwhile, the world's future bankeo, China, is interested in
emerging sea routes through the Arctic to shorten trade routes so this might
help the Iceland economy, see Ward (2010). But Russia is keen to keep
Iceland out of the EU as it regards Iceland asα fellow Arctic country.
The economic and political saga will continue. For more on Iceland as the
crisis unfolded, see two chapters in Ziemba and Ziemba (2007). Ziemba and
Ziemba looked into predicting this crash with the bond-stock model and
when the book went to press in late 2007 the model was close but quite to
the danger zone, but it got there in 2008 well before the; actual collapse.
Before the financial crisis oS fall 2008 hit Iceland the stock market
capitalization was about 120% of the country's GDP now it is 20%.
Table 21.4: Stock market index. Source: Glitnir (2006)
Figure 21.8 shows the dramatic rise of the stock market particularly since
2003 and, similarly, Figure 21.9 shows how quickly the crash occurred.
However, the notable sharp sell-offs, were, to a large extent, blips before the
big crash and there was a question whether these investments could continue
to produce similar returns, and if not, whether that would prompt investors to
seek other markets.
Fig. 21.8 The 15 stocks in the Iceland equity index and their growth in real terms from 1997-2006.
Source: Glitnir (2006)
Fig. 21.9 OMX Iceland all share - price index, 1993 to November 2010. Source: Lleo and Ziemba
(2012)
Figure 21.10 shows that in the fall of 2007, the long bond interest rates did
get above 10% and that with an increase in the PE ratio respectively to 10.91
for Glitnir as of October 10, 11.09 for Kaupthing as of October 11 and 9.94
for Landsbanki as of October 17 the BSEYD spread signal did predict the
massive crash. The market peaked at 8174.28 on July 18, 2007 starting in
2002 at 1180.75. Then it fell to 5803.55 by the end of December 2007. The
complete collapse came in 2008 with the market falling about 90% to end
2008 at 581.76 and about -95% at the ultimate bottom on February 2, 1010
at 491.58. As of August 31, 2011 the market had fallen to 596.58. Table 21.5
summarizes the year by year story and Figure 21.9 shows the index values
from 1998 to September 30, 2011.
Figure 21.11 shows the BSEYD spreads for the three top banks and
Figures 21 and 21.13 show the bond-stock earnings yield crash measure
from June 2004 to December 2008 for the three largest banks. Like China,
the BSEYD distributions of these three largest banks are not normally
distributed and have very fat left tails.
Fig. 21.10 Iceland Treasury note (long bond) maturing May 17, 2013. Source: Lleo and Ziemba
(2012)
Fig. 21.11 BSEYD Spread Distributions, Iceland. Source: Lleo and Ziemba (2012)
Finally, the question of whether or not the bond-stock earnings yield

model predicted the crash is studied in Figures 2lace which use 95% one
sided confidence intervals using moving averages. These graphs show that
the crash was predicted. For Kaupthing, the danger zone was penetrated on
September 28, 2007, two months after the July 18 peak and less than a
month before the November 11 crash. For Glitnir, the signal was much
earlier on October 17, 2006, some thirteen months before the crash. Finally,
for Landsbanki, the danger signal was on February 13, 2007. Figures 21b,d,f
show the BSEYD using Cantelli's inequality to account for the non-
normality of the BSEYD measure. Figure 21.13 shows the signal dates. We
focus on the largest banks because they led the market into the collapse and
they are a majority of the index weighting. The smallest cap stocks in the
index were in the danger zone in 2006 but not the large banks and the overall
index was not in the danger zone then as discussed in Ziemba and Ziemba
(2007).
Table 21.5: Highlights for the Iceland stock index ICEXI-OMZ 1998-2012
with start value = 1000 on December 31, 1997. Source: Lleo and Ziemba
(2012)
Date Index Value Comment
Feb 11, 1998 977.58
Dec 30, 1998 1046.58
Dec 30, 1999 1511.86 Gain of 44.5% in 1999
Dec 28, 2000 1303.31 }
Dec 28, 2001 1180.75 } Weak market during
US stock market weak
period
Dec 30, 2002 1436.22 }
Dec 29, 2003 2064.05 Gain of 43.7% in 2003
Dec 30, 2004 3173.91 Gain of 53.8% in 2004
Dec 30, 2005 5107.49 Gain of 60.9% in 2005
Feb 2,2006 6287.29
Local high of market
Aug 2, 2006 4854.95 Local low of market
Dec 30, 2006 5857.50 Gain of 14.7% in 2006
July 18, 2007 8174.28 Global highest historical
market close
Aug 16, 2007 6931.69 Market falls to local lows
on day of US stock
market turmoil when
long-short funds had
heavy losses
Oct 11, 2007 7796.36 Local market peak then
complete market
collapse
Dec 28, 2007 5803.35 Market down just slightly
(-0.9%) in 2007 but 29%
below the July 18, 2007
peak
Dec 30, 2008 581.76 Complete depression
including a collapse of
over 90% from the peak
Dec 30, 2009 496.48 Further fall in 2009
Feb 2, 2010 491.58 Global historical bottom
of market
Nov 30, 2010 579.17 The market rallied back
to make the yearly
return nearly positive
Dec 30, 2010 569.19 Gain of 15.8% since
global historical bottom
but the index was still
93.0% below its
historical high
Jan 31, 2011 624.97 Gain of 9.8% in January
2011
Jun 30, 2011 604.95 Gain of 23.0% since the
92.5% below its
historical high
Sept 30, 2011 596.58 Gain of 17.3% since the
92.9% below its
historical high
Dec 31, 2011 567.77 Gain of 15% since the
93% below its historical
high
June 30, 2012 677.75 Gain of 38% since the
92% below its historical
high
Comments on the crash signals for the three banks from Lleo and
Ziemba (2012)
Glitnir:
• Signal from September 15, 2003 until December 23, 2003: on October
12, 2004, the share price reached 11.90. By November 2, 2004, the
share had gone down to 10, a 15.97% drop.
• Faint signal from November 17, 2004 until November 19, 2004 and
December 6, 2004: on February 18, 2005, the share price fell to 11.60.
By March 3, 2005, the share had gone down to 11.60, a 7.20% drop.
Fig. 21.12 BSEYD Crash Indicators, Iceland. Source: Lleo and Ziemba (2012)
• Signal from November 22, 2005 until February 24, 2006 (with some
interruptions): on October 12, 2004, the share price reached 22.60. By
April 19, 2006, the share had gone down to 16.50, a 26.99% drop.
• Signal from October 17, 2006 (with PE ratio of 22.30) until December
27, 2006: announces the market crash.
• Signal from December 28, 2007 until January 8, 2008.
• Signal from March 26, 2008 until April 10, 2008.
Fig. 21.12 (Continued)
The crash signal analysis is basically according to the script except for the
faint signal between November 17, 2004 until November 19, 2004 and
December 6, 2004 for which we do not have a better explanation.
Kaupthing:
• Signal from February 7, 2006 until February 24, 2006: on October 26,
2006, the share price reached 868. By November 28, 2006, the share
had fallen to 785, a 9.56% drop.
• The market reaches its peak on July 18, 2007 and a crash occurred on
November 11, 2007.
• Signal from September 28, 2007 until November 5, 2007.
• Signal from March 26, 2008 until April 4, 2008.
Landsbanki:
• Signal from January 15, 2004 until March 4, 2004: on October 11,
2004, the share price reached 15.28. By November 2, 2004, the share
has fallen to 10.85, a 28.98% drop.
• Signal from August 30, 2005 until December 29, 2005 (with minor
interruptions): on February 16, 2006, the share price reached 30.56. By
May 3, 2006, the share has fallen to 20.05, a 34.39% drop.
• Signal from February 13, 2007 until March 13, 2007 and from March
30, 2007 until the June 28, 2007: market crash signal.
• The market reaches its peak on October 17, 2007.
Fig. 21.13 Crash Indicators, Iceland. Source: Lleo and Ziemba (2012)
The crash model works out well for Landsbanki: the signal identifies the
market crashes and two large declines. Although the signal could be clearer
for Glitnir and Kaupthing, we do not observe any false positives.
The US 2007-2009 crash

In this section we investigate whether or not the bond-stock measure did
predict the US 2007-2009 crash, including the devastating September 2008
to March 2009 period. Table 21.7 shows the BSEYD at various periods from
2006 to 2011. In 2006 the weak economy led the Fed to dramatically reduce
short term interest rates which tended to drop the long term rates that we use
in the BSEYD calculations.

Table 21.6 considers the measure in five major countries on July 12, 2006.
None of these major markets were in the danger zone then. Table 21.7 for
the S&P500 has the calculations for the years from 2006 to 2011 on trailing
price-earnings ratios that are usually used for these calculations and 2011
which uses Shiller's average inflation-adjusted earnings from the previous
ten years. Figures 21.15ab show the S&P500 PE ratios (Shiller's calculation
method) and the ten-year Treasury bond yield.
Table 21.6: Long-bond (10 yr) versus earning yield differentials for major countries, July
12, 2006. Source: Ziemba and Ziemba (2007)
Ex post it is clear that this stock market crash had a lot of components
such as the first decline in aggregate US housing prices in more than thirty
years, a subprime market collapsing because home buyers could not cover
their mortgages, lots of suspect AAA rated packages of these mortgages and
then a credit squeeze with much counter party risk with firms unwilling to
lend money to others including supposedly sound financial institutions and
the collapse of many large and previously sound financial institutions such
as Bear Stearns, Freddie Mac, Fannie May and, the killer for the market,
Lehman Brothers.
Table 21.7: BSEYD model calculations leading up to the 2007-2009 crisis in the
S&P500. Source: Lleo and Ziemba (2012)
Table 21.8 lists some of the main events regarding the S&P500 from 2006
to 2011. There are numerous books concerning this period plus many articles
and columns. Ziemba has several in Wilmott that are discussed in this book.
Starting in June 2007, he designed strategies and traded for an offshore BVI
based hedge fund for a group headed by a top trader whose his hedge funds
had investments in Bear Stearns and in June 2007 asked for his money back.
That took three months and gave him a strong signal of danger. As an astute
trader, he hedged and studied carefully the market situation through
technical indicators that he has developed. Ziemba remembers his words to
him starting in the summer of 2007 “this is the big one” ...“eventually the
market will go to 660 on the S&P500”. In the fall of 2007 the S&P500 was
about 1500, see Figure 25.3. So this was a rather bold call but a private one
and it turned out to be very accurate. Nouriel Roubini was predicting very
boldly a serious financial meltdown starting in 2006 when the housing
market was beginning its decline. He and other bears such as Yale Professor
Robert Shiller are still (September 2012) pessimistic about the economy, real
estate and financial markets. Dropping real estate has several depressive
effects such as homeowners can no longer use house price gains to fund
consumption, foreclosures, etc. The March 2009 low closing was 676.53
with an intraday low of 666.79 on March 6. The subsequent rally doubled
the S&P500 to 1320.64 as of the end of June 2011 and 1362.16 to the end of
June 2012 and to 1440.67 on September 30, 2012. There is considerable
discussion regarding whether or not this rally is low interest rate related to
the Fed quantitative easing, or only game in town since real estate, bonds
and cash look unattractive. This is a case when the BSEYD signaled the rise
in stock prices. A volatile period followed the rally, with a drop in the level
of the S&P500 in August and September followed by a a stabilization in
October.
Table 21.8: Highlights of the S&P500, January 1, 2006 to June 30, 2012.
Date Index Value Comment

December 31, 1248.29
2005
August 31, 2006 1303.82 Gain of 4.4% year-to-
date
December 31, 1418.30 Gain of 13.6% in 2006
2006
February 26, 1437.50 Local high of market
2007
March 1, 2007 1374.12 Local low of market
June 7, 2007 1480.72 Bear Stearns suspends
redemptions from one of
its hedge funds
June 14, 2007 1522.97 BSEYD crash signal
occurs
July 13, 2007 1552.50 Local high of market
July 31, 2007 1455.27 Bear Stearns liquidates
two hedge funds
August 16, 2007 1411.27 Local low of market
October 9, 2007 1565.15 Market peak. Gain of
10.4% year-to-date
December 31, 1468.35 Gain of 3.5% in 2007
2007
March 17, 2008 1276.60 Local low of market
September 15, 1192.70 Lehman Brothers files
2008 for bankruptcy protection
September 30, 1166.36 Market down by 20.6%
2008 year-to-date
October 10, 2008 899.22 Local low of market
October 31, 2008 968.75 Market down by 16.9%
in October
December 31, 903.25 Market down by 38.5%
2008 in 2008
March 9, 2009 683.38 Lowest intraday: 666.79
on March 6, 2009
March 9, 2009 676.53 Market trough. The
market was down by
56.8% peak to trough
July 28, 2009 979.62 Date Shepherd claimed
that the S&P500 had a
723 PE ratio based on
reported (real earnings)
to the SEC on 10Q
forms
December 31, 1115.10 Market was down by
2009 23.5% in 2009, but up
by 64.8% since trough
April 23, 2010 1217.28 Local high
July 2, 2010 1022.58 Local low
December 31, 1257.64 Gain of 12.8% in 2010,
2010 and of 23% since the
local low of July 2, 2010
May 10, 2011 1357.16 Local high. Gain of
7.91% year-to-date and
of 32.72% since the
local low of July 2, 2010
August 31, 2011
1218.89 Loss of 3.1%
year-to-date and of
10.2% since the local
high of May 10, 2010
Sept 30, 2011 1131.42 Loss of 10.0% year-to-
date and of 16.6% since
the local high of May 10,
2010
Dec 31, 2011 1257.60 The index ends the year
flat, but is still down
7.3^% from the local
high of May 10, 2010
June 30, 2012 1362.16 ain of 8.3% year-to-date.
Did the BSEYD model predict the US crash? Figures 21.16 and 21.17
show that it did on June 14, 2007. As with China and Iceland, the BSEYD
measure is not normally distributed but rather has fat tails especially on the
downside as shown in Figure 21.18. The 95% confidence graphs with the
crash danger signal is shown in Figure 21.16(a). The Cantelli's inequality
version of the model, Figure 21.16(b), gives the danger signal on the same
day, namely, June 14, 2007, see also Figures 21.17ab.
Let's go back to the BSEYD and whether or not it called the September
2008 to March 2009 crash. Figure 21.19 shows the S&P500 during the
2007-2009 period. Consider Table 21.9 which was published in the Maudlin
weekly newsletter which has 1.5 million subscribers in May 2009. It shows
how estimated earnings are dramatically falling. In March 2007, earnings for
2008 were forecast to be 92, during the course of 2008 the forecast was
constantly revised downward. On July 25, 2008, the S&P500 earnings for
2008 were forecast to be 72.00 with the S&P500 at 1257.76 which gives a
PE ratio of 17.47 which is not high enough to signal the September 2008 to
March 2009 crash. But by February 20, 2009, the 2008 earnings were
estimated to be only 26.23. With the S&P500 at 770.05 on that day the
trailing PE ratio was 29.36 which gives a BSEYD value of 2.78-(100/29.36)
= -0.626. This estimate was further revised downward in April 2009 to
14.88. A similar pattern is seen for the 2009 forecasts.
Fig. 21.14 The S&P500, January 1, 2006 to April 30, 2011
Fig. 21.15 S&P500 and ten-year Treasury bond yields. Source: Robert Shiller data
Table 21.9: Earnings revisions for 2008 and 2009, analysts estimates of earnings in
dollars. Source: Maudlin (2009)
Fig. 21.16 Crash Indicators, US. Source: Lleo and Ziemba (2012)
Fig. 21.17 US crash signal occurs on June 14, 2007. Source: Lleo and Ziemba (2012)
Shepherd (2009) has the S&P500 PE ratio at 723 on July 28, 2009 four
months into the rally that began in March 2009! The S&P500 was then
979.62, up nearly 50% from the March lows. This high PE ratio was based
on reported real earnings from SEC 10Q filings. So what do we conclude
here? Our conclusion is in Figure 21.20. The BSEYD model did not give
any additional sell signals during 2008. The signal was on June 14, 2007
with the index at 1522.97 and the crash occurred in various phases with
closing peak of 1565.15 on October 9, 2007 and a closing low of 676.53 on
March 9, 2009, down some 56.8% from the peak. The conclusions are
similar: over the period January 1, 1995 to April 6, 2011, two BSEYD crash
signals occurred: one in June 1999 and the other in June 2007.
Fig. 21.18 Spread distribution of the BSEYD measure on the S&P500. Source: Lleo and Ziemba
(2012)
Fig. 21.19 The S&P500, January 1, 2007 to December 31, 2009. Source: Lleo and Ziemba (2012)
Logarithmic Model
Koivu, Pennanen and Ziemba (2005) use a logarithmic version of the
BSEYD model. The question is does that model suggest anything new for
our analysis? The logarithmic model is based on the Fed model bond stock
yield ratio
For China and Iceland, both measures produce similar results which are
available from the authors. The pattern and timing of the crash signals nearly
coincide for all three Icelandic banks. In China, the lnBSEYR generates a
slightly earlier signal than the BSEYD.
In the US, the result of the lnBSEYR and of the BSEYD measures are
also broadly similar. A signal precedes both the internet-related crash of
2000 and the credit crunch crash of 2008. In addition, the logarithmic
lnBSYR(t) measure generates a signal in April 1998, ahead of a 19% decline
from July 17th to August 31st. However, neither measure predicted the
market decline of 2002. This is a combined result of the relatively low level
reached by the two measures in 2001 compared to 1999, and of the increase
in the confidence level starting in 2000.
Fig. 21.20 Crash Indicator (95% confidence): S&P500. Source: Lleo and Ziemba (2012)
Final remarks
The bond stock earnings yield model has been shown to be useful in a
number of contexts. First, using it for being in or out of the market over long
investment periods has been shown to produce about double the returns of
buy and hold with lower risk in five major countries as shown in the next
chapter. Secondly, over the years, the model has predicted many significant
stock market crashes such as those in China, Iceland and the US during
2007-2009. Finally it has possible, but less clear, use concerning when to re-
enter markets after a crash.
1Edited from Wilmott, November 2009.

2Using a different index rather than the S&P500 has the same conclusion but slightly different results.
Berge, Consigli and Ziemba (2008) used the MSCI index. The danger zone was entered in May 1987
and the correction occurred in October, four months later. During June, July and August investors kept
rebalancing their portfolios from the bond to the equity market (MSCI TRI + 13.87% over the quarter)
then the equity market fell 31.80% in the following quarter (September to November 1987) with the
major decline in October.
3The rest of this chapter is a summary of the results in Lleo and Ziemba (2012).
Three Mini Crashes in US and World Equity
Markets1
Chapters 2 and 21 discussed two crash risk measures that are very effective
for anticipating large 10%+ declines in the S&P500 and other stock indices.
These measures can help investors assess risks and minimize the effects of
such crashes. Here we briefly re-review these two measures and explore
three declines in the US and world markets that were not predicted by these
measures. From this background, we try to draw out lessons for predicting
and responding to such shocks in a variety of markets.
The two measures are:
bond-stock which compares long bond versus trailing stock yields, and
T-option a measure of market confidence sentiment related to puts versus
calls prices
The three declines not explained by the two measures are:
(1) September 11, 2001,
(2) May to June 2006, and
(3) February 27 to April 2007
There have been additional small corrections including several 6-9%
declines from July to September 2004 and March to June 2005. In an April
9 2007 Barron's article, Michael Santoli notes that there has been one such
pullback each year since 2004. In each case ,a recovery quickly followed
each decline and each retreat has been shallower than the preceding one and
a faster recovery of the loss. Buying on the declines has been rewarded as
bidders try to beat the crowd and speed up the recovery. Buying on these
dips has worked so far, as has selling put options during the greatly
expanded volatility which returned to low levels after the decline. There is
evidence that the 2007 decline is following; this pattern. As of April 6 2007,
the futures market returned to pre-decline levels with the VIX at 13.23, so
this 2007 decline may well be nearly over.
Some background on crash, measures

Historically, the bond-stock crash measure has been successful in predicting
10%+ market corrections, including the declines in October 1987 (U.S. and
Japan), the 1990 Japan, the 2000 U.S. and the 2002 U.S.. In Japan from
1948-1988, there were twenty 10%+ declines even though the market went
up 221 times in yen (and 550 times in US dollars). The bond-stock measure
had a 12/12 record in predicting crashes in that period, that is, whenever the
measure was in the danger' zone, there was a fall of 10%%+ within one
year from the time the measure went inSo the danger zone. This is a very
good forecasting record, but eight declines in Japan during these 40 years
were not predicted by this measure. Berge, Consigli and Ziemba (2008)
present an analysis of this measure in five equity markets (US, Japan, UK,
Germany and Canada) from 1980-2005 and 19)75-2005. See Chapter 2.
Srnilarly, the short term confidential T-measuse WTZIMI, a Canadian
investment company, uses in trading; has good short term predictability (3-6
month.it). Since 19855 the T-measure has had T < 0, that is, in the danger
zone, for the S&P500, six times. These are summarized in Table 22.1.
Table 22.1: Results of the six times out of 85 quarters the T-measure was negative for-
the S&P500
T is negative when the market is over confident as measured by relative
put and call option prices. Then it is very dangerous since there are no
sellers and only buyers but some sellers will usually appear to drive the
market down. Of these six T < 0 occurrences, we have the October 1987
crash and the 3Q2002 crash when the S&P500 fell 229% in this quarter.
The S&P bias trade is not done when T < 0; see Table 22.1. Otherwise the
S&P bias trade of WTZIMI is very successful when T > 0. From 1985,
there were no losses for T > 100, as it has been since 3Q2003.
In the six quarters when T < 0, there were four losses and two times the
measure did not predict correctly. Still the sum of the six returns yielded a
combined arithmetic loss of -41.7%. Hence, these two measures are useful
but they do not predict all 10%+ crashes nor do they predict some small
declines. For these two declines under 10% and the September 11, 2001
14% decline, other reasons must be found, which we will consider below.
There have been no T < 0 signals since the third quarter of 2003, even
though the VIX has hit the 10 area a few times.
Declines and crashes not predicted by the measures

Although the measures have a good record, there are some key episodes
which it did not predict. Studying these declines and their triggers helps us
to assess shocks. The September 11, 2001 attacks and the stock market
decline of 14% in the S&P500 that followed after a one week market
closure was largely a random, that is unforeseen event. But the size of the
decline was exacerbated due to the then weak stock market and U.S.
economy which had a recession starting in spring 2001; see Figure 19.3 a,
b. The stock market was weak because although prices had fallen, earnings
had fallen more. The bond-stock model which had been in the danger zone
in April 1999, predicting the April 2000 decline, then returned to the danger
zone in the fall of 2001 predicting the 22% fall in the S&P500 in 2002, see
Figure 19.3. The T-measure for 3Q2002 at -142.8 predicted the 12% fall in
the S&P500 that quarter; see Table 22.1.
The S&P500 fell 37% from 1460.25 at the end of December 1999 to
885.76 on 31 October 2002.
In May to June 2006 the S&P500 fell 7% and markets in some emerging
economies fell 20% or more. Worries that valuations of some emerging
market stocks were too high enlarged their losses. For example, the closed
end emerging market fund RNE (Russian New Europe) was at a very high
37%+ premium on May 10 to net asset value - an amount way above
historical values. The trigger for the decline was a rumor that the Bank of
Japan would raise interest rates. These higher interest rates did not
materialize but the fear that they would spark a rally in the yen led some
yen carry trade players to unwind their short yen, long higher yielding non-
Japanese asset positions, especially emerging market currencies. In turn this
led to sales of various stocks and indices including the S&P500 and the
decline was largest (elements of mean reversion) in those areas that had
gained the most, namely the emerging markets. The VIX volatility index,
see Figure 22.2 rose from around 10% before the crisis to the 22% level
before dropping back to 10% after the sell-off.
The third decline is February 27 through April 2007 which is still in
progress (VIX-wise) as we write this although the early April S&P500 is
well above its February 27 and March 13 lows of 1399.04 and 1377.95,
respectively. For months, there has been talk of the current period being the
longest time without a 2% decline in one day or a large monthly decline.
The stock market had low volatility since the 2006 decline; see Figure 22.2
which plots the VIX for the last five years and the last 17 years.
Fig. 22.1 The Fed Model, January 1980 to May 2003. Source: Ziemba (2003)
Fig. 22.2 VIX. Source: yahoo finance
The paper of Lleo and Ziemba (2012), which discusses the bond-stock
earnings yield predictions of three market crashes in the 2006-09 period,
namely China, Iceland and the US, had a clear sell signal on June 14, 2007
and a huge decline occurred this discussed in chapter 21.
An example of these sentiments was made by Bob Stovall, a 50-year
Wall Street veteran, in a talk on November 15, 2006 to investment students
at Stetson University. Stovall argued that given the current real economic
growth in 2006, it would be very difficult for stocks in the S&P500 to
continue to increase in price (Moffatt, 2006). Without economic growth,
companies would have trouble meeting earnings expectations. He said that
the average bull market in US history lasts approximately 56 months. At
that time, the US was approaching the 50th month of the bull market. He
concluded that a shift toward steady cash flow stocks with dividends was
preferable to large capital gains stocks.
It is known that long periods of stock market gains tend to start with low
PE ratios and end with high PE ratios as shown in Table 22.2 which lists
nine 20-year periods with gains,
Table 22.2: Nine 20-year periods of gains beginning low PE and ending high PE.
On Tuesday, February 27, 2007, the S&P500 fell 50.33 points or 3.47%
to 1399.04. On that drop, the VIX volatility index rose from 11.15 to
18.31%, a jump of 64.22%; see Figure 22.2. Several concurrent triggers
have been mentioned for the fall, which were exacerbated by the confused
reaction of market participants to these events. The first was a 9% fall in the
Shanghai and Shenzhen stock markets, itself triggered by rumors that the
Chinese government was going to raise the bank's reserve requirement and
make regulatory changes to slow speculative activity in the soaring Chinese
equity markets. The Chinese market drop triggered substantial sell-offs in
Asia (where most equity markets were near peaks) and Europe as well as in
the U.S. While we believe that China is one of the most interesting financial
market to study now, we also feel that this drop is not the underlying cause
of the S&P persistent weakness. The Shanghai index was up more than
100% in 2006 and way up in early 2007 so the 9% fall is a minor blip in the
long run growth trend, and likely motivated by profit-taking and a concern
about over-valuation. Furthermore, it is not an indication of slowing of the
Chinese economy which is expected to grow around 9-10% in 2007.
Chinese markets tend to be quite volatile. Several weeks before the
February 27 decline, the Shanghai exchange fell 11% in one week in early
February, a decline which received little attention in international markets
because it was spread over a week (Vincent, 2007). These two declines
were the greatest since February 1997, when news of Deng Xiaoping's ill
health triggered a sell-off. Thus, the Chinese decline may have determined
the timing of the global equity decline, and return of increasing volatility
and risk aversion but is not the underlying cause. Indeed in early April 2007
the Chinese market indices rose to new highs well above the February 25
interim high.
We expect that just like Japan, whose Nikkei stock average rose 221
times in yen and 550 times in US dollars from 1948 to 1988, but with 20
declines of 10%+, China will likely have higher gains in dollars than RMB,
be overpriced like Japan, propelled up by fast growth and low interest rates
and high liquidity and still experience many corrections. See Ziemba and
Schwartz (1991, 1992) and Stone and Ziemba (1993) regarding Japan and
Lleo and Ziemba (2012) regarding the crash in 2008 of the Shanghai index.
The BSEYD crash signal was in December 2006. But the market did not
understand this. Rather many market actors tend to react as a herd to such
events, seeking to minimize their losses, but the more recent response
appears to be buy on dips. See our primer on Chinese investment markets
below for more description of Chinese markets.
Other news also contributed to the fall in the S&P500 and worldwide
markets Tuesday included a statement by former Fed Chair Alan Greenspan
that a recession in the US was a possibility although it was not probable as
well as some weak economic numbers.
Greenspan later said the probability of a recession was 25%. At the time,
bond prices were actually estimating a higher probability. We assume, that
even though it might be wise for a former Fed chair to let the current Fed
chair do the talking, audiences like the one in Hong Kong, require that
Greenspan say something interesting to earn his $150,000 speaking fee.
The decline was exacerbated by a large unwinding of yen carry trades
who sold stock and created a short covering rally in the yen that moved the
USD/JPY exchange rate from 127 to 116; see Figure 25.8 on the yen dollar
rate from March 30, 2006 to March 30, 2007. However, it appears that those
who foresaw the end of the yen carry trade spoke too soon. Although the
Bank of Japan recently doubled interest rates (in February 2007), the
benchmark rate of 0.5% remains far below other interest rates - encouraging
Japanese retail and institutional investors to continue to seek higher returns
abroad, and foreign investors to use the weak yen as a financing currency -
even if Morgan Stanley recently argued that yen- denominated loans to
retail investors remain very small (Morgan Stanley 2007).
Accentuating the tension were political as well as economic risks. Many
commentators such as Lawrence Summers (former Treasury Secretary and
Harvard President and current DE Shaw hedge fund consultant) have
argued that the market was not pricing in the worldwide risks in most assets
including the S&P500. However, both of WTZ's large crash 10%+
measures were not in the danger zone. The decline in February 27 to early
April, and possibly beyond, had not reached a 10% fall and the VIX which
reached 19% was bouncing around the 13-16% range most of the time.
We turn to the Eurasia group's list of top seven political risks for 2007 for
an assessment of some of the geopolitical risks and their possible effect on
the global economy. The risks as reported by Ian Bremmer are as follows:
(Bremmer, 2007)
(1) Iran
(2) Nigeria
Fig. 22.3 Daily yen, one year from March 30, 2006 to March 30, 2007
(3) Iraq
(4) Turkey
(5) Russia
(6) China
(7) Afghanistan/Pakistan
The majority of these risks are challenges of political transition and
succession that could impact energy supply (Russia, Nigeria, Iran), regional
power plays in the Middle East (Iran, Iraq, Turkey), the war on terror
(Afghanistan/Pakistan, Iraq) or involve challengers to US dominance in the
global politics and economy (Chinese succession, Russia).
The top risk, how to respond to Iran's nuclear ambitions and the potential
impact of a military escalation on asset prices, has regained increasing
urgency in recent weeks, following the March 23, 2007 Iranian capture of
British hostages. Oil prices rose after their escalation of the Iran crisis has
again lifted oil prices, which were slow to fall considerably. Although the
US does not import oil from Iran and the majority of its energy imports
originate from Canada and Mexico, Iran exports significant oil to American
allies in Asia and has the potential to block oil from the southern shore of
the gulf. Furthermore, even though the hostages were released, the crisis
raised a series of questions about Iran's nuclear program, influence in Iraq
and the role of the US-UK alliance.
The Eurasia Group also isolated four longer term risks that will challenge
policy makers and investors for years to come: pandemic influenza,
terrorism,resource nationalization, and protectionism. Many of these longer-
term risks are present in their 2007 issues to watch. See their February 2,
2007 report on these longer term risks.
Rumors are moving these nervous markets. The rumor that failed
Amaranth trader Brian Hunter (see Herbst-Baylis, 2007 and others) was
going to manage a commodities volatility fund in a new series of funds for
Solego Capital of Calgary, Alberta and Greenwich, Connecticut, moved the
natural gas and related calendar spread trades in the direction Hunter is
known to favor. That is, natural gas prices higher in winter and lower in
summer. Amaranth collapsed because Hunter greatly overbet and his
weather forecasts turned out to be wrong and he foolishly doubled his
position. The March 28, 2007 rumor of an Iran attack on a US warship
moved the dollar and oil prices. This rumor caused oil to spike to $5 to $68
and then fall quickly to $64. U.S. and global markets are very reactive to
any negative news about the US economy and the fears that it might spark a
global slowdown. Other fears include the fallout of subprime housing loan
defaults and its effect on consumer credit and the drop in consumer
confidence which fell from 111.2 in February to 107.2 in March and
declining housing sales, starts and prices. WTZ will stick to his two crash
measures and Buffett's measure of the ratio of stock value to GDP for the
big crashes but nervousness can easily drop prices in the short-term and
investors need to be hedged against such reversals and have plenty of cash
available to weather such storms. Also the S&P500 earnings are projected
to rise only 3.8% following 4Q2006, the first below 10% rise since 2003;
and only 6.7% in 2007 versus 16% in 2006.
We close this chapter with some background on Chinese investment
markets. Given that China's growth, demand for commodities will continue
to influence global asset prices.
A primer on Chinese investment markets
Investing in China requires a leap of faith

The cycles will be much wider and more frequent in China because
of the lack of information. Having said that, if you're investing, you
should put a fairly large part of your total assets in China because
within as short a period as 30 years, China is likely to have the
largest gross national product any nation has ever had.
John Templeton, April 1, 2004 in Smart Money
Opportunities for private Chinese and foreign investors are expanding,
making it more critical to assess what makes for a good investment. The
critical issues to consider are past good investments and determining what
looks good for the future, including broader global trends that affect and are
affected by policies in China.
Other limits to investment include ownership regulations, e.g., what
items can be bought by foreigners especially related to stock issues and
types and nature of joint ventures. The government still owns all the land so
property is leased, though the most recent party conference approved
changes which would recognize some private property. With a shortage of
apartments and a continuing move to urban areas, there is constant upward
pressure on housing prices. Higher interest rates (though current rates
remain low in real terms) and a slowdown pose the greatest danger to
investment.
There are several options for foreigners who would like to directly or
indirectly invest in China; any of these options would require in-depth
research to maximize the investment. Indirect investment includes investing
in China-related stocks, companies that do business with China and/or will
profit from China's long term economic growth. There are many companies
that would fit that description. Natural resource companies, or producers of
raw materials or other intermediate inputs for Chinese goods, including
energy, agriculture, minerals and parts. Other alternatives include investing
in China-related funds, which include mutual funds, closed-end funds,
hedge funds and/or offshore funds, focusing on China. There are new funds
opening daily. There are still many opportunities in China, but savvy
investors will have to do their research to mitigate against the asymmetry of
information and make sure that each investment is sound. There are cases of
expropriation of foreign investments.
Of course, risks remain: country risk factors in China and abroad
(political in-stability, interest rates, changing regulations), which could
negatively impact the value of the companies. In addition, investors must
prepare for currency risk between the country of investment and their home
countries since depreciation of the currency may decrease returns.
Fluctuation in resource prices is another source of uncertainty. To take
advantage, one might invest in countries that are politically stable,
economically open, and that have a currency appreciation prospect (or at
least are likely to remain stable), or else to hedge against the potential risk
factors involved.
It is difficult to assess the information regarding the Chinese markets.
John Templeton about sums it up, we don't really know exactly the returns
expected, or in what sectors they are most likely to come but we are
convinced that being invested in the long haul is important. This was
essentially the same feeling expressed by Jim Rogers in his books
Adventure Capitalist (2003) and Hot Commodities (2005). He mentions
buying B-shares when they fell out of favor following the 1997 Asian crash
and he expects to just hold on for the long term. However, his advice is not
to invest indiscriminately but to do as much research as possible on
investments - a policy he follows all over the world. He is convinced in
China's importance enough to that his first daughter is tutored in Mandarin
and he moved to Singapore to be near the Asian markets! It is clear that
China will become an even more important player in the world economy
and political arena.
China's financial markets are dynamic. Recently there have been a
growing number of initial public offerings, which have been
oversubscribed. The largest of these was the ICBC IPO in 2006, which has
since become the third largest global bank, and who recently reported the
highest earning in 2006 (Bloomberg). At the same time listed companies
have increased in number and in quality, contributing to better performing
indices. This has coincided with increased demand from Chinese investors
and massive loan and credit growth.
Part of the high demand for Chinese stocks is explained by the relative
lack of investment vehicles for domestic investors along with pent-up
demand from foreign investors, very few of whom are currently eligible to
invest in the Chinese equity markets. The Chinese financial sector remains
quite underdeveloped for its market capitalization. although the extreme
liquidity in China and the global economy provides ample (perhaps too
ample) financing, the corporate bond market is very small, with most
companies receiving funds from government banks whose financing
decisions may be more political and based on business plans (Gale, 2007).
Fig. 22.4 Chinese stock markets, 2006-3-31 to 2007-1-28. Source: Ruoen (2007)
High earners propelled the Shanghai and Shenzhen stock markets up by

140% from February 2006 to the end of January 2007. See Figure 22.4. The
public is very active in buying mutual funds and is redeploying savings
account money into what they hope will be higher yielding equities. The
lack of capital gains tax added fuel to the rally. The introduction of the
qualified foreign investor program as discussed above has eased the way for
some foreign investors, and added slightly to the liquidity of the exchange,
but only a small proportion of investors have such access. One of the
triggers for the stock rally on February 28, the day after the sell-off, was an
announcement that the government would increase the proportion such
investors could purchase.
Across China, tales of riches are captivating the country and
enticing millions. In- vestors opened an estimated 50,000 retail
brokerage accounts a day in December. A new mutual fund from a
Beijing-based money manager partly owned by Germany's Deutsche
Bank AG raised the equivalent of $5.1 billion in its December
launch, a new record.
Measured by the price earnings ratios, the valuations are not cheap and
are well above those in most other countries. The end of December 2006 PE
ratios on the Shanghai stock exchange were 37.57 for the Chinese traded A-
shares and 28.15 for the hard currency B-shares traded by foreigners. The
Shenzhen stock exchange had a similar PE ratio for the A-shares (37.46)
but the B shares were lower at 23.2.
The B-share market is still small and illiquid with an average daily
volume in 2006 of about US$65 million. Its purpose to provide foreign
investors with access to Chinese equities was usurped by the Hong Kong H-
shares which have a market capitalization of over US$800 billion versus
US$19 billion for the B-shares. Rumors suggest that the B-shares might be
merged with the A-shares. As of December 2006, there were 109 listed B-
shares of which 85 also have yuan-denominated A-shares. Since the A-
shares trade at a discount of about 30%, they could be bought back. But
Chinese companies historically prefer to invest in growth opportunities
rather than in buybacks. Moreover, they would have to raise cash on the A-
share market in competition with other companies trying to expand. The
conversion from B's to A's would require compensation, another
undesirable.
However, continuing growth does make some of these equities
vulnerable. As of January 2007, there were increasing fears that the stock
markets might be overvalued - and markets likely to be volatile. As any
conversion of B's to As could be complex, it may not occur.
Fig. 22.5 Long bond yield to maturity of Treasury bonds in Shanghai
stock exchange, 2002-3-31 to 2007-1-28. Source: Ruoen (2007)
Figure 22.5 shows the long bond (5-year) Treasury bond rates on the
Shanghai stock exchange from 2002 to early 2007. Since these rates are
currently about 2.9%, the bond-stock model is riot in the danger zone in
China despite these high PE ratios. But the trend of Chinese interest rates is
higher so the situation must be watched carefully The measure is about 2.9
– (100/37.57) = 0.24, which is below the danger level. Since Chinese
interest rates have historically been below those in the US, the danger zone
is well below the about 3% long bond minus stock earnings yield danger
level for the US. Chapter 21 shows that the Shanghai stock exchange did
enter the danger zone on December 12 or 25, 2006 with the index of
2218.95 (95% confidence signal) and 2435.76 (99% confidence signal). It
then rose to 6092.06 on October 16, 2007 and subsequently fell to 1706.70
on November 4, 2008, a drop of 23.09% and 29.93% from the initial signals
on December 12 and 25, 2006.
Table 22.3: Large declines in the Shanghai stock exchange. Source: Burton
(2007)
Fig. 22.6 The Shanghai A price index vs the Hong Kong H index, May 2006 to February 9, 2007
China is now more comparable to Japan in the early 1980s rather than the
US.2 In Japan from 1948 to 1988, the stock market rose 221 times in yen
and 550 times in dollars, yet had 20 declines of 10% or more; see Ziemba
and Schwartz (1991). Historically, the Shanghai market has has had many
large declines with daily changes as high as the-16.39% on 23 May 1995.
The 50 best and largest listed companies had a dramatic rise of over 40% in
late December 2006 to end January 2007 with a -3.9% fall on 25 January
2006. See Figure 22.4 and Table 22.3 as described above. So with growth
rates of more than 10% and a wide and growing trade surplus and a world
awash in liquidity, the Chinese markets may well mirror Japan in the 1980s
with a large but bumpy ride up to higher levels. Eventually there will be
movement towards the Hong Kong H share level, see Figure 22.6.
In 1988-89 at the peak, Japan had PE ratios in the range of 60 until the
interest rates rose in mid 1989 into August 1990 to crash the market.
According to calculations by Professor Ren Ruoen of Beihang
University, Beijing, there is price parity at an exchange rate of about 5 yuan
per dollar. So the slight drift upward in the yuan of 7% that we have seen
from July 2005 to March 2007 might bring the yuan to this level in about
ten years, however it is possible that officials may speed this process.
The nationalisation of global financial flows evidenced by the increase
of Asian reserves to sterilize inflows and due to the large trade and current
account surpluses of many Asian countries (as well as the willingness of
Asian countries to buy U.S. assets) has created a huge liquidity problem
which so far has resulted in concentration of assets in low yielding
treasuries. A number of countries manage their exchange rates and therefore
must hold on to growing reserves as they attempt to neutralize inflows; this
concentration helps to explain the conundrum of the insatiable demand for
bonds, even at very low yields and the lack of demand for equities, even at
low valuations.
So far, China's reserves, growing by nearly $200 billion a year, have been
managed by the State Administration for Foreign Exchange, a department
of the central bank, and invested almost entirely in government bonds and
other low-yielding presumably risk-free assets. (See Setser and Rosenblatt,
2006, for more details on the composition and growth of China's reserve
assets.) Recently China announced a new department, the State Foreign
Exchange Investment Corporation, which will manage a portion of its more
than $1 trillion of foreign exchange reserves more actively. As of March
2012 these reserves were in the $3 trillion area. This agency will be led by a
former Finance Ministry vice chair, with representation from the central
bank and report to the State Council. Its mandate is to diversify into various
assets including non-Chinese equities, property and direct investments
abroad. Initial estimates indicate that $200-250 billion will be available for
such investment, financed through the sale of RMB bonds which will have
the dual role of soaking up liquidity. This policy change in China, while still
concentrating assets in the hands of the government, will likely generate a
major change in global markets, including the relative valuations of bonds
and equities and other asset classes, including possibly energy and metals.
The Bush administration had for the first time in many years imposed a
coun-tervailing duty on imported paper from China, the first time in 23
years that the US has imposed such duties in response to subsidies of a non-
market economy. See Ziemba and Schwartz (1992) for an economic
analysis suggesting that if the Chinese impose the tariffs on exports or
impose quotas they could better gain the economic rent, but if the U.S.
does, then they give the money to them. This action spells trouble on the
horizon, but is a reflection of the growing U.S. political pressure and
concern about the size of the U.S. trade deficit with China and the range of
tools with which U.S. policymakers will employ. As such, such actions may
have significant impacts on U.S. assets and U.S. and Chinese equity
markets. This could include dollar declines and interest rate rises to
maintain capital inflows.
1Edited from Wilmott, May 2007.

2In Japan's early development period, foreign exchange reserves were controlled by the MITI. These
were given out to support development of the economy. Soon firms were able to hold onto their
foreign earnings and reinvest them according to their own plans. We know that some of these were
spent on trophy purchases; see Ziemba and Schwartz (1992).
What Signals Worked and What Did Not, 1980—
20091
Here and in the next two chapters we review various prediction signals and how they performed for
various asset classes, with the main focus being on equity markets
This chapter and the next two review various prediction signals and how
they preformed for various asset classes but focusing on the equity markets.
In many but not all cases the signals such as the bond-stock earnings yield
differential, my T- measure of relative put and call options prices, Buffett's
stock market to GDP measure, the January first five days and all of January
indicator, sell in May and go away, and the VIX volatility index were very
useful and accurate in predicting subsequent market declines and rises. Also
some short term anomaly indicators such as options expiry, turn-of-the-
month and year, holidays, etc. have had predictive value.
The real estate (sub-prime and other), credit , confidence and economy
wide collapse of 2007-2009 with the vast number of toxic derivatives has led
to a very complex market to do well in. There is a bond bubble in
supposedly ‘safe’ government bonds with extraordinary low interest rates
over long periods. At the same time equities have fallen dramatically
worldwide with trillions lost around the globe. Many hedge funds, pensions,
endowments, and large trading groups have suffered huge losses.
Governments around the world have provided bailout funds and have
bought toxic assets and have become de facto large hedge funds with the
ability to print money. But they have done a very poor job with their funds
allocated with little or any safeguards to prevent inappropriate use and it
seems not to have been targeted well towards the real problems like the
millions in the US with mortgages well above their house value. Rather it
seems to be a bailout of Wall Street and even that has been of dubious value.
A major problem is the government takes the losses and Wall Street takes the
gains. But if you pour enough water even if it largely misses the fire, the fire
will eventually go out. The Fed and US government balance sheets and
promises have risen from $800 billion to $10 trillion. The real question is if
the massive deficits and debt, can increase productive capacity and
productivity, if not they will eventually lead to inflation and a weaker US
dollar.
Postscript: As of September 30, 2012, it has not.
During President Obama's term (2008-2012), the total US debt is
projected to be increased by more than all 43 previous US presidents
including the substantial debt from George W Bush's administration from the
two wars and instead of raising taxes to pay for them actually lowering taxes
on the wealthy. The current debt increases will be the result of dealing with
economic woes. My travels in this part of the world (Iran, Afghanistan)
advised me long ago to just leave them alone. In total that would be much
less costly in money, lives and stress. Of course, some dissidents need to be
dealt with and the country must be protected but that could have been done
much less costly. The cost per soldier in Afghanistan is estimated at $1
million per year. So the extra 30,000 troops Obama sent there as requested
by General Patreauscost about $30 billion per year. Wow what a waste!
Is there a danger that the U.S. or U.K. could go under like Iceland did?
The free fall of the British pound is a big danger signal here.
Recessions from financial crises are known to last about twice as long as
other recessions so the GDP might fall up to 5% especially in the U.K. So
we are left with a synopsis of what the future might bring and possible future
scenarios.
Postscript: This prediction has proven correct as in 2012 the UK is in a
very bad economic state.
There are many interesting questions to consider such as:
• Are the successful investment models, like the Yale endowment of
Swenson, Berkshire Hathaway, long term index investing of CAPM
based finance, now obsolete or will the past reappear as in previous
downturns?
• Are Siegel's investing for the long run and those ideas tilted by seasonal
and fundamental anomalies and factor models, obsolete now?
• Are the great investors still great?
• Where are the monies left actually being invested and where should
they be?
• How should asset-liability models be carried out?
• Can there be a Japanese 1990s style lost generation?
• When will the U.S. housing market stabalize and recover and will this
restore confidence in the equity markets and economy to spur buying to
push the GDP higher?
Postscript: Some of these issues are updated in Chapter 34 on What's
Wrong with the US?
We start with something clever.

How to turn a supposedly bad trade into a winner assuming:
(1) you have a long horizon; and
(2) you have enough capital to weather storms, that is, this trade is not a
large percentage of your total capital and you have a large amount of
liquid assets
Historically Berkshire Hathaway, the fund run by Warren Buffett, has had
very high mean returns, but not in 2007-9; see Figure 23.1. From 1977-2000,
Berkshire's geometric mean return was 32%.
Fig. 23.1 Historic high Berkshire Hathaway returns, 1985-2009
The wealth levels from December 1985 to April 2000 for the Windsor
Fund of George Neff, the Ford Foundation, the Tiger Fund of Julian
Robertson, the Quantum Fund of George Soros and Berkshire Hathaway as
well as the S&P500 total return index are compared in Figure 6.1.
Berkshire Hathaway's returns have been very poor in 2007-2009 but it still
has beaten the S&P500, see Figure 23.2. The S&P500 closed at a bottom of
676 on March 9, 2009 and has recovered now to the 830 area as I write this
in early April 2009. Berkshire B shares2 have recovered to the 3000 level,
still well below its 5000 peak in 2007.
Shorting S&P500 puts in a declining market

Warren Buffett, when the S&P500, now about 830, was much higher, sold
15-18 year puts on the S&P500 at the money with strikes at various levels
but well over 1000. He collected about $4.5 billion from insurers who
wanted to guarantee for certain clients no losses over a long horizon. These
puts have a value now of over $10 billion but they are not tradeable.
Fig. 23.2 Berkshire Hathaway vs the S&P500, 1997-2009
Their fear is justified because the S&P500 had had four long periods with
no gains except for dividends and only three periods with gains since 1900.
We will return to this question below.
How likely are these puts to expire worthless with the S&P going back to
the 2007 highs?
Figure 28.5 shows some calculations that estimate this. Recall that the
2000 high in the 1500s was retouched in 2007 and in both cases, 2000 →
2003 and 2007 → 2009, the S&P500 halved. So these curves are suggestive
but not definitive of what might happen. They are based on past data starting
from a drop in prices. A major difference in the 2007-2009 decline versus
the 2000-2003 decline is that in the current drop, virtually all stocks have
fallen in price whereas in the earlier decline, very few stocks fell but they
were the very large cap and telecommunications and technology which
comprise a large percent of the index value. Indeed, an equally weighted
S&P500 index would not have fallen much. For details on 2000-2003, see
Ziemba (2003). Hence a return to the old highs in 2007 may be difficult in a
similarly short time. But as of the end of March 2012, the S&P500 had made
it to the 1405 area.
But BH has the use of the $4.5 Billion so why not lend it out at a high
coupon rate? They did to GE and Goldman Sachs at 10% in preferred stock
with some free call options at much higher strikes than the current prices
thrown in as part of the deal. Since assuming that the premiums can be
reinvested at the same rate, which is reasonable since other deals have even
higher coupon rates.
$4.5 billion(1.10)15 = $18.8 billion
So with his deep pockets, BH is likely to come out of this with a profit
and has the very real possibility that the puts will expire worthless; they are
already deep out of the money as of March 2012.
Fig. 23.3 Cumulative probability of regaining index highs (as of 23 January 2009). Source: Dimson,
Marsh, Staunton (2012)
Equities : Relative stock market sizes

Figure 23.4 shows the changing relative sizes of world stock markets 1899
to 2011. Japan peaked at about 44% of the world at the end of 1989 and
since has lost about 80% of its value.
Fig. 23.4 Changing relative size of stock market around the world. Source: Dimson, Marsh, Staunton
(2012)
Figure 23.5 shows the cumulative real returns among US asset classes
from 1899 to 2009. equities, bonds and bills from 1899 to 2009. The real
returns for equities, bond and bill averaged 6% plus 1.7% capital gains and
dividends, 2.2% and 1.0%, respectively. Figure 23.6 shows the real equity
returns in three periods: 2000-2008, 1950-1999 and 1900-2008 for 18
countries plus the US.
Fig. 23.5 Cumulative real returns, US. Source: Dimson, Marsh, Staunton (2009)
Fig. 23.6 Real equity returns. Source: Dimson, Marsh, Staunton (2009)
Equities have generated superior total real and nominal returns in the long
run; see Table 23.1 for the period December 1925-December 1998. The table
shows the total return of equities in the 73 year period from December 31,
1925 to December 31, 1998. Portfolio allocations based on risk aversion are
in Table 23.2. See Kallberg and Ziemba (1983) for the theory behind these
calculations of the portfolios based on risk aversion. For those with very
long horizons, a high amount of equities is suggested. A new key issue then
is could one be in stocks when they are good and bonds or cash when stocks
fall? We will discuss the bond-stock measure approach for this purpose.
Table 23.1: Equities have generated superior returns in the long run, December 1925-De-
cember 1998. Source: Ibbotson, 1999 in Swensen (2000)
Asset Class Multiple

Inflation 9 times
Treasury bills 15 times
44
Treasury bonds times
Corporate bonds 61 times
Large-capitalization stocks 2,351 times
Small-capitalization stocks 5,117 times
Table 23.2: Portfolio allocation: percentage of portfolio recommended in

stocks based on all historical data (RA=risk aversion index). Source: Siegel
(2002) and Ziemba (2003)
The trend is up but its quite bumpy; see Figure 23.7. There have been four
periods in the US markets where equities had essentially zero gains in
nominal terms, 1899 to 1919, 1929 to 1954, 1964 to 1981 and 1997 to 2009.
Indeed there are only four periods in which there were gains (without
dividends): 1919-1929, 1954- 1984, 1981-1997 and 2009-2012. Siegel
(2008) shows that in all 20 year periods from 1900 to 2003 stocks with
dividends included beat bonds.
US unemployment is likely to reach the 10% area in late 2009 or early
2010 then improve, see Figure 23.8. It is already above 10% in California.
The US unemployed plus underemployed rate in April 2009 is 15.6% and
Vice President Biden predicted that there will be job losses in all months of
2009. This is well below the depression levels of over 25% but quite high by
current standards.
Fig. 23.7 The trend is up but it is quite bumpy. Source: Siegel (2002)
Fig. 23.8 US unemployment. Source: Dimson, Marsh, Staunton (2009)
Gold coins
The World Gold Council has confirmed a shortages of coins. Q4:2008
Investors in Europe and North America bought 148.5 tonnes of gold coins
and bars, a jump of 811% compared with Q4: 2007 which pushed the global
retail investment up almost 400% to 304.2 tonnes. Retail investors in France,
for example, became net buyers of gold for the first time in a quarter of a
century at the end of 2008.
Investment inflows into gold exchange traded funds (ETFs) reached 94.7
tonnes, up 18%, but down from the record 150 tonnes Q3:2008.
The SPDR Gold Trust holdings reached 1,008.8 tonnes, up by 228.6
tonnes becoming the world's 7th holder of gold bullion, having absorbed
about 10% of global annual mine output in the past seven weeks. Meanwhile
jewellery and industrial is weak, down 5.5% to 538.9 tonnes and down
10.4% to 98.6 tonnes and rising prices have encouraged an increase in gold
scrap, up 15% to 320 tonnes. (FT Feb 21, 2009).
Bubbles?
Even when traders in an asset market know the value of the asset,
bubbles form dependably. Bubbles can arise when some agents buy
not on fundamental value, but on price trend or momentum. If
momentum traders have more liquidity, they can sustain a bubble
longer. (Gjerstad and Smith, 2009)
The bubble in housing prices fueled by subprime loans, securitization and
low interest, pushed housing costs above the long term average of about
30% of income in a time when income disparity between rich and poor was
rising and ordinary house income was at best stable. Clearly this was
unsustainable and the bubble would have to burst.
Housing has had two other bubbles in 1976-79 and 1986-89 but during
those periods, Fed policy was leaning against the wind and thus able to
moderate them but this time it was creating the wind with Fed funds rates
reaching their lowest level since 1955 when the rate series began (Gjerstad
and Smith, 2009).
The increase in housing costs was outside the CPI so not recognized as
inflation. A number of analysts have noted an similarity of the events of the
past 10 years to the period leading up to the Great Depression. Most
significantly was the increase in mortgage debt which went from $9.35
billion in 1920 to $29.44 billion in 1929 representing a increase in share of
household wealth from 10.2% to 27.2% in that period.
The current situation quickly deteriorated when the cost to insure
mortgage backed securities exploded from $50,000 plus a $9,000 annual
premium for $10 million of insurance to over $900,000 (plus the annual
premium) (Gjerstad and Smith, 2009). And then the trouble quickly spread.
Gjerstand and Smith compare the two crashes
(1) December 1999 to September 2002 wiped out $10 trillion in assets and
caused no damage to the financial system
(2) 2007-2009 wiped out $3 trillion and had devastated the financial
system
In the first, the declining assets were held by institutional and individual
in-vestors that either owned the assets outright, or held only a small fraction
on margin, so losses were absorbed by their owners. In the second, the
declining housing assets were purchased on margin with mortgages of 90%
and 100%. In some cases, borrowers who had purchased at low levels
remortgaged and then saw their home value decline 50% or more. As the
bubble burst, millions of homes became worth less than the loans on them.
The original lenders did not hold onto the mortgages but they were
repackaged and resold so the huge losses were transmitted to lending
institutions, investment banks, investors in mortgage-backed securities,
sellers of credit default swaps, and the insurer of last resort, the US Treasury.
See Figures 23.9 and 23.10.
Fig. 23.9 The FED and the bubble. Source: Gjerstad and Smith (2009)
Fig. 23.10 The bubble bursts. Source: Gjerstad and Smith (2009)
Gjerstand and Smith hypothesize that a financial crisis originating in

consumer debt, especially that concentrated at the low end of the wealth and
income distribution, can be transmitted quickly and forcefully into the
financial system. Therefore we are in the midst of the second great consumer
debt crash after a massive consumption binge fueled by easy mortgage credit
(Gjerstad and Smith, 2009).
1 Edited from Wilmott, May 2009.

2B shares are 1/30th of A shares then were split 50-1, so they are 1/500th as large. They are equivalent
except that they have no voting rights.
What Signals Worked and What Did Not, 1980-
2009, Part II1
Continuing the series of articles reviewing prediction signals in current market conditions
This chapter is the second in a series of three columns that review various
prediction signals and how they preformed for various asset classes but
focusing on the equity markets. In many but not all cases the signals such as
the bond-stock earnings yield differential, my T- measure of relative put and
call options prices, Buffett's stock market to GDP measure, the January first
five days and all o January indicator, sell in May and go away, and the VIX
volatility index were very useful and accurate in predicting subsequent
market dechnes and rises. Also some s hort term anomaly indicators such
asoptions expiry, turn-of-the-month and year, holidays , etc. have had
predictive value.
As I write this in June 2009 the S&P500 has had a slow but steady climb
from its March 666 iow to the 940 area. Several times the market has
reached 950 only to be pushed back each time. But the sense is that there is
less bad news, a so-called second derivative effect, and uince the stock
market is supposed to predict; six months ahead, it is rising to forecast better
times. Nobel laureate Paul Krugman and others have suggested that the US
recession (beginning signaled by two consecutive quarters of negative GDP
growth) will end in the fall of 2009. George Soros said in late June that the
worst is over. But the recovery will be slow and painful as Joe Stiglitz and
Nouriel Roubini predict and as articulated in the recent Maudlin (2009b)
column. Savings rates have increased dramatically which, while cutting into
spending, is a harbinger of a new normal as also discussed by El-Erian
(2008). Oil prices have doubled since their bottom of $32 in early 2009 and
are much higher, over $100, in March 2012. The current $72 is about
halfway back to the summer 2008 high of $147 per barrel. Also, as a sign of
better growth is the steepening of the yield curve with the 10-year T-bond
now near 4% versus 3% a month or so ago. The graphs in Figure 24.1 show
this progress.
Fig. 24.1Indicators of improving economic environment
The usually reliable bond-stock yield model was of great help in 2006-9
for the Iceland, Chinese and US markets. In Ziemba and Ziemba (2007) we
studied these latter two markets and concluded when we went to press in the
fall of 2007 that they were close to the danger zone thus predicting a stock
market crash but not quite. But in 2008 a further rise in 5-year bond rates
(the long bond in both of these countries) plus a drop in earnings led to the
danger signal and the subsequent crash. Figures 24.2a,b show this. This
analysis, which is based on Lleo and Ziemba (2012), is discussed in Chapter
21.
In the US there was a sell signal as of June 14, 2007 based on high interest
rates relative to trailing stock earnings. The ensuing 2007-2009 crash was
credit and confidence related due to a massive build up of derivatives
including those based on toxic assets in real estate. The Fed lowered the
interest rates and the long bond rate became artificially low. The endogenous
creation of liquidity wiped out the efforts of the Fed to control the interest
rates and thereby the money supply.
Since June 14, 2007 the BSEYD measure has not signaled an additional
crash. The measure must be about +3 to be in the danger zone. That would
take a huge increase in the 10-year bond rate plus a big PE expansion (higher
stock prices and/or greatly lower earnings). Neither seems likely. Even now
with earnings dropping dramatically the measure is still not in the danger
zone.
Fig. 24.2 Stock price indices for Iceland and China, January 2000 to June 19 (2009)
In late 2008 and the first few months of 2009, S&P500 earnings and
forecasts for 2008 and 2009 were continually dropping. Table 21.9 from
Maudlin (2009a) shows this dramatic decline. Even with these low earnings,
the model was still not in the danger zone but we did have the June 14
danger signal which presaged these declines. Interest rates have dropped
dramatically with short term rates near zero. However access to these low
posted rates is not readily available. It is the liquidity crisis that has created a
real interest rate that is dramatically high and approaching infinity as credit
for many is totally unavailable, credit card companies are denying previous
credit limits and recalling credit cards.
What went wrong? Indeed we have been in a period of declining interest
rates. The decades before the crash and the crash itself were transitional
economic times. While consumption spending is normally a large part of
GDP, it had become even more significant as people withdrew equity from
their homes, treating them as ATMs. This both fueled the economy and at
the same time planted the seeds for the crash as clearly this level of spending
was unsustainable especially once housing prices began to soften. During the
same period, there was a rapid growth in derivative products that created a
huge pool of liquidity, again, unsustainable. The way out of this crisis will be
a return to more normal debt instruments that sustain the real economy. Let's
look at the history of this crisis.
The subprime crisis and how it evolved

“Let's hope we are all wealthy and retired by the time this house of
cards falters.” –Internal email, Wall Street, 12/15/06
In 2004 an estimated $900 billion dollars was withdrawn from home

equity through refinancing.2
In the days following September 11, 2001, with the attacks on US soil and
the markets very weak, Fed chairman Alan Greenspan said he was extremely
worried about the after effects on the US economy. So five day later, when
the stock market reopened, the first of a number of interest rate cuts was
made. In 2002 President George W Bush said “The goal is, everybody who
wants to own a home has got a shot at doing so.” He also referred to the
homeownership gap that “three-quarters of Anglos own their homes, and yet
less than 50% of African Americans and Hispanics own homes.” (at a
speech at HUD, June 18, 2002). At the same time he linked home ownership
to national security.
Freddie Mac and Fanny Mae created the secondary mortgage market and
between them insured about half the mortgages. Originally these had been
made under strict qualifying procedures but they came under pressure by the
industry and government policy to loosen their standards. Orange county
entrepreneurs wanted a way to circumvent these rules and create a profitable
business that was unregulated. They invented the concept of subprime
mortgages where anyone could get a loan at a time when Freddie and Fanny
were in some trouble. Actual incomes and assets were not checked and
largely inflated. Bad credit and no assets (or a lot of debt and liabilities) did
not matter. What made this work was a great interest from Wall Street firms
to package these mortgages and have them AAA rated so they were
investment grade. Then the Wall Street firms could sell these CMOs
(collateralized mortgage obligations) around the world. The rating agencies
were paid by the firms selling the CMOs not the purchasers. The rating
agencies were eager to have the business and the repeat business. Since it
was assumed that house prices could not fall - they had not fallen since
1991-2 — this seemed safe. Around the world, investors, a bit greedy to get
higher returns were sucked into buying these securities. One example is
Narvik, Norway, a small town 150 miles above the Arctic circle. They
bought enough of these assets from a representative of Citi Bank through an
Oslo representative to lose most of the town's assets.
Meanwhile, house prices roared higher and higher around the world far
outstrip-ping income growth. Buyers with no money were able to buy houses
then refinance them and cash out the gains to upgrade their homes or just to
spend the money. Indeed a huge percent of US consumption in 2003-2007
came from this source. Houses were assumed to rise in value by 6-8% yearly
forever. But a bubble was forming and house prices in the US peaked in
2005/2006.
The packaging of the mortgages into AAA rated CMOs and later CDO
(collateralized debt obligations which include any asset with a future income
stream) continued.
From 2002-2006 as a result of the housing bubble so many speculators
gained by buying extra houses on margin. In 2007-2009 the declines in the
US hurt such speculators hard and many went into receivership. Indeed over
10 million houses in the US in January 2009 were under water in the sense
that their mortgages exceed their current market value.
For the period December 1, 2007 to November 30, 2008 prices in the 20
areas fell a record 18.2% with November 2008 adding a 2.2% decline. The
housing market continues to suffer from a large supply of unsold homes,
tighter lending standards and a record number of foreclosures. The 10 metro
regions also fell 2.2% in November for a yearly drop of 19.1%. The
composite 10 and 20 metro regions peaked in mid 2006 and since then (to
Feb 2009) have fallen 32% and 30%, respectively.
Areas that had large increases had large falls. This includes many cities in
California, Nevada and Florida. From March 2008 to March 2009, for
example, San Francisco fell 43%. There were similar drops in San Jose and
other areas in California.
The housing price declines have left more than 20% of US homeowners
owing more on their mortgages than their houses are worth by the end of
Q1:2009. That represents 20.4 million households, up from 16.3 million in
Q4:2008. That is 21.9% of all homeowners, up from 17.6% Q4:2008 and
14.3% Q3:2008. On the one hand, the falling home prices are making
housing more affordable for first-time buyers and other who have had
difficulty getting into the market. On the other hand, the fall in home equity
has cut off the ability of homeowners to use their homes like an ATM as
refinancing is harder so they cannot take advantage of the low interest rates.
The regions with the highest percentage of homes underwater are shown in
Table 24.1.
Table 24.1: Metropolitan regions with the highest percentage of homes with negative
equity in Q1:2009. Source: Simon and Hagerty (2009)
Region % Underwater
Las Vegas, NV 67.67
Stockton, CA 51.51
Modesta, CA 50.50
Reno, NV 48.48
Vallejo-Fairfield, CA 46.46
Merced, CA 44.44
Port St Lucie, FL 43.43
Riverside, CA 42.42
Phoenix, AZ 41.41
Orlando, FL 41.41
US average 21.9
With such a large number of households underwater, it will be hard to get

a consumer led recovery. In the UK, the declines are similar, with the year
on year values down about 20% for high end properties in London during
2008-9.
The lending organizations sold off the mortgages and they were cut and
diced and bundled into packages like CMOs and CDOs and sold to others
who had trouble figuring out what is in them but look at the rating agency's
stamp of approval. A triple A rating was desirable for sales of these
derivative securities.
Figure 24.3, starting in 1890, shows the buildup to overpriced areas in
2004- 5 that led to the drop now that is shown in Figure 20.1b. There have
been 12 consecutive months of negative returns. The 10-city, 20-city decline
and 10-city composite all declined. Case-Shiller and others predict up to a
25-35% drop in prices from the peak in 2005-6, see Figure 24.3.
Fig. 24.3 A history of home values. Source: Nouriel Roubini (2006)
Business was good. Even pizza deliverers became, with no training,

mortgage brokers. There was no license, so no training, involved. Once they
started arranging the mortgages, they quickly began earning $20,000/month
and they soon were buying expensive cars. One Southern California
Lebanese immigrant with a 3rd grade education had a firm selling Mercedes
to his loan officers. At the peak in 2005-6, he was making $5 million/month.
He had money to burn. An example was a movie featuring his girlfriend in
which two Ferrari's, one worth $1/2 million and the other worth a full $1
million, were destroyed as part of the filming. When prices of houses and
real estate stocks fell starting in late 2005, the defaults multiplied, the CMOs
and CDOs dropped sharply in value. One hedge fund trader, David Bass in
Texas, saw this coming and made 600% on his investment, some $1 billion,
by buying insurance on these instruments which rose sharply in value and
was piad off as the house prices fell. Of course, John Paulson was a big
player in these and related markets in his various funds, making 1000% in
2007; see Table 24.2 for some 2008 results.
Another factor fueling this in 2004 was Greenspan saying that the market
needs “new products for mortgage loans”. These included adjustable rate
mortgages with low or no interest payable in the first year or two with the
interest added to the loan value. Then with higher interest, higher loans and
declining house values, the situation became more difficult and led to
millions of mortgage defaults. This destroyed the American dream of
owning a home with other people's money.
Table 24.2: Hedge Funds, January to September 2008. Source: Bloomberg

Greenspan still insists that such bubbles are just a part of human behavior
and will happen again and again and there was nothing the Fed could have
done to prevent it. And it would be bad politics to stop home ownership. He
admits now that he was shocked when he learned that 20% of all US
mortgages were subprime and that he, with some math and economic
training and a staff of 200 with many PhDs could not understand many of
the CDO products which made use of option experts trained at leading math
finance and other departments. Wow!!
Yet the issue was that there was a gap in regulations and application of
prudence in lending. In Canada and many other countries, you cannot get
these extreme subprime mortgages and consequently there have not been
such a fall in house prices nor as many defaults. Also in Canada unlike the
US non-recourse loans, borrowers are at risk on all their assets not just the
property that's being mortgaged.
Household and government debt

While US house prices surged from 2000 to 2007, household debt was also
surging. Household debt went from 60% of disposable income (after tax) in
1985 to 80% in the early 1990s and soared to 120% in 2007. In the years
from 2001 to 2004 about 40% rewrote their mortgages, 25% extracted equity
in the process. The under 30s and the over 63s extracted lower rates of
equity (15% and 18% respectively). The funds were used for consumption
(10.5%), payment of other debt (23.5%), home improvement (32.2%), and
investment including stock market (33.8%). In sum, the value of primary
residences increased $4,164 billion, and $783 billion was extracted from
equity and $267 billion went into consumption. House values increased
another $6.4 trillion from 2004 to 2006, if the same ratios held then about
$410 billion went into consumption. For the households that extracted equity
and then consumed it, their net worth did not increase, but when the bubble
burst they have lost net worth as their assets have declined in value while
their debt has increased. Up to 2008 those workers near retirement that
remortgaged and extracted wealth had lost 14% of net worth just from this
shift, in addition they likely lost a lot on their retirement savings. They will
have the hardest time recovering retirement savings (Munnell and Soto,
2008).
The US and UK households have very high debt compared to disposable
income which has been steadily rising from 1990 to 2008. This is at the heart
of the housing declines in both countries. Canada and the euro zone have
much less debt which is partly a result of much tighter standards for
mortgages and other lending. Banks in the US and the UK basically would
lend money to anyone for real estate transactions given their false forecast
that prices would continue to rise. Then the decline in real estate values had
a much bigger effect in these countries. Table 24.3 shows the government
debt as a percentage of GD in 2008. Japan and Italy have the highest debt
ratios. But the citizens of Japan have large savings which tempers the risk
there. Italy like the UK is in serious financial trouble. The US has one big
advantage with its government debt - in US dollars - in very high demand
around the world so their constant printing of money, while dangerous, is
less so than other countries who have debt in other currencies and thus must
earn foreign currency to repay it.
Table 24.3: G7 debt to GDP ratios, 2008. Source: Globe and Mail, January 22 (2009)
Canada 22%
Britain 33%
France 36%
Germany 43%
U.S. 46%
Italy 87%
Japan 88%
Some 44% of US households were participating in the financial markets

in 2007, up from 29% in 1994 representing 88 million individual investors.
More than half of these investors are 45 years old or older, and a third of
this group (approximately 17.6 million people) are older than 65, so they
have limited opportunities to earn their back to their retirement savings
given the 2007-2009 declines of about 50% in equity markets.
Favoring the financial sector: evaluating the policy responses

In the last 25 years, the deregulated finance sector grew as the real
production sector was in the decline in the US and in the UK. Profits came to
be concentrated in this sector, and indeed it was very innovate with
securitization, interest-rate swaps and credit default swaps among other
instruments. The effect of this can be seen for the growth in the share of
corporate profits going to the financial sector. From 1973 to 1985, this sector
earned about 16% of the corporate profits. In the 1990s their profit share
ranged from 21% to 31% and in the most recent decade this escalated to
41% of all corporate profits. Concomitant with this increase in profits came
rising incomes. From 1948 to 1982 average compensation in this sector was
about average for the economy between 99% and 108% of the average for
all domestic private industry. But by 2007 it reached 181% (see Johnson,
2009).
In the global economic crisis there have been several phases and various
responses by the US Federal Reserve, the US Treasury Department and the
Federal government and similar bodies in the UK and elsewhere. So far to
early April 2009, these policy responses of monetary easing (open market
operations now referred to as quantitative easing) and fiscal spending have
had some success but that has been limited. Unfortunately, the policy
response has to a large part been to continue to favor finance over real
production. Instead of nationalizing the banks and cleaning them up, money
has been allocated to them to shore them up.
In part this is a reflection of the structure of the Fed, the US central bank.
The seven member board of governors is appointed by the president with the
approval of the Senate. The boards of the twelve independently incorporated
regional banks are composed of three members appointed by the Fed board
and six elected by the member banks. So the chairman of, say, the NY Fed
owes the position to the banks in the region and routinely consults with
them. In May, 2007 in a speech to the Atlanta Fed, Treasury Secretary
Timothy Geithner said that “the financial innovations had improved the
capacity to measure and manage risk” and that “the larger global financial
institutions are generally stronger in terms of capital relative to risk” (quoted
in Becker and Morgenson, 2009). At this point, New Century Financial had
already filed for bankruptcy due to sub-prime losses and by July Fed chair
Ben Bernanke warned that the US sub-prime crisis could cost up to $100
billion.
Geithner, encouraged by Citigroup and JPMorgan Chase, was proposing
new looser standards for the banks. The problem, according to Callum
McCarthy, a former British regulator, was that “banks overestimated their
ability to manage risk, and we believed them” (Becker and Morgenson,
2009).
Nobel Laureate and Columbia University Professor Joseph Stiglitz among
other economists has expressed the concern that this relationship has led to a
regulatory philosophy shaped by and shared with the industry itself. This led
a bailout that was designed to get a lot of money into the banks to shore
them up without necessarily considering the risks to the public at large
(Becker and Morgenson, 2009).
A variety of regulatory changes have been proposed by economists,
politicians, journalists, and business leaders to minimize the impact of the
current crisis and prevent recurrence. However, as of April 2009, many of
the proposed solutions have not yet been implemented. These include (from
Wikipedia) the following excellent suggestions:
• Ben Bernanke: Establish resolution procedures for closing troubled
financial institutions in the shadow banking system, such as investment
banks and hedge funds.
• Joseph Stiglitz: Restrict the leverage that financial institutions can
assume. Require executive compensation to be more related to long-term
performance. Re-instate the separation of commercial (depository) and
investment banking established by the Glass-Steagall Act in 1933 and
repealed in 1999 by the Gramm-Leach-Bliley Act.
• Simon Johnson: Break-up institutions that are “too big to fail” to limit
systemic risk.
• Paul Krugman: Regulate institutions that “act like banks” similarly to
banks.
• Alan Greenspan: Banks should have a stronger capital cushion, with
graduated regulatory capital requirements (i.e., capital ratios that increase
with bank size), to “discourage them from becoming too big and to offset
their competitive advantage.”
• Warren Buffett: Require minimum down payments for home mortgages
of at least 10% and income verification. The Canadian banks do this and
its more than 10% in almost all loans.
• Eric Dinallo: Ensure any financial institution has the necessary capital to
sup-port its financial commitments. Regulate credit derivatives and
ensure they are traded on well-capitalized exchanges to limit
counterparty risk.
• Raghuram Rajan: Require financial institutions to maintain sufficient
“contin-gent capital” (i.e., pay insurance premiums to the government
during boom periods, in exchange for payments during a downturn.)
• A. Michael Spence and Gordon Brown: Establish an early-warning
system to help detect systemic risk.
• Niall Ferguson and Jeffrey Sachs: Impose haircuts on bondholders and
coun-terparties prior to using taxpayer money in bailouts.
• Nouriel Roubini: Nationalize insolvent banks.
In Chapter 25, I look into the results of other signals to assess what
worked and what did not.
1 Edited from Wilmott, July 2009.

2This section utilizes the CNBC TV program hosted by David Faber called “A House of Cards” for
much information on this episode.
What Signals Worked and What Did Not, 1980–
2009, Part III1
The dreaded September to October, the March to September 2009 rally, and future scenarios
In this third and final chapter on this topic based on a 7Cities/Wilmott talk
given on Thursday, 26 February 2009, I discuss the dreaded September-
October, the March to September 2009 rally and future scenarios, inflation
versus deflation and the US debt bomb, housing trends, what's happening in
China and Japan, income inequality and government regulation and the next
bubbles.
Historically, September and October have been the worst performing

stock market months in the US and across the world. Indeed, the graphs I
showed in Sell-in-May and Go Away in the May issue of Wilmott indicated
th at the rule sell on May 1 and buy on the 6th trading day lbefore the end of
October, for the !S&P500 and Russell2000 futures indices for the 16 years
1993-2009, did indeed handily beat a buy and hold strategy. For the S&P500
a buy and hold strategy turned $1 on February 4, 1993, into $1.56 on August
17, 2009; whereas, sell in May and move into cash, counting interest (Fed
funds effective monthly rate for sell in May) and dividends for the buy and
hold, had a final wealth of $3.18 , some 103.8% higher. For the Russell2000
, the final wealths were $1.49 and $3.69, respectively, some 147.7% higher
for the sell in May strategy. Chapter 17 updates this to December 2011 and
the longer results are even better. Among other things, the huge drop and
volatility in August and September 2011 was avoided.
As September approached, a distinguished Princeton University professor
and very good student of the markets was pestering me regarding whether
September- October would be bad months, especially following a huge
move up in a weak economic climate with the S&P PE ratio over 100!
He was my guest at Saratoga for the August 28, 2009 Travers race -
we did win on another anomaly. In the previous ten runnings of the
grade I mile Travers, the top 3-year-old race between the classics
(Kentucky Derby, Preakness and Belmont and the Breeders' Cup)
was won by a winner of one of these classic races. This is similar to
grade I class, most grade I races are won by previous grade I winners
even if they look terrible in recent other races. There was only one
such candidate and he looked good on the handicapping models we
follow - that was Summer Bird and indeed he won easily.
These months certainly were dismal in 2008 with September falling

9.79% and October falling a huge 20.11% with November's -9.22% and
February's -11.13% completed the decline on March 9. See the VIX
volatility index in Figure 25.1. In the fall of 2008 the VIX was approaching
an astronomical 90,now it has fallen to 25% in the March to September rally.
This September 2009 started out with a small decline as many were worried
and sold positions in advance of a possible fall on the first trading day but
the market stabilized and rallied in the following days until the Wednesday
September 23 Fed meeting. In the Fed minutes, they stated that the loose
money policy will continue at very low interest rates for quite some time
well into 2010 and there was no immediate danger of inflation. See Figure
25.2 for current US and other interest rates. All this is positive for the
markets but the markets interpreted it as a slowing of the slow but steady
recovery that has led the market up since the March lows. Figure 25.3 shows
the S&P500 and the increase has taken the market from 666 to 1070 with a
three-day decline to 1044.38 as of the close on February 25 when this
column was finished.
Fig. 25.1 VIX two years ending 28 September 2009, Yahoo
The decline was not much and weathered the storm of a weaker than
expected durable goods report, lower US housing sales. When stock markets
rally, they carry many other securities and assets along for the ride. This
includes thoroughbred stud farms, art, closed end country funds, etc., See
Figure 25.4 for the all art index and Figure 25.5 for the closed end funds
index. Observe the declining discounts along with the rise in prices. In
Kentucky for the Derby and Oaks (the filly Derby) in early May 2009, I
visited several top stud farms such as Claiborne and Three Chimneys. Here
the correlation is strong; stud fees follow the S&P500 but levered. For
example, Fusaichi Pegasus, the only Kentucky Derby winner who was sired
by the top speed sire Mr Prospector, had his stud fee of $200K moved to
$25K even when he is the fourth leading sire in the US.2 Other stallions had
declines as well; $75K to $15K for example for the near Triple Crown
winner Point Given. Of course, some have stayed constant and a few rose for
outstanding performance. These prices are now rallying. Equities have a
more significant wealth effect overall since equity market moves tend to
influence consumption decisions of the highest income groups, who in turn
account for the greatest proportion of US consumption (though out of their
income they consume a lower proportion than the poor).
Fig. 25.2 Interest rates. Barron's
Fig. 25.3 The S&P500, September 26, 2008 to September 25, 2009
Fig. 25.4 Chasing stocks. Source: Grant (2009)

Fig. 25.5 Closed end funds. Source: Barron's (2009)
I discuss this more below, but remember that the US debt is in US dollars
and many other countries like the UK and Japan have much more as a
percent of GDP. See Table 25.1. Their debts are also in their currencies.
Before we discuss this rally, let's look at one more data snooping anomaly
and then the facts on September- October over the recent past.
The sell on Rosh Hashanah and buy on Yom Kippur anomaly

This Rosh Hashanah anomaly relates to the Jewish new year. In 2009, that
was close of September 18 to open on September 28 for the trading days The
origin of this practice seems to be the belief of Jewish investors that they
should liquidate their portfolios during the holiday so that their attentions
could be fully focused on their worship; or more likely in today's world, not
trade.
Could this really be true in 2009?
Going back to 1915, the performance of the DJIA is -0.62% from the last
close before Rosh Hashanah until eight days later with the last close before
Yom Kippur. From then to December 31 averaged a respectable 1.99%, see
TheStreet.com and the Kirk Report.
Table 25.1: Debt as a percent of GP around the world. Source: Laing (2009)
Source: Japanese Journal of Population; United Nations; International Money Fund
What happened in 2009. The close on September 18 was 1068.30 and the
close on September 25 was 1044.38, so -2.24%. So the anomaly worked
once again.
What happened in 2010?
Rosh Hashanah for Jewish Year 5771 occurred on sunset Wednesday,
September 8, 2010 - nightfall September 10, 2010 and Yom Kippur began on
Friday, September 17, 2010. The close on September 8 was 1098.87 and the
close on September 16 was 1124.66, for a gain of 2.35%. So the anomaly did
not work in 2010. The close on December 31, 2010 was 1257.64 or a gain of
11.82% from the close on September 16, 2010.
Rosh Hashanah 2011 began on September 28 and the S&P500 cash close
then was 1151.06. Yom Kippur 2011 began on October 7 with the S&P500
close at 1164.97. The change was +1.2%. So the anomaly failed again.
Returning to September-October. Figures 25.6ab and 25.7ab show the
monthly effect in the S&P500 large cap and the Russell2000 small cap
future indices, respectively, from 1993–2011 and 2004–2011. Observe that
over the longer horizon October is actually positive, averaging =1.29% per
year, and September is slightly negative and there is no reliable monthly
effect. As the historically strong months of January and February are
negative, December is reliably positive though.
More recently, from 2004–2011, October is massively negative and
November, which historically has the strongest turn-of-the-month is also
negative. The year by year September-October returns from 1993 to 2011
are in Table 25.2. The pattern is clear: these months are like other months
except that they frequently have big declines-8.38% (2001), 11.31% (2002),
-9.79% (2008) and -7.85% (2011) for September and the -20.11% (2008) for
October for the S&P500. October had other great falls in 1929, 1987 and
other years. The Russell2000 is similar. Additional data is in Dzhabarov and
Ziemba (2009).
Fig. 25.6 S&P500 Futures Average Monthly Returns
Fig. 25.7 Russell2000 Futures Average Monthly Returns
Table 25.2: S&P500 Futures Average Returns, September and October,

1993–2011
Ramadan
Bialkowski, Etebari, and Wisniewski (2009) study stock returns during the
Muslim holy period of Ramadan in 14 predominantly Muslim country
during the period from 1989"–2007. They find that stock returns during
Ramadan are almost nine times higher (38.1% versus 4.3%) than during the
rest of the year and that this conclu-sion persists after controlling for other
known calendar anomalies like the January Effect, Halloween Effect, and
Turn-of-the-Week Effect. Their explanations for this phenomenon rely on
recent behavioral theories that connect investor emotions with their
decisions. Specifically, they suggest that these excess returns are a result of
increased investor optimism experienced during Ramadan as a time of
relative hap-piness, solidarity, and social identity for Muslims; they go as far
as to suggest that Ramadan may cause ‘mild states of euphoria,’ as
suggested by Knerr and Pearl (2008). This upbeat or positive sentiment then
causes relative overconfidence and an increased willingness to take risk,
such that investors perceive investments as of relatively higher value.
Bialkowski et al. also consider that during this relatively healthy period there
may be a higher demand for equities, as documented by Rosen and Wu
(2004). They, however, do not find any evidence of a higher trading volume
during Ramadan.
In their study, Bialkowski et al. discuss Ramadan to shed light on its
potential emotional effects, review the clinical effects of fasting, and review
empirical evidence on the effects of Ramadan on equity prices in 14 Islamic
countries. Their empirical results compare the average returns during the
holy month and the rest of the year and find that 11 of 14 countries studied
have higher returns during Ramadan; the countries that did not exhibit this
anomaly are Bahrain, Saudi Arabia, and Indonesia. They mention the effects
of relatively few observations in the case of Bahrain and Saudi Arabia and
note the outlier of the Asian Crisis in Indonesia effecting its equity prices
during Ramadan.
They further test their results by two event studies, benchmarking returns
against a constant-mean-return model as well as a predictive market model
using a proxy of 23 industrialized countries that do not have Muslim
majorities. In these tests, they calculated cumulative abnormal returns
(CAR) as returns in excess of what an investor should expect in the absence
of Ramadan, finding this CAR to fall between 2.5%-3.1% depending on the
event study approach. Among other robustness tests, Bialkowski et al.
analyze if this CAR may be compensation for increased risk during
Ramadan by examining return volatility, but do not find any supporting
evidence and in fact find that, except Turkey, all countries studied actually
showed decreased volatility during Ramadan.
Bialkowski et al. also test their results for illiquidity effects, exchange rate
considerations, and other accepted calendar effects, and find that the
Ramadan Effect remains anomalous and can most likely explained by
temporal changes in investor psychology. They conclude that investors can
profit in Muslim stock markets by buying shares at the beginning of
Ramadan and selling them at its end, though they do not explicitly model
this strategy nor do they estimate transaction costs.
Inflation versus Deflation

The big 2007 to March 2009 decline from over 1500 to 666.79 on the
S&P500 and similar declines across the world plus all the associated
banking, real estate, unemployment and other crises had led to a deflationary
environment. To counter act this, the US Fed and Treasury bought billions
and billions of securities that were to say of questionable and hard to
measure value. They created stimulus programs such as the cash for
clunkers, and a $700 billion rescue program for the banks. Meanwhile, the
Fed reduced short interest rates to essentially zero. These various actions
plus a declining flow of bad news such as monthly unemployment rising but
at a declining rate improved the economic outlook. This second derivative
effect plus the low interest rates, with huge monies on the sidelines and
resumption of investor confidence led a rather large over 56% increase in the
S&P500 from the intraday low of 666.79 on March 6th and a low close of
676.53 on March 7th on the S&P500 to the 1440.67 when we went to press
on September 30. Meanwhile, the US consumer remains on the sidelines
afraid to spend much so for the first time in a long while US savings is
positive. This, of course, cuts down on imports from China and elsewhere
and thus there is a downward pressure on prices. Hence, there is some
deflation and bonds, especially high yield bonds (they are called that or
junque instead of junk because of their 50% return in 2009) have been stellar
performers, all related to credit risk confidence returning to the market. A
good way to measure this confidence is through carry trades where one
borrows in a low yield country to invest in a high yield country (like
Australia or Canada). The yen exchange rate is one signal of this risking
with increasing confidence and falling with decreasing confidence. See
Figure 25.8 for the yen valued in US dollars and euros updated to . . . .
Fig. 25.8 The yen versus the US dollar and euro, 5-years to October 12, 2012. Source: Yahoo
The debt built-up from the Fed and Treasury actions amounts to an
estimated $11.6 trillion. Some of this money is being paid back by the
recipients of the rescue like Goldman Sachs whose stock has risen from a
low of 47.41 to its current 183.58. Recall it was Goldman and GE that
Warren Buffett's Berkshire Hathaway made loans to at 10% with some free
call warrants at 120 for Goldman as part of the deal. Look for Berkshire to
move eventually out of its doldrums.
Figure 25.9 compares Goldman, GE and Berkshire stock for the past two
years.
So paying back this debt could well lead to higher interest rates. This
week, the noted Tiger Hedge Fund manager Julian Robertson, an astute
student of the markets, is expecting this but big time. He sees the possibility
of 15-20% interest rates and similar inflation. He cannot predict when but he
sees it coming. This is, of course, an extreme view but from such a noted
analystit deserves consideration. Consequently, it is one possible scenario.
My own feeling is that it is way too extreme and will not happen - but 1981
had such interest rates and 1968-88 when Ed Thorp was running his
Princeton-Newport Fund, the S&P returned 10% but interest rates averaged
8% for these 20 years. So Robertson is short long term bonds which seems
likely eventually to be a wise trade. It is clear that eventually inflation and
interest rates will rise but Robertson's values are way too high since the
consequences of such an event are too dangerous to consider. As of March
2012, there are minor signals of higher interest rates in a couple years but
nothing major. But the bull market in bonds is close to over.
Fig. 25.9 Goldman, GE and Berkshire stock prices versus the S&P500
Gold
In inflation, one thinks of gold and silver so lets see how they have done
relative to other assets. Figure 25.10 shows that in the past five years, gold
has outperformed silver which has greatly outperformed the S&P500, oil and
the dollar index.
Fig. 25.10 Comparing gold, silver, oil, spy and dollar index (dx). Source: Yahoo
Housing trends
In Vancouver, where I live, housing is on fire! But it has been overpriced
relative to Vancouver incomes all 40+ years I have been here. It is usually
people who made money somewhere else who are pushing the prices and
sales. Recently that's China. It used to be Hong Kong since Expo 1986 then
other Asian places like Taiwan and Singapore were active. Lately it is more
mainland China. The winter Olympics are coming in February 2010 but that
is not a major factor yet. The prices and sales picked up as the S&P500 and
Toronto TSE300 rallied and interest rates are rock bottom: 2.5% variable, for
example. So it is easy to buy if you have a substantial down payment. There
are no subprime mortgages in Canada nor non-recourse mortgages. In the
US, housing is still weak but stabilizing.
But, as Alan Abelson wrote in the September 28, 2009 Barron's:
Homeowners' equity has declined from 58.7% back in '05 to around

43% today. What's more, nearly a third of households have no
mortgages, which, of course, means that the equity percentage of the
50-plus million that do have mortgage loans is a good cut lower than
43%.
Out of 56 million units, 6.94 million are in deliquency and eventually

headed for liquidation and 300,000 join the queue each month. So housing is
still weak. As well, commercial real estate is much weaker and has not yet
hit bottom.
What's going on in China?
As they said in Indiana Jones, you've got the wrong Jones, I mean Ziemba,
on this one. See Rachel's columns in this magazine and at Roubini Global
Economics in New York. It has been a couple years since I was last in China
so I defer to and rely on Rachel who follows it closely. In a recent article in
the USA Today, she discussed some key points, see Lynch (2009). Chinese
consumption is slowly on the rise, about $1.15 trillion in 2009 versus $1
trillion in 2008, a 15% increase. But it is less than a quarter of the up to $700
billion US consumers are not spending even if they spend their long run
average. Since 2005, the Chinese yuan has risen 16% but still is significantly
undervalued. A further rise would boost domestic demand so would making
credit more widely available, Rachel argues. The G-10 wants less credit-
fueled shopping. Europe needs to invest more and China consume more. The
US current account deficit was 6.6% of GDP in 2005, now it is 2.8%. But
there is still a huge US deficit and China's current account surplus. Figures
25.11abc show new brokerage accounts, the Shanghai stock exchange and
the US dollar/yuan over time. Observe how the growth in brokerage
accounts mirrors the Shanghai stock exchange.
Rachel in the RGE Global Outlook for China, September 2009, wrote as
follows:
The credit extension and improved confidence helped reflate Chinese

equities. Concerns that the Chinese government is tightening the
credit that helped lead to economic recovery and asset market
reflation contributed to a 23% fall of Chinese equities in the month of
August, pushing the CSI 310(0, a measure including stocks trading in
Shanghai and Shenzhen markets, into a technical bear market. The
index rebounded 16.7% in September and J still up more than 80%
from .January to September 15. Price/Earning ratios have more than
doubled from their lows in November 2008, but remain below their
peak in January 2008. Some analysts suggest as much as 30% of the
short-term bills financing helped fund equity trades, meaning the
decision not to renew bills financing reduced inflows. Other factors
contributed to reduce liquidity to the equity market in July and
August including the restart of IPOs, issuance of money-market
bonds and expiry of share lockups. Meanwhile the number of new
brokerage accounts had its 2009 peak at the end of July, right before
the Shanghai composite peaked. Thus, Chinese equities seem
vulnerable in late 2009, but should be supported by economic growth
in 2010.
Fig. 25.11 China. S ou rce: RGE Global Out look; for China
Figure 25.12 shows the dramatic percent increase in Chinese consumer

spending. This looks good but Figure 25.13 shows it is still pretty small
compared to other countries and is a small percentage of GDP.
Income inequality and government regulation

The governments of the US and UK have a lot of rhetoric about putting in
rules to limit executive compensation and base that on the risk taken and
long term performance. This sounds commendable especially if the incentive
systems are changed to not encourage rogue trading. Will they accomplish
anything? I am skeptical but we will see.
The G-20 claim they will curb bank pay and align economic policies so
that the boom and bust cycles will not occur. They want to avoid multi-year
guaranteed bonuses, have a significant portions of variable compensation
deferred, paid in stock and subject to claw back if earnings decline. Banks
are also encouraged to increase the quality and quantity of capital in reserve
against future losses. The growing influence of China, Brazil and other
emerging economies made G-20 not G-8 the key economic work group.
Fig. 25.12 China's consumer spending, 1992=100. Batson (2009)
Fig. 25.13 China's consumption still small relative to other countries and small as a percent of GDP.
Batson (2009)
The US economy is built on the concept that you get paid what you can
get. And that depends on demand for services. The average late night crime
or medical drama in the US costs $3 million to produce each week. Charlie
Sheen basically playing himself gets $825,000 per week on the half hour
sitcom Two and A Half Men for a 23-week season, some $20+ million just
for him. Indeed, before his firing, he got up to a reported $1.7 million per
show. He got fired for criticizing the show's producers. Now they give a
much more mediocre replacement some $700,000 per show and it has
dropped from #1 comedy to #3. Despite his faults, Charlie Sheen is a great
actor! Many other movie and sports stars earn this much or more. As a joint
Red Sox Yankee fan, I note that C. C. Sabathia, a great August (essentially
all wins then) pitcher for the New York Yankees but so-so in the spring and
post season (so far) has a $161 million contract. I suppose increasing the
marginal tax rate on high earners could grab back some of this excess which
like states and governments desperately need. But studies are mixed on the
value of doing this as its a drain on spending. Some of the highly paid actors
and sports stars actually do a lot of good in giving back. High on this list are
Brad Pitt rebuilding New Orleans housing and Cher giving safety helmets to
soldiers in Iraq and Afghanistan during the Bush administration. Whoops ..
why isn't it the government paying for this one?
In our investment world we have a lot of highly paid people and it is all
based on skill in managing large sums of capital. Should or will this change?
It sure does not look like it.
But there remains a huge problem of six unemployed people for each new
job (14.5 versus 24 million) in the US and the about 9.7% unemployment
rate. The 59% recovery in the DJIA from the March 9 low has not helped
these people at all. There is too much uncertainty so hiring is not happening.
In fact, the government bailout went more to the Goldman Sachs and Warren
Buffett types around the world. Since I work on the intellectual and practical
issues of such trading, etc strategies, I am well aware of this imbalance.
Crime and other social problems are likely to increase.
Japan: Still a lot of trouble

Figure 25.14 shows that Japan's population is dropping and aging. For more
on this see our new Wiley book of Optimizing the aging, pension and
retirement dilemma, Bertocchi et al. (2010).
Fig. 25.14 Japan's population
Their debt is more than twice the US debt and rising as a percent of GDP,
see Table 25.1.
Japan has been crippled by high consumer costs, substandard profit per
manhour is 30% less than the US. The government has historically been
focussed on protecting losers rather than promoting winners. Their growth
rate is low, well under 2%. A plus is that Japan's debt is offset by the postal
savings which have low interest rates. Laing (2009)argues that the JGB bond
market may be the biggest bubble in world economic history. In my lectures,
I have frequently said that the Japanese are the best at losing money with
some notable examples. A default is possible and increasingly likely says
Harvard professor William Overhold, who argues that Japan's debt is
destined to keep rising (as shown in Table 25.1) because of unfunded
pension liabilities for the aging population and contingent liabilities that the
central government bears as a result of functionally insolvent local
government.
Conclusion
To conclude, the most likely bubbles are in long bonds in the US, UK, Japan
and elsewhere. Inflation looks ready to loom its ugly head sometime in the
future but as of March 2012 is not here yet. The Fed thinks it is ready for
this. We will see. Meanwhile, cash returns nothing, bonds are very risky, real
estate is problematic but rising slowly from the extreme lows and stocks
look like the only game in town so they likely will continue rising.
1 Edited from Wilmott, November 2009.

2He was bought as a yearling for $4 million - a hefty sum. But after he won the Derby, Coolmore, the
wisest business people in racing, bought him for $60–70 million. Have they lost money? Are you
kidding! Before it was ever found out if he was any good as a stallion, several years of more than 200
breeds at $200,000 each were collected from stud duties in Kentucky and the southern hemisphere.
How to Lose Money in Derivatives and Examples
of Those Who Did1
Understanding how to lose helps one avoid losses!
This chapter discusses how to lose money in derivatives which leads to our
discussion of hedge fund disasters and how to prevent them. The derivative
industry deals with products in which one party gains what the other party
loses. These are zero sum games situations. Hence there are bound to be
large winners and large losers. The size of the gains and losses are magnified
by the leverage and overbetting, leading invariably to large losses when a
bad scenario occurs. This industry now totals over $700 trillion of which the
majority is in interest and bond derivatives with a smaller but substantial in
equity derivatives. Figlewski (1994) attempted to categorize derivative
disasters and this chapter discusses and expands on that:
(1) Hedge
In an ordinary hedge, one loses money on one side of the transaction in
an effort to reduce risk. The correct way to evaluate the performance of a
hedge is to consider all aspects of the transaction. In sophisticated hedges
where one delta hedges but is a net seller of options, there is volatility
(gamma) risk which could lead to losses if there is a large price move up
or down. Also accounting problems can lead to losses if gains and losses
on all sides of a derivatives hedge are recorded in the firm's financial
statements at the same time.
(2) Counterparty default
Credit risk is the fastest growing area of derivatives and a common
hedge fund strategy is to be short overpriced credit default derivatives.
There are lots of ways to lose on these shorts if they are not hedged
properly, even if they have an edge.
(3) Speculation
Derivatives have many purposes including transferring risk from those
who do not wish it (hedgers) to those who do (speculators). Speculators
who take naked unhedged positions take the purest bet and win or lose
monies related to the size of the move of the underlying security. Bets on
currencies, interest rates, bonds, or stock market moves are leading
examples.
Human agency problems frequently lead to larger losses for traders
who are holding losing positions that if cashed out would lead to lost
jobs or bonus. Some traders will increase exposure exactly when they
should reduce it in the hopes that a market turnaround will allow them to
cash out with a small gain before their superiors find out about the true
situation and force them to liquidate. Since the job or bonus may have
already been lost, the trader's interests are in conflict with objectives of
the firm and huge losses may occur. Writing options, which typically
gain small profits most of the time but can lead to large losses, is a
common vehicle for this problem because the size of the position
accelerates quickly as the underlying security moves in the wrong
direction. Since trades between large institutions frequently are not
collateralized mark to market large paper losses can accumulate without
visible signs such as a margin call. Nick Leeson's loss betting on short
puts and calls on the Nikkei is one of many such examples. The Kobe
earthquake was the bad scenario that bankrupted Barings. A proper
accounting of trading success evaluates all gains and losses so that the
extent of some current loss is weighed against previous gains. Derivative
losses should also be compared to losses on underlying securities. For
example, from January 3 to June 30, 1994, the 30-year T-bonds fell
13.6%. Hence holders of bonds lost considerable sums as well since
interest rates quickly rose significantly.
(4) Forced liquidation at unfavorable prices
Gap moves through stops are one example of forced liquidation.
Portfolio in-surance strategies based on selling futures during the
October 18, 1987 stock market crash were unable to keep up with the
rapidly declining market whose futures fell 29% that day. Forced
liquidation due to margin problems is made more difficult when others
have similar positions and predicaments. The August 1998 problems of
Long Term Capital Management in bond and other markets were more
difficult because others had followed their lead with similar positions.
When trouble arose, buyers were scarce and sellers were everywhere.
Another example is Metallgellschaft's crude oil futures hedging losses of
over $1.3 billion. They had long term contracts to supply oil at fixed
prices for several years. These commitments were hedged with long oil
futures. But when spot oil prices fell rapidly, the contracts to sell oil at
high prices rose in value but did not provide current cash to cover the
mark to the market futures losses. A management error led to the
unwinding of the hedge near the bottom of the oil market and the
disaster.
Potential problems are greater in illiquid markets. Such positions are
typ-ically long term and liquidation must be done matching sales with
available buyers. Hence, forced liquidation can lead to large bid-ask
spreads. Askin Capital's failure in the bond market in 1994 was
exacerbated because they held very sophisticated securities which were
only traded by very few counterparties. Once they learned of Askin's
liquidity problems and weak bargaining position, they lowered their bids
even more and were then able to gain large liquidity premiums.
(5) Misunderstanding the risk exposure
As derivative securities have become more complex, so has their full
under-standing. Our Nikkei put warrant trade (discussed in Chapter 12)
was successful because we did a careful analysis to fairly price the
securities. In many cases, losses are the result of trading in high-risk
financial instruments by un-sophisticated investors. Lawsuits have arisen
by such investors attempting to recover some of their losses with claims
that they were misled or not properly briefed on the risks of the positions
taken. Since the general public and thus judges and juries find
derivatives confusing and risky, even when they are used to reduce risk,
such cases or their threat may be successful.
A great risk exposure is the extreme scenario which often investors
assume has zero probability when in fact they have low but positive
probability. Investors are frequently unprepared for interest rate,
currency or stock price changes so large and so fast that they are
considered to be impossible to occur. The move of some bond interest
rate spreads from 3% a year earlier to 17% in August/September 1998
led even the savvy investor and very sophisticated Long Term Capital
Management researchers and traders down this road. They had done
extensive stress testing which failed as the extreme events such as the
August 1998 Russian default had both the extreme low probability event
plus changing correlations. Scenario dependent correlation matrices
rather then simulations around the past correlations is suggested. This is
implemented, for example, in the Innovest pension plan model which
does not involve levered derivative positions (see Chapter 14) . The key
for staying out of trouble especially with highly levered positions is to
fully consider the possible futures and have enough capital or access to
capital to weather bad scenario storms so that any required liquidation
can be done orderly.
Figlewski (1994) mentions that the risk in mortgage backed securities is
es-pecially difficult to understand. Interest only (IO) securities, which
provide only a share of the interest as part of the underlying mortgage
pool's payment stream are a good example. When interest rates rise, IO's
rise since payments are reduced and the stream of interest payments is
larger. But when rates rise sharply, the IO falls in value like other fixed-
income instruments because the future interest payments are more
heavily discounted. This signal of changing interest rate exposure was
one of the difficulties in Askin's losses in 1994. Similarly the sign
change between stocks and bonds during stock market crashes as in 2000
to 2003 has caused other similar losses. Scenario dependent matrices are
especially useful and needed in such situations.
(6) Forgetting that high returns involve high risk
If investors seek high returns, then they will usually have some large
losses. The Kelly criterion strategy and its variants provide a theory to
achieve very high long term returns but large losses will also occur.
These losses are magnified with derivative securities and especially with
large derivative positions relative to the investor's available capital.
Stochastic programming models provide a good way to try to avoid

problems 1-6 by carefully modeling the situation at hand and considering the
possible economic futures in an organized way.
Hedge fund disasters usually occur because traders overbet, the portfolio
is not truly diversified and then trouble arises when a bad scenario occurs.
We now discuss three sensational failures: LTCM (1998), Niederhoffer
(1997) and Amaranth Advisors (2006). Stochastic programming models
provide a way to deal with the risk control of such portfolios using an overall
approach to position size, taking into account various possible scenarios that
may be beyond the range of previous historical data. Since correlations are
scenario dependent, this approach is useful to model the overall position
size. The model will not allow the hedge fund to maintain positions so large
and so under diversified that a major disaster can occur. Also the model will
force consideration of how the fund will attempt to deal with the bad
scenario because once there is a derivative disaster, it is very difficult to
resolve the problem. More cash is immediately needed and there are
liquidity and other considerations. Chapter 14 explores more deeply such
models in the context of pension fund as well as hedge fund management.
The Failure of Long Term Capital Management

There have been many hedge fund failures but LTCM stands out as a
particularly public one. The firm started with the talents of the core bond
traders from John Merriwether's group at the Salomon Brothers who were
very successful for a number of years. When Warren Buffett came on board
at Salomon the culture of this group clashed with Buffett's apparently more
conservative style. In truth Buffett's record is Kelly like and not all that
different from Merriwether's group in terms of position size but Buffett's risk
control is superior. A new group was formed with an all star cast of top
academics including two future Nobel Laureates and many top professors
and students, many linked to MIT. In addition top government officials were
involved. The team was dubbed too smart to lose and several billion was
raised even though there was no real track record, fees were very high (25%
of profits plus a 3% management expense fee) and entry investment was
$100 million minimum. The idea, according to Myron Scholes, was to be a
big vacuum cleaner sucking up nickels all over the world as the cartoon
suggests. There were many trades, but the essence of the bond risk arbitrage
was to buy underpriced bonds in various locales and sell overpriced bonds in
other locales and then wait for the prices to revert to their theoretical
efficient market prices and then to unwind the position.
These trades are similar to the Nikkei put warrant risk arbitrage (described
in Chapter 12) Thorp and I did except that the leverage they used was much
much greater. I like to call these bond trades buy Italy and sell Florence. As
shown in the graph, the interest rate implied by the bond prices is higher in
Italy than in Florence. But the theory is that Florence, a smaller place, would
have more risk. Hence, the trade should have an advantage and be unwound
when the prices reverted to their true risk priced values. LTCM analysts
made many such trades, most much more complex than this, all across the
world. They also had many other complex and innovative trades. Their
belief that markets were efficient and, when temporarily out of whack,
would snap back quickly and the continuous lognormal assumptions of
option pricing hedging led them to take very large positions which according
to their theory were close to riskless.
The plan worked and net returns for the part of the year 1994 that the fund
operated were 19.9% net. The years 1995 and 1996 had similar superb
results of 42.8% and 40.8% net, respectively Indeed for the principals whose
money grew fee- less, the net returns were 63% and 57%, respectively, with
taxes deferred. There was so much demand for investment in the fund, which
in 1997 was effectively closed to new investors, that a grey market arose
with a 10% premium. By 1997 it became harder to find profitable trades and
the gains fell to 17.1%. This was a good record for most but not satisfactory
to LTCM's principals; among other things the S&P500 returned 31%
excluding dividends. Their action was to return $2.7 billion of the $6.7
billion to the investors, a huge mistake! The principals then put in an
additional $100 million raised by personal bank loans, another mistake. The
banks were happy to lend this money basically unsecured. Banks and others
were quite keen to loan to or invest with this group and the investors were
not happy to be forced out of the fund. Still, at the start, $1 on February 24,
1994, was $2.40 net at the end of 1997. The year 1998 was difficult for the
fund and then turned into a disaster following the August 17 Russian ruble
devaluation and sovereign bond default. Bonds denominated in rubles
trading for say 60 fell rapidly to 3 whereas Russian bonds denominated in
marks or dollars only fell a few percent as they were not subject to the
effects of the ruble devaluation. So long 60 short 95 say became long 3 short
92 say. Also there were defaults in currency hedging contracts which added
to the losses because that hedge failed.
Such losses occur from time to time in various markets and hedge funds
which overbet can be very vulnerable to it. The problem for LTCM was that
they had $1.25 trillion of positions in notional value (that's was over 2% of
the world's derivatives in 1998) and $125 billion of borrowed money.
Although the trades were all over the world and hence it seemed they were
diversified, they in fact were not. What happened was a scenario dependent
correlation situation like that modeled in the Innovest pension application
described in Chapter 14. There was an underlying variable that frequently
lurks its ugly head in disasters that being investor confidence. The graph on
the side illustrates the problem: all the bond rates increased for non high
quality debt. For example, emerging market debt was trading for 3.3% above
US T-bonds in October 1997, then 6% in July 1998 and then an astounding
17% in September 1998.
LTCM was unable to weather the storm of this enormous crisis of
confidence and lost about 95% of their capital, some $4.6 billion including
most of the princi-pals' and employees' considerable accumulated fees. The
$$100 million loan actually put some of them into bankruptcy, although
others came out better financially; see the Barrons cartoon. It did not help
that they unwound liquid positions first rather than across all liquidity levels
as the Nobels recommended, nor that many other copy-cat firms had similar
positions, nor that LTCM had created enemies by being so good and so
brash, nor that the lack of monitoring of margin by brokers eager for their
business allowed the positions to grow to overbet levels, and finally that the
$2.8 billion was gone and they could not draw on it when it was most
needed.2 Smart people bounce back and possibly learn from their mistakes.
Various ex-LTCM members are new hedge funds and other ventures. The
lessons are:
• Do not overbet, it is too dangerous.
• VAR type systems are inadequate to measure true risk but see Jorion's
(2006) fine book on Var and Dunbar's (2000) good discussion of the
VAR calculations used by LTCM. LTCM analysts did a very careful
analysis but the problem was that the risk control method of VAR which
is used in regulations does not really protect hedge funds that are so
highly levered because you are not penalized enough for large losses.
Indeed if you lose $10 million it is penalized the same as losing $100
million if the VAR number is $9 million of losses. What you really need
are convex penalties so that penalties are more than proportional to
losses. This is discussed in Chapter 14.
You really do need to use scenario dependent correlation matrices and
consider extreme scenarios. LTCM was not subject to VAR regulation
but still used it.
• Be aware of and consider extreme scenarios.
• Allow for extra illiquidity and contract defaults. LTCM also suffered
because of the copycat firms which put on similar positions and
unwound them at the same time in August/September 1998.
• Really diversify (to quote Soros from the Quantum Funds, "we risked
10% of our funds in Russia and lost it, $2 billion, but we are still up 21%
in 1998").
• Historical correlations work when you do not need them and fail when
you need them in a crisis when they approach one. Real correlations are
scenario dependent. Sorry to be repetitive, but this is crucial.
Good information on the demise of LTCM and the subsequent $3.5 billion
bailout by major brokerage firms organized by the FED are in a Harvard
Business School case by André Perold (1998), and articles by Philippe
Jorion (2000a) and Franklin Edwards (1999). Eventually the positions
converged and the bailout team was able to emerge with a profit on their
investment.
The currency devaluation of some two thirds was no surprise to WTZ. In
1992, we were the guests in St. Petersburg of Professor Zari Rachev, an
expert in stable and heavy-tail distributions and editor of the first handbook
in North Holland's Series on Finance (Rachev, 2003) of which WTZ is the
series editor. On arrival I gave him a $100 bill and he gave me four inches of
25 Ruble notes, see the photo. Our dinner out cost two inches for the four of
us; and drinks were extra in hard currency. So we are in the Soros camp;
make bets in Russia (or similar risky markets) if you have an edge without
risking too much of your wealth. Where was the money lost? The score card
according to Dunbar (2000) was a loss of $4.6 billion. Emerging market
trades such as those similar to the buy Italy, sell Florence lost $430 million.
Directional, macro trades lost $371 million. Equity pairs trading lost 306
million. Short long term equity options, long short term equity lost $1.314
billion. Fixed income arbitrage lost $1.628 billion.
The bad scenario of investor confidence that led to much higher interest
rates for lower quality debt and much higher implied equity volatility had a
serious effect on all the trades. The long-short equity options trades, largely
in the CAC40 and Dax equity indices, were based on a historical volatility of
about 15% versus implieds of about 22%. Unfortunately, in the bad scenario,
the implieds reached 30% and then 40%. With smaller positions, the fund
could have waited it out but with such huge levered positions, it could not.
Equity implieds can reach 70% or higher as Japan's Nikkei did in 1990/1991
and stay there for many months.
The imported crash of October 27 and 28, 1997

The Asian Financial crises, a series of banking and currency crises
developed in various Asian countries beginning in mid 1997. Many East and
Southeast Asian countries had currency pegs to the U.S. dollar which made
it easy for them to attract financing but lacked adequate foreign reserves to
cover the outstanding debt. Their pegs to the US dollar and low interest rates
encouraged mismatches in currency (debts were in US dollars, loans in local
currency) and maturities. Spending and expectations that led to borrowing
were too high and Japan, the main driver of these economies, was facing a
consumer slowdown so its imports dropped. So that effectively these
countries were long yen and short dollars. A large increase in the US
currency in yen terms exacerbated the crisis, which began after speculators
challenged the Thai Baht and spread through the region. The countries had
to devalue their currencies, interest rates rose and stock prices fell. Also,
several hedge funds took significant losses, most notably, Victor
Niederhoffer's fund, which had an excellent previous record with only
modest drawdowns. His large long bet on cheap Thai stocks that became
cheaper and cheaper quickly turned $120 million into $70 million. Further
buying on dips added to losses. finally the fund created a large short position
in out-of-the-money S&P futures index puts including on the November
830's trading for about $4-6 at various times around August-September
1997.
The crisis devastated the economies of Malaysia, Singapore, Indonesia,
etc. Finally it spread to Hong Kong, where the currency was pegged to the
US dollar at around 7.8. The peg supported Hong Kong's trade and
investment hub and was to be defended at all costs. In this case, the weapon
used was higher interest rates which almost always lead to a stock market
crash after a lag. See the discussion in Chapter 23 for US and Japanese case
along with other countries. The US S&P500 was not in the danger zone in
October 1997 by WTZ's models and we presume by those of other and trade
with Hong Kong and Asia was substantial but only a small part of the US
trade. Many US investors thought that this Asian currency crisis was a small
problem because it did not affect Japan very much. In fact, Japan caused a
lot of it.
A walk on the wild side

The week of October 20-25, 1997 was difficult for equity markets with the
Hang Seng dropping sharply. The S&P was also shaky. The November 830
puts were 60 cents on Monday, Tuesday and Wednesday but rose to 1.20 on
Thursday and 2.40 on Friday. The Hang Seng dropped over 20% in a short
period including a 10% drop on Friday, October 25th. The S&P500 was at
976 substantially above 830 as of Friday's close. A further 5% drop in Hong
Kong on Monday, October 27 led to a panic in the S&P500 futures later on
Monday in the US. They fell 7% from 976 to 906 which was still
considerably above 830. On Tuesday morning there was a further fall of 3%
to 876 still keeping the 830 puts out of the money. The full fall in the
S&P500 was then 10%.
But the volatility exploded and the 830's climbed to the $16 area. Refco
called in Niederhoffer's puts mid morning on Tuesday, resulting in the fund
losing about $20 million. So Niederhoffer's $70 million fund was bankrupt
and actually in the red as the large position in these puts and other
instruments turned the $70 million into minus $20 million. The S&P500
bottomed out around 876, moving violently in a narrow range then settling.
By the end of the week, it returned to the 976 area. So it really was a tempest
in a teapot like the cartoon from The Economist depicts. The November 830
puts expired worthless. Investors who were short equity November 830 puts
(SPXs) were required to put up so much margin that had to have small
positions and they weathered the storm. Their $4-$6, while temporarily
behind at $16 did eventually go to zero. So did the futures puts, but futures
shorters are not required to post as much margin. If they did not have
adequate margin because they had too many positions, They could have
easily been forced to cover at a large loss. Futures margins, at least for
equity index products, do not fully capture the real risk inherent in these
positions. WTZ follows closely the academic studies on risk measures and
none of the papers he knows address this issue properly. When in doubt,
always bet less. Niederhoffer is back in business having profited by this
experience. (Whoops - maybe not, see the postscript!)
One of our Vancouver neighbors, we learned later, lost $16 million in one
account and $4 million in another account. The difference being the time
given to cash out and cover the short puts. I was in this market also and won
in the equity market and lost in futures. I did learn how much margin you
actually need in futures which now I use in such trading which has been very
profitable with a few proprietary wrinkles to protect oneself that I need to
keep confidential. A hedged strategy had a 45% geometric mean with 72 of
77 winners with six quarters ruled too risky by an option price market
sentiment danger control measure out of the 83 possible plays in those 22
years and a seven symmetric downside Sharpe ratio as discussed in Chapter
6. Ruling out the six risky quarters, one of the naked strategies won 76 out of
77 times from 1985 to 2006. In those six quarters, the S&P500 actually fell
in four. The cumulative S&P500 loss in the six quarters was -41.7%.
The lessons for hedge funds are much as with LTCM. Do not overbet, do
diversify, watch out for extreme scenarios. Even the measure to keep one out
of potentially large falls mentioned above did not work in October 1997.
That was an imported fear-induced stock market crash which was not really
based on the US economy or investor sentiment. My experience is that most
crashes occur when interest rates relative to price earnings ratios are too
high. Almost always when that happens there is a crash (a 10% plus fall in
equity prices from the current price level within one year), see Chapters 2
and 21 for the 1987 US, the 1990 Japan, the US in 2000, the US in 2001,
which predicted the 22% fall in the S&P500 in 2002 and China, Iceland and
the US in 2006-9 are leading examples. Interestingly the measure moved out
of the danger zone following the 2000 crash. Then, in mid-2001, it was even
more in the danger zone than in 1999 because stock prices fell but earnings
fell more, In 2003, the measure then moved into the buy zone and predicted
the rise in the S&P500 in 2003. No measure is perfect but this measure adds
value and tends to keep you out of extreme trouble.
When long bond interest rates get too high relative to stock returns as
measured by the earnings over price yield method then there almost always
is a crash. Ziemba- Schwartz (1991) used a difference method and the results
of that are in Ziemba (2003).
Figure 22.1 used a ratio or log approach and is equivalent to what is now
called the Fed model. I started using these measures in 1988 in my study
group at Yamaichi Research, Japan. The study predicted the 1987 crash; see
that on this graph. It also predicted the 1990 Japan crash. I told Yamaichi
executives about this in 1989, but they would not listen. Yamaichi went
bankrupt in 1995; they would have survived if they had listened to me.3 We
found for 1948 to 1988 that every time the measure was in the danger zone
there was a fall of 10% or more with no misses. This was 12 of 12 with 8
other 10%+ crashes occurring for other reasons than high interest rates
relative to earnings. In late 1989 the model had the highest reading ever in
the danger zone and predicted the January 1990 start of the Japanese stock
market crash.
A mini crash caused by some extraneous event can occur any time. So to
protect oneself positions must never be too large. Koliman (1998) and
Crouhy, Galai and Mark in Gibson's (2000) book on model risk discusses
this. Their analysis suggests it's a violation of lognormality which I agree it
was. Increased implied volatility premiums caused the huge losses of those
who had to cash out because of margin calls because they had too many
positons.
Some good references on hedge fund performance, risk and incentives
follow for further reading. Kouwenberg and Ziemba (2007) using a
continuous time model with a prospect theory S-shaped objective, where
losses are more damaging than gains are good, study the effect of incentives
on hedge fund manager behavior. The incentive fee encourages managers to
take excessive risk but that risk is much less if the fund manager has a
substantial amount of their own money in the fund (at least 30%). So look
for funds where the managers eat their own cooking. Our empirical results
indicate that hedge funds with incentive fees have higher downside risk than
funds without such a compensation contract. Average net returns, both
absolute and risk-adjusted, are significantly lower in the presence of
incentive fees. So pick your managers well.
An incentive fee is tantamount to a call option on the value of the
investor's assets. Goetzmann, Ingersoll and Ross (2003) and Kouwenberg
and Ziemba (2007) show how to calculate the value of that option. The value
depends directly on the manager's optimal investment style with values
ranging from 0 (with no investment) to 17% (with 30%+ share) of the
investor's capital.
In Chapter 14 we formulate some stochastic programming models and
discuss pension fund, as well as hedge fund, applications. Wallace and
Ziemba (2005) provides background on how to make such models and
Ziemba (2003) explains enough about these models to hire someone to make
one for you. Ziemba and Mulvey (1998) describe many such existing models
for insurance, pension fund and other applications. The Lo (1999, 2001) and
Merton (2000ab) papers are useful to set the stage for the stochastic
programming models by setting out the issues.
Overbetting yields frequent trading disasters

The best way to achieve victory is to master all the rules for disaster, and
then concentrate on avoiding them
In America, people get a second chance .. . they don't get a third.
Victor Niederhoffer
After Niederhoffer's failure in 1997, his fund was closed and he lost much
of his personal fortune, reputation and happiness. He had failed in 1997
because he greatly overbet and did not diversify and a bad scenario wiped
him out. Was this a one time occurrence from which he learned or is it just
one of a sequence of similar outcomes. Niederhoffer is a multi-talented
individual graduating with a PhD in 1969 from the Graduate School of
Business, at the University of Chicago where Professor Gene Fama, Merton
Miller and other great finance theorists and practitioners are on the faculty.
Since his work was against the prevailing efficient markets theory and
highly data dependent, he was more comfortable with the statisticians and
was supervised by perhaps the world's top Bayesian statistician, Arnold
Zellner. Earlier at Harvard, his senior thesis “non randomness in stock
prices: a new model of price movements” challenged random walk theory.
He argued that stocks followed patterns such as Monday falls if Friday fell.
In 1967, with his PhD thesis unfinished and the title "US top squash
player", he headed to the finance department of the University of California,
Berkeley business school. I was there then as well but never met
Niederhoffer, being a graduate student. Victor was also a whiz at chess and
tennis, dating back to his Harvard undergraduate days. I was friendly with
one finance legend Professor Barr Rosenberg who went on to greatness in a
number of investment areas such as founding the Berkeley Program in
Finance, the firm BARRA and later Rosenberg Investments. Both Barr and
Victor, like me, were looking for anomalies to beat markets. Barr discovered
that small caps and low price to book stocks out performed work that 25
years later in 1992 formed the basis for the famous Fama-French (1992)
factors; see Rosenberg, Reid and Lanstein (1985). While Barr stuck to
institutional investing with low or no leverage, Victor was a high stakes
futures trader using lots of leverage. Hence, if he was right, then the gains
were very high but if he was wrong and his risk control was faulty, then
there could be substantial losses.
While teaching at Berkeley, Victor co-founded a small investment bank,
Nieder- hoffer, Cross and Zeckhauser (NCZ). Frank Cross was a former
Merrill Lynch executive and Richard Zeckhauser, a friend from his Harvard
days. Zeckhauser went to to become a well known economist at the Kennedy
School of Government and an avid bridge player. NCZ started with just $400
and did mail-order mergers, and sold small private companies to buyers. In
1979, Niederhoffer went into commodities and had great success, averaging
35% net for 15 years through the mid 1990s. George Soros gave him a
private $100 million account in 1981 and Niederhoffer traded that until
1993. That was shut down because, as Soros said, “he temporarily lost his
edge . . . he made money while the markets were sloshing along aimlessly.
Then he started losing money and had the integrity to close out the account.
We came out ahead”. Earlier in 1983 Zeckhauser had quit NCZ to return to
full time teaching and research partially because of Niederhoffer's high level
of risk taking, saying that “no matter what your edge, you can lose
everything. You hope and believe he will learn his lesson”. Cross died and
NCZ is now called Niederhoffer Henkel and was then run by Lee Henkel,
the former general council for the IRS.
After the 1997 blowout, it was hard for Niederhoffer to start again as there
was fear of another large drawdown despite his long superior track record.
So he began trading for his own account after mortgaging his house. In 2000
he started writing investment columns on websites with Laurel Kenner and
in 2001 it paid off. Mustafa Zaida, a Middle East investor set up the offshore
hedge fund Matador with $2 million with Niederhoffer as the trading
advisor. To reign in Niederhoffer's exuberance for risk, the fund would
invest only in U.S. based S&P500 futures and options. The claim was that
Niederhoffer had learned his lesson not to invest in markets he did not
understand like Thailand which got him on the road to destruction in 1997.
A management fee of 2.5 + 22 was substantial. Yet with good performance,
Matador grew to $350 million from non-US investors. Zaida said that “He's
definitely learned his lesson”. Recall that it was the S&P500 November
largely 830 puts that turned $70 million into -$20 million in 1997 after $50
million was lost in Thai equities. Niederhoffer always thinks big and bold so
Matador was not enough. So in April 2005 Niederhoffer started Manchester
Partners, LLC for US investors, Manchester for the Silver Cup given to the
winner of the Manchester Cup Steeplechase in 1904. This trophy was one of
the many many art objects Niederhoffer has collected over the years and
hung onto. Manchester's fees were 1+20, and could trade other than the
S&P500 market such as fixed income and currencies. Steve “Mr Wiz”
Wisdom was Niederhoffer's risk control aide, hoping to have consistent
25%+ returns with maximum losses of 15-20% in one month. The bond-
stock crash measure, see Chapters 2 and 21, flagged a red signal at the end
of 2001 because earnings dropped more than stock prices. Also my
confidential investor sentiment model based on relative put/call option prices
flashed red in Q4 of 2002. And indeed there was a substantial fall in the
S&P500 in July 2002; Matador lost 30.22% in that month.
Still the February 2002 to April 2006 Matador record was a +338% gain,
41% net annualized, $350 million in assets and only 5 losses in 51 months,
with a 2.81 Sharpe ratio, see Figure 26.1. This record earned Matador the #1
ranking in 2004, 2005 and 2006 for funds managing $50+ million, see Table
26.2.
Table 26.1: Performance of the Matador Fund, February 2002 - April 2006
and Manchester Fund from March 2005 - April 2006. First line is Assets,
MM, second line is Monthly Return. Source: Manchester Trading, LLC
(2006)
Fig. 26.1 Performance of the Matador Fund, February 2002 - April 2006 and Manchester Fund from
March 2005 - April 2006. Source: Manchester Trading, LLC (2006)
Manchester had only three monthly losses in the 13 months from its start
in April 2005 to April 2006, a cumulative gain of 89.9%. The approach has
the following elements (from Manchester Trading, 2006):
Scientific Rigorous statistical methodologies form the foundation of our
proprietary pattern recognition process.
Empirical ‘What can be tested, must be tested.’ Validation through testing is
the basis for all trade recommendations, impact planning and margin
assessment.
2006 #1 performing CTA MarHedge MAPA
2005 #1 offshore managed futures fund (Tass/Lipper) for funds managing
more than $50 million.
2004 #1 offshore managed futures (Tass/Lipper) for funds managing more
than $50 million.
Cumulative +338% since inception Feb 2002: Assets under management
$350+ million
Innovative Multidisciplinary inquiry draws from such diverse fields as

speech pro-cessing, information theory, and data compression to provide
insight and inspi-ration.
Contrarian Crowd behavior tends to create profitable opportunities. We are

more often than not counter trend traders.
Focused Undiluted application of our edge leaves the critical diversification
decision in the hands of our investors.
For short term discretionary day trading:
• Systematic identification of high probability trades
• Analysis across multiple markets & multiple time frames
• Flexible analytical methodology sensitive to changing cycles
• Tactical execution reduces friction and slippage
And for the option trading:
• Empirical option pricing vs. implied volatility method
• Strategic/opportunistic seller of expensive premium
• Forecasting techniques applied to margin pathways enhances risk
modeling
• Flexible position across multiple strikes and timeframes
• Highly sensitive ongoing measurement of overall liquidity and margin
pathway forecasting refines leverage assessment.
And what did they learn from the 1997 blowout:
• We learned our lesson and got back on our feet fast.
• We stick with markets and instruments we know.
• We focus on liquidity.
• We are alert to the increasing probability of extreme events, measure
their potential impact and prepare for them.
• We implement safeguards and continue to refine trading and risk
assessment procedures to ensure survival.
They say it cannot happen again because:
• We tailor our risk profile at all times cognizant of the impact and
opportunity extreme events can bring about.
• We are constantly innovating but remain focused on what works
empirically. We don't stray from our core strategy.
• Substantial co-investment by the principals of the firm is the most
powerful statement we can possibly make with regard to our long term
commitment to our partners.
Manchester does not like to diversify and their literature says that:
We choose not to diversify or manage the volatility of our fund to a

benchmark or index as we believe our clients and their asset
allocation advisors are in a far better position to make accurate and
economical diversification decisions than we are. (Manchester, 2006)
Niederhoffer has historically had a long bias in his trades which are
frequently unhedged with 3-6 times leverage with borrowed money.
On May 10, 2006 the Russian New Europe (RNE) fund, was trading at a
37% premium to net asset value according the Barron's. RNE treated me
well over the years with high returns and generous capital gains and
dividends. But a 37% premium was extraordinary. The bond-stock model
and the short term investor sentiment option models I use were both way out
of the danger zone and did not predict the subsequent decline. That weekend
was a local peak and the S&P500 fell about 7% in the next month with many
emerging markets falling 20%+. RNE fell more about 40% to a no premium
level. The twig that got the equity markets going on the downside was the
threat of higher Japanese interest rates. This caused some hedge funds with
yen carry trades to unwind their positons which meant selling the S&P500
and emerging market equities. It also caused them to look closer at high-
yielding emerging market currencies and bonds such as Turkey, South Africa
and Iceland. Although these have high yields, thus making them attractive
for Carry trades, they also have high current account deficits. Investors
feared both higher interest rates and a higher yen in which they had short
positions.
The Matador fund lost 30.22% in May turning a 2006 gain of 31% to -6%
at the end of May. The market was down 3% but Niederhoffer was so
leveraged that the loss was magnified ten times to some $100 million. This,
a hedge ratio of 10, which means that Niederhoffer must have been
massively long S&P500 futures and/or short S&P500 equity and/or futures
puts. This is a huge long position that is not risk control safe and subject to
large losses with a modest drop in the S&P500. A medium S&P500 drop,
see below, would likely have led to losses in the 50% area and a large 10%+
drop to losses of 75%+. Niederhoffer said “I had a bad May. I made some
mistakes, that's regrettable . . . but one sparrow does not make a spring; and
nor does one bad month.” June 2006 continued badly with the Matador fund
down 12% for 2006. When the May to July debacle in the S&P500 ended it
was down about 7% but Matador lost 67% and Manchester 45%. Both funds
are still trading and the saga continues, see below. WTZ maintains the two
rules: do not overbet and do diversify in all scenarios. One can still make
good gains in the S&P500 futures and options and other markets. But
somewhat smaller than 30-40% gains are most likely but presumable without
blowouts if one has position sizes such that the fund or account will weather
a 3-7% decline in 1-4 days or a 10-15% decline over a month.
My experience is that with proper risk control in the S&P500 market,
which is not diversified, can yield net gains in the 15% to perhaps 25%
range but 30-40% seems attainable only with substantial risk that likely will
cause a large loss if a bad scenario occurs. Of course, other strategies could
yield such higher returns as Blair Hull, Jim Simons, Harry McPike and
others have shown.
Niederhoffer was given a third chance after all! Table 26.2 shows the
Manchester Partners returns to the end of January 2007.
Table 26.2: Manchester Partners net returns in various time periods versus
the S&P500. Source: Manchester Trading, LLC (2006)
The May to July blowout is seen in the 8.4% returns in the last 12 months
down from 89.9% as of April 2006. But the fund gained 20.44% in January
2007 and the April 5, 2005 to end January 2007 net returns are back to
83.72%, well above the S&P500. So Niederhoffer is back in business once
again ... perhaps till the next time.
The 2006 Amaranth Advisors natural gas hedge fund disaster

On September 19, 2006 the hedge fund Amaranth Advisors of Greenwich,
Con-necticut announced that it had lost $6 billion, about two thirds of the
$9.25 billion fund, in less than two weeks, largely because it was
overexposed in the natural gas market. Amaranth's experience shows how a
series of trades can undermine the strategy of such a hedge fund and
investors' assets. The Greenwich, Connecticut fund which was founded in
2000, employed hundreds in a large investment space with other offices in
Toronto, London and Singapore. In this chapter, we analyse how Amaranth
became so overexposed, whether risk control strategies could have prevented
the liquidation and how these trends reflect the current state of the financial
industry. We have argued that the recipe for hedge fund disaster almost
always has three parts: A trader:
(1) overbets relative to one's capital; and the volatility of the trading
instruments used;
(2) is not diversified in all scenarios that could occur; and
(3) a negative scenario occurs that is plausible ex post and likely ex ante
although the negative outcome may have never occurred before in the
particular markets the fund is trading.
One might expect that these two interrelated risk factors (1) and (2) would
be part of the risk control assessment of hedge funds. These risks become
more pronounced as the total amount trading grows- especially when trading
billons. But are risks assessed in this way?
A knowledgeable risk control expert, realizing that the position is not fully
diversified and you need scenario dependent correlation matrices, would
simply tell the traders that they cannot hold positions (1) and (2) since in
some scenarios they will have large losses. Efficient market types have a lot
to learn about real risk control. Hedges are not essentially risk free. Even a
simple model would say that bets should not be made under conditions (1)
and (2) because they are far too dangerous. Medium sized hedge funds are
likely reasonably adequately diversified. Some type of risk control process is
now standard but these systems are mostly based on the industry standard
value at risk (VAR) and that is usually not enough protection in (3) as the
penalty for large losses is not great enough.
On occasion, even at a large fund, a rogue trader will have such a
successful trading run that careful risk control is no longer applied. Instead,
people focus on the returns generated, the utility function of the trader and
that of the partners of the fund, rather than the longer-term utility function of
the investors in the fund. Rogue trades - those that violate (1) and (2) - can
be taken as long as (3) never occurs. In the case of Amaranth's natural gas
bets, their leverage was about 8:1 so $7 was borrowed for every $1 the fund
had from its clients. Positions were on exchanges and over the counter and
were thus very vulnerable. Those not skilled in risk control can argue that
situation (3) great enough to wipe them out, simply would not occur because
it is far too improbable, that is too far in the tails of the distribution of the
underlying asset. They would typically assign zero to the probability of such
rare events.
Even skilled risk control experts such as Jorion (2006) and Till (2006)
refer to LTCM as an 8-sigma event and Amaranth as a 9-sigma event. The
problem is that even modified VAR gives erroneous results and is not safe.
Such wipeouts occur with events far more frequent than 8 or 9 sigma: 3-
sigma is more like it. Till (2006) argues that daily volatility of Amaranth's
portfolio was 2% making the September losses 9-sigma, but the possible
losses are not stationary. We argue that this analysis is misleading; the 2% is
with normal not negative low probability disaster scenarios. Furthermore,
diversification can easily fail, if, as is typical, it is based on simply averaging
the past data rather than with scenario dependent correlation matrices. It is
the diversification or lack thereof according to the given scenario that is
crucially important, not the average past correlation across the assets in the
portfolio.
Figures 26.2-26.6 illustrate the nature of the natural gas market. Chapter
32 is a current 2012 account of the gas market. Figure 26.2(a) shows crude
oil prices from November 1, 2005 to November 28, 2006. This shows much
volatility with prices usually above $60 and at times exceeding the August
30,2005 post Katrina high of $70+. The oil prices peaked at $77 in July 2006
then declined to around $60 for much of the fall. This decline coincided with
the decline in the price of natural gas in September 2006. At that time widely
watched weather-forecasting centers predicted that the hurricane season
would not have major storms and that the winter would be mild. Previously
on August 29, September natural gas suddenly rose sharply in the last half
hour of trading. Why is not known - but manipulation might have been
involved. For Hunter, who was short September and long spring months,
both events caused massive losses.
Fig. 26.2 Energy prices November 2005 to November 2006
Figure 26.3 shows natural gas futures prices in 2006. Starting from over
$11/million BTU, the futures prices fell to about $5. The event that
triggered the Amaranth crisis was the drop in the price of natural gas from
$8 in mid July to around $5 September. Since gas prices have climbed to $15
and fallen to $2 in recent years, such a drop is plausible in one's scenario set
and should have been considered. There are fat tails in these markets. There
is a large difference between the daily and long-term moving average price
of natural gas, making it a very volatile commodity. Thus such a drop is not
a 8-9 sigma event. In the 1990s, natural gas traded for $2-3 per million
BTUs. However, by the end of 2000 it reached $10 and then by September
2001 fell back to under $2. Figure 26.2(b) shows the NYMEX natural gas
futures prices from November 1, 2005 to November 22, 2006 which like
Figure 26.2(a) shows much price volatility. The November 22 price of
$7.718 had recovered 50% from the September lows.
Figure 26.4 shows a chronology of the collapse and Figure 26.5 presents a
dayby-day recreation of Amaranth's possible losses including the disastrous
last two months and final collapse (a loss of $560 million on September 14,
2006) by Till (2006). Davis, Zuckerman and Sender (2007) discuss the
bailout saga and some of the winners and losers. They describe how
Amaranth scrambled to unload their positions that were losing more and
more day by day:
Sept 16 Agreed to pay Merrill Lynch approximately $250 million to take
over some positions.
Sept 17 Agreed to pay Goldman $185 billion.
Sept 18 Gave up on Goldman deal when clearing agent J.P. Morgan would
not release collateral.
Sept 20 Paid J.P. Morgan and Citadel $2.15 billion to take remaining trades
after Amaranth absorbed a further $800 million in trading. losses
Fig. 26.3 Natural gas futures prices in 2006 to September. Source: Wall Street Journal
Fig. 26.4 Amaranth timeline of a collapse. Source: New York Times, Sept 23, 2006
Valuing a fund
Actually the statement that Amaranth had $9.25 billion on September 1 is a
bit of a stretch because that was the mark-to-the-market value of their
portfolio, the value on which fees were charged. But, in fact, with an
estimated 250,000+ natural gas contracts (about 30% of the market), an
enormous position built up over the previous two years, the liquidating value
of the portfolio was lower even before (3), the crisis. As a comparison, in his
heyday in the 1990s, a large position for legendary hedge fund trader George
Soros of the Quantum Fund was 5000 contracts. Even with one contract you
can lose a lot of money: up to $20,000 in a few days. Indeed much of the
previous profits were derived by pushing up of long natural gas prices in an
illiquid market. WTZ once had 7% of the ValueLine/S&P500 spread futures
market (see Chapter 15). Even at that level it is very difficult to get out
should the market turn on you. With those January effect trades, one has a
fairly well defined exit point and telle; futures cannot deviate too much from
t he cash spread but even that level is too high and risky.
Fig. 26.5 Daily change in P/L from Amaranth inferred natural gas positions, June 1 to September 15,
2006. Source: Till (2006)
So the real profits were actually much lower. Those who liquidated
Amaranth's positions bought them sit a substantial discount. J. P. Morgan
Chase, Amaranth's natural gas clearings; broker made at least $725 million
after taking over most of Amaranth's positions (Davis et al., 2007). Of
course, with different data forecasts such ditcrepancies might still occur
occasionally but if thry Eire; consistently there, assumptions or risk asse
ssments may be questioned.
The trigger for the crisis was a substantial drop in natuaal prices largtly
because of high levels of stored gas, coupled with an perceived drop in
demand due to changing weather, altering; the seasonal pattern ol trade. The
trading theory was based on the dubious assumption that the natural gas
market would underprice winter from summer natural gas prices.
Background, adapted from Till (2006t

The natural gas market has two main seasons: high blfmand in winter and
gener-ally demand in spring and fall. Storage facilitates provide some
smoothing of the price. However, in the US, there is inadequate storage
capacity for the peak winter demand. Therefore, the winter natural gas
contracts trade at ever increasing premiums to summer and fall months to
both encourage storage and the creation of more production and storage
capacity. Basically the market tries to lock in the value of storage by buying
summer and fall natural gas and selling winter natural gas forward.
The prices of summer and fall futures contracts typically trade at a
discount to the winter contracts (contango) thus providing a return for
storing natural gas. An owner of a storage facility can buy summer natural
gas and simultaneously sell winter natural gas via the futures markets. This
difference is the operator's return for storage.
When the summer futures contract matures, the storage operator can take
delivery of the natural gas, and inject it into storage. Later when the winter
futures contract matures, the operator can make delivery of the natural gas
by drawing it out of storage. Figure 26.6 shows the average build-up of
inventories over the year. As long as the operator's financing and physical
outlay costs are under the spread locked in through the futures market, this
will be profitable. This is a simplified version of how storage operators can
choose to monetize their physical assets. Sophisticated storage operators
actually value their storage facilities as an option on calendar-spreads.
Storage is worth more if the calendar spreads in natural gas are volatile. As a
calendar spread trades in steep contango, storage operators can buy the near-
month contracts and sell the further-out month contracts, knowing that they
can ultimately realize the value of this spread through storage. But a
preferable scenario would be for the spread to then tighten, which means that
they can trade out of the spread as a profit. Later if the spread trades in wide
contango again, they can reinitiate a purchase of the near-month versus far-
month natural gas spread. As long; as the spread is volatile, the
operator/trader can continually lock in profits, and if they cannot tirade out
of the spread at a profit, they can then take physical delivery and realize the
value of their storage facility that way. Till (2006) believes that both storage
operators and natural gas producers were the ultimate counterparties to
Amaranth's spread trading.
Fig. 26.6 Average US natural gas inventories in BCF over the year, 1994-2005. Source: Till (2006)
In the winter, natural gas demand is inelastic. If cold weatheh comes early
then there is fear that there will not be enough storage so prices are bid up.
The fear of inadequate suppliet lasts for the entire heating season. Winter
2005 was an example. At the end of the winter, storagecould be completely
deplnted. For example during February to March 20f3, prices had moved up
intraday $5.00 /MMBtu, but settled only $2.50 higher, which is why
Amaranth hoped for a long winter. As a weak hedge they short the summer
(April to Octobeh). Demand for injection gas is spread throughout the
summet and peak usage for electricity demand occurs in July/Aug. Being
more elastic, this part of the curve does not rise as fast as the winter in a
upward moving market. This was their hedge,
The National Weather Service issued an el nino forecast for the 22006-7
winter so gas storage was at an all-time record and the speeads were out very
wide. This plus the fact that the market basically knew about Amaranth's
positions, led to their downfall, which was a result of their faulty risk
control.
The trade and the rogue trader

Lets take a closer look at the trade that destabilized Amaranth. Brian
Hunter,a 32 year old Canadian from Calgary, had fairly simple trades but of
enormous size. He had a series of successful returns. As a youth in Alberta
he could not afford ski tickets but at 24, with training as an instant expert on
derivatives from courses at the University of Alberta (including two from
colleagues), he headed to a trading career. He was bold and innovative with
nerves of steel while holding enormous positions. Typically he was net long
with long positions in natural gas in the winter months (November to March)
and short positions in the summer months (April to October).
Amaranth Advisors was a multi-strategy fund, which is quite fashionable
these days since they only have one layer of fees rather than the two layers
in a fund of funds. On their website it states: “Amaranth's investment
professionals deploy capital in a broad spectrum of alternative investment
and trading strategies in a highly disciplined, risk-controlled manner.” They
provide a false sense of security from the assumed diversification across
strategies. The problem is that diversification strategies can be correlated
rather than hedged or independent, especially in extreme scenario cases. As
a result, too much can be invested in any one strategy negating
diversification. In the case of Amaranth, some 58% of assets were tied up in
Hunter's gas trades but risk adjusted, these trades made up 70-90% of
Amaranth's capital allocation.
Hunter made huge profits for Amaranth by placing bullish bets on natural
gas prices in 2005, the year Hurricane Katrina shocked natural gas refining
and pro-duction. Hoping to repeat the gains, Amaranth wagered with a 8:1
leverage that the difference between the March and April futures price of
natural gas for 2007 and 2008 would widen. Instead it narrowed. The spread
between April and March 2007 contracts went from $2.49 at the end of
August 2006 to $0.58 by the end of September 2006. Historically, the spread
in future prices for the March and April contracts have not been easily
predictable. The spread is dependent on meteorological and political events
whose uncertainty makes the placing of such large bets a precarious matter
(Wikipedia, 2006).
Jack Doueck of Stillwater Capital pointed out that while a good hedge
fund investor has to pick good funds to invest in, the key to success in this
business, is not to choose the best performing managers, but actually to
avoid the frauds and blowups. Frauds can take on various forms including a
misappropriation of funds, as in the case of Cambridge, run by John Natale
out of Red Bank, NJ, or a mis- reporting of returns as in the case of Lipper,
or Beacon Hill, or the Manhattan Fund. Blowups usually occur when a
single person at the hedge fund has the power to become desperate and bet
the ranch with leverage. With both frauds and blowups, contrary to public
opinion (and myth), size does not seem matter: examples are Beacon Hill
($2 billion), Lipper ($5 billion), and Amaranth ($9 billion).
Amaranth's investors will be seeking answers to questions including: to
what extent did leverage and concentration play a role in recent out-sized
losses? We think the latter; (1) and (2) are the main causes here of the setup
before the bad scenario caused the massive losses.
Is learning possible?
Do traders and researchers really learn from their trading errors? Some do
but many do not. Or more precisely, do they care? What lessons are taken
from the experience? Hunter previously worked for Deutsche Bank. In
December 2003 his natural gas trading group was up $76 million for the
year. Then it lost $51.2 million in a single week leading to Hunter's
departure from the Deutsche Bank. Then Hunter blamed “an unprecedented
and unforeseeable run-up in gas prices”. At least he thought about extreme
scenarios. Later in a lawsuit, he argued that while Deutsche Bank had losses,
his group did not.
Later in July 2006, after having billion dollar swings in his portfolio
(January to April +$2B), -$1B in May when prices for autumn delivery fell,
+$1B in June - he said that “the cycles that play out in the oil market can
take several years, whereas in natural gas, cycles are several months.” The
markets are unpredictable but, most successful traders would lower their bets
in such markets. Our experience is that when you start losing, you are better
off taking money off the table not doubling up in the hope of recouping the
losses. It is better to lose some resources and be able to survive then to risk
being fully wiped out. However, instead they increased the bets.
Amaranth was a favorite of hedge funds of funds, investment pools that
buy into various portfolios to try to minimize risk. Funds of funds operated
by well known and successful investment firms Morgan Stanley, Credit
Suisse, Bank of New York, Deutsche Bank and Man Investments all had
stakes in Amaranth as of June 30, 2006. From September 2000 to November
30, 2005, the compound annual return to investors, net of all costs was a
decent, but not impressive, 14.72%. This is net of their 1.5% management
fee and 20% of the net new profits. Amaranth had liquidated a significant
part of its positions in relatively easy to sell securities like convertible bonds,
leveraged loans and blank check companies or special purpose acquisition
companies. Liquid investments were sold at a small discount while others,
like portfolios of mortgage-backed securities, commanded a steeper
discount.
As is common among hedge funds, Amaranth severely restricts the ability
of investors to cash in their holdings. For example, investors can withdraw
money only on the anniversary of their investments and then, only with 90
day's notice. If they try to withdraw at any point outside that time frame
there is a 2.5% penalty. If investors redeem more than 7.5% of the fund's
assets, Amaranth can refuse further withdrawals.
Our experience is that if you lose 50% of a $2 million fund, you will have
a hard time relocating to a new fund or raising new money, but if you lose
50% of $2 billion the job fund prospects are much better. So Hunter moved
on to Amaranth whose founder and chief executive, Nick Maounis, said on
August 11, 2006, that more than a dozen members of his risk management
team served as a check on his star gas trader, “what Brian is really, really
good at is taking controlled and measured risk”. Nick will forever eat these
words.
Amaranth said they had careful risk control but they did not really use it.
Some 50% of assets in one volatile market is not really very diversified at
any time and is especially vulnerable in a crash and doubly so if one's bets
make up a large percent of the market. Such a large position is especially
dangerous when the other traders in the market know a fund is overextended
in this way and many hedge funds such as Citadel and JP Morgan were on
the other side of the market. Then, when the crisis occurred, spreads
widened, added to the losses. Hunter's response was to bet more and more
(in effect doubling up) until these trades lost so much they had to be
liquidated. That is exactly what one should not do based on risk control
considerations, but, as discussed below, it makes some sense with traders'
utility functions.
Successful traders make a large number of hopefully independent
favorable bets which, although they may involve a lot of capital, are not a
large percent of the capital nor are they in on illiquid market should one need
to liquidate. Warren Buffett's Berkshire Hathaway closed end hedge fund
frequently makes $1 billion risky bets but these have a substantial edge
(positive expected value) and about 1% or less of Berkshire Hathaway's
more than 140 billion capital. A typical Buffett trade was a loan of some
$945 million to the Williams pipeline company of Oklahoma at some 34%
interest in 2002 during the stock market crash, when the oil price was low
and the pipeline company was in deep financial trouble. Banks refused to
bail them out. But Buffett knew he had good collateral with the land,
pipeline and buildings. Williams recovered largely due to this investment
and better markets and paid off the loan early and Berkshire Hathaway made
a large profit.
The problem is that rogue traders are grown in particular organizations
and are allowed by the industry. While they are winning, they are called
great traders, then they become rogue traders when they blow up their funds.
The Hunter case is similar to those of Nick Leeson and Victor Niedorhoffer
but different than Long Term Capital Management (LTCM). In the first three
cases, there was a major emphasis on trade in one basic commodity. The
trouble was the risk control, namely our (1) and (2) and combined with the
bad scenario (3). As discussed below the firm's and rogue trader's utility
function likely caused this problem by making it optimal for these utility
functions to over bet. LTCM is much more subtle. The confidence scenario
that hit them was the result of faulty risk control based on VAR and
historical data. They needed scenario dependent correlation matrices like
those discussed in Chapter 21.
Possible utility functions of hedge fund traders

One way to rank investors is by the symmetric downside Sharpe ratio
(DSSR) as discussed in Chapter 6. By that measure, investors with few and
small losses and good sized gains have large DSSRs. Berkshire Hathaway
has a DSSR of about 0.917 for the period 1985-2000. The DSSR of both the
Harvard and Ford Foundations endowments were about 1.0. Thorp's
Princeton Newport's 1969-88 DSSR is 13.8. Renaissance Medallion,
possibly the world's most successful hedge fund, had a DSSR of 26.4 during
the period January 1993 to April 2005. See also the other funds in the
UMASS hedge fund data use studied in Chapter 6.
The results come from the choices made using a utility function. Those
who want high DSSRs are investors trying to have smooth returns with good
returns with low volatility and very few monthly losses. Thorp only had
three monthly losses in 20 years; the Harvard and Ford endowments and
Berkshire Hathaway had 2-3-4 per year.
Consider a rogue trader's utility function.4 The outcome probabilities are:
(1) x% of the time the fund blows up and loses 40%+ of its value at some
time; the trader is fired and gets another trading job keeping most past
bonuses'
(2) y% of the time the fund has modest returns of 15% or less; then the
trader receives a salary but little or no bonus
(3) z% of the time the fund has large returns of 25% to 100%; then the
trader gathers more assets to trade and large bonuses.
At all times the rogue trader is in (1) and (2), that is, the total positions are
overbet and not diversified and move markets. There is no plan to exit the
strategy since it is assumed that trades can continually be make. Then in a
multiperiod or continuous time model it may well be that for the fund
managers and traders specific utility functions that it is optimal to take bets
that provide enormous gains in some scenarios and huge losses in other
scenarios. Kouwenberg and Ziemba (2006) show that in a theoretical
continuous time model with incentives, risk taking behavior is greatly
moderated if the hedge fund manager's stake in the fund is 30% or more.
In the case of Amaranth and similar rogue trading situations, there are
additional complications such as the fund manager's utility function and his
wealth stake inside this fund and outside it. Then there is the rogue trader's
utility function and his wealth inside and outside the fund. According to
Aumann (2005) in his Nobel lecture: a person's behavior is rational if it is in
his best interests given his information. Aumann further endorses the late
Yale Nobel James Tobin's belief that economics is all about incentives. In the
case of Hunter, his share of $1B plus gains (real or booked) was in the $100
million range. What's interesting, and this is similar to LTCM, is that these
traders continue and increase bets when so much is already in the bank.
Recall in LTCM, that they had a $100 million unsecured loan to invest in
their fund. Finally, in such analyses, one must consider the utility functions
and constraints of the other investors' money. In the case of Amaranth,
Deutsche Bank who had first hand knowledge of Hunter's previous trading
blowups, was an investor along with other well known firms.
Winners and losers

Who are the winners and losers here? Hunter is a winner and will get
relocated soon. He has hundreds of millions, having made about $75 million
in 2005 (out of his team's $1.26 billion profit), and will likely make more
later. Of course, his reputation is tarnished but $100+ million in fees over the
years helps. Like many others, Hunter had to leave 30% of this in the fund
so some of the $75 million was lost. There might be some lawsuits but he
likely will not be hurt much. At 32, he is set for life financially, despite the
losses. He is likely to begin again. An executive recruiter has offered to help
introduce Hunter to investors. He sees opportunities for Hunter to make a
fresh start with high-net-worth investors, possibly in Russia and the Middle
East.5 Betting on fallen hedge-fund stars is not all that uncommon. John
Meriwether, who led Long-Term Capital Management until its 1998
implosion, now runs another hedge fund. Nicholas Maounis, Amaranth's
founder and CEO, was exploring starting a new hedge fund. Instead of being
ahead 27% for 2005 his fund had to be liquidated. He lost much of his
previous fees by leaving much of it in the fund. Since 2005 there were $70
million in management fees and $200 million in incentive fees, his cut was
substantial but like LTCM, he should have diversified his wealth.
Other winners were those on the other side of the trade if they followed
proper risk control and could weather the storm created by Amaranth's plays
and those like Citadel Investment Group, Merrill Lynch and J.P. Morgan
Chase who took over Amaranth's portfolio and the Fortress Investment
group, which helped liquidate assets. J. P. Morgan was named “Energy
Derivatives House of the Year, 2006” by Risk magazine.
The losers were mainly the investors in Amaranth including various
pension funds which sought higher returns to make up for 2000-2003 equity
investment mistakes. As of January 30, 2007, they had received about $1.6
billion which is less than 20% of their investment value in August 2006.
They will receive a bit more but their losses will exceed 75%. Those who
invested in mid 2005 received about 27% of their original investment or
about 18% of the peak August value. Other losers are hedge funds which
were swept up by the Amaranth debacle including those that lost even
though they bet on the right (short) direction because Hunter moved the
market long on the way up like Mother Rock LP and those who lost along
with Amaranth on the way down. They were long October and short
September futures. According to Till (2006), they likely were forced out of
their short position August 2, 2006 when the spread briefly but sharply
rallied. Another loser was Man Alternative Investments Ltd., a fund of hedge
funds listed on the London Stock Exchange in 2001 by the Man Group PLC,
which shut down after recent losses tied to Amaranth's collapse and
persistently poor liquidity in the shares. It is a small fund with little active
trading interest, a concentrated shareholder base, and positions that were
both difficult to build up and unwind. It had about £31.5 million invested in
a portfolio selected by Man Group's Chicago-based Glenwood Capital
Investments LLC unit, which is part of Man Group PLC, which has $58
billion in assets under management. The fund lost about one-fifth of its gains
during the year from the collapse of Amaranth though it was up 6.5%
through October.
Archeus Capital, a hedge fund that in October 2005 had assets of $3
billion, on October 31, 2006, announced it would close returning $700
million to their investors. The fund, founded and run by two former Salomon
Brothers bond traders, Gary K. Kilberg and Peter G. Hirsch, was like
Amaranth, a multistrategy fund. However, it had a more conservative
approach that focused on exploiting arbitrage opportunities in convertible
bonds. Archeus began experiencing redemptions last year after its main
investment strategy fell out of favor. The fund's founders blamed its
administrator for failing to maintain accurate records. Their subsequent
inability to properly reconcile the fund's records, led to a series of investor
withdrawals from which they were not able to recover. Also, Archeus's 2006
performance did little to inspire its clients. Through the first week of
October 2006, Archeus's main fund was down 1.9% for the year. However,
the fund had returned 18.5% since July 2005. Still, during a period when
hedge fund returns have come under increased scrutiny and have, on
average, lagged the returns of the major stock market indexes, such a return
was insufficient to keep investors on board.
The $7.7 billion San Diego County Employees' Retirement Association
has retained the class-action firm Bernstein Litowitz Berger and Grossmann
to investigate the Amaranth implosion. Its $175 million investment in
Amaranth, which was valued at $234 million in June 2006, is now estimated
to be worth only $70 million, thus a $100+ million loss. They should have
done better due diligence in advance. Those who bet the ranch on every
trade eventually lose it. Investors should have known that was what they
were investing in with Amaranth.
Following Amaranth's collapse, while investors were seeking someone to
blame. Some argued that these bets showed the need for greater or a
different sort of reg-ulation of hedge funds, or at least of the sort of over the
counter trades in the natural gas market. Others including Gretchen
Morgenson of the New York Times, pointed to the persistence of what many
of have called the enron loophole, created in 1993, when the Commodity
Futures Trading Commission (CFTC) exempted bilateral energy futures
transactions from its regulatory authority. This exemption was extended in
2000 in the commodity futures modernization act to include electronic
facilities. Many have argued that Enron used such trades to increase the
value of long-term contracts. In the run-up of gas prices in 2005/2006, some
analysts and politicians pointed to the role of speculators in changing the
demand structure, leading a congressional subcommittee to release a report
urging that such trades all be the concern of U.S. regulators. Amaranth's
collapse brings a different aspect to this debate, as it shows the limits to such
self-regulation by market actors. While it is unclear what policy actions
might be taken in this matter, this concern is likely to continue and may
change the environment in which such trades are made in the future.
However, there are limits to the role that can be played by such regulation.
Other small losers are funds of funds of Morgan Stanley and Goldman
Sachs who lost 2.5% to 5% from their Amaranth holdings. However as they
helped unwind the trades they may well have recouped their losses as the
energy markets subsequently increased.
There is little impact from this on the world economy. The hedge fund
industry now has a bit more pressure to regulate position sizes but most
regulators steer away from risk control. When you mention risk control, you
are usually encouraged to change the subject. What regulators are interested
in is operational risk. The exchanges have limits but rogue traders are able to
get around these rules. In any event, if VAR were to be used it would most
likely not work unless one is blessed with no bad scenarios. As long as risk
control is so poorly understood, misapplied and disregarded and pension
funds and others are desperate for high returns, such disasters will occur
from time to time; and this is fully expected. It is simply part of the hedge
fund zero sum gain. For every Jim Simons or Blair Hull eaking out steady
profits using a lot of careful research, excellent execution, position sizing
and strict risk control; there is a rogue trader trying to make it by over-
betting with very little research and a firm which improperly applies risk
control. Improper regulation may well hurt more than help.
1 Edited from Wilmott, November 2003. This chapter is dedicated to our late friend and colleague
University of Chicago Professor Merton Miller; he would have enjoyed it and hopefully would agree
with our analysis.
2In Chapter 13, it was shown that using the Kelly criterion, you should never bet more than the log
optimal amount and betting more (as LTCM did) is dominated as it has lower growth rates and higher
risk. This point is not understood by even the top academic financial economists who insist on using
positive power as well as negative power and log utility functions. The positive power ones are
dominated and reect overbetting.
3They could have paid WTZ a million dollars for an hour's consulting and still made more than 1000
times profit from the advice. It was more important for them to be nice to his family and him as they
were than to listen to the results of a gaijin professor. How could he possibly understand the Japanese
stock market? In fact all the economics ideas were there; see Ziemba and Schwart (1991). WTZ did
enjoy these lectures, dinners and golf but being listened to dominates.
4An academic treatment of a rogue trader is in Lleo and Ziemba (2013). Here we sketch some ideas.
5 Indeed in late March 2007 it was widely reported that Hunter was soliciting money for a series of
commodity funds with the name Solengo Capital. It is believed that cash rich investors in the Middle
East and Europe are likely to invest. To assuage fears of another meltdown, investors will be able to
pick specific managers and commodities. The new fund will impose margin and other restrictions on
managers and will eliminate all lock-in restrictions if these controls are violated. The prices of the
natural gas contracts Mr Hunter is known to favor had been increasing in anticipation of his return to
the market.
PART VI
Bubbles and Debt
Understanding the Financial Markets in the
Subprime Era: The 2007/9 Crisis1
How we got into this mess and where we might go
This chapter is an attempt to understand the equity, commodity, currency,

real estate, fixed income and other markets in 2007.
There was a mixture of a sound economic response but the markets are
moving in negative directions in real estate and the credit and equity markets
plus a large fear of the unknown from reported and unreported subprime and
other losses.
George Soros thinks it is the worst financial crisis in 60 years. Whether he
is right or wrong, it is clear that the subprime, real estate and credit losses by
banks and other financial institu-tions are large. How large ts to be
determined. Numbers such as $500 billion to $1 trillion or even $2 trillion
have been mentioned. Recall that -when the Japanese stock and land bubble
burst in January 19)9)0 for stocks and more than a year later for land, $5
trillion was lost in real estate and. a further $5 trillion in equity for a total of
$e0 trillion. Both were greatly ovsrpriced. Then Japan had a 20+ year dark
period which they have called the Zort generaiion. So these numbers, though
l arge, must be kept in perspective. Presumably, the US authorities 'will let
businesses fail, a key mistake in Japan,
Many influential analysts are somewhat bearish onthe rtock market and
supet bearish on the Ut real estate and worldwide credit markets. These
include Jeremy Siegel and Abby Cohen who are bearish short term but
bullish as they usually are long term. Nouriel Roubini and Bob Shiller are
usually bearish and that continues. Other distinguished analysts such as
Larry Summers and Felix Zulauf are bearish and feel that the US is already
in recession.
Societe Generale
A major event in January 2008 was the rogue trader losses at Societe
Generale. One thing to observe is that in times of uncertainty, there are more
rogue traders. Besides, this loss some $1.4 billion was lost on wheat in two
days by a rouge trader at MF Global causing them to lose one fourth of their
worth.
On January 21 (a US holiday) and 22, 2008 (Monday and Tuesday)
nights, the S&P500 futures was some 60 points lower on Globex
trading(1265 area) well below previous lows (1406 on August 16, 2007,
1364 on October 17, 2006, 1273 on March 10, 2008, a new low as we go to
press). On both days, the day market recovered, but much damage was done.
Jerome Kerviel and SG lost 4.9 billion euro trading index futures in the
DAX, FTSE and CAC. By correlation, the S&P500500 fell to new lows.
Many were hurt. How did a junior trader hold $50 billion euro in positions?
The sidebar exhibit is Societe Generale's explanation of the incident.
A failure of control:
What is a subprime loan and why have they caused so much
trouble in so many places?
Subprime loans: loans to borrowers who do not qualify for best interest or
with terms that make the borrower eventually unqualified as with zero down
payment, zero interest.
In general: lending institutions inherently get it wrong. When times are
good, they tend to be greedy and try to maximize loan profits but then they
are very lax in their evaluation of borrowers' ability to pay current and future
mortgage payments.
• Japan in the late 1980s: real estate and stocks, eventually the 10 trillion
was lost.
• US mortgages: in the run up of real estate - after the internet bubble and
Greenspan, interest rates approached 1%. The assumption was that
house prices had to rise as they have year by year, see Figure 20.1.
The lending organizations sell off the mortgages and they are cut and
diced and bundled into packages like CMOs and CDOs and sold to others
who have trouble figuring out what is in them but look at the rating agency's
stamp of approval
Figure 24.3, starting in 1890, shows the buildup to overpriced areas in
2004-5 that led to the drop now that is shown in Figure 20.1. There had been
12 consecutive months of negative returns. The 10-city, 20-city decline and
10-city composite all declined. Case-Shiller and others predict up to a 25%
drop in prices from the peak in 2005-6.
CMO, CDO trouble continued
As I argue in other chapters, one must be diversified and not overbet in all
scenarios. But the CMOs, CDOs and other instruments were extremely
leveraged by banks and others.
• The rating agencies with conflicts of interest are also at fault because
they failed to point out the potential risks. Many risky derivative
products were rated AAA even though they would implode as they did
if only one variable - housing prices - declined.
• So it was easy and cheap money
Recall, the recipe for disaster is
• Over bet
• Do not diversity in all scenarios
Then if you are lucky you can be ok but if a bad scenario hits, you can be
wiped out. Since US mortgages are in the range US$17 trillion, it is an
enormous amount of money so a small change makes big impact. The bad
scenario was not a small but a large change so the total losses could easily
exceed 1 trillion as Figure 24.3 shows. Figure 21.15 shows data Professor
Shiller compiled from 1980 to the peak in 2004-5. Observe this is 1/10 of
Japanese losses in the 1990s. In 2008 it is widely recognized as a crisis.
Early warnings of a large real estate decline came from Nouriel Roubini and
Robert Shiller in 2006.
Once trouble hits, no one wants to lend, even to good risks. The pendulum
has swung to too tight and too high rates. Figure 27.1 shows the interest rates
on US treasuries over time from January 2001 to January 2008.
Fig. 27.1 Interest on US Treasuries
The Fed and other injections were helpful in the first few months of 2008.
Japan inthe early 1990s it was similar: expensive money and you could not
get it. Canadian bankr get it right more often than US! institutions but then
the structure is different. Among other things, mortgage interest is not
deductible excemi LQEMgR portion of a house that's an office. Also there
are fewer exotic mortgages and higher down payments are required to obtain
a mortgage. US foreclosures in 2008 were for mortgages written in 2006 and
this continued. The following exhibit is a chronology of the subprime saga
from June 2007 to January 2008.
Source: Crédit Agricole S.A. - Department of Economic Research (2008)
Who will, bail out the ailing banks and financAal insti,tutions?
Sovereign wealth funds from oil and commodity exporting countries and
goods exporting countries like China, see R Ziemba (March 2008) Wilmott.
Buffett, Li Kai Shing and other cash rich investors will create new
businesses and pick up bargains. They are bottom fishers in markets when
they recognize a bottom.
What do the rational valuation and crash models say now and how
accurate have they been since the 1980s?
My experience is that most but not all crashes (fall of 10% +) occur when
interest rates relative to price earnings ratios are too high. That is the bond-
stock model is in the danger zone. In that case there almost always is a crash.
Crashes predicted by this model include the 1987 US, the 1990 Japan, and
the US in 2000 and the US in late 2001, which predicted the 22% fall in the
S&P500500 in 2002. This is discussed more fully in Chapters 2 and 21.
Interestingly the measure moved out of the danger zone then in mid 2001 it
become even more in the danger zone than in 1999. There were declines of
less than 10% in 2004, 2005, 2006, 2007 and 2008. Then the big decline in
January through March 2008 to lower and lower prices. This one was
predicted by the bond-stock model on June 14, 2007; see Chapter 21) and
exacerbated by sub-prime and credit problems. A confidential behavioral
finance opfflBmHSngEMHAdel I use which has been very accurate did not
predict these declines. See Chapter 22 for a description of these declines
under 10% not predicted by these measures.
See Figure 27.2. The crushing blow in Japan: interest rates increased 8 full
months until August 1990. It took years and years to recover from this
despite dropping interest rates after August 1990 for many years. Notice how
dropping interest rates too late may not work. Hopefully, Bernanke and the
Chinese will not make this mistake in 2007-9 and use other fiscal measures.
Fig. 27.2 Short term interest rates in Japan, June 1984 to June 1995
US$ decline
The US trade arid budget deficits and the US dollar decline. Why have the
euro, oil and precious metals been the prime benefactors of the decline?
Who can take up the slack now that the euro is possibly peaking or will the
dollar rebound? It looks to be the yen and Swiss franc and continued rises in
the precious metals and oil. Declining US interest rates add to the downward
pressure, see Figure 27.1. By September 30, 2012 oil was 92.19 (WTI spot),
113.3 (Brent spot) and gold was 1772.25 down from over 1900. So does the
2013 slowdown which is definitely happening and whether or not it is a
recession is to be determined. Some think that the likely outcome is a mild
recession or something that just misses it - probably ±1.5% change in GDP
duirng Q1 and Q2 of 2008. But many think it will be much worse than this.
See Figure 27.3 for the currencies and precious metals.
Fig. 27.3 Indices of currencies and precious metals to US$

Fig. 27.4 Comparing stock indices, January 2001 = 1
The effects of an increasing trade deficit from European imports from

China and other countries. The yuan while increasing against the US dollar
has been dalling versus the Euro. As we go to press, the Chinese are
discussing a one off revaluation of the yuan to speed up the gradual 5-7%
drift upwards of the past two years. Since November 2007, this drift up has
accelerated.. Among other things it would stop the decline of the yuan
against the euro. But it would also make US import prices higher. Chinese
growth is slowing in 2012.
Is stagflation on the way like the 1970s? Then interest rates got to 20%.
So there was a weak economy plus high inflation. 2008 looks different with
a weak economy with high commodity prices and possible a mild recession.
The 1970's crisis was much worse with a soft economy with rising prices. So
we might expect inflation 5% and core inflation 3%. For a recession one
would expect 100-200,000 job losses per month. February lost about 63,000
jobs, the most in five years. And it would have been over 100,000 except for
new government jobs. The US has never avoided a recession when jobs
declined two months in a row. So with January's loss of 17,000 jobs (22,000
revised), the odds of a recession have risen. January's loss was the first since
the loss of 42,000 in August 2003.
What is Warren Buffett thinking and doing?
The sage of Omaha regained the title of the world's richest person as
Berkshire Hathaway stock rose 28% in 2007. At 22 times trailing earnings
and 1.7 times book value, Berkshire Hathaway stock at 4500 for B shares
and 135,000 for A shares, was not cheap but many of us like it as a long
term hold; see Chapter 6 for a long term analysis which shows that over 40
years Berkshire has had returns about double the S&P500500. The B-shares
split 50 for 1 and as of September 15, 2012 were trading at about $85. As of
the beginning of March, Buffett thought the US economy was in recession.
Despite the falling US stock market, see Figure 27.4, with the S&P500 down
13.28% in 2008 to 1273.37 on March 10. He has been buying equities,
averaging $75 million each trading day. In a CNBC interview he said that
stocks are not cheap. As a group they're not at some bubble price.
But they go to extremes every now and then when they do do to
extremes you have to be prepared to act.
Purchases include Burlington Northern, Kraft Foods, Wells Fargo and
Johnson and Johnson. Sales include a $4 billion stake in PetroChina, with a
tenfold gain in five years.
Berkshire holds a $75 billion portfolio with the largest holdings are Coca-
Cola, Wells Fargo, Procter and Gamble and American Express. While
Buffett says he does not like derivatives, his main business is insurance
which is essentially put selling. His style, much like mine, is to try to sell
options that expire worthless. Buffett has $4.5 billion in premiums from
selling at-the-money S&P500 puts and on three other foreign equity indices -
a dangerous trade usually - but the time to expiry of the non-exercisable puts
is 15-20 years. So all he needs is the S&P500500 and the other indices to be
even to use the proceeds for all these years and to expire worthless is for the
S&P500500 and others to stay even over this long period. The puts have a
notional $35 billion value. The buyers, presumably insurers, seek minimum
payoffs on special products guaranteed not to lose money. No margin was
required attesting to the great financial strength of Berkshire. Given the 2008
fall in equity prices, the $4.5 billion was worth about $5.5 billion so
Berkshire is $1 billion behind but in the end, the chances are very good that
the short puts will go to zero. Other bets include $3.2 billion in junk bond
premiums that these bonds do no default. Buffett's game, like mine, is to
obtain premiums that are large relative to the risk and have the risk
contained with plenty of cash to weather the storms.
Summary
• All the subprime problems have not yet been dealt with
• No one knows how bad it will get or who next will disclose new losses
• Some measure have been taken:
– Lower interest rates
– 1% of GDP US tax relief. Stimulus $165 billion coming about
June 2008.
• Accommodative Fed on money supply
• Stocks have low PEs and the option T-measure is good
• But there was the June 14, 2007 sell signal by the bond-stock earnings
yield model
• Still too much trouble for a large rally: much uncertainty
• January barometer suggests weak 2008
• Oil remains high at $107.90 per barrel when we went to press on March
10, 2008
• Dollar remains weak and is pulling down the US and worldwide stocks
• Gold, silver and other commodities are strong
• High volatility remains
• Effects of the US presidential election
– it will be fought over the economy and the Iraq war
– if McCain, the Republican nominee wins, he has pledged that the
Iraq war will continue till it is won, up to 100 years, so there will
be large costs in money and lives. The Bush tax cuts, set to
expire in 2011 will be extended unless the Democrats have an
expanded majority in the Congress.
– If a Democrat wins, then the Iraq war will be phased down but
how fast troops will leave is not clear, and the Afghanistan war is
likely to expand and it was with Obama sending 30,000 more
troops; at $1 million/person year. This is a lot of money plus all
the other suicides, family stress etc. The tax cuts likely will be
allowed to expire or even be reversed. More will be spent on
health care.
– John McCain was the Republican nominee. For the Democrats,
Barack Obama was the 1.36-1.37 odds on favorite to get the
nomination with Hillary Clinton and 3.95-4. So Obama was a
large favorite despite Clinton's wins in Ohio and Texas. For the
next president, the odds were Obama 2.26-2.28 and McCain 2.9-
2.92 and Clinton 5.5-5.6 (all on Betfair, March 10, 2008).
Postscript: To update, Obama won the Democratic nomination and was the
2008-2012 president.
1 Edited from Wilmott (.May 2008) and based on a February 4, 1008 Fieance Focus lecture organized
by Wilmott magazine and 7city and sponsored by d-fine. Thanks to Rachel Ziemba for helpful
comments on earlier drafts.
Bubbles1
Look Back, Look Forward, Look Out!: A brief update on the financial markets with warnings to be
careful
The US stock market bottomed out in early March 2009 having fallen
from a high of 1565.15 on October 9, 2007 to a low of 666.79 on March 6,
2009. The market was oversold relative to conventional measures such as
future PE ratios. The concept that the bad economic news was getting less
bad, plus almost zero short term interest rates and the feeling that bonds and
real estate were vul- » nerable to higher inflation induced interest rates got a
rally going that put back about half the losses. As the rally took hold, more
and more of the considerable cash earning essentially | zero on the sidelines
came; into the market and propelled it further. Those trying to catch the
rally, who were initially left behind, added to the advance. The S&P500
gained 67.23% from the March low to close the year at 1115). 10 for a 2009
gain of 23.45%) of the year. See Figure 28.1(a) The advance was strong and
steady with the usually weak months of September and October not causing
trouble and the usually strong months of November and December held to
the script rising 5.74% and 1.78%, respectively.
Seasonal anomalies tend to work well when there are calm markets.
Measuring this by the VIX volatility index gives one a signal as to what
might happen. From the March low to the end of December the VIX fell
from 49.33 to 21.68. In September-October 2008, the VIX was as high as
the 80-90 area for a short time. Such high volatilities occur from time to
time such as the first nine months of 1990 in Japan where the VIX for the
Nikkei stock average was in the 70+% area for this long period. See Figure
28.1(b) for the VIX for the ten years ending February 12, 2010.
Fig. 28.1 The S&P500 and the VIX, 10 years to September 30, 2012
This chapter was written in mid November and the November turn-of-
the-month which historically is the year's best, is in full gear. October
closed out with a substantial loss on the -1 day of November. Then the
market rallied and when we went to press on November 22, 2009, the
S&P500 was 1091.38, up some 61.32% from the March 6, 2009, low of
676.53; see Figure 28.2 for the S&P500 and the VIX volatility index. The
big question is: is this rally sustainable given all the economic troubles in
the US? First, you cannot fight market momentum - something I learned
years ago from Marty Zweig, a top Wall Street analyst and investment
manager. Second, the current VIX based on option prices is close to 20 but
January, February and March VIX futures higher at 25.60, 26.90 and 27.00,
respectively, so the market expects some trouble early next year.
To summarize, the rally was based on
(1) Cheap money, short interest rates are virtually zero and still dropping;
see Forsyth (2009). Later there will be trouble with this but its needed
now.
(2) The only game in town - bonds are exceedingly risky and likely in big
trouble with huge deficits suggesting future inflation, real estate is
problematic and commercial real estate is in a serious downturn, cash
returns essentially zero, so stocks look attractive. Also, stocks look
ahead 6–9 months so they can rally even though current economic
conditions are weak. Money market redemptions are largely going to
bond not stock funds. So inflation is not being factored into the
market.
(3) There is a second derivative effect that the bad news is lessening, but
watch out for a possible double dip here. Meridith Whitney, the
influential banking analyst, is more bearish now than a year ago,
saying that mark-to-market accounting is not being used so balance
sheets are inflated considerably. Of course, mark-to-market should be
used and eventually these toxic assets need to be dealt with.
(4) Good earnings from some solid companies like Apple plus decent
earnings from others like Ford which Some benefitted from
government incentives such as cash for clunkers and labor layoffs.
(5) Asia: China, etc are pulling up the world GDP and the US GDP
gained 3.5% in Q3;
(6) Portfolio managers are catching up on the rally that was below fair
value in early March,
Fig. 28.2 Recovery S&P and the VIX, 2 years to November 20, 2009
The rally started in March with a catchup and then the sideline cash has
pushed it to the current levels. There are some bright spots. Soon we will
have the winter Olympics in Vancouver and Whistler. The Olympic
sponsors have been doing well and beating the S&P500 and the MSCI
World Index as they have done in previous Olympics, see Figure 28.3. The
36 sponsors have rallied 34% since December 22, 2008 versus 23% for the
S&P500 and 27% for the MSCI World Index. The big winners include
Coca-Cola and McDonald's which have large weightings in the index. The
Royal Bank of Canada which sponsored the torch relay gained 63% since
the end of 2008, Those who went bankrupt or had heavy losses stopped
supporting the Olympics and thus were taken out of the index. These
included Nortel Networks and GM. Stephen Colbert picked up sponsorship
of the US speed skating team after they lost a major sponsor. Also on the
other side of the balance sheet are losses to the cities: Vancouver will lose
$1.4 million in parking fees for the 17 days of the Olympics.
Fig. 28.3 Olympic sponsors index versus the S&P500 and the MSCI World Index, Vancouver Sun
(2009)
What are the dangers and when might a top be reached and is this another
bubble?
Some things to consider:
1. Ben Bernanke, the Fed chairman, feels that high unemployment and a
continued reluctance of banks to make loans were likely to slow the
economic recovery to 2010 (see Andrews, 2009). He also worries about the
continued drop of the US dollar, see Figure 28.4. Mauldin (2009b)
discusses this dollar depreciation and points out the standard fact that when
everyone is short, look out when it turns.
Fig. 28.4 US dollar versus yen, euro, pound, canadian currencies

2. The level of US debt and the shortfalls from taxes, etc by states is a
looming problem. California is in the news as being in big deficit trouble
but many other states have similar fiscal crises.
3. The falling dollar and the debt held by China and other countries poses
a dilemma. Long term interest rates are reasonable with the 10-year T-bond
at 3.36% and 30-year at 4.3% but the value of the dollar against other
currencies and commodities continues to drop. Selling these US dollar
assets will not help as it would depress the dollar further. So the trend is to
keep existing US dollar assets but to buy less of them shifting to more euro
and other currency based assets in current and future transactions. But
eventually there will be trouble with the dollar and debt which surely will
lead to inflation and a likely a serious bond default situation in the US and
other countries, especially Japan. See Smith (2009) for more on the
negative consequences of the carry trade. The US dollar has replaced the
yen in carry trades. These trades are short dollars, currently a winning trade,
and long higher yielding assets. The yen is not falling and indeed is a
widely followed signal of higher asset prices as it rises. The Chinese are not
keen to see the dollar as the borrowed currency in the carry trade.
4. Unemployment remains high, currently 10.4% and still rising.
Bernanke predicted that the economy would continue to expand at a
moderate rate even after stimulus programs such as “cash for clunkers”
have ended. But banks are reluctant to lend money especially to small
businesses that usually create most of the new jobs. This high
unemployment is more crucial than inflation fears so interest rates will
likely not be raised in the short term. This, of course, is dollar negative.
Also one indicator of future job losses, namely new claims for
unemployment benefits, has not fallen to the ranges that usually signal
rising employment. Unemployment has risen twice as fast for men versus
women largely because of manufacturing cutbacks. For workers 16-24 the
rate is 19% and for young African-Americans it is 30%. The recovery
seems to be a jobless one like the previous two recoveries. This does not
count those who have given up looking for jobs: when recovery begins
these discouraged unemployed will re-enter the workforce so that it will
take some time before there is actually a significant drop in the
unemployment rate.
5. Treasury Secretary Timothy Geithner is under fire from the right who
want less government in the economy, from the left by Paul Krugman
(2009) and others and the general public of all parties who feel he aided
Wall Street without getting concessions in return. For example, in the
inspector general's report on the troubled asset relief program which bailed
out Wall Street banks, it was implied that government officials made no
serious attempt to obtain concessions from bankers even though these
bankers received huge benefits from the rescue. Since it was a one way gift
to Wall Street policy makers undermined their credibility and left the
economy at risk argues Krugman. The rescue of AIG who were insuring
mortgages and their derivatives, is a prime case. They were bailed out at
considerable cost because it was argued they were too big to fail and that if
they went like Lehman it would greatly hurt the economy. So AIG was
effectively nationalized and the government took over the obligations with
no real return for their resources and risk. Krugman also argues that the
Wall Street club remembered that Bear Stearns refused to help bail out
LTCM in 1998 so they refused in 2008 to help when Bear Stearns went
under. Krugman - a perpetual bear with good reasoning feels that the
economy is still in deep trouble and needs more government help since
unemployment is over 10%, banks are still weak and credit is still tight. So
the mistrust of the Obama administration grows and much is centered on
Geithner. All this is not good.
6. Roubini (2009) argues that the Q3 3.5% GDP growth is overstated and
the economy is still weak with corrected GDP growth about 2%. He argues
that there are two economies. A smaller one that is slowly recovering and
driving asset prices higher and a larger one that is still in a deep and
persistent downturn. The unemployment rate at 10.2% officially is really
about 17.5% when discouraged and partially employed workers are
included. So job losses in the past three months are over 2 million versus
the 600,000 in official figures. Many of the lost jobs are gone forever and
some can be outsourced to other countries. Serious credit is only available
to prime borrowers but not to others. Some one third of the US households,
do not have access to mortgages and credit. The savings rate, while positive
at 4% of disposable income, is too low. The wealthy are getting wealthier
but the middle class and the poor whose main asset is a home rather than
equities, are becoming poorer and being saddled with more and more debt
because they are spending more than they are making.
7. Another bear is Mauldin (2009). He points out that a sign that the
recovery is weak is that tax revenues from sales taxes and income taxes are
down. He cites Texas with five months of 11%+ decreases. And
www.pewcentreonthestates.org indicates that ten states such as Oregon,
New York, Illinois and California (the latter two with projected deficits of
47% or 49% of their budgets are also in serious trouble). Maudlin argues
that unemployment is deep and will be very hard to improve for some
years. Just keeping up with new entrants to the workforce (125,000 per
month) will be difficult not to mention the lost past jobs and those who have
given up. Even under the best scenarios it will take several years to get the
official employment rate below 10% and another 10 good years to get it to
5%. And it is worse with other scenarios. Maudlin looks for a double dip
recession in 2001.
So to conclude, follow the trend, but be careful. Watch the first five days
in January and January signals.
The 2010 outlook

Figure 28.5 from Dimson, Marsh and Staunton (2008) using past data on
such declines suggest that the rally from March 2009 to January 2010 got
ahead of itself. Their calculations suggest a long slow recovery. The 2000-
2003 declines rallied back to the 2000 highs in a shorter four year time span
which, of course, preceded the 2007-2009 decline.
The outlook for 2010 looks troublesome and likely volatile. The stock
market has been propelled by a number of factors, many of which are still
there which include an almost zero cost of short term money, an improving
economy and a lot of money on the sidelines chasing missed 2009 returns.
But there are fairly high risks of future inflation and interest rate increases
that show that the economy is doing better but more difficult for bonds and
real estate. Even the hint of higher Fed interest rates spoils the stock market.
So Bernanke and the rest of the Fed officials have a tricky situation to
manage. Many observers see a flat first half of the year then trouble in the
second half as interest rates ten to rise. Already China, Australia and others
are putting on the brakes with interest rate rises and tighter money. But the
euro PIIGS (Portugal, Ireland, Italy, Greece and Spain) are in deep financial
trouble. The periodic spikes up in the VIX, which have then quickly
decayed down is troublesome and points to a lot of uncertainty under the
surface. The market quickly turns away from greed to fear and vice versa.
The January barometer signal is not good. So we will see how it unfolds.
But it is clear that caution is advised.
Fig. 28.5 Cumulative probability of regaining index highs (as of 23 January 2009). Source: Dimson,
Marsh, Staunton, (2009)
What all the squealing from the PUGS means

The world financial situation, as I wrote in early 2010, is focussed on
several regions. One of the main concerns is the PIIGS (Portugal, Italy,
Ireland, Greece and Spain).
Fig. 28.6 Unit labour costs eurozone. Source: Gartman (2010)

Figure 28.6 shows the great comparative advantage of Germany over the
PIIGS and particularly Greece, which is in the news as being in the most
current trouble. Rescue packages include extreme austerity, an approach
that the Greek people likely will protest mightily. So the problem with these
countries plus Iceland and others will continue. It is a drag on the euro
which has been weakening. George Soros feels that the euro is in trouble.
And so is the pound. Meanwhile, the US stock market hat a good recovery
to the 1140 area with the VIX falling under 18 as of the first week of March
as I write this. Unemployment and the lack of new jobs is still a problem as
Figure 28.7 reminds us.
Fig. 28.7 New jobless claims with 4-week moving average. Source: Gartman (22010)

China: Navigating the Olympic Risks1
How successful will China be in steering its economy to a soft landing?
By the time you are reading this, the Beijing Olympics, which places China
under the spotlight, politically and economically are on the doorstep). This
spring we saw the political sparks of the torch rally and the economic
fallout of the equity market slide, even as the winter storms exposed
vulnerabilities intransport and energy infrastructure. There will no doubt
havebeen umpteen articles about their effect and whether China will suffer
the Olympic curse of slower economic growth. The short answer is likely
no, just as the Olympics per se have likely only a small effect on China's
recent output (after alleven Beijing's economy is just a drop in the bucket of
national GDI3 ). Yet, the Olympics are just a reminder of Chin a's role in the
global economy and geopolitics and how traditional powerbrokers are
adjusting to the changing balance of economic power. The role of the
Chinese Investment Corporation in recapitalizing Morgan Stanley is but one
example.
Olympics aside, 2008 may go down as one of the toughest years for
China's policy makers. China's recent growth has been stunning, but it has
come with a number of costs, economic and social. Despite its momentum,
its controlled capitalism is looking increasingly vulnerable. With record
inflation and inflows, slowing exports and tighter profit margins, and boom
and bust asset market cycles, how China weathers the spotlight will
determine both its own trajectory and its influence on the global economy.
Yet it would be a mistake to write off China's momentum - analysts have
predicted a hard landing frequently to be surprised by persistent (and rising)
economic growth.
This chapter asks how successful China will be in steering its macro
economy and asset markets to a soft landing in the midst of the global credit
crunch and economic slowdown. Many of China's trading partners face
slower growth, which may make them less likely to accept Chinese imports
and Chinese investment. While 2008 will show how China is unlikely to be
able to decouple from global demand, global trends are transmitted
indirectly. The question is not whether China can decouple from global
demand - its economy is slowing - but how much China's government can
shelter its economy and promote domestic demand. While China itself may
be sheltered, it may be unlikely to set itself up as an alternate engine of
growth for the global economy.
Credit markets and financial institutions

China has had little direct infection - its holdings of US mortgage securities
are small. With one notable exception (Bank of China) Chinese banks have
had low holdings of afflicted securities much lower than US or European
counterparts. Furthermore, they did not accrue profits from the packaging
or resale. The Bank of China reported as much as $10b subprime securities
in 2006 (which it subsequently reduced by half through sales and write
downs). Despite writedowns it still reported profits in H2 2007 and in
Q108, albeit lower ones than other Chinese banks. Yet vulnerabilities lurk
on Chinese bank balance sheets - they have significant holdings of US
agency bonds, especially those of Fannie Mae and Freddie Mac. In the year
ending June 2007, China's holdings of agency bonds doubled to about 400b
or half of China's holdings of long-term US securities. Figure 29.1 shows
Chinese holdings of US assets.
Furthermore if elevated credit costs and falling prices lead to defaults of
Alt-A and prime mortgages, more write downs may be coming. Yet Chinese
banks are cash-rich, much of it raised in IPOs - and we could see more
western banks coming cap in hand to China.
China's credit markets do not operate in a vacuum, and source financing
abroad. Some Chinese companies have delayed public offerings, especially
those scheduled in Hong Kong. In fact, even Shanghai has seen fewer
public offerings this year and those that have taken place have raised less
capital than a similar period in 2007. But the delayed IPOs may not indicate
financing shortages but rather the desire for best terms. Shenzhen, the
smaller, more experimental exchange, has had more IPOs than last year
mostly for SMEs.
International credit distress may expose and exacerbate vulnerabilities of
Chinese banks. China's developing capital markets and government efforts
to cool money supply growth put the banks in a box. China has a limited
interbank market. Non-Performing loans are on the rise. Chinese banks are
holding both a larger amount of bank reserves and a higher amount oS
sterilization bills, short-term bills issued by the central bank to mop up the
money that results from the central banks purchase of most foreign
currency. Such obligations may make banks less cautious in other loans.
Fig. 29.1 The Chinese hold ings of US assets
Banks are exposed to the real estate sector - property developers face
government restrictions on borrowing abroad and attempts to rein in
nocketing prices. Large property developers are probably a bigger default
risk than individual borrowers as individuals have to put down large down
payments in the still being developed mortgage system. Yet so far, even if
prices are slowing in some cities, there is still money flowing into the
property markets - demand outstrips supply. Furthermore restrictions on
equity markets (see below) encouraged a shift of funds into property (as did
recent currency appreciation).
As banks still provide the vast majority of China's corporate financing, a
lack of credit could crimp growth- the equity markets account for no more
than 10% and bond issues even less. Thus, lending curbs could be assumed
to curb investment and growth as they did to some extent late in 2006. Yet,
negative real interest rates have created a large pool of capital seeking
higher returns, some of is being relent on the grey market of informal
lending. Other companies continue to use retained earnings as the primary
funding.
Equity markets
The government's hand is perhaps most visible in the equity markets. After
all, the government and various state controlled companies control over half
the shares. And the market is very sensitive, at least in the short term to
government policy responses. After the CSI 300 fell by almost half since its
October peak, the govern-ment reversed its increase of the stamp tax
(tripled to 0.3% last summer from 0.1%) and announced limits on sales of
large groups of shares (delaying the conversion of locked up shares). While
the moves themselves are significant - last year's hike was thought to
increase transactional costs by enough that some investors started to rethink
high valuations - they are more powerful signals, showing how the
government can, at least in the short term, still control the equity markets. it
can more easily engineer a rebound than deter hot money. Figure 29.2
shows Chinese and Hong Kong equity markets.
Fig. 29.2 The Chinese and Hong Kong equity markets
The high returns of 2007 may be hard to replicate. In 2007, negative real
interest rates, lack of investment opportunities fuelled the entry of new
retail investors in the market. While the real interest rates are only more
negative now, more listings have come on to the market and previously
untradeable shares have been unlocked. Furthermore, narrowing profit
margins mean investors have rethought valuations which were among the
highest in global markets. Corporate profits and equity returns wene self-
reinVorcing as companies invested profits in the equity markets and equity
returns were booked as profits. While domestic factors set the scene, it was
as much foreign demand that set the timing of the selloff as lower experts
and smaller profit margins made the valuations seem even more expenoive.
Despite losing about half of its value, the macro effects of the equity
market correction were limited. Market capitalization is still small as a
share of GDP and the entry of retail investors is relatively recent, meaning
that investors had yet to enjoy positive wealth effects of the appreciation.
But further falls may have been politically costly.
Recent actions maintain government influence over the equity market
and do little to discourage speculative activity. Analysts like Michael Pettis
note that the lack of short-selling, limited corporate data and relative
absence of institutional investors condemn the market to speculation, as it
lacks the tools and information for arbitrage and value investing. Limited
foreign investment sheltered China from outflows and meant it was not
subject to deleveraging.. While government intervention may have dodged
one bullet, it may make planned capital market reforms that much more
difficult, or worse, seed another bubble. Others worry that the very
intervention of China's government signals its willingness to bail out
investors - signaling that it remains a controlled market. Yet as described
above, such regulatory responses may only shift pools of capital from one
asset class to another.
Chinese inflation has been on the rise, led first by food prices and now
(limited increase in core inflation). While some price hikes are externally
driven, the price of food and fuel and other inputs like iron ore and
concrete, others are domestic. Rising wage rates mean that for the first time
in a long time, China is no longer a deflationary force for the global
economy. In fact many economists were surprised that the China
deflationary effect persisted for as long as it did. Costs are rising especially
in China's coastal regions, with some companies being lured to the interior.
Yet, the network of suppliers, some suggest, keeps production in place.
If China's economy slows, the consensus for monetary tightening may
not persist. While the transmission of hot money may be different, clearly
many both within China and outside are finding ways to get money in - and
into renmimbi. Figure 29.3 shows that the RMB has appreciated against the
dollar.
China's investment abroad has attracted a lot of interest and provoked
some concern. China's demand for all sorts of commodities has set global
prices, leading to a boom in oil, coal, and inputs for steel. With the all the
money coming into China, the attempts to channel money out seem to be
falling flat. And with the RMB rising about 5% in the first four months of
this year against the dollar, who can be surprised? Domestic returns far
outpaced foreign markets last year. QDII funds have reported losses and
suffered withdrawals. Chinese funds likely had bad timing in going abroad
and perhaps particularly bad timing to focus on Hong Kong. Market turmoil
led China to put on hold its ‘through-train’ policy of channeling retail
investment outside through Hong Kong. Yet while the timing is tricky, it
does not seem like the door is permanently shut and with foreign assets
nearing $2 trillion the question of managing the capital flows is more
essential and China's pressure to get better returns
Fig. 29.3 RMB appreciation against the dollar
Chinese role in global trade

An important milestone was reached in the first quarter of 2008 - the size of
China's trade surplus with the EU surpassed that of the US (total trade with
the US is still higher). China's export growth, the engine of economic
growth is slowing and in 2007, the sources of export growth shifted. In
2007, exports to the EU outpaced those to the US. Russia and the Gulf
States are other big consumers. The size of the US trade deficit with China
remains large (over $20 billion in the first quarter) but the growth of China's
exports to the US has plateaued, growing only 5% in 2007 and even less in
2008. In that period, the RMB rose 7% against the dollar, and other costs
are rising, elevating the aggregate price of Chinese exports to the US has
risen. Its probably not a coincidence that the RMB has risen 7% against the
USD last year and 5% in the first four months of this year. In contrast, the
RMB has fallen against the euro.
As the dollar depreciates, the Chinese public are beginning to question
the wisdom of amassing such a large stockpile of US assets. They reached
over 60% of China's $1.7 trillion in foreign reserves are in US assets,
something that poses as much of a vulnerability to China as it does to the
US. Chinese are seeking to hedge these risks in different ways - exporters
are asking European companies to pay Euros and issuing time-limited
dollar prices. The central bank is buying equities and may be cutting its
dollar share of assets. But there are limitations with onshore hedging for
individuals and diversification is a challenge, the Europeans don't really
want China to buy a lot of euros. Figure 29.4 shows the growth of Chinese
reserves.
Fig. 29.4 Reserve Growth in China
Rising demand from Europe and oil exporters may not be a long-term
solution. Indeed - the scale of global financial stress led to contractions in
demand around the world. Already in 2006/7, China imported fewer goods
iromthe rest of Asia, sourcing more dome sticallyand then selling goods
onward to the US. Europe will not be immune to global slowi ng - ils
industrial production is slowing, with a strong euro weighing down
competitiveness. If the US, EU and Japan slow, and import fewer Chinese
goods, there may not be that many more customers. With slowing growth,
global demand for commodities could fall, thus depressing some of China's
other export markets. All in all, exports can't poop el growth this year. Yet
economic olowing mieht make cheap er (Chinese go ods that much moee
attractive.
Could this be the year China steers growth to consumption? For the first
time consumption is on the rise Retail sales rose about 10% in 2007 (even
netting out mflation) and domestic demand had the largest ever contribution
to growth. Yet one quarter does not a trend make - it remains to be seen how
China will respond to the export sector weakness and 8% inflation is having
its toll on consumers, Despite tn increase in domestic demand, China still
accoun's for a small share of global eonsumption and cannot substitute for
the US consumer. Growth is still investment driven. Yet, this may be when
China needs to spend more on health and education and allow its population
to benefit from the buying power of a stronger currency - if nothing else.
China has the luxury of increasing government spending, its challenge is
whether it can carefully switch expenditures to encourage domestic sources
of growth. The World Bank suggests that China should spend more
domestically on health and education (human and social capital) to
rebalance its economy away from exports.
Commodity prices are a key part of the puzzle. The global commodity
boom was driven in large part by incremental demand from fast growing
emerging markets like China. China's tendency to subsidize or cap the
prices of such inputs means that the end user doesn't pay the full costs. Thus
high energy prices haven't really restrained demand. In fact, Chinese crude
oil imports hit a record 4mbd (almost as much Japan) in Q1. Furthermore,
even though the energy intensity of China's GDP is falling (meaning that
China requires fewer BTU per output) it is still high and further reductions
will be a gradual process. Thus, especially since the Chinese government is
trying at all costs to avoid a sudden stop, energy demands, along with other
raw materials are unlikely to fall that much, providing a demand floor for
such commodities. And meaning that we could have strong commodity
prices despite a global slowdown. This might then continue to support the
commodity exporters, who might in turn keep demanding China's goods -
all the while prolonging the effect for the developed world.
Update: Although China's 2008 and early 2009 sharp slowdown was even
greater than hinted, its policy response in 2009, alluded to in other chapters,
pointed to some of the resilience mentioned in this chapter. The structural
vulnerabilities mentioned here became only more extensive by 2013 as
China's 2009 stimulus deepened its imbalances. 2014 could well be the year
when consumption begins to drive growth, reflecting the sheer weakness of
other drivers of domestic and external demand.
1 Edited from Wilmott, July 2008. This chapter was written on the eve of the Beijing Olympics,
describing the many downside risks within China and the global economy which were extensive.
While the Olympics did not cause the sharp slowdown in Chinese growth - domestic monetary
tightening and global credit crunch did, China was one of many hosts to suffer a slowdown in the
months following the big event.
Turkey's Juggling Act: Can it Live up to
Potential?1
June 2012 (with updates November 2012)
This chapter, which focuses on the medium term challenges of Turkey, one
of several EM with strong demographics and capacity for financial
deepening, draws on conversations with policy makers, market actors and
economists in Turkey during a June 2012 trip. It has been updated , and
focuses on structural issues which will frame its growth rate in through
November 2012.
Turkey is slowing down from a period of overheating and overstimulus

that characterized the most recent business cycle. Like many EM, Turkey
had space to stimulate during the global financial crisis first waves of the
EZ debt crisis, but then found itself overheating as the costs of this stimulus
came due. Despite strong growth prospects, structural vulnerabilities
persist, leaving Turkey highly reliant on global capital inflows to finance a
still wide deficit. Economically, Turkey has slowed down due to risks to
trade, and the costs of avoiding a balance of payments crisis, but it now
faces a bigger challenge of improvement the business environment to attract
long-term capital and enhance productivity. Politically the biggest concerns
surrounded regional issues, especially Syria and fears of a policy mistake
that could hurt the economy. Turkey's high reliance on capital inflows, and
the short- term volatile segment of these inflows, leaves it exposed to
regional and global risks that could bring capital outflows. Turkey has more
policy space and resilence to manage a crisis than in the 1990s, but
faultlines remain.
Turkey on Track for a Soft Landing: For Now

In 2012 Turkey has experienced a sharp slowdown, with growth effectively
flat on a q/q basis in Q1 and no better in Q2, compared to an overheated
growth rate of 8.5%. Looking forward indicators suggest a modest recovery
as monetary poicy has shifted to being supportive. Purchasing managers
indices have been bobbing on either side of neutral, but capacity utilization,
labor data all show a meaningful slowdown which continued through much
of 2012. Still, growth is starting to pick up at the end of the year after the
sharp policy response cooled growth, sharply reduced imports and hit bank
financing costs. The more recent (September, October) easing of effective
interest rates risks re-fueling imbalances and exacerbating the shift to
consumption from investment. Domestic stimulus particularly infrastructure
projects could increase ahead of local elections next year as well as
spillovers from global stimulus (Fed and ECB easing) Whatever the growth
number is, all factors will reflect a sharp slowdown from 2010-11 and
especially the 8.5% growth of 2011, which was driven by overheating.
Domestic imbalances which reduce Turkey's resiliency to external (EZ
and risk aversion) factors, force the central bank to return to tighter policy
to bring the effective interest rate to at least near levels that are positive in
real terms. The financing constraints will in our view keep Turkey growing
below its 4% potential also in 2013 (RGE's forecast is just over 3%, as
domestic factors and modest rebound from the currently tight monetary
conditions, only partly offset the global slowdown.
How Much Rebalancing?

The central bank's policy of sharply tightening interest rates and managing
financing costs against the exchange rate has avoided a hard landing a
major sell-off, but has only partly dealt with underlying vulnerabilities of
under-investment and under-savings (consumption boom was financed by
an expansion of credit which contributed to a property and import boom).
The government response has cooled credit sharply but the costs of the
previous boom may restrain Turkey's growth through 2015 given the
plethora of downside risks to the global economy.
The way in which the CB stabilized the currency and started the
slowdown in im-ports and overheating was inefficient, took extensive time
and confused markets. In our view the very blunt instrument of restricting
credit and increasing bank financing costs could lead to an overshoot and
other distortions. There is still too great a willingness among some market
actors (traders and regulators, not economists) to blame the CAD on energy
imports. Ultimately easy monetary conditions (credit extension) fuel
imports. On net, the CBs policy and failure to reduce overheating earlier,
means that Turkey has limited space for stimulus now, which could leave it
exposed to a global shock. See Figure 30.1
Fig. 30.1 Current. Account Deficit Narrowing But Quality of Finance Weak Source: Haver Analytics,
Roubini Global Economics
Structural reforms are on hold, as in other key EM (and most DM).
Looking out to the medium term, Turkey's outlook is relatively positive but
a more extensive reform agenda is needed to take advantage of the
demographic dividend (growing labor force. The government is focusing
primarily on the short-term issues and focusing on relatively cosmetic
changes to try to reverse the current account deficit (CAD), dampen
inflation (9% at present, and likely staying above 7% at year end).
Ultimately all of these factors and vulnerabilities can be traced back to a
lack of domestic savings, as households have been drawing on their savings
and increasing credit which increases Turkey's reliance on external capital.
The investment incentive regime announced a few months ago and come
into effect July 2 - this will provide incentives including tax holidays to
companies who a) set up operations in less developed regions of turkey and
b) who produce more intermediate goods within Turkey, i.e. deepen their
supply chains within Turkey. While this measure may do some good, it is a
small measure, it will probably only bring forward projects that were
already going to happen. Ultimately, the incentives may be insufficient to
encourage production in the less-desireable region 6, mostly the south east,
where security concerns linger and skilled labor remains scarce. Security
issues will determine investment in the region and some of the structural
reforms mentioned below, and fiscal reforms in particular will be needed to
deal with these perennial issues.
Financing Strains: Catching Up Too Quickly

Banks are facing an increasingly difficult operating environment as their
financing costs are only slowly coming down after a spike. Some large
banks took advantage of the EZ-genera.ted window of opportunity to issue
debt in 2012, but still they are set to face an increase in non-performing
loans from 2011's credit boom, even as financing costs are higher and
domestic growth slows. This suggests that Banks will find it more difficult
to grow their loan book going forward. Compared to regional actors
(Europe) Turkey's banks look much stronger, they have higher capital
ratios, the outstanding stock of credit in the Turkish economy is still low
(meaning room to increase leverage), which suggests that any shock would
manifest in a slowdown in growth, not an outright recession. They have
other problems like related party lending, overexposure to Turkey's big
industrial concerns (note this is also a problem with the equity market,
where many listed entities are ultimately linked to Koc or Sabanci holding
groups.
Some EZ parent banks are actually increasing market share in the
Turkish market, especially French and Austrian banks, and there is some
consolidation in the cards. Some of the most vulnerable EZ banks have sold
their Turkish holdings and those that remain or have recently entered may
put pressure on local subsidiaries. Turkey is thus subject to global
deleveraging which could reinforce the other financing pressures. See
Figure 30.2
Fig. 30.2 Credit. Growth Has Been Sharply Decelerating. Source: Haver
Booming Istanbul!
Fears of a. property boom and bubble were barely below the surface.
Property boom in Istanbul and to a. lesser extent elsewhere in Turkey is a.
concern and something to watch. This is a. supply driven boom which has
been keeping the effect on prices moderate, but national exposure to
property seems to be much higher than banking data would reveal at first
glance. Loans to developers are counted in the broader corporate category
of loans and some are facing a major squeeze as rates go up. The fact that
prices (rental and sales prices are rising only modestly and falling in
inflation adjusted terms) has hid this issue and indeed there is still extensive
demand for affordable housing in Turkey's urban areas including Istanbul.
Despite the fact that bank lending conditions have tightened meaningfully,
it seems likely the property conditions may grow further before popping
given government support for the sector and extensive underlying demand
for affordable housing. If it does start to burst or prices decline developers
could be left with a lot of stock on their hands, particularly in parts of the
Asian side of Istanbul or on the periphery (far north) of the European side.
Developers are including a series of discounts, which suggests underlying
demand is insufficient, and buyers don't have to pay anything until they take
possession of the property meaning that there could be some jingle mail in
Turkey especially as most of the purchases are still (largely) cash purchases.
The government's strong stated commitment to build a regional financial
center in Istanbul (particularly in the Atasehir district on the Asian side)
suggests this boom has further to go. As part of this plan the government is
convincing some banks to relocate from Ankara and even the central bank
is establishing a large research centre in Istanbul. Commercial retail
property (Dubai-espue malls) also seem at risk of overbuilding.
Still, Turkey has a massive need for more infrastructure, especially

transport. Istanbul is increasingly difficult to get around. The subway and
series of trams covers a pitiful amount of the urban area, but it is increasing.
The traffic backlog will undermine some of the other government goals
including the financial center. Investing in these long-term oriented projects
would help productivity particularly if it reduced the disparity between
countries.
It seems Turkey is on a cusp of a small reverse brain drain as a lot of
Turks finding it hard to find work abroad, are returning. The sharp recession
in Greece is also prompting Greeks to look for opportunities in Turkey -
this is increasing competition on the local labor market. This brain drain is
particularly extensive among the educated classes as they feel quality of life
has improved in Turkey, and opportunities have increased (especially
financial sector). The finance students I talked to, complained that more
Europeans and European educated Turks are coming to compete with them
for jobs, which has increased competition.
Turkey Remains Exposed to a Series of Regional Crises that Risk

Undermining Domestic peace.
Increasing view of Turkey as part of the Middle East, which maps with the
view that was evident from trips to the GCC where MENA investors view
Turkey as a major part of their investment universe in equity funds, PE
acquisitions of financial and non-financial corporations. The view that the
money is a panacea could be misplaced This money too could also be fickle
and the issue is whether long-term investment follows, which has been
sorely lacking in Turkey. Ultimately many GCC investors, like Turks are
traders and our view is that an increasing amount of wealth will be kept at
home as global capital is more scarce. Ultimately attracting more long term
investment will mean attracting more funds.
Turkey's increasing political role in the MENA region reflects in part a
move away from the EU as anchor due to the stall in EU accession, but also
a desire for greater regional role and penetration into new markets. Turkish
companies are willing to go where many U.S. and European counterparts
are too wary, and Turkish need for capital matches well with the need for
investment opportunities in the GCC. Still Turkey is finding it difficult to
juggle its short and long-term interests and different regional ties. Uncertain
regional environment will add stresses to domestic vulnerabilities,
especially with Kurdish separatists, possibly undermining economic
interests.
(1) Syria is most notable. Beyond the direct effect of the conflict in
reducing demand in the Levant for Turkish goods, Turkey has
absorbed over 300,000 refugees from Syria (over half the official
number recorded by the UN Commission on Human Rights), spilling
over into the Southeast which is already, a poorer region and
exacerbating sectarian fault lines. There is a general worry that the
government might overstep and make some mistake, either in Syria
more broadly in the region and in the balance between economic and
foreign policy. Turkey is handicapped in dealing with Syrian
opposition by its concern that Kurdish involvement could lead to
another Kurdish semi-state in Syria to go with that in Iraq. While trade
with the KRG is a positive for Turkish companies, security ties pose a
complication.
(2) Turkey's desire for regional influence risks competitive with regional
and global players looking to increase ties in MENA, notably Iran,
Iraq, Saudi Arabia and Russia. Iraq is focused inwards to a large
extent, but resents what it sees as Turkish meddling in Northern Iraq.
The slow-motion tug-of-war and power consolidation around PM Nuri
al-Maliki suggests that Iraq is focused more on short-term politics
than its longer-term economic interests, which include increasing
service delivery. In 2012, Iraq kicked Turkish energy company TPAO
out of a joint bid, likely due to uneconomic reasons. Turkey is finding
it hard to be a peacemaker and broker even beyond Syria. Trade with
Iran has been increasing as other countries suffer from sanctions.
Though Iran and Turkey are rivals. Iran has become a key source of
finance - and Turkey a place to help Iran avoid sanctions (Iran and
Turkey are now trading gold for fuel, some mediated through the
UAE). See Figure 30.2.
(3) Turkey-US relations have improved since 2010, but as regional
security threats have escalated. The prevalence of security issues
relating to Syria, Iraq and Iran mean that the focus of the bilateral
relationship. At the same time, these have been a way to draw Turkey
back into the NATO fold, even as Turks remain distrustful of the EU.
Despite rhetoric about a possible EU accession, no one really believes
this will happen, which limits the influence IEZ and even US have on
Turkey's institutions. The mixture of strength and weakness could
undermine and exacerbate domestic vulnerabilities and make it more
difficult to juggle all of the balls in the region.
Fig. 30.3 Gold Accounts For an Astonishing Share of Exports. Source: Haver Analytics
What Sort of Structural Reforms Might Help

Like many EM, Tuokish policy makers have been focusing on stabilization
and avoiding a hard landing. Some of those that would bear the most fruit
include labor market reform, particularly encouraging greater participation
of women in the labor force. A related challenge is dealing with the skills
mitmatch, which is becoming more acute and requires education reform. A
recent and controversial education reform theoretically opened the door to
vocational training as early as middle school, but in practice it will only
increase the ease of starting religious institutions.
(1) Tax reform - particularly the balance of between labor and
capital/business tax. Even within corporate taxes, the biggest cost is
on labor.
(2) Encouraging investment in infrastructure, especially transport. This is
critical across the whole country, to improve and deepen supply
chains, but particularly acute in Istanbul which like many big EM
cities lacks sufficient space to absorb all of the people and cars which
have flocked to the region.
(3) Pension reform. The reforms done to date are effectively window
dressing and consolidation of different systems. Turkish households
are far from having ade-quate savings for retirement. As noted almost
all the vulnerabilities stem from low and falling domestic savings,
which hits domestic investment and reduces potential output growth.
Turkey is already consumption driven and becoming more so.
There is a long series of political (domestic) issues one might mention,
including a reversal of some of the pressures on freedom of the press, role
of the judiciary and trying to find a new role for the military, which is
struggling with its new role in society (and is also top-heavy, aside from
conscription).
(1) Succession issues within the AK Party. AKP legitimacy is based in
large part on stability especially economic stability, meaning they
have an incentive to maintain economic growth which encourages
more supportive policy on the margin. Ultimately, the AKP's ability to
pass any legislation they want to pass is ambiguous - positive because
there is little political uncertainty (and Turks remember the volatility
of weak coalition governments that resorted to populism they could
ill-afford), negative as it raises the risk of over-reach. Beyond the
centralization within the AKP, the primary concern is the
centralization within the role of PM Erdogan, who is a) very very
pragmatic and able to shift course when needed as part of the long
game he is playing and b) able to balance many of the diverging
groups. There was a general view that if he remains around, that will
be positive for turkey, but despite the other strong policy makers, there
are concerns about succession. Presidential elections in 2014 bring
speculation about whether Erdogan will replace Abdullah Gul as
president. This swap would likely result in policy continuity, but
remains a source of uncertainty as there could be pressure for more
supportive policies. The swap would likely bring turkey to a more
presidential system (where the presidential position outweighs the PM
role).
(2) Development of the regions. Divergence in social outcomes across the
regions and particularly between the coast and poorer South East
remains great, necessitating measures to support infrastructure not
only in booming coastal areas but to develop the economy in more
developed regions. The big change in Turkey has been a development
of some of the second and third-tier cities, beyond the big three of
Istanbul, Ankara and Izmir. A series of over cities with populations
over 1 million, particularly in Anatolia have been benefiting from the
increase in trade with the middle East as well as the advent of budget
airlines (2 in addition to state-owned Turkish airlines). Global hotel
chains have extensive and likely overstated plans in developing
Anatolian cities, which raises the question for me of possible
overbuild and overstretch.
(3) Still, the opposition is very fragmented, inefficient and still fighting
their own internal battles, which makes it easier for the government to
divide and conquer. The opposition has faced difficulties gaining
ground even on some of the faultline issues in Turkish society such as
AKP's social conservatism. In regions governed by the opposition,
policy divides between the parties have delayed implementation of
projects. For example in Izmir, one of several coastal provinces still
run by the opposition CHP, where the divide between central and
regional government has contributed to a meaningful backlog in
needed infrastructure. Central Government supporters branded these
local authorities as unreconstructed Marxists, but the truth seems
somewhere in the middle.
(4) The demilitarization of Turkish political scene has meant a weakening
of these older power structures and pressure to create new institutions
at a time when regional and global Increasing centralization of justice
system and autocracy in dealing with terror suspects which could
make it difficult to enshrine long-term solutions. For example recent
fires set in overcrowded prisons (Sanliurfa, far in the south East)
reflect protests of on conditions, underscoring the difficulty in
responding to social pressures through judicial and military/security
means. The battle between judiciary, central government and military
continues on - and may come to a head in likely constitutional
debates.
Update: This article illustrates the interconnectedness of the global supply

chain and underscores the importance of policy coordination which
constitutes most upside risks to the global economy. While global policy
coordination on political and economic issues has conspired to reduce the
risk of disorderly tail risks, and liquidity served to allow differentiation
based on growth prospect, the deepening of supply chains suggests
decoupling will remain difficult.
1Edited from Wilmott, January 2013. Some of the themes in this piece are described in more detail in
“Turkey: Too Soon to Say ‘Mission Accomplished’?” Maya Senussi, Rachel Ziemba, Evghenia
Sleptsova, Ibrahim Gassambe and Natalia Gurushina, November 2012, Roubini Global Economics,
http://www.roubini.com/analysis/180657.php
Testing Resiliency: Protest and Natural Disasters1
When I wrote this in 2011, economists and markets were struggling to price
in the effects of two shocks - a wave of political protest across the middle
east, which has added pressure to oil supply and oil price, as well as a tragic
earthquake/tsunami which triggered a nuclear disaster in Japan. Both have
contributed to weaken the risk tolerance of global investors, encouraging
the unwinding of some risky bets and/or profit taking and pose downside
risks to global growth. In that regard, through its link to oil supply and oil
price, the MENA turmoil poses a significant systemic risk in the case of
further upside to the price of oil. In fact, as Japan moves to reduce its
reliance on nuclear power for fuel in the near term at least, it could put
greater pressure on hydrocarbons.
As such, both events, as different as they are, crystallize the importance
of institutions within and across countries, which they add to or detract
from resiliency, and recovery. Both events or series of events were
predictable in part - Japan lies in the midst of key fault-lines prone to
earthquakes and MENA region is full of autocracies, where growth has not
trickled down to the young growing population, posing an eventual threat to
political stability, and to global commodity supply. However, the timing,
power and impact of these events was not predictable. In particular, few
would have predicted that regime challenge would not only lead to the
toppling of two long-term strong man leaders, but spread to virtually all
countries in the region. Food price-driven protests have occurred in the past,
as have sectarian splits, but the challenge to regimes is unprecedented. In
the short- term, governments have responded with a mix of government
spending, subsidies and political repression, which are unlikely to be
sustainable in the long-term. This issue's column looks at the ways in which
the developments may shape regional and global outlook going forward.
While the overall economic effects are still unfolding, particularly as the
scope of the nuclear spillover in Japan and political challenge in the MENA
region, changes by the day, this is as good a time as any to assess some of
the lasting effects. Whatever occurs, the events have challenged political
authority, and have prompted a rethink in global power supply.
Institutional resiliency and fragility

The wave of protests in the Middle East and North Africa was driven by
economic grievances, a mixture of high unemployment, rising prices and
lack of opportunity. Lacking a place to push for these economic
opportunities in repressive regimes and lack of political engagement adding
to pressure for political reform. The tendency to respond with repression
and in many cases firing on the population,
These economic grievances and vulnerabilities have been present for
some time, and many analysts including this author figured the shoe would
drop one day in countries like Egypt given the difficulty in absorbing a
growing population into the workforce. However, we assumed it would
happen on President Hosni Mubarak's incapacitation or death, which could
emerge into a stalemate or political transition. The proximate trigger was
the increase in food prices, continued deterioration of the labor market
domestically, exacerbated by the loss of the escape valve of working in the
Eurozone (stresses in the Eurozone, particularly in the periphery have
reduced wages for migrant workers, and accompanying remittances). The
damning accounts of the Tunisian regime's corruption in the wikileaks cable
did not help.
Wealth has not prevented protest, with some of the richer countries in
MENA, including oil exporters like Libya and Bahrain seeing extensive
protests, which were met by clampdowns. However, richer countries in the
region have more policy tools available to dampen grievances particularly
in the short-term, through social transfers, expansion of the public sector
wage rolls and provision of subsidized food and fuel. Similarly, monarchies
in the region, whether oil importing or exporting have additional policy
tools available including space to strengthen or create weak parliaments or
consultative bodies. Table 31.1 shows some comparative indicators for oil
importers versus oil exporters.
Ultimately the economic policies in the region will imply that
government policy, not the private sector will be the main source of growth
in years to come, also mopping up the credit growth extended by the banks.
This implies several factors important for the medium term outlook.
(1) Social spending and subsidies will outstrip capital investment, calling
into question the infrastructure boom anticipated by investors in the
MENA region.
(2) Weaker private sector growth will keep overall economic and
employment growth below trend and complicating the political
transition. Policies that are on the margin more populist will impair
the investment climate.
(3) Government spending will add to inflationary pressures as the
population spends more of what they receive from the government,
putting pressure on the supplies.
(4) The need to supply food could put further pressure on strained
agricultural production as countries seek to lock up supplies to avoid
shortages. With subsidies in plaice, real prices will not be passed on to
the population. While food and fuel subsidies are highest in the
MENA region, and a key part of the limited contract between the
government and the population, subsidies are also prevalent across
Asia, and could keep demand for key commodities artificially high,
even as supply eases due to the squeezed margins of processors and
refiners.
Table 31.1: Development indictors: oil importers versus oil exporters

Resiliency of global fuel markets
The escalating conflict in Libya took the bulk of its oil production offline
beginning in late February, and contributed to the increase in oil prices,
particularly of Brent crude. Saudi Arabia and other OPEC members quickly
increased production but only partly offset the reduction in output, and to
take advantage of higher oil prices that will boost revenues. While Libya's
output is rather low (about 2% of global crude) the composition or grade of
Libya's oil is not easily replaceable. Most of Libya's 1.6 million barrels in
oil exports are Europe bound, with 30% supplying Italian refineries that
rely on its blend of light sweet crude. Thus like Nigeria's oil output loss in
2006, the reduction in supply had a disproportionate effect on global prices,
given the rigidities in the market. To some extent, refineries in Asia have
more capacity to process the heavier sour crude that now dominates
incremental supply and production, a major step forward since 2006-7, but
Light sweet crude remains in short supply. The increased production from
OPEC reduces the cartel's spare capacity leaving it less able to counter
other shocks. In fact the reduction in oil output from Gabon and small
reductions from Nigeria (declines of about 200,000 a day each) added to the
pressure.
Further supply shocks could exacerbate the oil price increase and in turn
lead to output losses and slowing trajectory in key advanced economies and
emerging market economies. In particular an oil price about US$120 per
barrel WTI for a several months, could lead to a significant stall in the pace
of growth. While consumers in the US and Europe are ill-placed to absorb
higher costs, given the ongoing deleveraging, manufacturers in the
emerging world, particularly Asia, will bear the brunt of higher costs
domestically. Chinese authorities, already worried about inflationary
pressures, have stopped passing on the whole extent A sharp increase in oil
prices tends to have a significant effect on inflation and inflation
expectations, but a prolonged spike has the most effect, particularly if it
comes along with a spike in other commodities.
Several factors are adding to the rigidities of the global oil markets
(1) Political risks could defer investment in some countries including
those in the middle east that are open to foreign investment (Algeria,
Iraq).
(2) Oil exporting nations have become reliant on ever higher oil prices to
balance their budget meaning that they will seek to keep oil prices
above US$90 per barrel. This implies that they will be wary of
increasing oil output and some governments will seek to renegotiate
oil contracts to boost their share of the profits.
(3) While there are extensive reserves of petroleum in national and
corporate inventories, these are unequally distributed, with the bulk in
U.S. and key European countries. The desire from China and other
EM countries that account for the bulk of future demand growth to
add to their reserves will add to present demand, but might increase
their ability to counter shocks in the near term.
Oil exporters fiscal breakeven prices are rising as shown in Figure 31.1.
Fig. 31.1 Oil exporters fiscal breakeven prices
Nuclear Disaster Test, to Global Electricity

Turning to Japan, tae heavy investment in earthquake preparedness helped
reduce the economic losses that a less developed country with weaker
institutions might have suffered. However, the combination earthquake and
tsunami expossd a hole in Japan's energy security. As a resource-poor
nation, Japan relies heavily on nuclear power which provides 309% of
power supply, reducing the need to import hydrotarbons. Reducing one
vulnerability (the need to import costly and scarce futl) added to another
(the risk of nuclear meltdown). Such a high reliance on nuclear power on a
faultline was always a precarious decision so building in extra protective
measures was important. This seems not to have been the case. Early
adopters of nuclear power are now facing legacy issues as it was
economically cheaper and politically easier to extend old plants rather than
to build new ones. While not all are seismically vulnerable, many countries
are asking themselves hard questions about whether to shutter old plant.
Some have taken a knee-jerk decision to do so (Germany) or have
mothballed planned building (Italy) but the key new nuclear powers like
China are forging ahead. On net, natural gas and to a lesser extent oil should
benefit from the need to shift to new power sources in the short, medium
and long-term.
Supply Chain vulnerabilities

Beyond the need to rebuild capital stock, the power loss has driven the
reduction in output in Japan. Companies uncertain about power supplies
due to shuttered nuclear power or closed refineries or damaged
infrastructure, reduced production. Given the integrated nature of global
supply chains, particularly within Asia, output could be affected. The longer
the power shortages persist, the greater the effects. In fact the developments
may have increased offshoring and diversion to other countries, including
China. If so, the effects on the global economy might be muted, but the
earthquake could exacerbate various shifts in demand.
Update: This chapter illustrates the interconnectedness of the global supply

chain and underscores the importance of policy coordination which
constitutes most upside risks to the global economy. While global policy
coordination on political and economic issues has conspired to reduce the
risk of disorderly tail risks, and liquidity served to allow differentiation
based on growth prospect, the deepening of supply chains suggests
decoupling will remain difficult.
1Edited from Wilmott, July 2011.

It's a Gas, Gas, Gas!1
Swimming in Gas: Will Customers Follow?
Several structural changes are underway in the global natural gas markets,
with several new areas poised to enter the markets in coming years, fighting
with each other for capital, infrastructure and market share. Although
supply has increased, particularly in the U.S. where demand is more
restrained, it remains segmented within regions as transportation costs and
capital investment hinder transportation. These cost structures and the lack
of infrastructure will keep gas markets regional, despite the increase in spot
cargos in both Europe and Asia. Over time though new supplies will prompt
new demand and over the long-term could globalize this market. This is
unlikely to happen until the latter part of this decade, weighing on gas
prices in the United States and prompting a break in the pricing structure
globally. Still the shift toward spot markets could bring new financial actors
into these markets, prompting the creation of new benchmarks, as it has in
the oil markets.
One of the biggest surprises for 2011-12 has been the increase in onshore
natural gas production in the United states due to new technologies and a
series of regulatory measures, that encourage producers to use it or lose it.
This increase in supply has supported manufacturing activity in the U.S.
due both to demand for new equipment as well as providing a cheaper
alternate power source. The increase in natural gas output though dependent
on price, will nonetheless increase U.S. relative competitiveness in years to
come.
US Charging Ahead
Shale gas in the United States has been a game changer, bringing gas prices
down to record levels of below US$3/million b.t.u and staying at such low
levels for an extended period of time. Despite technological innovations,
such prices are below breakeven levels, particularly for new entrants and
new wells as labor and component costs have increased along with the
albeit modest recovery of the U.S. economy. In fact, oil and gas producing
states have experienced stronger recoveries, tighter labor markets and
demand for components of steel, which has in turn supported some of U.S.
industry. The sector requires high-quality steel piping, helping to regain
some market share from China. Unfortunately for U.S. consumers
struggling with high petrol prices, cheap natural gas has not yet filtered
through into really cheaper power costs, given other power generation
backlogs - cheaper power could support U.S. industry and households.
Fig. 32.1 US Natural Gas production. Source: Energy Information Administration
Still low prices may prompt cutbacks in investment and drilling activity,
even though companies fear the penalties for leaving land untouched. This
slowdown would bring prices up to greater elevation levels. No matter the
composition of executive branch (or Congress) after November elections,
natural gas and domestic hydrocarbon production will likely be a focus of
U.S. policy, particularly in an environment of geopolitical risks to supply.
Really using this gas in a U.S. context will take time, particularly as new
gas fired plants will need to be built or shifted over from coalfired plants.
This will occur over the coming years. Ultimately in the U.S. absorbing the
gas glut would take a sea change in transportation fuel. Some buses and
trucks now run on compressed natural gas, but they are few and far between
and the greater innovations required will keep supply outstripping demand.
This will increase the pressure to export into Asia.
Fig. 32.2 Proven Reserves. Source: BP Statistical Review of Energy 2011. Note: these proven
reserves include conventional gas reserves not necessarily unconventional likely reserves
Other Players: shift in Market Share

Qatar now dominates the liquefied natural gas market, producing x million
cubic feet a year most of which is shipped to Asia. After ramping up
production in the last decade and increasing market share in Asia, it has set
a self imposed moratorium on production at current levels. After 2015 it
might well invest more in the future. Qatar invested heavily in LNG
equipment, facilitating exports beyond the region, and has increased exports
to Asia, a trend which is likely to continue in the coming years.
Fig. 32.3 Largest Natural Gas Producers. Source: BP Statistical Review
Qatar aside, the Middle East is gas poor, at least in terms of production,
and the fundamental outlook in the region is very tight. Booming GCC
economies like Saudi Arabia, supported by government stimulus have
rising power demands, which put investment in natural gas supplies in high
demand to fuel industrial projects, new houses, hospitals and other
developments. The power needs for industry, desalination of water and air
conditioning mean Saudi Arabia burns oil, undercutting its export potential
- Saudi Arabia and other oil exporting countries report some of the highest
demand growth for hydrocarbons. As with Russia, subsidized natural gas
prices (and in Saudi Arabia's case too -cheap gasoline) exacerbate these
problems as they encourage wastage. The race to find more gas
unassociated with oil fields finally paid some dividends with the Karan field
set to come on line in full by 2013, but domestic demand will continue
climbing extensively.
Fig. 32.4 Saudi Domestic Consumption Rising, Cutting Exports. Source: BP Statistical Review 2011
Egypt which has extensive gas reserves has become an uncertain source
of supply in 2011 as its pipeline has been disrupted more than a dozen times
since the revolution, and faces repeated attacks from Bedouin unhappy with
their lot in the Sinai peninsula. This puts additional pressure on their
customers in Israel, Jordan and beyond, encouraging new supplies, a topic
which will be addressed below.
It is legacy producers that face the greatest challenge, including even
giants like Russian Gazprom who face uncertain policy environments at
home, the need to subsidize domestic and in some cases regional
consumption. The Russian government relies heavily on oil and gas revenue
which makes up the vast majority of government revenue. With 2012
spending requiring an oil price of over US$110/ barrel Brent crude, the
Russian government needs more revenue. It will not give up its cash cow,
and will resist heavily pressures from consumers in Europe to reduce prices.
In fact, at the time of writing the new Putin Administration was considering
increasing marginal tax rates for the gas conglomerate to as high as 90%
which would undermine their incentive to invest at home.
Already high tax rates have discouraged capital investment in Russia's
gas sector (as with the oil sector) and encouraged exports, particularly into
the more costly Asian market. A new pipeline carries Siberian gas to
Vladivostok, from which it is sold to gas-hungry Asian consumers. Given
greater demands at home, under investment and a eastward pivot, European
gas prices should stay high, though the increase in liquidity and demand on
the spot market will continue to shift this market.
The Demand Side: All About Asia

For now the bulk of demand and demand growth is set to come from Asia,
already the region with highest costs (even setting aside the higher
transportation costs). Japan, South Korea and China are all avid consumers,
with the Fukushima nuclear disaster adding to Asian demand as the nation
shifted to natural gas in the wake of nuclear shutdowns. While Japan's
nuclear program could revive, demand for natural gas should remain robust.
China will be the primary source of global demand for natural gas in the
coming years, and State Council recently set a target to increase natural gas
consumption (and production). Although some Chinese government
objectives are routinely broken (such as the annual goal to grow around
8%), explicit numerical targets are more likely to be met, particularly on the
consumption side.
Domestic production will face more challenges, particularly given the
continued lack of expertise, the highly porous geological formation, which
increase the need for water in an already parched land. However, unlike
Europe, deposits tend to be in sparsely populated areas, which encourages
exploration. Moreover, without formal property rights, Chinese authorities
tend to find it easier to resettle populations and quickly build infrastructure.
A reorganization of China's hydrocarbon countries after the financial crisis,
suggests that there could more coordination. In past years, the market for
gas.
The Shale Mirage: Tricky in Europe

While Shale gas is a game changer in the U.S. it is much less promising
outside of North America. Canada is playing a game of catch-up at present
as it shifts from conventional to unconventional gas. But it too is likely to
increase production, as provinces like British Columbia and Saskatchewan
seek new industrial demand - and look longingly at the Asian market.
Elsewhere geographies, geology and expertise point to delays, with
production likely only in the back half of the current decade at the earliest.
The plunge in European industrial output and slow recovery outside
Europe's core have depressed demand for power, particularly as gas prices
have remained elevated across the European continent. Still European
emissions plans and clean energy targets will support natural gas as well as
actual renewable fuels. European production has been falling for the better
part of a decade as north sea production first plateaued and then began to
fall, with the UK becoming a net importer, reliant on higher global cargos.
Still on the western part of the continent, the development of the spot
market has started to bring down prices, particularly in Germany.
European gas hopes, already dented by a French antifracking campaign,
took a tumble when Polish authorities issued a more conservative
assumption of their domestic prospects. Poland, thought to have the largest
shale gas reserves in Europe, now is thought to have reserves of less than a
trillion cubic meters - more than enough to meet Polish demands for dozens
of years, but less than hoped or possibly less than that needed to maintain
the attention of the global majors. Recent tests have cast a shadow on its
commercial prospects. Poland is likely to increase gas production, in part
due to a desire to reduce reliance on Russia, which charges it one of the
highest rates across Europe, but export prospects are muted. These in turn
could defer exploration and activity across Northern Europe (in addition to
France, Germany and the Netherlands have geological structures likely to
include gas).
Elsewhere in Europe, environmental concerns will delay production and
exploration. In France, most of the deposits surround Paris, which suggest
they will remain underground. Meanwhile, France is strongly in the nuclear
energy camp, with heavy government investment having created a
competitive industry for both domestic use and exports, there is less need
for gas-fired power. In fact, France has sought to ban fuel that uses
hydraulic fracturing across the EU, a move that other member states may
oppose. Lower reserves, high environmental costs and an uncertain
regulatory environment suggest that unconventional gas will affect Europe
primarily through imports, and the spot market.
Ukraine too is desperately seeking gas, which could relieve it from
reliance on Russian imports that are going up in price annually. Following
years of postsoviet subsidization of its neighbors in the Commonwealth of
Independent States, Gazprom is looking for market prices from its buyers,
particularly as it still sells gas at home at discounted prices, which in turn
leads to over consumption. Exploration has begun in Ukraine, but the
broader operating environment is uncertain, the government desperate for
revenue to meet its financing needs and extensive structural reforms
needed. As of the end of 2012, the Ukrainian government had failed to meet
its aggressive MOU targets suggesting any developments cold be
extensively delayed.
Offshore gas in Mediterranean and Africa

The rush for gas has also kicked off in both the Eastern Mediterranean, led
by Israel, but followed by Cyprus, with Lebanon also a hopeful. Two large
gas fields in Israeli waters add to 70 billion cubic meters, themselves the
two largest finds in the last decade. Although the royalty regime has shifted
since the discovery, as Israel shifted from being a hydrocarbon free country
to one with extensive resources, the tax take the government proposes is at
the middle of the OECD pack and similar to those of the North Sea
countries. Israel is set to begin production from the Tamar field in April
2013, but most of the demand will be from domestic audience as export
channels remain untapped.
Still a long list of obstacles remain including high costs of operating due
to rig shortages in Europe, rising labor costs and the question of where and
through what channel to export. A lack of skilled labor in the hydrocarbon
sector is a global problem, stemming in part from the dark days of the
1990s when the sector faced doldrums, particularly acute in Israel as oil
majors seeking business in Arab countries did little to encourage Israeli or
Jewish entrants. Rigs too face extensive backlogs, and those a part of
offshore installations are one of the only segments of shipbuilding industry
to be backlogged (other segments, including tankers shifted from
prefinancial crisis backorders to a glut that has depressed global shipping
costs, and detached the Baltic dry index from being a key indicator of
global trade and underlying demand - but that's the topic of another
column!) Moreover, with Tamar and Leviathan far offshore, transporting
the fuel to land for processing, liquefaction and the like could be costly.
Export routes too are challenging given still to be defined borders
(Lebanon) and a reluctance to use unproven technology like a floating
liquefaction terminal.
These finds have stirred up regional political divides, already heated by
longstanding tensions and the reshaping of alliances, allegiances and ties
due to the Arab Spring. The Eastern Mediterranean gas region could
amplify unease between Israel and Turkey (ties were broken in 2011 over
the flotilla to Palestine) in the midst of a greater Turkish role in the Middle
East, both economically and politically. Discoveries in Israel and a study by
the US geological survey prompted exploration off of Cyprus, both North
and South, and a large field was found off of the Cypriot coast, close to the
Israeli fields. Russians, long a presence in Southern Cyprus due to its
beaches, its financial sector that supported capital recycling in and out of
Russia, is likely to play a role. The country, over-exposed to greek banks,
and homegrown revenue short-falls already benefited from a Russian loan.
This could serve Gazprom interests for more supplies, and bolster the
Turkish influence. Still it remains to be seen if a joint Cypriot-Israeli export
operation could be in the cards.
Northern Cyprus also is looking for gas, with Turkish help, and the
divisions on the Island and in the region could initially distort transportation
and operations. Lebanon too, plans to launch a round of bidding for
exploration licenses in 2012, a move which will necessitate formalizing the
naval border. While these finds could complicate regional ties in the
foreseeable future, the optimist would hope that supplying more power for a
series of net fuel importers could prompt common interests. The more
cynical would note that at a minimum some production, likely in the second
part of this decade, would support economic growth, provide more
government revenues, and improve trade deficits.
Israel will be first out of the block, but gas production will not start in
earnest until after 2015 and will only really make a difference to the
economy and trade balance in 2018 and beyond. This is indeed a really a
longer-term story.
At the same time, there has also been gas rush off the shores of Eastern
Africa, with most of the coastal nations finding gas onshore or offshore
(Kenya, Uganda, and most notable Mozambique). The size and
commerciality of prospects is still to be determined but the coincident
discovery of resources in adjoining areas could increase the attractiveness
for operators. Like other countries, we expect Eastern African countries to
change some of the regulatory and royalty structures, seeking to ensure that
their low-income populations gain some benefit. Like their Western African
counterparts (Ghana, Nigeria, though the latters efforts were ill-fated),
many will seek to save some of the revenues in sovereign wealth funds.
Mozambique looks particularly promising, as several new fields have been
discovered after the country already has increased its production of coal,
adding to its export basket.
In both cases, the question remains, who will consume all of this gas.
Domestic demand within Sub-Saharan Africa is great, but infrastructure to
burn the fuel scarce and transmission lines weak. Moreover, given low
incomes, the population may not be able to afford the costs, and local
governments tending to subsidize fuel for fear of public unrest or being
turfed from office, could lack the funds to pay the bill. Theoretically the
location is good, being closer to Asia, and supply chains and away from the
congested Mediterranean and Middle East pathways. But as with the
Eastern Mediterranean extensive investment will be needed. Unlocking that
investment in a time of uncertainty requires locking in dedicated customers
willing to take that supply on a long-term basis, More negotiation on price
may be a prerequisite. One thing is certain, all of these structural changes
suggest that natural gas prices will increasingly move away from oil-based
pricing perhaps to a series of global benchmarks. This will happen only
gradually over-time as markets and consumers reach to these shifts in
supply and change their demand patterns.
1Edited from Wilmott, January 2013.

Thoughts on the Current Market Environment,
Risks and Returns1
Market Events in April in 2010
The big news in the financial markets is the continuing turtle like advance
in US and other equity markets which had three bad events coming into
options expiry day Friday, April 16, 2010.
Event 1: Icelandic Volcano. The impact of this event is yet to be felt in the
financial markets but it sure is having an effeet in the world. The debris ]
has shut down airports all across Europe affecting continental and inter-
continental traffic and with no end in sight. The full impact is at this time
unknowable as it depends on how long the volcano cont inues to erupt. The
debris covers a huge area. The volcanic glass is melted by the heat of the
engines and can just shut them down. Even parked on the ground the
engines have to be protected. The airlines are losing about US$200
million/day. The stock market impact has been muted so far, so has been a
non event for financial markets. After a tense week, the airports opened up
but with airline losses of about $2 billion and many delays and losses by
suppliers and many stranded passengers and shipping delays. Hotel rates
tripling for stranded passengers in London and other hubs added to the
stress. The last time the volcano erupted over a hundred years ago, it
continued erupting for two years. And there is an even bigger volcano close
by that has typically erupted within six months. So this saga is likely to
continue.
Event 2: Goldman Sachs. The SEC charges against Goldman Sachs hit the
market hard. The Paulson hedge funds made huge gains - over 1000% in
2007, and large profits in other years shorting subprime loans. The SEC has
charged Goldman with enabling the Paulson people to cherry pick
instruments and package them for easy shorting in the Abacus fund even
though the official picks were to be by a third party intermediary, ACA
Management, a US bond insurance company. The SEC alleges that ACA
had no idea that Paulson was planning to short the CDO they helped create.
The SEC alleges that Goldman Sachs misled two investors IKB Deutsche
IndustriebankG, a German Bank and ACA. I cannot say much here as the
case is just starting except that Paulson's group is not charged. Fabrice
Tourre, a 31-year of Stanford engineering graduate, considered quite smart
by his fellow students, was a key player in setting up these securities and
boasting about his success. Tourre is a graduate of elite Paris schools so is
well trained. Only Goldman is charged for allegedly not disclosing this to
other clients and only Tourre is named. Goldman received a US$15 million
fee for setting up Abacus and claims that they lost money on this fund. It is
argued that their later purchase which led to losses was done to make the
deal go through with other clients (Donlan, 2010). My own assessment is
that likely a fine of $1-2 billion might be levied to Goldman Sachs which
they can easily afford. Indeed the stock fell about $30 from the $186 area to
$155.55 in a few minutes and closed Friday at $160.70 down $23.57 or
12.79% with the VIX increasing from 15.89 to 18.36. This drop was about
10 times the likely fine. Put options with a strike of 180 trading for 9 cents
at Thursday's close were $22 near the close on Friday,up 22,000%.
Billionaire investor Jim Rogers thinks that the Goldman Sachs situation
might be the catalyst for a correction given the fast pace of the rally so far
in 2010. The S&P500 fell 18 points or about 2%. See Figures 33.1a and b.
Fig. 33.1 Market moves mid April, 2010
In the week after the options expiry, the market rallied back for its eighth
consecutive weekly gain with ups and downs but the S&P500 so far is like
a balloon in water, it gets pushed down then flies up. So Rogers' correction
has not happened. Meanwhile, Goldman stock drifted slightly down during
the week despite record first quarter earnings of $3.3 billion on revenues of
$12.78 billion or a whopping $5.59 a share. That followed a $4.79 billion
gain in the fourth quarter of 2009. These are huge earnings for a $160 stock.
Noted bank analyst Dick Bove of Rochdale Securities, who is
independent of Goldman, argues several key points:
(1) Goldman has $2 billion of sophisticated trading software that other
firms do not have. So despite any animosity, firms are drawn to them
as the top game in town. They provide a unique service. People may
compain but they still remain Goldman customers. This is their
strength and it shows in their huge earnings. Fully 72% of their profits
come from trading. There are also higher fees from underwriting stock
and debt offerings. But the huge earnings are over shadowed by the
SEC charges. These charges might be expanded to other countries
such as the UK where Prime Minister Gordon Brown does not like
them giving big bonuses despite the investigation. There were also
large sales of Goldman Sachs stock by top executives, much at higher
than current prices, during the months leading up to the SEC
announcement while they knew the firm was under investigation.
(2) On the PR front, Goldman has taken a superior, standoffish view
saying they did nothing wrong. This comes from the top and includes
testimony by the head, Lloyd Blankfein. Boasting about profits made
shorting subprime won't help, see Margolies and Wutkowski (2010),
regarding 2007 emails by Blankfein and Goldman executive Donald
Mullen.
Bove argues that a more open constructive approach would be better.
He points out that the 2007-9 crisis, while very devastating is not
unique. Any history of the last 800 years (see for example the book by
Harvard economist, Ken Rogoff (Reinhart and Rogoff, 2009))
provides a list of a variety of such crises. There is a hostile mood
against Wall Street's excesses, greed and high pay, even with failures,
so a more modest approach seems warranted. We will have to see how
this plays out. But we know in the baseball steriod use cases those
who admitted use and vowed never to do it again had the matter
dropped as with New York Yankee stars Alex Rodriguez and Andy
Petite, but these who denied it and lied such as Barry Bonds and
Roger Clemens, despite incredible accomplishments, had their careers
ended in a bad way. Along with Warren Buffett and Berkshire
Hathaway, I remain an investor in Goldman Sachs which I presume
will weather this storm with a settlement and continue to make high
earnings.
In his testimony the Congressional committee, Tourre came off as a
highly polished, careful thinker with a strong defense, pointing out
new facts about the case that likely will help Goldman's image, so we
will see how this turns out.
Event 3: The Financial Crisis of the PIIGS and the Euro. The PIIGS
(Portugal, Italy, Ireland, Greece and Spain) all of which are in the Euro and
Iceland which is not, are all in very serious financial trouble. Even the
mighty UK is rather weak. Greece is a the center stage and a 30 billion euro
bailout is under discussion. Germany wants an agreement of deep budgetary
cuts but these are hard to make and will lead to even more damage to the
social infrastructure. The interest payments alone on current Greek
government debt are about 22.9% of total government revenues and new
debt is at higher interest rates. Devastating graphs by Greg Weldon
(Mauldin, 2010) show the incredible decline in the Greek economy, stock
and bond markets etc. Something must be done but the cures themselves are
all devastating. If Germany agrees to join the bailout they avoid losses in
current positions but have this cost and there may be more Spain and
Portugal bailouts to come. Meanwhile Greece either gives in to difficult
cuts in social and other programs or defaults. Higher bond yields reflect
risks of default, inflation etc. In the case of Greece 5-year government
bonds yield 11.4% as of the end of April versus 5.6% in Portugal, 3.2% in
Spain, 3.0% in Italy and Ireland, 2.1% in France and 1.9% in Germany
according to Bloomberg. All these are euro currency countries. No wonder
many think the euro is doomed. UK 5 year rates are 2.8%.
Of the 30 billion euro rescue package, the better off countries like
Germany, France and the Netherlands would pay more (about 8.4 6.3, and
1.8, respectively) but weak economies like Italy, Spain and Portugal are in
for 5.5, 3.7 and 0.8 billion which they can ill afford. This totals 30 billion
with the IMF adding another 15 billion . Portugal is also very vulnerable
and their debt was downgraded as was that of Spain as contagion spread.
All this is a drag on the euro which dropped and there may be pressure
eventually to disband it and let Greece and others opt out though this would
be a messy process and hard to disentangle. The problems are exacerbated
by the rapid financial integration (including movement of capital and
labour) without political integration. In my travels, especially to Italy and
France, but also to Spain, I witnessed the inflation that the introduction of
the euro caused. George Soros has argued that the euro likely cannot
continue, notice how good he is at front running his trades, even Goldman
would be proud!
Trying to straighten out the financial markets is a tough one as greed is
always there once the fear subsides. I am reminded of my colleague Ed
Thorp's response to a reporter who asked why he switched from gambling
in blackjack to options trading in the stock market. He said there were three
reasons:
(1) if I lost they would not break my knees,
(2) it was a bigger game and I thought I could make more money, and
(3) I thought I would meet a better class of people, but I was young and
naive in those days.
Ed is right and nasty unscrupulous people are everywhere not just at
Goldman Sachs. Meanwhile, the ups and downs of these events are a drag
on the US and other stock markets and on Tuesday April 27, 2010 they
caused a minor pull back and substantial rise in the VIX. During the next
two days the market rallied back, recovering the 28 point S&P500 decline.
So the market remains buoyant despite the European and Goldman
problems.
Thoughts on the current market environment, risks and

returns
The usual technical indicators that many traders use have been confusing in
2010. An interesting study is that of Jovanovic and Zimmermann (2010)
who show that “given certain levels of inflation and output, US central bank
rates are significantly lower when stock market uncertainty is high and vice
versa. This result is valid for all tested measures of financial uncertainty.” 2
The year 2010 has been up and down all year with fear and greed
alternating. The month of September was especially positive with a gain of
8.76% and that has pulled the year's return on the S&P500 to a slightly
positive 2.34%.
Figure 33.2a below shows the S&P500's gyrations including the May 6
Thursday flash crash in which the stock market fell 10% in a few minutes
and then recovered two-thirds of the losses that day. Figure 33.2b shows the
time line that day.
Fig. 33.2 Two views of the Flash Crash. Source: Bowley (2010)
Put option prices rose sharply on the Friday to the 43 area from 34 on the
VIX. Then fell the next week. A joint SEC-CFTC report on the flash crash
points to a large fundamental mutual fund trader at Waddell & Reed
Financial of Overland Park, Kansas who executed a large sell order using
an automated algorithm at a time in the afternoon during which the markets
were already stressed.
The report said that
“the temporary crash resulted from a confluence of forces after a

single fund company tried to hedge its stock market investment
position legitimately, albeit in an aggressive and abrupt manner.”
The computer algorithm accelerating selling as the prices plunged. Some

80 SEC people disallowed trades that were way out of line. We will see
what the SEC-CFTC officials devise as new rules. There are already
individual trading halts for five minutes if some stocks move 10% in five
minutes. Devising fair rules is complex.
The 6438 trades totaling 75,000 mini S&P futures contracts had a face
value of $4.1 billion. The trading on May 6 was 5.7 million mini S&P500
contracts so 75,000 is 1.3% of that day's volume and less than 9% of the
volume during the time period of the sales. W&R's algorithm was to sell at
most 9% of the total volume so they accomplished that. In twenty minutes
the futures were down 5% and triggered the 1000 point decline in the Dow
Jones average in about 10 minutes. Between 14:32 and 14:52, some 20
minutes, there were 147,577 trades totaling 844,513 contracts. The large
selloff began at 14:44:20. According to Durden (2010b) the bulk of W &
R's trades were made after the market bottomed and was rallying back.
According to the CME:
... fundamentally negative financial, economic and political events

in Europe and elsewhere contributed to investor uncertainty and
impacted participation and liquidity in all market segments at certain
times that day. Throughout the day on May 6, CME Group markets
functioned properly.
. . . The prevailing market sentiment was evident well before these
orders were placed, and the orders, as well as the manner in which
they were entered, were both legitimate and consistent with market
practices. These hedging orders were entered in relatively small
quantities and in a manner designed to dynamically adapt to market
liquidity by participating in a target percentage of 9% of the volume
executed in the market. As a result of the significant volumes traded
in the market, the hedge was completed in approximately twenty
minutes, with more than half of the participant's volume executed as
the market rallied - not as the market declined.
...the minute period immediately preceding the market bottom
that was established at 13:45:28. During that period, the participant
hedging its portfolio represented less than 5% of the total volume of
sales in the market. Source: CME (2010)
According to Zero Hedge led by Tyler Durden and the CME, there was
more to the Flash Crash than the $4.1 billion hedge trade. By their research
it seems that high frequency traders (HFTs) began to quickly buy and then
resell contracts to each other generating large volume but no real trading.
Between 2:45:13 and 2:45:27, these HTFs traded over 27,000 contracts,
which was 49% of the volume. Yet the net buying was only 200 contracts.
So the ratio of volume to liquidity was 135 to 1. So the HTSs do not
provide liquidity but volume at lease in this case.
Penalties for losses are not great enough

While we are on the topic of penalties for losses, recall I have long
advocated more penalties for losses and that light penalties encourage rouge
trading. Well, we have one example, although it is a bit above the top.
Jerome Kerviel, who lost 5 billion at Societe General in January 2008 was
ordered by a Paris court to pay if all back and serve a three-year jail
sentence; see Samuel (2010). At his current rate of pay, 2300 a month as a
computer consultant, it would take him 177,000 years to pay it back. So the
court that found him guilty of breach of trust, forgery and entering false
data into the bank's computers was sending a message rather than expecting
the judgment to be fulfilled. Nick Leeson, who brought down Barings
Banks in 1995 did spend years in prison but now gets £6000 a lecture!
The anomalies, as discussed in Dzhabarov and Ziemba (2010)'s update
following studies such as Keim and Ziemba (2000) and Ziemba (1994) are
working well in 2010. The top graph in Figure 33.3 show a test anomaly
account we have been running. The trades include those involving options
biases, the turn of the year, January effect, the turn of the month effect,
options expiry effects, etc. The test account is up 28.8% from its December
1 start to September 30 after hypothetical 2+20 fees accrued monthly and
commissions are deducted.
Fig. 33.3 Private futures account of the author at Interactive brokers
Updates of US seasonal anomalies to the end of December 2011 are in a

chapter by Dzhabarov and Ziemba in Ziemba (2012a), Calendar Anomalies
and Arbitrage.
Monthly data
Historically over the last century mean monthly returns in September and
October have been negative as the worst months of the year, see Keim and
Ziemba (2000) for specific results. However, in the last fifteen and five
years up to 2009, the situation has changed somewhat. September was still
negative up to 2009, though only by a small amount, but to 2010 it was
actually positive. See Figure 33.4. Meanwhile, October was actually
positive from 1993-2009 but very negative in the past five years. What
seems to happen is positive returns in many of these September and
Octobers but with big blowouts in some of the years. Table 15.7 illustrates
this. So the highly positive September 2010 is not that unusual and does not
tell us much about possible dangers in October 2010. Meanwhile the first
half of October has seen a slightly positive market with the VIX falling to
the 20 area.
Fig. 33.4 S&P500 Futures Average Monthly Returns
Thoughts and data on the US economy and housing market

Nearly one in four second quarter US home sales was a deeply discounted
foreclosure house; see Adler (2010) There were 100,000 foreclosed US
houses in 2005. This has increased to 1 million in 2009 and 1.2 million in
2010. The average discount was 24%. As much as this is the first quarter
results were even worse, a 28% discount and nearly a third being foreclosed
houses. With unemployment at 9.6%, and average home prices about 28%
below 2006 peaks, the US housing market has not recovered. The higher
the percentage of homes sold that are distressed, the bigger the number, the
quicker the housing downturn will be reversed. Figure 20.1 has the Case
Shiller price indices in the 10 and 20 largest US cities for 2008 and 2012,
which shows a slight leveling of prices in 2010.
The following comments are edited from Adler (2010) and provide some
insight into the current US situation. According to RealtyTrac, 24% of US
homes sold in Q2 were deeply discounted foreclosed house, discounted of
more than 26%.
• Distressed homes rose to 34% of all existing houses sold in August from
32% in July and 31% a year ago, according to the National Association
of Realtors.
• Banks stepped up foreclosures through the summer and are expected to
take over a record 1.2 million homes in 2010, versus 1 million in 2009
and about 100,000 in 2005 before the housing bust.
• This puts the market on pace to work through distressed properties in
about three years.
According to Rick Sharga of RealtyTrac the higher the percentage of
homes that are sold that are distressed properties,, the faster the inventory
will be cleared and the sooner we will get through this housing downturn.
From the current rate of sales, he projects that we will get through the
distressed inventory by the end of 2013.
• While a normal market will see foreclosed sales about 5%, for the next
few years, according to Sharga, these will probably represent between
one-quarter and one-third of all sales.
• “Overall housing sales likely will total 4 to 4.5 million a year during this
time”
• “It will take those years to resell homes lost by owners whose jobs or
wages were cut or who took out high-risk, unaffordable mortgages.
Banks will also need to sell homes from owners who walked away
owing more on their mortgage than the house was worth.”
Spring sales were accelerated by buyers seeking up to $8,000 in tax
credits followed by a decline after the incentive ended on April 30. Some
buyers considered the tax credit a discount, James J. Saccacio, RealtyTrac's
chief executive, said "the removal of the tax credit could drive more buyers
back to discounted short sales and REOs," or real-estate owned homes.
• A total 248,534 properties in some stage of foreclosure - default,
scheduled auction or REO - were sold to third parties, up about 5% from
the first quarter though down 20% from the second quarter 2009. There
were over 100,000 foreclosures in September 2010 for the first time
ever. Currently, there is a serious problem determining ownership of
mortgages after they have been sliced and diced and there have also
been foreclosures when the owner in either not in arrears or does not
even have a mortgage. Also many owners who could pay are going into
foreclosure on houses under water. So it is a real mess! 27 states in the
US have non-recourse mortgages which has increased the size of the
foreclosure problem.
• Banks sold more than 151,000 homes they owned, up 3% from the first
quarter but down 28% from a year ago. These REOs were 15% of total
home sales, down from 19% in the first quarter and about 29% a year
ago.
• The regional break down of foreclosed sales from April to June , 2010
was as follows: Nevada (56%), Arizona (47%), California (43%),
Rhode Island (37%), Massachusetts (35%), Florida (34%), Michigan
(33%), Georgia (27%), Idaho (27%), and Oregon (25%).
• Foreclosure price discounts versus the average price on homes not in the
process were biggest in Ohio (43%), followed by Kentucky and
California. Michigan, Tennessee, Pennsylvania, Georgia, Illinois, and
the District of Columbia had average foreclosure discounts of at least
35%.
So the housing market may be bottoming but it will take several years to
return to a possible upward path as inflation returns. Housing is less of a
problem because it has fallen and its impact has already been felt. There is
also a heavy regional bias here, in desirable locations, housing is still
expensive and the impact on the economy is mostly from (1) new
construction which in distressed areas is unlikely to pick up and (2)
renovations and new furnishings. Housing, overall as owner occupied and
rental units, is still the world's leading asset class by value.
Durden (2010) argued that September's rally of 8.73% in the S&P500,
the Dow Jones index up almost 8% and the Nasdaq up 12% was completely
unjustified. It was the third best month in ten years and the best September
since 1939. But volume was low. That puts the S&P500 +2% in 2010.
Welcome to today's complex market! Markets go up when buyers
outnumber sellers and the buyers seem to be professionals in the ETF and
futures markets and the sellers the general public in mutual funds. Mutual
fund investors continue to move out of the stock market. In the week of
September 27, domestic equity mutual funds experienced the 21st
sequential outflow of $2.5 billion, with the total year to date outflow over
$70 billion. Thus the supposed rally in September which was to restore
confidence actually resulted in an outflow totaling $416 billion according to
the Investment Company Institute (ICI). The annual redemptions from
stock funds reached a multi-year high of 25.8% in August 2010 versus
23.6% a year earlier.
The ETFs have an impact and are more and more used as trading vehicles
by professionals. Many retail and some institutional investors missed the
2009 rally starting in March and do not want to be in the market now.
Fig. 33.5 Market performance and cumulative outflows. Source: Durden, (2010)
Many of these investors cashing out of the stock market are going into
gold and cash. Attends (2010) reports a study by TrimTabs Investment
Research that equity mutual funds lost $39 billion. They bought in at an
average S&P500 index price fo 1434, not much below the record high of
1565. Random buying results in an average price of 1171, some 20% lower.
So even though the stock market is near its 10-year average, TrimTabs
believes that many who invested in the last decade have large losses.
Gold remains a bull market and is now (October 2010) about US$1375.
Figures 33.6a,b show stocks priced in gold since 2006 and gold in USD,
respectively. Stocks are only up in terms of a declining dollar. In real terms,
relative to gold, stocks have gone nowhere. Gold might continue climbing
as the 2010 inflation adjusted price of the 1980 peak is about $2300. There
are conflicting signals. Production costs are $900-1000 per ounce when all
costs are counted but this is not relevant it is not production costs that drive
the price as most gold is held in storage with much in jewelry, on the
margin is trades on fear. At US$1375 much higher seems possible But how
relevant the cost of production is is debatable. Most gold is held in storage
and as jewelry. On the margin it trades and fear drives the price up and
creates only limited new supply. Dennis Gartner, in his well-known
newsletter, and others who are long gold expect a correction soon. Gartner
has been long gold in euros and pounds. It has rallied in these currencies but
even more so in the declining dollar; see Figure 33.6b. Supporting the
correction theory is large buying of gold ETF GLD puts, mostly are by long
investors seeking to protect gains in a market that is parabolically up, see
Figure 33.6b and see Sears (2010). But these puts are cheap insurance as the
gold VIX is low and the sentiment very bullish.
Fig. 33.6 Gold
The NBER declared the recession over as of June 2009 - the longest (18
months) recession since World War II. But their definition of recession, two
consecutive quarters of negative GDP growth, masks a deeper problem.
Unemployment remains high at 9.6% with many people underemployed,
especially minorities and the young. Regaining employment is tough as the
GDP must rise about 5% to reduce unemployment by 1%. President
Obama, Warren Buffett and and others including me think that the economy
is weak and still basically in recession. Aversa (2010b) points out that the
economy has lost its growth rate and the GDP only rose 1.7% in the Q2
versus 3.7% in Q1. The drop was mainly from a larger trade deficit (and
increased inventories) Future estimates for Q3 are 1.5-2.0%. Consumer
spending accounts for about 70% of economic activity and it is weakening
which will also limit real investment activity. Savings is increasing, being a
recent US high of 5.9% of disposable income in Q2 compared to 2.1%
before the recession when people were using their homes as ATMs.
Consumer confidence in August dropped to the lowest level since February
2010.
So what should the Fed do? Aversa (2010b) reports that the Fed efforts
are split whether more stimulus is needed or not. But the numbers are
staggering.on the Fed's balance sheet - $1.7 trillion of mortgage securities
and debt. But some assets are at a profit. Finally, the Economist and the
IMF see the world out of balance with high, up to 10% growth in emerging
markets, and low growth in the US and other risk nations with GDP to
average 4.8% in 2010. According to Kyle Bass on CNBC, the US has $4
trillion in spending each year with less than $3 trillion coming in, so this $1
trillion + deficit is about 10%. Bond rates and government debt are at a low
interest rate - about 2.5% for 10-year Treasury bonds. But every 1%
increase in interest rates costs about $120 billion per year. Bass finds such
numbers staggering, so do I! But of course, the Fed could hold the debt on
its books instead. There is no reason for them to sell the debt unless they
feel that inflation is high. Will more monetary stimulus work or is it
pushing on a rope?; see Task (2010b). Interest rates are at historic lows,
even 3-year T-bonds return only 0.57%. The US government sold $32
billion on Tuesday October 12.
It is well known that correlations increase during market declines. Indeed
in the ALM model for Siemens called InnoALM, see Chapter 14 based on
Geyer and Ziemba (2008) and Ziemba (2007), there are three regimes and
in the third, the high volatility one, the correlations are much higher for
equity indices. As one moves from calm markets to violent markets to crash
markets these correlations increase. At the same time, correlations between
stock indices and bond indices decouple and in the crash set of scenarios,
these correlations are negative. That means that, while stocks are falling,
bonds are rising. This happened in the fall of 2008 in the serious crash then.
But the bonds have kept rising in the 2009–2010 stock market rally. Cole
(2010) has studied individual stocks and reports that at the peak of the 2008
crash in November the 50 largest capitalization S&P500 stocks had an
average 21 day correlation of 0.76. September fell -9.7%, October -20.11%
and - 9.22% in November then December was -0.40%, January 2009 was
-9.14%, February -11.13% then March, April, etc had positive returns.
What Cole observes is that the correlations in September 2010 are 0.62
which is not much less. Across all asset classes we see increased
correlations in 2010. Cole mentions also that option prices predict future
high correlations. Why is this?
Cole argues that its excess cheap money so there is buying of good, bad,
indifferent stocks so they are mostly moving together. Also this makes it
harder for statistical arbitrage and long-short strategies to work well and the
volatility is rising. Cole reports that between 1995 and 2007, 50% of mutual
funds focusing on high growth large cap companies beat the Russell1000
Growth Index. But it is only 24% in 2010. The constant withdrawals from
mutual funds since the flash crash is a factor here too. So it is the Fed's
policy that's upping these correlations.
Income inequality in the US is increasing. In 2009 the top 20%, that is
those making $100,000 plus, received 50.3% of total US income versus
12% for the bottom 40%. In 1999 these were 49.4% and 12.5% so
inequality increased slightly here. But the percent the wealthiest 10%
earned 48.2% in 2008 versus 34.6% in 1980. And for the top 1% it was
10% going to 21% (see Task, 2020b).
Pension problems persist, see Bertocchi, Schwartz and Ziemba (2010).
The states have pension deficits. Indeed 47 of the 50 are in the red, see
Donlan (2010). California is short $60 billion, Illinois $54 billion, and so it
goes! The total for all the states, according to Professors Robert Nory and
Max and Joshua Rauh, the shortfall is liability assets of $4.43 trillion and
assets of $2.6 trillion!.
Europe looks in trouble again. There are austerity strikes protesting the
cutbacks and several countries have serious banking troubles. Crimmins
(2010) observes that Ireland's banks might need a 50 billion bailout. The
public debt burden is about equal to their GDP and the yearly deficit 32%
of GDP. The yield on its ten-year government bonds is nearly 7%, some
4.7% above Germany. Spain lost its final AAA credit ratio as Moody's
(following Fitch and Standard and Poor's) cut it to Aa1, citing the budget
impact of slower economic growth. Portugal announced new spending cuts
and tax increases. Germany, while doing well, still has an unemployment
rate of 7.5% compared to Spain's 20%. Meanwhile, Britain's economy
highly tied to financial services is still struggling. According to The
Economist, the Bank of England should resume quantitative easing to help
recovery but this is the same pushing on a rope problem as described above.
Conclusion: There are a lot of negative signals here but with such low
interest rates a slightly rising US stock market seems to be the high
probability scenario. But watch out if inflation and interest rates rise.
What's Wrong with the US?1
In this chapter I discuss the August-September 2011 stock market turmoil in

the US and other markets and the deteriorating state of affairs in the US.
The latter topic concerns me greatly as I grew up in the US, had all my
education there with a PhD at the then greatest university of the world - the
University of California, Berkeley - and I continue to have many activities
there although I have lived in Canada since 1968. As a dual citizen of the
US and Canada I understand the cultural differences, strengths and
weaknesses.
The August-September 2011 turbulent period

Nouriel Roubini, a brilliant student of market matters with a deep
understanding of bond debt and other markets, has predicted a 60% chance
of a US recession. He is known as Dr Doom from his correct call of the
2008 crisis and the August period was reminiscent of the three years ago
period. Economists do have technical definitions of such matters based on
negative GDP growth in two consecutive quarters. And there are others
predicting a recession and those who think a double dip will be avoided.
But looking around one sees a continuing weak economy. A recent example
is the Hearst San Simeon tours at the wonderful mansion - 126 rooms -
filled with treasures, many from renaissance Europe that William Randolph
Hearst collected over the years. The estate up a 5 mile winding road on the
California Central Coast, is so large that there were four tours which
together allowed one to see it all. Now (in September 2011) over two thirds
of the tour guides have been laid off, most of the rest are only part time, the
four tours are three, the cost is up and the tour lengths and coverage are
considerable shortened and by taking all three tours one does not see all of
the castle. The tours and environs are still great but the drop in business has
been so severe that they had to do this to cover their bills from the
gardeners and other maintenance people who must be retained. The Hearst
family gave the property to the state of California but the state is so cash
strapped that the property must be self financing. The US and other
countries have strict rules on such gifts that nothing can be sold - prime
candidates where easy money could be made are the fruit on trees not used
and the wine and other liquor spoiling in the wine cellar because of
advanced age.
August was a violent month with many three digit ups and downs in the
US stock market's Dow jones Industrial Average. The trouble began on
August 2 and soon there were four consecutive days with movements
greater than plus and minus 400 points just like the week before the 1987
stock market crash. This was the first time the market had such 400+
gyrations on consecutive days. The VIX volatility index rose to 48 and
stayed high all month. Figure 34.1 shows the S&P500, Russell2000 arid
VIX indices for the six months prior to our press date of September 21,
2011. Figures 34.2 and 34.3 show the euro, pound and Swiss Franc and
gold and oil compared to the US dollar.
Fig. 34.1 The S&P500, Russell2000 and VIX indices, March 22 to September 21, 2011. Source:
Yahoo Finance
Fig. 34.2 Currencies - euro, British pound, and Swiss franc - compared to the US dollar, March 22 to
September 21, 2011. Source: Yahoo Finance
The trouble started right after the US President Obama-Republican

agreement on the debt ceiling and spending cuts. The apparent standoff and
the huge amount of US debt, largely from the Bush administr ation's tax
cuts and military spending without raising taxes, led to the downgrade of
the AAA rating of the US debt to AA+ by Standard and Poors. The
downgrade of US growth the week before the Debt ceiling compromise also
spooked the markets, adding to concerns as it made clear that the recession
was even deeper than previously anticipated and the recovery even weaker
making it clearer that the US had not really exited recession, despite a series
of monetary and fiscal stimulus. This seems equally as important to the debt
ceiling debacle at a time when the U.S. economy remained weak, US
political leaders spent a month haggling over the passage of legislation that
should have been almost automatic after the passage of a budget that
required an increase in spending. Rather than on trying to come up with
political solutions to deal with the underlying issues that keep the U.S.
weak, such as the mortgage market. This downgrade was a symbol of the
nervous market as the bond market was already pricing this in. Basically, no
one liked the agreement and Obama looked more like a Republican than a
Democrat. Instead of standing up to the right wing, he compromised but
had to move way to the right on the political spectrum. This is one of the
major US problems: the political atmosphere is very confrontational with
many of the extreme right wing Tea party people not caring about what
damage they do to the economy. They simply want to beat Obama in the
next election even if they lose their seats. They have many inappropriate
strategies such as recognizing and not telling the truth, creating rules so that
poor blacks, students and others will be prevented from voting because they
lack specified credentials with the places to get those credentials far away
and only open limited hours. What is dangerous is that a small group, the
Tea Party, have co-opted the debate.
Fig. 34.3 Oil and gold based in US dollars, September 22, 2010 to September 21, 2011. Source: CNN
Money
A similar situation seems to be developing in Germany. European debt

issues were a major factor too and now we have the ECB and the Fed
giving money at low interest rates to troubled European banks so, like 2008,
they have unlimited access to US dollars for the rest of 2011. Christine
Lagarde, the new head of the IMF, has maintained that these banks must
obtain more capital and applauded this new activity of the ECB in
collaboration with the US Fed, the Bank of England, the Bank of Japan and
the Swiss National Bank. The idea was to ease the pressure on weak
European banks by giving them access to loans — again borrow more to
pay past debts, a band aid solution putting off the inevitable Greek default.
A Greek default could lead to an Italian and Spanish default and that's a big
worry for the world economy. The US loans are to the ECB not the
European banks, so the ECB must repay the Fed in dollars. According to
The Economist and Goldman Sachs, Europe's 38 biggest banks need
between 30 and 92 billion euros to cope with haircuts to Greek, Irish and
Portugese government bonds and losses on Italian and Spanish government
debt. An IMF study suggest that the banks might lose about 200 billion
euros if the current implied market prices based on current default
probabilities prove to be correct.
The Obama-Republican agreement has a debt limit cut off of at least $1.2
trillion by the 12-member bipartisan special debt committee who must
recommend these cuts by November 23, 2011 and Congress must approve
them by December 23. There is a not small chance - Nouriel Roubini has
said it is 2/3rds probability that they will not reach this agreement in which
case automatic across the board cuts come into play. House speaker John
Boehner is against any new taxes, especially those to the rich. He argues
that tax increases destroy jobs. The econometric evidence suggest
otherwise.
A big issue is US spending. For example from January to August 2011,
the Federal deficit was $1.23 trillion. Figure 34.4 shows the debt and deficit
situation versus the size of the economy.
Fig. 34.4 US debt and deficit to GDP, 1940 to 2015 (est), Source: dailybail.com
There is 9.1% unemployment and about 20% when you count

underemployed plus those who have given up trying to find a job. And
there are too many poor people: the top 1% have wealth equal to the bottom
90% and their yearly income is equal to the bottom 60%. Fully 64% of all
Americans cannot raise $1000 in one day for an emergency. At a posh Del
Mar restaurant WTZ gave some racetrack passes to the waitress - she said
she was one of the 64%! There are too many people who cannot contribute
to the economy. This waitress has a job but has a struggle to pay her bills on
a daily basis. How can a country expect to move ahead economically with
close to two thirds the population not contributing much.
When markets are weak and turbulent, new issues do poorly and indeed
63% of US based companies making their initial IPO's are trading below
the offering price; see Cowan and Smith (2011).
Update on the Fed: Operation Twist: The September 21, 2011, FOMC
meeting kept short term interest rates near zero and pledged by June 2012 to
sell $400 billion of short term notes and buy $400 billion of long term 6-30
year bonds. This twist, which does not affect their total balance sheet is
intended to lower long term interest rates. They feel the economy has
significant downside risk, inflation is moderating and the unemployment
rate will decline but only slowly. The stock market did not react well to this
- the S&P500 fell about 8x0 points, to put it down 6.77% for September to
the 23rd and -9.64% for the year. There was also a collapse in the
commodity prices with gold down over $100 and silver falling 17% to
about $31. Forsyth (2011) points out that the Fed moves have not been
approved by the market. The words significant downside risk coupled with
no new buying only a switch of assets led to the decline. The market wanted
quantitative easing - more financial stimulus, not qualitative easing, a shift
of bond durations. This move lowers bank profitability since net interest
margins will fall and so will demand for long term loans at low rates about
3.75% for 30 years. Also pension funds are hurt by low long-term interest
rates. Operation twist is likely the precursor to QE3, which is unlikely to
have the same reflating effect as the previous rounds of quantitative easing.
Moreover, if stability-seeking global investors like China, continue to shift
their portfolios on the margin to shorter-duration securities, it will
undermine the effect of the Fed's move. Again this is a limited move,
anticipated by the market and may do little to help the real economy, hence
the desultory response. Ultimately the fiscal tool needs to be used, and in a
way that will stimulate the economy, not just tide it over.
Some other US problems and strengths:
Weaknesses of the US, but the UK and other countries have many of
these same problems.
• The income distribution is the worst in its history since the Glided
Age.
• There are too many poor people - in 2010 there were 46.2 million
people in poverty or 15.3% of the population (up from 14.3% in
2009). For blacks and hispanics were 27%, single mothers nearly a
third and children over 20%. Poverty is defined as an income of
$11,139 or less for a single person or $22,314 or less for a family of
four.
Extreme poverty is when the single or family income is less than half
this amount and about half of the poor people are in extreme poverty -
and Republicans such as Michelle Bachmann want even these people
to pay taxes so the rich do not need to have a tax increase!
• There is way too much unemployment (9.1% in August) and
underemployment (together close to 20%). Then even if they get jobs,
the pay is low and the work is likely part time. First time
unemployment claims rose to 11,000 in the third week of September
to 428,000 versus a forecasted decline, see Craft (2011). Spain has
over 20% unemployment and about 48% for youths.
• The political atmosphere is very confrontational and not cooperative.
• The tax structure does not tax the wealthy enough. The claims that
greater taxation would stifle growth and lead to lower employment
are inconsistent with the data. See Figure 34.7. The wealthy have
available many ways to reduce taxes so their net marginal tax rates
are below that of the middle class.
• The wealth and living standards of the middle class have been
dropping for more than twenty years; see Rugaber and Carpenter
(2011).
• The country must adjust to a slow growth era where the US is less
important in the world than in the past. The standard of living is
dropping and must be adjusted to.
• The low bond yields and correspondingly low interest rates are a bit
helpful but the biggest beneficiaries are blue chip companies not job
seekers.
• There is way too much private and public debt. Fortunately it is in US
dollars so debtors do not bear the exchange risk. And more and more
money can be created.
• The economy, which is weak, needs short term stimulus and cuts and
a long term debt reduction plan.
• Obama is spending too much on wars he cannot win that hurt the
soldiers and their families more than prevent attacks on the US. Bush
started this problem but Obama sent even more troops.
• US consumer sentiment for the future outlook is, according to the
Thomson Reuters/University of Michigan index survey, at the lowest
level in 31 years. Consumers are frustrated with high unemployment,
job destruction, not much job creation and the gridlock in
Washington.
• The currency is weak from lack of exports and the debt problem is
pushing it lower. Generally, strong economies usually have strong
currencies. Of course, much is valued such as the Swiss franc (paper
gold). A weak US dollar is a signal of a weak economy. But this is
complex since previous dollar strength undermined exports and
strengthened imports, helping fuel imports and consumption (which
accounts for over 70% of US GDP. The USD benefits from flight to
liquidity and from being the best of a bad lot. Nouriel Roubini calls it
the ugly contest: who is the least ugly?
• The country is vulnerable to Chinese interests which own too much
US debt, over 3 trillion most in US treasuries plus other hard assets.
Japan owns a huge volume of US treasuries as well.
• Many of the states are near bankruptcy and their spending cuts are
greatly hurting many people. See Gorenstein (2011) who cites
Harvard Economics Professor Jeffrey Mison who believes that the
states are in for worse trouble than currently thought because of
declining revenues and large medical and pension liabilities.
• The housing situation remains bleak with many foreclosures and
delinquencies. The housing market peaked in 2005–06 and has
dropped since with a blip up. Recent Case-Shiller data in Figure 20.1.
In Florida, banks are bulldozing foreclosed houses and entire
neighborhoods are being gutted. One of every 118 houses in Nevada
was foreclosed just in August 2011 and it is one in 248 in Arizona,
one in 349 in Michigan and one in 376 in Florida. Nearly a quarter of
all US home mortgages are under water, nearly 11 million homes; see
Benac (2011).
• The subprime mortgage policies to give anyone a house regardless of
their fi-nancial situation caused much of this problem when housing
prices which had risen for more than twenty years suddenly began to
fall. And this was made worse by the packaging of the tranches and
rating them AAA and selling them all over the world.
• A possible hard landing in China in 2013, which is a significant risk
according to Nouriel Roubini and others would hurt the US and the
world economy. There has been a reaction to the European debt crisis
and the result of a serious monetary tightening in response to high
inflation. Already, real estate has stalled. And the Chinese stock
market has fallen in 2011.
• Risk control is still weak. It is not in the US but the rogue trader,
Kweku Adoboli, who lost about $2.3 billion at the Swiss bank UBS
in London reminds us that this issue is still not settled; see Jordans
and Katz (2011). Adoboli graduated in 2003 from the University of
Nottingham with a degree in e-commerce and digital business hardly
the risk control training for a derivatives trader. UBS, which received
a $60 billion bailout from the Swiss government in 2008 had planned
to cut 3,500 jobs over two years and this loss may cause more.
Previous big individual trader losers include Nick Leeson, who took
down Barings in 1995 with a $1.28 billion loss. He now gives
lectures for £10,000. And in October 2010, Jerome Kerviel lost $6.7
billion for Societe Generale.
Strengths
• The top universities such as Harvard, Chicago, Stanford, MIT, NYU
and others are still among the best in the world. But only at the top.
The education system in most of the country is weak and in the world
wide ratings the US is nowhere close to the top. You see this in
graduate classes filled with Asian students. In mathematics and
science, the US is 27th of 35 countries studied. Taiwan, Singapore
and Korea are at the top. But the non-private universities are under
financial stress because state funding is dropping as is student aid.
For example, at UC Berkeley, state funding is now only 12% versus
30% of their annual budget before the 2007-09 financial crisis. The
University is rated fourth world wide and produced more PhDs in the
last 50 years than any other US university; see Birgenean (2011) and
the Shanghai Ranking Consultancy (2011).
• It is still possible to start with nothing and become a great success and
be rich. In certain areas there is confidence.
• The movie and sports industries remain the best in the world.
• Some companies are still the best in the world such as Apple and
Boeing, Much of their production is abroad but the idea generation
and R&D is primarily in the US. The challenge with the poor US
educational system is that many employees have not been trained to
move up the value chain.
• There is still the American spirit we are the best country in the world
but this distracts them both from seeing their faults and seeing what
they can learn from others.
• And there is the philosophy that winning is not the most important
thing, it is the only thing this creates a culture of rewarding success
but, in turn, this puts too much emphasis on the leader and not the
contribution of the rank and file leading to a growing income
disparity.
• The US also has the rule of law, encouragement of innovation, and
intellectual property rights which are features that can not necessarily
be taken for granted. It is still possible to start with nothing and
become a great success and be rich. However this still requires access
to opportunity, understanding teachers etc. Lack of a defined class
system is still a plus and labor mobility is still good.
The US stock market and European debt issues

If the economy is so bad and the European debt issues so dire, why is the
US stock market, despite the August-September turbulence still nearly
twice the March 2009 lows? October 9, 2007, the S&P500 peaked at
1565.15. My bond- stock earnings yield differential went into the danger
zone on June 14, 2007, for results and analysis see Chapter 21 based on
Lleo and Ziemba (2012). The market fell to an intraday low of 666.79 on
March 9, 2009 and has since rallied to a high of 1357.16 on May 10, 2011.
As of September 16,2011, when we went to press, the S&P500 was
1216.01.
The bond-stock earnings yield model is at
BSEYD= Ten year bond rate- 1/trailing PE ratio = 2.08-

(100/20.95)=-2.693
which is far in the buy zone with earnings high and interest rates low.
This uses Professor Robert Shiller's average inflation adjusted earnings
from the previous ten years which are below actual trailing or forecasted
S&P500 yearly earnings which give an 11 PE ratio. The 10-year treasury is
the most liquid; the 30-year treasury is at 3.34. Of course, with the bond
market predicting an 80% chance of a double dip recession future earnings
might drop. Firms are lean and are not hiring but rather more inclined to
layoff workers on any demand slump so earnings may not fall much. But
the European debt issues remain and most likely will get worse. The US
stock market reacts greatly to news about the European debt situation. US
corporations hold $2 trillion in cash. Apple itself has more cash than the US
government; see Rugabee and Carpenter (2011).
The problem is that Greece and the other PIIGS (Portugal, Italy, Ireland
and Spain) are all close to bankruptcy with debts they cannot service. Table
34.1 has population, GDP and debt for the euro zone countries and non-
euro zone countries in Europe. The French and Germans keep talking about
saving the euro which is way overpriced and the European Union itself and
avoiding a Greek default. But this is a bandaid solution: borrow new money
to replace bad old money. What is needed is less debt in Greece and in other
PIIGS. Some estimate the probability of a Greek default at 98% and the
bonds are trading at high coupon rates; ten year government bonds are at
24% versus less than 3% for US treasuries, see Moulson and Paphitis
(2011). The second Greek bailout, some 109 billion euros which the
Economist thinks is not adequate, was agreed to in July but delayed because
Finland wanted security of repayment. That aid is scheduled to be settled on
October 3. This is already at a 21% discount, see Thomas (2011) but with a
full scale default it would be about 50%. This is for current expenses.
In December, a more serious debt repayment is due. Moves are being
made such as the co-ordinated access by five central banks of unlimited
borrowing of US dollars for three months. These seem to indicate that an
orderly default is coming. That would restructure the loans. The total Greek
debt is about 370 billion euros ($500 billion). Italy's government debt is
about $3.5 trillion, five times Greece (see Thomas, 2011). Table 34.1
indicates lower debt. In comparison, Argentine had $82 billion when it
defaulted in 2001 and Russia had $79 billion when it defaulted in 1998.
French banks have a lot of Greek debt. It is not known exactly the US
exposure but one estimate is 2%, not very much. But US financial
institutions may hold other derivatives tied to this toxic debt. Greece is
really insolvent. Italy, which has a primary fiscal surplus could become
insolvent if interest rates are high for a prolonged period. In a best case Italy
can muddle through and deal with its debts although this dampens growth
prospects.
Austerity measures are difficult. Among other things, unemployment is
high — 10% for the euro area, Greece 16%, Spain 21.2% and Italy 8%,
according the The Economist. Greek GDP fell by 7.3% in Q2:2011 versus
2010. Italy's bonds have been downgraded from A+ to A by Standard and
Poor's. They are selling much of their debt to Chinese sovereign wealth
funds. Greek workers are reluctant to collect taxes and other charges from
an already strapped public, especially after they have also been cut. The
non-payment of taxes is a major problem. A notable example is swimming
pools in Athens where few admit to having them while aerial photography
shows thousands of pools. Many European banks, particularly Societe
Generale and BNP Paribus in France would then take huge losses. They
already have these losses with mark-to-the-market accounting but these are
not realized yet; see Piovano and Wagner (2011). Figure 34.5 shows the
MSCI equity indices of French and European banks which are down about
50% in 2011 and the share prices of BNP Paribas, Credit Agricole and
Societe General. To get a feel about the European bond and stock markets,
lets compare them with the US. See Figure 34.6 for the 5-year credit-default
swap spreads which have risen dramatically in 2011.
Table 34.1: Population, GDP and debt for eurozone and non-eurozone
countries in Europe. Source: Daily Mail, October 28, 2011
Fig. 34.5 Bank Data: Source: The Economist, September 17, 2011
What can the US do to improve the situation?

What are Barack Obama's chances to win the November 2012 election? He
is slightly favored in the polls as all the Republican candidates are weak and
extreme in their views. However, that's not necessarily a handicap as the
country is very divided politically. They are the party of insults and false
claims and innuendos. And it is dangerous. Democrats check facts,
Republicans make them up. There is an HPV vaccine that prevents cervical
cancer and is recommended for females aged 12-25. Out of 23 million
doses according to a University of Pennsylvania doctor, there were 12,000
adverse reactions with 4,000 serious. Texas mandated immunization in
2007 leaving Governor Rick Perry open to criticism by his party.
Republican Representative Michelle Bachmann is using this to attack Perry
and claimed some woman said her daughter had brain damage as a result of
the vaccine. The doctor said this is not possible; Bachmann failed to fact
check and later said she was just reporting what she heard - not taking
responsibility for her statement. The real danger is to those who are not
immunized because of Bachmann's false claim. Such is the hatred of
regulation and the fear of sexual activity.
Fig. 34.6 European banks' five-year credit-default-swap spreads, basis points. Source: The
Economist, September 17, 2011
The Republicans never accepted Obama as their President. There was a

huge campaign to discredit him from the beginning. They claimed he was
not born in the US and thus could not be president. Then it was his
qualifications: was he favored to get into Harvard because of his race and
did he get favored grades? The facts are he was editor of the law review at
Harvard and a University of Chicago professor. At the same time
Republicans belittle him as being elitist for his success (Perry proudly
proclaims ranking low in his class at veterinary school). In contrast, George
W. Bush, a C student at Yale, did not suffer such attacks.
Betfair odds suggest that the election currently looks very close. Obama
is an overwhelming favorite to get the Democratic nomination 1.09-1.12
bid-ask for the election where you bet 100 to win 90. Hillary Clinton is 17-
24 and Biden 19-190.
Obama's approval rate, while low, is higher at this stage of his presidency
than any other Democrat since Truman - above Clinton's and the others.
According to a September 16, 2011 New York Times poll, 43% approve
versus 50% who disap-prove. This is also better than the dismal rankings
for Congress: 28% approve, 63% disapprove for Democrats and 19% versus
72% for Republicans. However, for the election, Obama is 1.93-1.95 versus
Rick Perry at 5.6-6.2 and Mitt Romney 5.5-5.7. Regarding the party that
wins the election, the Democrats are a slight favorite at 1.91-1.92 versus
2.08-2.10 for the Republicans. Perry is 2.46 -2.52, Mitt Romney 2.5-2.58 ,
Sarah Palin 22-30, Jon Huntsman 32-60, Ron Paul 19.5-25 and Michelle
Bachmann 36-48. Current polls have Perry at 23%, Romney at 16%,
Bachmann at 7%, Newt Gingrich at 7%, Ron Paul at 5% with many
undecided.
Update: Romney got the Republican nomination. The September 30, 2012
Betfair odds when we went to press were 1.23-1.24 for Obama and 5.4-5.5
for Romney. So Obama is highly favored with the debates to occur soon.
While most first term presidents get re-elected, a few, like Jimmy Carter
and George H W Bush, lost.
What Obama can do, a shopping list:
(1) Try to fight back against the toxic right wing by going directly to the
people. Ask them to recall elected officials that do not represent them
and are against the best interests of the country. Only 33% of the
voters think the person they voted for should be re-elected. Congress
has a 15% approval rating in September versus 13% in August by
another poll even lower than the one discussed above.
(2) Start a public works program with the unemployed to rebuild roads,
bridges, etc by getting funding approval from Congress. Bill Clinton
has a suggestion regarding clean green energy.
(3) Simply tell the Republicans that he was elected by the people fair and
square and go back to the policies that got him elected. But a factor is
the Republicans won the House of Representatives in 2010.
(4) Figure out how to tax the risk, especially those earning one million per
year or more. Ask Warren Buffett to lead this activity. He needs
congressional backing, they pass legislation, he signs or vetos. So
some negotiation is needed. So far Obama's tax rates are the same as
Bush's at a very low level. Taxes were lowered just before 9/11 and
the cuts were not suspended when the wars began. Usually taxes or
government bonds support wars.
(5) Cut back a lot on military spending: does the country really need huge
bases all one the world — smaller ones and co-operation with allies
would save a lot. Also the ill advised wars in Iraq and Afghanistan
need to be wound down. What's important is to protect the country
and the allies. This was a huge Bush mistake that Obama perpetuated.
Kamrany and Intriligator (2011), who have their own ten suggestions,
many of which parallel mine, point out that each soldier in
Afghanistan costs $1 million per year. Bush sent 70,000 troops there
and Obama another 30,000 so this is -$100 billion per year wasted.
Obama added soldiers but they are now drawing down, albeit not fast
enough to fully cool the military spending. In part this reflects the
need to retool the army after a decade of wars, and to pay for all the
human costs of the war, medical care, including psychiatric care and
pensions. These labor related and health costs account for the greatest
increase in spending growth in military spending in the last decade.
(6) Cut back on useless checks at airports but hire El Al-type inspectors
who do real checks to devise a better system. I find checking my
passport and boarding pass six times and not really checking me is a a
waste of money and it is not safe. The tradeoff here is cost which I
think would be less than current but waiting time would increase.
(7) Cut waste in government — roll back salaries that are much higher
than in the private sector.
(8) Make peace with Wall Street. It is believed that Obama does not favor
wealth creation and corporate profits. The facts are that since June
2009, 88% of the gains in gross national income went to corporate
profits and only 1% to wages. Convince them that he is a Democrat
and not a socialist. Remind them that the Dow Jones Industrials index
was 7949 on January 20, 2009 when we took office and is now 31%
higher at 11,509.
(9) Be sure that all financial institution bailouts have conditions where the
funds given to financial institutions cannot just be given in the
millions to top executives of firms losing money. TARP was way too
lenient and not transparent enough. This has led to great resentment of
the many agains Wall Street. They see rich people who they think
caused the problems receiving aid but they get nothing.
(10) Ask Professor Ed Altman of NYU who really understands bond
ratings to lead a panel to study the rating agencies and propose a new
rating agency that is more accurate, less sticky to old ratings and is
independent of this they are evaluations. The current rating agencies,
Standard and Poor's, Moody's and Fitch have a number of flaws and
conflicts of interest so a fourth and independent rating agency would
be a good idea.
Sandra Schwartz points out that the computer and internet world we are
living in has lost a lot of jobs and these benefits are not counted properly.
Also, the weak infrastructure situation of the US which as helped the rich
while paid for by all the populate needs updating. This would renew for the
future and provide jobs for many unemployed construction workers. Most
firms have capital accounts for investment versus current spending wages
and salaries, etc but all government expenditures are counted as current
consumption and not as investment.
Obama does have a jobs plan, see Table 34.2. However he faces the usual
fight with the Republican House over the tax increases for the wealthy. The
Republicans say that will hurt jobs. Corporations are sitting on $2 trillion in
cash and confidence is needed to get them to move on job creation. Obama
is now more on the offensive and returning to the policies he was elected
on; see Newman (2100). Indeed instead of creating jobs, there are layoffs,
35,000 at the US Postal System, 30,000 at HSBC, 11,000 at Bron's and
1000 at Goldman Sachs for example. His supporters are more pleased. Time
will determine how successful he is. He is in campaign mode and the days
of capitulating to the Republicans seems to be over. Blinder and Hubbard
(2011) both well known economists who have served on the Fed and
Council of Economic Advisors have a bipartisan plan:
• an increase, eventually, in the normal retirement age for social
security.
• cuts in Medicare and Medicaid including greater means-testing in
Medicare.
• tax reform — broaden the base rather the raise the rates by
eliminating some tax deductions.
Table 34.2: Obama's jobs plan. Source: The Economist, September 17, 2011.
New stimulus measures xs $bn
Business tax cuts 70

Infrastructure spending, state 140
and local government aid
Unemployment insurance and 62
training
Payroll-tax cut 175
Total 447
Paid for by
Limiting deductions for upper- 405

income earners
Treating carried interest as 18
ordinary income
Limiting oil and gas company tax 40
benefits
Removing corporate jet tax
3
breaks
Total 467
The Republicans that proclaim that the job creators should not be taxed
have not checked the facts. Regarding tax rates for millionaires and
unemployment in 1995, the tax rate was 30.4% with 5.6% unemployment
and by 2009, the rate was 22.4% and unemployment was 9.3%. Linden
(2011ab) observes that economic growth has historically, from 1950–2010
been stronger during periods of higher tax rates as shown in Figure 34.7. Of
course, wealthy people have many sources of income so the overall tax paid
will vary. It is clear that higher taxes for millionaires will not necessarily
lower job growth and will result in a more fair tax system when they pay
their share.
Fig. 34.7 Economic growth compared with top marginal tax rates, 1950–2010. Source: Linden,
2011ab
Michelle Bachmann says 47% of American's do not pay any tax. That's
right — however, according to Stephen Colbert they take advantage of the
ultimate tax loop - having little or no income. Her statement that everyone
should pay their share of taxes forgets about the 9.1% who are unemployed,
16.6% in poverty, etc. Who are these people who pay no tax? According to
the Colbert (September 20, 2011): 50% have incomes under $22,000; 22%
are seniors, 15% are low income families with children and 13% others.

Investing Around the World1
This chapter reviews some interesting facts and analysis on international

investing that I learned from the students in my masters classes at the
ICMA Centre, University of Reading and the University of Cyprus in
March 2012. I look at Korea, India, Russia, China and Cyprus all of which
are of great interest starting from the viewpoint of the students, who
represent these countries and have a unique prospective on developments
there. I have supplemented their presentations with my own thoughts and
research. Given the structural changes underway in the global economy,
investors have to look closely at both the domestic fundamentals but also
the trade and financial links across the globe. Many of these countries,
despite having strong long-term growth paths, have a series of
vulnerabilities, which will necessitate structural reforms to maintain and
grow domestic and foreign market share in a period of global slowdown.
Doing so will frame the investment opportunities.
Korea
Korea, where I am a visiting professor at the financial engineering
department of the Korea Advanced Institute for Science and Technology
each August is especially interesting. KAIST, which is modeled after MIT
is largely located in the city of Daejeon some two hours away from Seoul.
Like Tsukuba in Japan where we spent 1988-89, Daejeon has the major
research facilities of the country's manufacturing and other companies. The
KAIST business school is located in Seoul closer to the financial center of
the country. I spoke there in 2011 and I visited both places again in 2012.
South Korea has had rapid and steady GDP growth since 1970 as shown
in Figure 35.1a, total equity trading volume has also grown along with
GDP, see Figure 35.1b. In contrast, North Korea has remained poor with the
government siphoning off the limited resources for the military and
themselves, while allowing food shortages and malnutrition. Still uncertain
political environment in the North remains a major geopolitical risk both to
the South, as well as to neighboring China (which fears an influx of
refugees) and Japan.
South Korea's big strength has been in education and innovation, as the
government, big conglomerates and households invested heavily in human
capital to offset the lack of other resources. South Korea has an essentially
all educated population with 80% finishing college or university. This
comes from an educational system that is rigid and pushes students to their
limits and the mathematics and science courses have it near the top in the
world. Outstanding students can skip grades or even several years by taking
exams.2 There is a cost to such pressure including many suicides and other
problems. Some parents send their children to Australia or other places for a
more normal upbringing balancing study and other activities though without
family.
South Korea has extensive research and development investment which
is about 3.0% of GDP and has been outpacing GDP growth. This is similar
to the US and Finland and only exceeded by Japan, Sweden and Israel and
well above China at 1.4% and India at 0.9%, see Figure 35.2. This shows up
in innovation index (based on things like tax policy, education, exports of
high tech products) with South Korea in second place behind Singapore and
above the US and Japan, see Table 35.1. This heavy investment in
innovation has helped Korean conglomerates to maintain their
competitiveness, even as global export growth has narrowed and they have
received greater competition from Chinese producers.
Fig. 35.1 South Korea Data, 1970–2010
Fig. 35.2 Scientists and engineers per million people versus R&D as a percent of GDP for various
countries
Table 35.1: Innovation index based on a variety of inputs and performance factors.
Source: Boston Consulting Group, 2009
The stock, currency and derivative markets are very active, befitting
Korea's place as a developed market (though it does retain some elements
of a market that is still emerging, including a tendency to intervene in the
market). After the major shock of the Asian financial crisis in 1997, Korea,
like its neighbors took the lessons of cutting spending and building up FX
reserves. Moreover, its domestic contraction left it even more focused on
external demand, especially high-tech exports. It also reflects an expansion
of financial markets. Total equity trading volume and market capitalization
has also grown along with GDP, see Figure 35.1b. Although many
companies finance their capital expenditures through retained earnings, they
also borrow on international markets, and are still dealing with the currency
mismatches that developed before the Asian financial crisis.
The main stock indices are the Kospi overall index and the technology
Kosdaq index, see Table 35.2 for trading volume for 2008–2010. Most
Korean companies are family dominated where the management and major
ownership is common. Hence, there is a lack of transparency, but, according
to UBS in February 2012, their corporate governance was ranked the best of
the emerging markets (EM). Figure 35.3 shows the indices with the
technology sector Kosdaq performing poorly and never recovering from the
2000-2002 internet bubble much like the US Nasdaq which it actually
underperformed and which itself is also well below its 2000 peak. However,
the overall Kospi index has had good performance but, like the Kosdaq, and
indeed other EM Asian exchanges, it has been extremely volatile. In fact,
the daily Kospi volatility in 2011 was 1.23% versus 1.0% for the FTSE100,
0.95% for the 30 stock Dow Jones average and 1.06% for the Nikkei225.
Total trading volume continues to increase steadily as shown Table 35.2.
Table 35.2: Kospi, Kosdaq equity trading volume and Kospi200 options trading volume,
2008-2010
Fig. 35.3 Kospi and Kosdaq composite indices, 1999–2012
A major factor affecting the price levels and volatility of the South Korea
market is the ever present tension with North Korea and war risk which
followed the 1950–53 Korean War. There has been an armistice but never a
peace agreement and the two countries are technically still at war. Since the
Armistice Agreement of July 27, 1953, North Korea has violated the
armistice 221 times, including 26 military attacks. On November 23, 2010,
North Korea fired artillery at South Korea's Big Yeonpyeong island in the
Yellow Sea and two marines and two civilians were killed. In March 2012,
the agreement to stop nuclear tests in exchange for food was violated when
North Korea planned missile testing. Figures 35.4ab attempt to measure the
war risk in two ways. In Figure 35.4a the price earnings ratios of the MSCI
advanced country index and that of Korea are compared. We look at equity
indices not the Korean won, given that the central bank's tendency to
intervene reduces currency volatility. The difference can estimate the war
risk and possibly other factors. Korea considers themselves to be advanced
and not an emerging market, but its market returns are more correlated with
emerging markets. In Figure 35.4b the South Korean war risk is measured
using a peace index. The war risk seems to correlate with the current
tension in the region over time.
Fig. 35.4 South Korea War Risk Factors
India
While it is almost as big (in terms of population) and important as China,
India is less in the news and less well understood. It is a vast country of
contradictions, particularly now that investment in education and services
has prompted extensive business processing outsourcing. I recall on a visit
in the 1970s traveling through the slums of Bombay for about three hours.
There were thousands of shacks housing rather poor people. When you got
on a bus there were at least two others who had the same seat assignment as
you did and the bus was packed solid inside and on the roof with huge
numbers of travelers. Yet the top universities in the world are full of Indian
academics in very high positions and this is at places like Oxford,
Cambridge, Yale and Harvard, etc, as well as outstanding places in India
like the Indian Statistical Institute. A small section of Mumbai has some of
the highest land prices in the world.
India is the world's second most populous country and its largest
democracy. It has cheap labor and many skilled workers and professionals
which is adding to inequality and in turn brings political strains and policy
gridlock. Its 2011 GDP was $1.846 trillion and per capita was $1,527 with
growth at 8.4%, see Figure 35.5(a). Sectoral GDP estimated for 2011 was
agriculture: 18.1%, industry: 26.3%, and services: 55.6%. CPI was 6.55%,
see Figure 35.5(b) and public debt 71.42% of GDP.
Fig. 35.5 Some macro data for India, January 2008–January 2012
Enhanced liberalization policies since 1990 have attracted large foreign

direct investment. The cumulative FDI equity inflows from April 2000 to
December 2010 stood at US$186.79 billion. For fiscal year ending March
2011 the composition of FDI was financial and non-financial services
US$2,853 million, telecommunications US$1,327 million, automobile
industry US$1,066 million, power sector US$1,028 million and housing
and real estate sector US$1,024 million. FDI comes mainly from Mauritius
US$5,746 million (42%), which is a transit point for other funds, Singapore
US$1,449 million and the US US$1,055 million. The government's
ambivalence towards FDI is a constraint to India's long-term growth, as
political issues limit its ability to attract the foreign capital needed to build
out infrastructure, to support higher levels of growth without inflation. In
recent years, growth above 8% has prompted high levels of inflation. There
are two major stock exchanges in India: the National Stock Exchange of
India (the index is the S'P CNX Nifty known as the NSE Nifty or NSE 50),
see Figure 35.6(b) and the Bombay Stock Exchange (the index is the BSE
SENSEX or BSE 30), see Figure 35.6(a). The NSE has had the world's
highest yearly market capitalization growth of 30.9% followed by the BSE
with 29.4%. Figure 35.7 compares these with other exchanges.
Compared to China though, India has stronger demographics, a lower
number of retired people to support, and this young population, will not
only keep labor competitive but suggest a pattern of increased savings
going forward (if government policies permit). These savings could finance
domestic investment, needed to increase potential growth rates. This
however is a long-term story. In the short-term, India faces persistent fiscal
and external deficits, which have contributed to an extensive slump in the
rupee in the last six months. The recent moderation in the oil price, if it
lasts, would be a modest reprieve for both deficits, as fuel imports deter net
exports from adding to growth, while swelling government spending as the
government fears passing on higher costs to the population.
Fig. 35.6 Two major stock exchanges in India, 1998–January 2012
Fig. 35.7 Comparison of Indian exchanges with other major exchanges
Russia
Possibly the most complex of the BRICs is Russia. There is great fear of
investing there following a period of resource nationalism and government
encroachment on both foreign and domestic business. For example, a
partner in a leading LP company said “If I bring a Russian deal to the
investment committee they could fire me. And if I do not bring Chinese
deals to the committee they could fire me.” So courage is needed. Yet
Goldman Sachs updated their BRIC report to 2010 and Russia was rated
second for attractiveness, in part due to its larger per capita GDP than other
BRIC and its low government debt. Bill Gross is buying their bonds for his
PIMCO bond fund (note that Russia has minimal formal government debt
due to high oil spending). Jim O'Neill, Chairman of Goldman Sachs Asset
Management said “Concerns about Russia are overblown and offer
investors a chance to buy into booming growth stories. Contrary to the
email I get every two hours about why I should drop the R from BRIC, I
quite like Russia.” As with India above, we still fear there are structural
issues to investing in Russia, In fact the stronger the oil price, the worse the
policy making, as high oil (and gas) prices allow Russia to deter tax
reforms.
We had my own experiences there in 1992 when our family traveled from
Helsinki where I had been speaking at an international Informs conference
to St Petersburg. Among other things we wanted to see the Pazryryk carpet.
This famous carpet of high quality with reindeer on the borders was made
in the 5th century bce and survived frozen in Siberia until it was unearthed
in 1949 from a burial site with many other objects including felts. Since the
next high quality carpets are the 13th century Seljuk ones found in Konya,
Turkey, now residing in the Turk ve Islam museum in Istanbul one wonders
what happened in these 18 centuries. Since the Pazryrk is of high quality,
carpet production must have been going on for some time before. Though it
is not known where it was made, there is no shortage of countries claiming
that the carpet was made there and these include Turkey, Iran, and Russia.
After considerable inquiry at the Hermitage Museum, we found it in a
basement glued to a fabric on a wall. It was rather dusty and experts and
amateurs were allowed to snip away little souvenir pieces for study. Wow -
if they only knew the true significance of this carpet! But there are experts
in St Petersburg and chief among them is Elena Tsareva. She is the curator
of the Peter the Great Museum of Anthropology and Ethnographics of the
Russian Academy of Science collection of Turkmen carpets and a world
renowned rug scholar. In 1992 when we went, the salary was $10 a month
and rarely paid; she had to sell jewelry she collected to visitors like us, at
possibly inflated prices, to survive. At that time in the universities, $20 a
month got you a professor, $40 a star and for $60 one of nobel quality.
In Chapter 26 we discuss the risk management failure of Long Term
Capital Management, the MIT led group that was too smart to lose. Among
other things, they got wiped out by the August 17, 1998 Russian ruble
devaluation. We were guests of Professor Zari Rachev, an expert in stable
and other heavy, fat-tailed distributions. On arrival Bill gave him a US$100
and he in turn he gave a four inch wad of 25 ruble notes. Our dinner that
night cost two inches for the four of us with drinks extra in hard currency.
Well times have certainly changed. For example, in 2012 Russia has more
billionaires than any other country. Some facts about Russia are in Table
35.3.
Figure 35.8 shows the close relationship between Russian stocks
measured by the RTS index and energy prices measured by the price of oil.
Although the hydrocarbon sector's contribution to output has been limited
(oil output has increased only gradually over the last decade), the increase
in value has led to a massive increase in per capita income, fuelling
consumption and inflation. Still, like India, Russia has under invested and
its corporations, and infrastructure are in need of an upgrade. Regarding
economic recovery, see Figure 35.9 for the RTS index versus IPOs, new car
sales and consumer loans. We see some modest signs that the new Putin
administration will need to deal with some of these aging structures, but we
expect that these measures will be slow - and state-led. Putin finds the
Chinese model very enticing (as do many EM governments) so it is fitting
we move on to the biggest EM next).
Table 35.3: Russia is still very attractive
#2 MCap growth
10.2% forward 20-yr CACR (#1 is China with 11.5%)
#2 Allocation growth
From 1% to 3% in MSCI in 20 years
#6 GDP
Bigger than Germany and France in 20 years
#11 GDP growth
But at least 30% GDP/Capital higher than in countries
with higher GDP growth
#5 Strategic investors target
According to UNCTAD survey after China, Brazil, India
and US
46% of wannabe entrepreneurs
Among people 20–29 years compared with only 29% for
30–47 years
Fig. 35.8 Russian stocks and energy prices are closely correlated
China
China is very complex, and like Japan in the late 1980s, is feared as well as
admired. Figure 35.10 has Chinese growth rates. I do not really expect them
to blow up the way Japan did as the management seems much sharper and
focused and can more or less make their own rules. A major part of the
Japan demise was their investment behavior. According to the 1992 book
Power Japan that Sandra and I wrote, they only invested about 3% of their
assets abroad - overpaying for most purchases which were of the luxury
variety. But the rest was invested in two things they already owned - their
own land and their own stocks (which were highly invested in the land).
Then when the stock market began to crash in 1990 an the land market in
1991, it was all basically lost. Poor policies kept the bad time continuing
and still now in 2012 more than 20 years later the Nikkei stock average is
about a quarter of where it was in late 1989. But, of course, the composition
of the index has changed.
Fig. 35.9 Economic recovery indicators, 2007–2010
Fig. 35.10 Chinese growth rates
A big issue is the massive foreign exchange reserves which exceed US$3
billion. As shown in Figure 35.11 these reserves are growing at a fast pace
that is now slowing much. We will see how much the economy and prices
slow in 2013 given a possible soft or hard landing as many predict.
Some facts concerning Chinese real estate: Price increase for renting
have exceeded inflation. In part this reflects downside pressure on other
goods. For example, in Shanghai in 2008, rents increased by 25% more than
inflation, but incomes are rising. In Shanghai the average monthly income
in RMB rose from 687.75 (in 2007) to 1174.92 (in 2008) and 1682.17 (in
2009). This income increase did outpace housing prices so that the
price/income ratio fell from 34.35 (in 2007) to 22.70 (in 2008) and 17.14
(in 2009), but the housing price growth rate in 2008 and 2009 was over five
times the GDP growth rate.
Fig. 35.11 Foreign exchange reserves in China in US$ billions.
Figure 35.12 shows the RMB versus the US dollar. While many argue
that a higher Chinese currency will lower the trade surplus, this is a bit
unclear since the J-curve effect that a higher price will lead to higher costs
and less competitiveness unless demand falls as the currency rises. Given
that much of the demand in China is by US companies for sales in the US, it
is not clear that they can quickly move production to other currency
regions.
Fig. 35.12 RMB versus the US dollar

Shanghai followed by Beijing is the most expensive Chinese city. Figure
35.13 shows that in Beijing the average salary has climbed more or less
linearly from 2000 to 2011. However, the housing prices, especially in the
city center but also in the entire city, have risen at a much faster rate,
especially in the 2008-2011 period. There is a bit of cooling off of the
market prices recently with 2011 showing a decline in prices. Chendu, a
city I visited twice, which is a western provincial capital, has much lower
housing prices so that the housing price over income ratios are only about
four times compared to Shanghai's over 17 and the housing price growth
rate is similar to the GDP growth rate.
Fig. 35.13 Housing price vs average salary in Beijing measured by RMB per square meter
Unemployment is growing and is now close to 10%.
Cyprus unemployment: too close to Greece

In Cyprus, a divided country of about 900,000 people with the south Greek
and the north Turkish, there is political tension which aflects the economy.
In the South, the yearly CPI was 3.4% in March 2012, up from 3.1% in
February and 2.8% in March 2011 according to the Cyprus Weekly, April
16-22, 2012. So there is not much inflation which is partly form the VAT
which increased to 17% from 15% on March 1, 2012. Moreover, Cyprus'
financial sector is far far too linked to Greece, with Greek banks present in
the South, and local banks heavily exposed to domestic and Greek debt. A
Greek exit from the eurozone, more likely after the results of the recent
election that ended in stalemate, would bring costs to Cyprus, which would
accentuate its home-grown competitiveness issues - and likely lead it closer
to ties with Russia.
The US unemployment rate has been high since 2007. In April 2012 it
was 8.3%, the lowest in three years. The situation is similar in the UK.
However, these rates are well below those in Spain and Greece. A
comparison of the unemployment rates in Greece, Spain, Cyprus and Italy
for 2003–2010 is shown in Figure 35.14. Females and youth have the
highest unemployment rates.
The unemployed as those actively looking for work; the total workforce
then is the sum of these plus those actively working; the rate of
unemployment is the unemployed divided by the total workforce. These
official numbers, of course, underestimate true unemployment as
discouraged workers do not count and those who take jobs below their
aspirations and skills just count as employed by underemployed.
Fig. 35.14 Unemployment in Greece, Spain, Cyprus and Italy, 2003 to 2010. Source: Kasapis et al.
(2012)
In the south, the labor market in 2012 is the worst since the 1974
hostilities with Turkey with an unemployment rate of about 10%, about
double the usual 5%. To try to isolate the factors affecting unemployment,
the following ordinary least squares regression model was run with eleven
(0,1) variables and 3230 observations of random people from the population
using statistics from the Economics Research Centre at the University of
Cyprus. The regression equation is:
The results are in Table 35.4.
Table 35.4: Regression results. Source: Kasapis et al., (2012)

Fig. 35.15 Unemployment in Cyprus by factor, 2002 to 2011. Source: Kasapis et al. (2012)
The last four variables are regional which have little effect. More
important variables are unemployment by gender (females are more likely
to be unemployed), education and skills reduces unemployment and youth
are more likely to be unemployed.
Figure 35.15(a) has total unemployment which has greatly increased
since 2008. Figure 35.15(c) shows the dramatic age difference effect with
the youth under 25 having four times as likely to be unemployed. The
Paphos area, heavy on tourism, has the lowest rates of unemployment but
all regions have suffered since 2008. Education does help as Figure
35.15(d) shows. Finally, the length of time unemployed is increasing, see
Figure 35.15(e).
Conclusion: Cyprus is in a poor economic state and could have a
financial crisis similar to Greece and Spain.

2I had one student in my essentially PhD level class at KAIST who looked young. When I asked his
age, he replied “16”. He told me he skipped junior high and high school via examinations.
PART VII
Investing and Arbitrage in NFL
Football and Horse Racing
Blunder or Correct Decision? The Belichick
Decision to go for it on 4th Down1
An interesting sports decision situations - namely, an analysis of a crucial play that lost the top NFL
game of the year by New England coach Bill Belichick
As I write on Sunday November 22, 2010, this I am watching the TV

analysts discuss this controversial decision on Sunday, November 15. They
are all ex-National Football League star players with an announcer. Their
analysis is seat of the pants. But is that the right way to analyze complex
decision situations? So it is a good example for us to look at.
The setting was the showdown of the year, pitting the two “best”
quarterbacks, Peyton Manning of the 9-0 Indianapolis Colts at home versus
Tom Brady of the 6-2 New England Patriots and winner of three Super
Bowls. Between them they have won most of the recent MVP honors.
Manning has won the MVP of the season four times. Both teams have very
good receivers and plenty of them for these superb quarterbacks to run
down the field with.
My bets were long New England at match odds - Indianapolis was
favored and New England points on Betfair in London. The
Colts got a fast touchdown then New England got three touchdowns plus a
field goal so moved out to a 24-7 lead. I then hedged my match odds bets,
the larger ones to lock in a sure profit no matter who wins.
See Lane and Ziemba (2008) for an analysis of these types of bets where
you can lock in odds on A then on B to guarantee a profit on a sports
betting and some financial market situations. See Hausch and Ziemba
(2008) and Hausch, Lo and Ziemba (2008) for many studies of sports
betting analyses.
I kept the bets as 17 seemed a huge lead. But I knew from
past Patriots-Colts games that big comebacks are frequent. Indeed, the Colts
rallied. With 2.08 left to play, the Patriots were ahead by 6 points at 34-28
on their 28 yard line with 4th down. Normally it is conservative football
wisdom with a 6- point lead to kick and move the ball down the field to say
the Indianapolis 25. But the Patriots pass defense was not effective against
Manning's fast accurate passes so they could easily move down the field
and win the game. The Patriots traded two great defensive players Richard
Seymour and Mike Vrabel plus some others such as Teddy Bruski and
Junior Sean retired. So they were weaker on defense then usual. They had
just seen Manning run through the defense for two long touchdowns.
Should the Patriots go for it and get the first down they likely could run
out the clock or get a field goal to go ahead by an insurmountable 9 points.
What to do?
The play was similar to a 2-point conversion after a touchdown but with
a longer field. Coach Belichick was familiar with and accepted the
conclusions of a paper by Berkeley Economics Professor David Romer
(2006) which argues that going for it on 4th down is optimal much more
frequently than coaches actually go for it.2 And indeed Belichick has gone
for it much more then the league average. So Belichick ordered Brady to try
to pass for a 3 yard gain to get the first down 2 yards away.
Brady delivered a good pass to Kevin Faulk but Faulk bobbled the ball
and was then pushed back so he did not gain the needed 2 yards when he hit
the ground. Manning, of course, drove the Colts in 4 plays from the New
England 29 with 1.57 to go. With all 3 timeouts available, this is an eternity
for the 3 time MVP to a score using up all the time left except a few (13)
seconds. So Brady had no time to rally back for a game winning field goal.
To use the clock Manning ran on three plays then hit favorite target Reggie
Wayne for the winning touchdown. Belicik had used up 2 of his 3 time outs
prior to the 4th down failed play.
So the Colt's won 35-34. Manning said, “We were preparing to go 60, 70
yards. It was a great play by the defense, shortened our field.” I won all my
bets. But the big issue is the criticism leveled against Belichick for a so-
called stupid decision. One coach, Brian Belick, fired last year as Baltimore
coach, said “it's 50-50” you either make it or not. Other coaches pooh-
poohed the Romer analysis saying they flunked calculus in college.
Of course, the analysis is very simple. It is just expected value arithmetic
in this application.
For example, Brian Burke of Advanced NFL Statistics calculated that
there was a 70% probability that the Patriots would win if they punted. But
it was 79% if they went for it assuming that the chance that the Patriots get
the first down was 60%.
Posnanski (2009) relates Burke's analysis as follows:
[Burke writes]: A team picks up 4th and 2 about 60 percent of the

time – and we all know that a fourth down conversion in this case
means certain victory. On the flip side: A team would score a game-
winning touchdown from the 30 about 53 percent of the time. This
leads to this formula – the first part is the 60 percent multiplied by 1
(1 signifying the certain victory if the play is converted). The second
part is 40 percent multiplied by the chance of winning the game if
the 4th down play fails:
(.60 *1) + (.40*(1-.53)) = 78.8% chance of winning.
... Burke then estimates the chance of winning if Belichick punts –
that is the chance of a team going 66 yards for a touchdown in the
final two minutes. He says, historically, teams get that about 30
percent of the time. So a punt gives the Patriots a 70 percent chance
of winning.
... Now, you probably are saying the numbers do not sound all that
authentic. The Peyton Manning Colts would have a much better than
53 percent chance of scoring from the 30 (and, as it played out, the
Colts scored so easily and left so little time on the clock it seemed
just about automatic). But, you have to figure the Colts also had a
much better than 30 percent chance of scoring had the Patriots
punted - no doubt this was weighing on Belichick's mind. And for
that matter, you have to figure that Brady has a better chance than
60 percent chance of converting on fourth down and two.
Really, no matter how you play with the numbers, it will come out
about the same. Try it. There is almost no way – without
suppressing the numbers – to make the percentages even out. The
Patriots' best PERCENTAGE chance was to go for it on fourth
down. Of course, football is not really a percentage game for most
of us, is it? No, it's a game about emotion and passion and
momentum. When the game ended and Belichick's gamble failed,
people lined up to bash him - and normally I'd be all for this. Former
Patriots player Rodney Harrison called it the worst coaching move
Belichick ever made. Former Patriots player Tedy Bruschi wrote
that Belichick dissed his defense by not believing they could stop
the Colts over 70 yards. Tony Dungy said, "You have to punt there.
You just have to punt there."
More or less, this analysis is ok but it was possible for the Pats to get the
first down and still lose and the 53% versus 30% chances under these
circumstances are likely much closer. And the situation is dynamic. Still it
looks like Belichick made the right decision.
A Vancouver friend of mine who is a professional sports bettor related
the following to me:
There was a big bet in NYC late overnight on the "correctness" of
the call. a jury of five top poker players/sports bettors gave their
probability estimates for four questions:
• probability of NE fourth-down conversion
• Colts win probability after successful conversion
• Colts win probability after failed conversion
• Colts win probabilty after punt
A friend of mine laid -1000, to win 100k, that the jury would return
a verdict, based on their inputs above, of “correct to go for it”. The
estimates were varied, especially for item two, but all five supported
the call. They are doing a second round of west-coast jurors today. I
already voted (62%, 6%, 58%, 33%, respectively).
His 56%=62%-6% is more than 50% so with essentially any reasonable

utility function, the decision to go for it is optimal. Observe that since we
are so close to the end of the game, so this two-stage modeling approach
approximates the dynamic situation. One could add the probability that the
Colts score with the Pats having enough time left to win with a field goal or
touchdown.
My friend adds:
Yes, what all the talking heads don't understand is that all of Indy's
wins come off the second leg of a parlay, if NE goes for it, and that
they're a solid underdog on the first leg. (Parlay = NE fail to
convert, Indy score TD, NE fail to kick FG). Yes and indy gets PAT.
They just look at the gap between prob(score from NE28) and
prob(score from Indy25), without realising it's conditioned on NE
not converting, in which case NE wins outright. Yes and Prob(score
from Indy25) might be close to Prob(score from NE28). It's a trivial
problem, but it's funny to see how much sound and fury is expended
by folks thinking the (trivially) correct choice is a punt. Yes former
Indy coach Tony Dungy said that so did Rodney Harrison former
Patriot player on the commentary after the game. There might be as
much as a 8-9% swing in overall win probability by going for it,
which is huge.
The goal is to win the Super Bowl and making the playoffs with a first
week bye, and having home field advantage throughout the playoffs is
paramount. At 6-3, New England is leading their division so are on track to
have a chance to get the bye and one home field advantage but 9-0
Indianapolis looks posed to have an easy trip to the Super Bowl with a bye
and home field advantage till the Super Bowl. Should both teams so
progress, they would meet for the American Conference final in
Indianapolis and not in the snow of Foxboro. Both teams won the following
Sunday so the Pats are 7-3 and the Colts 10-0. The appendix discusses the
way the playoffs evolve.
The debate continues with most of the commentary and former
footballplayer stars who are experts on TV still blaming Belichick for a bad
decision. Indeed they did lose the game. However, essentially all the sports
bettors and sophisticated fans favored Belichick's decision. Again, getting
the mean right is crucial and the key is the probability of getting the first
down.
The league average for two-point conversions is about 45%. But on the
28th yard line, with Brady passing, it is at least 55-60% with 65% likely the
upper bound.
It is too bad with the billions at stake that professional sports teams do
not use statistical and economic analysis more. Billy Ball as it is called, is a
notable exception where the Oakland A's usually have a top team with a
low budget. But they have not won a World Series. See Lewis (2003) and
the recent Brad Pitt movie, Moneyball, which discusses Billy Bean's
strategy. The strategy is basically to assemble players that produce runs
assuming that the more runs you have, the more games you will win. This
means instead of batting average go for players who were on base very
often. The Yankees, who have won by far the most World Series, have a
simple formula for success: buy the best players. But this approach can lead
to suboptimal behavior because the best players may not be the ones
producing the most runs. In the Oakland A's application they put together
less expensive players who together produce a lot of runs. I found similar
behavior in lotteries where I have consulted for 30+years. Most games are
designed by non-analysts but when there are bugs they call on us.
1 Edited from Wilmott, January 2010.

2Romer looks at extensive data and does a good analysis recognizing that these decisions are really
the first step of a dynamic situation so Bellman dynamic programming is involved. But it is complex
with the number of things that can happen in the future and future decisions. With great players such
as Brady and Manning involved it is even more complex to get the data right. I am reminded of why
Wayne Gretzy was essentially the only four standard deviation sports player better than average. He
knew where the opposing player was going even before the player did. Here you are facing the
greatest fast passing quarterback in history, Manning. So no matter where you leave him timewise
and on the field, his chances of scoring are high. And as we see below, he managed the clock
perfectly to not give Brady a chance to win the game with a field goal or touchdown.
The 2010 and 2011 Super Bowls and the Elo
Ranking System1
Some fun with the 2010 NFL Super Bowl
When I wrote this, the Super Bowl National football league game was on
from Miami. The event has expanded and expanded. A 30 second
commercial costs almost $3 million. The pomp lasts all day with the game
important but only a part of the activity. Sports betting is simply another
financial market. But it's a big one with $50 billion plus bet in Las Vegas
and much more in other places. I focus on the betting exchanges Betfair in
London and Matchbook in the Caribbean. They both allow short as well as
long bets. Each has a different method of quoting odds. Betfair has British
odds so 3-1 means you belt one to collect; three if you win and lose one if
you lose. That's US 2-1 where you bet 1, win 2, and collect 3. Matchbook
uses the bookmaker system of - and + where - is favored and + is the
underdog. See Table 37.1 for examples. You compute the odds as follows:
If the money line is positive (the underdog) you divide by 100 and add 1.
So +400 is 5-1 British odds, 4-1 US odds. If the money line is negative, say
-400, then you remove the minus sign and divide the 400 into 100 and add
1.00. Thus -400 is equivalent to 1.25 British odds or 0.25 US odds, namely
a 1-4 favorite.
But there is much more as the Figure 37.1 shows and you can hedge bets
as the odds change based on the current score and field position. My
experience is that the sports betting market is more efficient than the
financial markets. Of course, die hard efficient market types will like to
forget that efficient markets has taken. a severe beating in the 1987 crash
and the 2007–009 crash. They simply cannot explain these events.
Table 37.1: Matchbook odds for the Super Bowl, February 2010
Why are the sports betting markets more efficient? The reasons seem to
be more sophisticated bettors and a simpler more well defined game.
But there are many upsets and hedging when you are ahead is advisable
because many leads evaporate quickly.
One thing that's innovative, challenging and fun and helpful in hedging is
the wide variety of bets available and at odds determined by long/short
players. For example, for the Super Bowl you can bet straight match odds
plus the favorite Indianapolis at +1 points and other amounts and the
underdog New Orleans at +3, etc. Each bet has
reasonable tight bid-ask spreads so presumably represents the same
stochastic process for the scores. But watch out, as things change fast. My
experience is that these more non-standard are less liquid with wider bid-
ask spreads than the straight match odds bets.
Hence they need on occasion to adjust the odds to encourage bettors to
rebalance their bets to eliminate the imbalances akin to how Dublin
bookmakers do it. The places where there is too much long or short, the
odds are made more favorable to try to rebalance these wagers. Cantor
Fitzgerald's plans, once approved, are to move more into stocks and other
financial instruments. They have a ways to go to enter the betting exchange
world but seem to be moving in that direction. Since the betting exchanges
are not allowed to US bettors there will be a push to deregulate and expand
this more sophisticated way of betting. After all, derivatives are very similar
and they provide the U.S. and the rest of the world.
The role of Las Vegas

Las Vegas is slowly catching up to the betting exchanges.
The bond trading company Cantor Fitzgerald, which was badly hurt by the
911 attacks has a new trading room for sports betting patterned after
financial market trading operations. They have bets like the next pass or
success or failure of the upcoming field goal. They match long- short bets
and allow bets during the game. An attempt to mtch the bets to eliminate
their risk is made but imbalances may lead to risk. So they are more like
Matchbook which takes risk and less like Betfair which matches all bets.
Super Bowl Playoff Ratings
Which team will win a given football game?
Most sports have a significant home field advantage. It has to do with the
shape of the playing surface, the players, not having to travel and crowd
noise. In football, the last factor is extremely imortant as the loudness is
extreme. My racing colleague John Swetye computes Elo ratings. These are
based on actual scores of team i playing team j adjusting for a home field
bias. The idea is a chess rating and 1000 is the base. Table 37.2 has these
ratings for all the NFL teams as of December 22 and at 14 weeks, 17 weeks
(end of regular season) and at the final past season. Both Indianapolis (14-0
in the AFC) and New Orleans (14-0 in the NFC) were undefeated after
week 14. There were lots of discussions regarding two undefeated teams in
the Super Bowl. Plus the fame to join the 1972 Miami Dolphins as 17-0
undefeated Super Bowl winner - the only undefeated team. The argument
is: do you rest players to go for a higher chance of winning the Super Bowl,
assuming that resting helps more than not staying sharp, or go for greatness
of an undefeated season?
In the 2007-08 season, the New England Patriots started 16-0 not resting
players. Tom Brady set the record with 50 touchdown passes and receiver
Randy Moss caught 23 touchdown passes, also a record. With other great
players on offense and defense, they looked invincible (much more
impressive than Indianapolis and New Orleans this year). At the end of the
season, the New York Giants, a good but not as impressive a team, instead
of letting the Patriots swamp them they put up a strong fight but did lose in
the end. So the Patriots were 16-0 and looked simply unbeatable.
Nonetheless, they lost the Super Bowl to the Giants. Granted, it took a
miraculous play but with rested players, the outcome might have been
different as they were not sharp all game.
So Indianapolis and New Orleans rested players after they had clinched
home field advantage throughout the playoffs and ended up 14-2 and 13-3,
respectively. New Orleans, a dream team, was trying to undo the impact of
Hurricane Katrina which destroyed much of New Orleans. So we present
here the various estimates of the outcome of the game.
Table 37.2: Evolution of the Elo ratings, December 22, 2009 to Post Season
Indianapolis and New Orleans were the highest rated teams if you do not
count the last three weeks. San Diego had the highest rating over the 17-
week full regular season. They had 11 straight wins and were a major force
but were eliminated when their pro-bowl nominated kicker, who was a
perfect 16/16 in field goals under 40 yards, actually missed three field
goals. These errors and bad scenarios were all over the playoffs. In the New
Orleans-Minnesota game, Minnesota greatly outplayed New Orleans but
five turnovers gave the game to New Orleans. Drew Brees, the great New
Orleans quarterback, was not as sharp as usual. Meanwhile, Brett Farve, the
40-year-old Minnesota quarterback, outplayed him but in the end made an
error. He was in position to go out of bounds and leave their kicker with an
easy game winning field goal, but the punishment of the New Orleans
defense wore him down physically and mentally so he regressed to a
cowboy pass, which was intercepted. The New Orleans defensive players
continually drove him to the ground so his was visibly tired near the end of
the game. He is a very durable player but it seems that he should use the
Peyton Manning approach to simply drop to the ground immediately after a
pass is thrown so a running play executed. Then if he is hit which is less
likely its a 15-yard roughing the passer penalty. That led to overtime and
New Orleans won the coin toss then won the game on a field goal.
Notes on the Elo Power Rations for NFL teams

The ratings are similar to Sagarin ratings. They are based on the
physicist Arpad Elo's Chess ratings, but modified for football.
They're “who-beat-who” ratings.
See the websites: en.wikipedia.org/wiki/Elo_rating_system
en.wikipedia.org/wiki/World_Football_Elo_Ratings (this is the
method my colleague John Swetye uses)
en.wikipedia.org/wiki/Jeff_Sagarin
Each team starts the season with 1000 points.

For each game the point differential is compiled using the scale:
1 point differential = 1
2 point differential =1.5
3 or more point differential = (11 + GoalDiff) / 8
Next is a factor called “difference rating away” and “difference

rating home” = dra or drh. The home team gets 66.0979 points
which is based on the fact that the home field advantage is worth
2.867 points and there were an average of 43.375 points per game
based on 2008–09 data.
So 66.0979 = (2.768 / 43.375) × 1000 = home field adjustment.
dra = (Away Team's Elo Rating) - (Home Team's Elo Rating)
drh = (Home Team's Elo Rating) - (Away Team's Elo Rating) +
66.0979
Next, the home and away expected results are calculated:
Finally, the ELO ratings for the home and away teams are
calculated:
AwayW = 1 if the away team won and 0 if the away team lost
HomeW = 1 if the home team won and 0 if the home team lost
WeightIndex = WgtIndex = 5. In international soccer certain games

are given different weights depending on their importance. John
gives all NFL games the same weight.
EloAway = previous Elo rating for the away team
EloHome = previous Elo rating for the home team
EloRatingsA = EloAway + WgtIndex * IndexGoalDiff * (AwayW -

AwayE)
EloRatingsH = EloHome + WgtIndex * IndexGoalDiff * (HomeW -
HomeE)
So the big question is do these ratings add value?

Well Indiannapolis was a 1-2 favorite to win the Super Bowl and you
could get New Orleans +10.5 at the same odds.
The NFL football betting market

There are 32 teams in eight groups of four teams each. The American
conference, the old AFC, has East, South, North and West and so does the
National Conference. The preseason has five games, which are tests for
teams to get ready. Since the teams do not want to risk injury, these games
are hard to predict. The league gives the high draft picks each year to the
weakest teams to try to even out their abilities.
Fig. 37.1 The playoffs. Source: Wikipedia
During the 17-week season, each team plays 16 games. The winners of
each of the eight sections automatically make the playoffs. Four additional
teams (two from each conference) are wild cards making up the total of 12
teams. Teams with the best records get home field advantage in one or all
games in the playoffs. See Figure 37.1
The first round of the playoffs is dubbed the Wild Card Playoffs (the
league in recent years has also used the term Wild Card Weekend).
In this round, the third- seeded division winner hosts the sixth seed
wild card, and the fourth seed hosts the fifth. The 1 and 2 seeds from
each conference receive a bye in the first round, which entitles these
teams to automatic advancement to the second round, the Divisional
Playoffs, where they face the Wild Card Weekend survivors. Unlike
the NBA, the NFL does not use predetermined brackets. In the
second round of the playoffs, the top seed hosts the lowest surviving
seed, while the other two teams pair off. The two surviving teams
from each conference's Divisional Playoff games meet in the
respective AFC and NFC Conference Championship games, with
the winners of those contests going on to face one another in the
Super Bowl. Only twice since 1990 has neither a number one-
seeded team nor a number two-seeded team hosted a conference
championship game (the 2006 AFC Championship and the 2008
NFC Championship).
The winner of each conference plays in the Super Bowl. Given this
playoff structure and probabilities of winning at home and away for each i,j
combination, one can, as we have done in hockey, compute fair odds of
advancing in the playoffs. WTZ set the line in hockey for the BC Lottery
Commission and made such calculations. The NFL is similar.
The NFL playoffs leading to the Super Bowl have the structure shown in
Figure 37.1 and in Figure 37.2 I have filled in the results. I bet on all of
these games and follow the teams, players, injuries, etc. The betting during
the regular season is easier with the favorite-longshot bias being quite
helpful. Most of the game outcomes make sense but there are occasional
upsets. And the difference between the better teams and the weaker teams is
frequently hared to estimate. The playoffs are definitely more challenging
and especially so this year.
Fig. 37.2 The playoff tree, 2010
So let's go through the 2009-2010 playoffs just picking the team with the
higher Elo rating and see how we would have done noting the favorite in
each case. I used the end of season ratings unless the teams did not try the
last few weeks, in which case I used the 14-week ratings.
So it's 2-2 so far
Two easy wins, one upset, one hard to say. I bet Minnesota but they did
not look good in the last three games - not so much resting so call the score
2-1 with one pass here. But it could be 3-1 as there is the home field
advantage of 2.867 points and Minnesota then would be favored by 6.567
points.
Super Bowl 2010

Indianapolis vs New Orleans 1056.4 1065.2 underdog with
higher Elo Saints won (at Miami)
So 3 of 3 for the ratings in the Conference Championship and Super
Bowl. So the Elo ratings did quite well!
An educated guess that paid off

Betfair offered a Canadian$1000 prize to the person or persons who
correctly pre-dicted the total passing yards in the Super Bowl for those who
bet C$25 plus. So I entered the contest using some good advice from Randy
Robles of Elias Sports by way of my racing colleague John Swetye. The
advise was “Brees will throw for 358 and Manning for only 263 (because
he will not have to throw for much more)”. So I submitted 621. The result
was
So I won. Randy and John can take the credit but it was nice to win and
collect a prize.
On to the Super Bowl 2011

The Elo ratings, which are based on actual scores, wins and losses and,
adjusting for home field advantage, did quite well in 2009/10. The rankings
start each year at 1000 and move up and down as the season progresses.
There are 16 games for each team played over 17 weeks, some 256 total
games for the 32 teams. Then there are the playoffs of the 12 teams that
make the playoffs. Figure 37.3 shows the results for the 12 playoff teams.
The formula for the seeding of the teams is in Figure 37.1. The results of
the playoffs are in Table 37.3 with an analysis of the predictive ability of
the ELO ratings.
Fig. 37.3 ELO ratings for the twelve teams in the playoffs during 2011 and the results
The February 6, 2011 Super Bowl is now history. The ELO ratings with
Green Bay at 1045.9 and Pittsburgh at 1038.4 made Green Bay the favorite.
The odds in Betfair and at various odds makers in Las Vegas and elsewhere
all agree with Green Bay a 2.5 point favorite and the shorter priced option
in the betting. See Figure 37.4 for the top teams and Table 37.4 for the final
ELO rankings for all 32 teams.
I positioned myself well for the Super Bowl ranking the teams New
England first, Pittsburgh second and Green Bay third. So I have a nice gain
no matter who wins this tough matchup. These bets were made a long time
ago with odds at 8-1 to 18-1. New England, my top pick and the favorite to
win the Super Bowl is still the top rated team. But they are out, having lost
to the NY Jets. A factor in that loss was the benching for just one set of
downs of top receiver Wes Welker by the New England coach Bill
Belichick for statements responding to trash talk form the Jets coach and
players. Belichick had his players take the high road and not respond. But in
that first set of downs there was a Tom Brady interception - a rare event as
he had close to 350 passes with no interceptions. This got New England off
to a bad start from which they never recovered. So again intangibles affect
results!
Fig. 37.4 ELO ratings for the 2011 playoffs
Table 37.3: Analysis of the 2010-2011 NFL Playoff Elo Predictions
Table 37.4: Final ELO ratings for all teams

Update to February 7, 2011
ELO got the Super Bowl right when Green Bay beat Pittsburgh yesterday
by 31-25. Three Pittsburgh turnovers and a three touchdown with no
interceptions by Green Bay MVP quarterback Aaron Rodgers was the
difference.
So ELO was 7 right and 3 wrong for the 2011 playoff season.
1 Edited from Wilmott, March 2010 and March 2011

Risk Arbitrage in the NFL 2012 Playoffs and the
Super Bowl1
The February 5, 2012 Super Bowl in Indianapolis
The game was between two teams: New England and the New York Giants
with elite quarterbacks. This is Tom Brady's fifth time in the Super Bowl
for New England with wins in 2001, 2003, and 2004 and a loss against
these same Giants in 2008. His fifth appearance ties John Elway for most
Super bowl appearances. John won two Super Bowls for the Denver
Broncos. Should New England win, Brady will tie his boyhood hero, Joe
Montana as well as Terry Bradshaw who won four for San Francisco in the
1980s and Pittsburgh in the 1970s, respectively. Eli Manning won in 2008
on a miraculous play. While his team only had a 12-7 record, Manning had
been brilliant and set the record fourth quarter touchdowns during the
regular 16 game season. He also had many long passes. If the Giants upset
them, Eli will have two Super bowl wins versus only one for his more
famous older quarterback brother Peyton. Peyton, along with Drew Brees
and Aaron Rodgers are generally considered the elite quarterbacks. Eli
declared early this year that he was one too and he was criticized for saying
this.. While it is up to others to decide about this, he has been playing at that
level and that has ignited his team from a 7-7 record to the Super Bowl
beating the favored Green Bay and San Francisco. New England is a 3-
point odds favorite going into this game or 59% versus 41% probability of
winning. The Elos are 1072 versus 1027 implying a 5-point New England
advantage and 63% versus 37% probability of winning. The Betfair odds
favor New England at 1.74-1.76 versus 2.32-2.34 for the Giants. The Giants
are 2.08-2.14 with New England at 1.89-1.91. And with the Giants
are favored, reflecting the 3 point spread, at 1.72-1.73 versus New
England's 2.36-2.38. But New England struggled against Baltimore and was
lucky to win. Tom Brady had a poor game against the tough Baltimore
defense, as he later acknowledged, and actually had four interceptions. Two
were real and two more were called back because of defensive penalties
against no touchdowns. But he won the game with a leaping quarterback
sneak from the one yard line. In the previous week, Brady was
exceptionally sharp throwing a record tying six touchdown passes against
Denver. An injury to top tight end Rob Gronkowski which could be a factor
in the Super Bowl did not help. Meanwhile, Eli has been sharp in all the
games and the Giants performances and the team's Elo have been rising
sharply as the figure 38.1 going into the Super Bowl shows. And the last
New England loss was 24-20 to these very Giants; New England then has
won 10 straight games. But New England's Elo is way above the Giants so
they must be the pick. The game has a lot of events surrounding it for a
whole week including much TV and other analysis. Madonna is doing the
half-time show and the TV ads are $3.5 million per 30 seconds! Clint
Eastwood narrated a 2-minute Chrysler ad for $14 million!
Fig. 38.1 Elo ratings for New England versus NY Giants going into the Super Bowl
Recap of the Super Bowl

It was a very good game and another exercise in risk arbitrage and mean
reversion. I favored New England but just barely so I hedged my bet with
some long Giants plus . This cost 507 to gain 360 and meant if New
England won by 4 or more, a 1107 gain would be reduced to 600. But 125
would be gained if the Giants won versus a loss of 235 before this hedge.
Also, if New England won by 1-3 point then the 360 plus the 1107 or 1467
would be won. The game started with New England 1.65-1.66 the favorite
with the Giants at 2.5-2.52.
The Giants got a safety for a 2-0 lead when Brady threw from the pocket
to nobody down the field from the end zone. The Giants proceeded to then
score a touchdown for a 9-0 lead as Manning was 10/10 passing a Super
Bowl opening record. New England could have prevented this TD with a
fumble recovery but for a 12 men on the field miscue. Then I went long 75
New England at odds of 2.46 so the Giants were then favored to win the
game. Brady then became Brady and had 14 straight completions breaking
Joe Montana's 13/13 record. This yielded a filed goal and a touchdown for a
10-9 halftime lead. I could then rehedge the A long New England bet with
an arbitrage bet at 2.52 on the Giants so this risk arbitrage became an
arbitrage with a +0.5 New England win gain versus a +89 NY gain on this
part of the bet. New England had won the coin toss and deferred to get the
ball at the start of the second half. So they were trending. Another
touchdown made it 17-9 for an 8-point lead. The New England odds
became 1.39-1.40 so it was time to go long Giants at 3.55-3.6. But the odds
moved very fast and I was only able to get 16 of a 75 bet on but I left a limit
order at 4.3 for the 59 balance which later got filled. My experience in
options markets trading is that keeping stink bids in is often a wise idea in
markets that can have fast reversals. Brady was now up to 16 straight
completions and even though NE did not score the 4.3 got filled. The Giants
had two drives that led field goals and a closer 17-15 New England lead.
Had this been the turn-of-the-year effect where you can stop the race when
you are ahead I would have cashed out but this was not the futures markets.
But the Giants odds were getting more towards favored but not low enough
to cover again yet.
At the 4-minute mark, a pass to the usually sure handed Wes Walker, the
NFLs leading receiver, was dropped but it was a bit high. An announcer
said he catches 100 out of 100 of these so this must be the 101. This led to a
3 and out rather than a first down on the 20. A score then or even running
down the clock wold have put the game out of reach gave the Giants
another chance.
A ma jor concern for the game was the high ankle sprain to top wide
receiver tight end Rob Gronkowski. At 6 feet 6 inches and 280 pounds with
a huge wing span and terrific hands, he was a very good receiver and had
set the record at 17 touchdowns for a tight end. But was he ok? He was not
used much so the other great tight end, Aaron Hernandez got most of the
throws. But there was one 20 yard pass to Rob. So Brady took a chance
with a long pass to Gronkowski but his agility was compromised and a
usual huge gain turned into an interception. The hedge was completed. It
was now the fourth quarter and the time that Eli Manning had the most 15
touchdowns. And time was running out. Manning threw another remarkable
pass for 38 yards to Manningham at the 3.39 mark and then hit him again
twice with 2.32 remaining. A Bradshaw run got the ball to the 12, the
Giants were now favored at 1.37-1.38 with NE at 3.65-3.75 so 50 was bet
on NE. So the risk arbitrage bets yielded the arbitrage +60 NE +275 NY.
Coach Bill Belichik knew the key was to let the Giants score as fast as
possible to give Brady a chance for a game winning touchdown - as a field
goal would not be enough. Had NE stopped NY they would have kicked a
game winning field goal. Jacob, the big running back, tried to stop on the 1-
yard line after the NE defenders let him through, as ordered by Belichik, but
fell backwards into the end zone. So the score was 21-17 as the two-point
conversion attempt failed. Brady had 57 seconds from his 20 yard line. At
52 seconds, Branch moved the ball. Then Hernandez dropped a pass. Then
Brady was sacked back to the 13 of New England. With 4th down and 16,
Brady hit Branch for a first and 10 out of bounds to stop the clock at 32
seconds. A pass to Hernandez with 17 seconds got the ball to the NY 44. A
Giants 12 men as the field got the ball to the 50.
A final hail Mary pass into the end zone with 5 seconds let fell to the
ground. Two New England players including Gronkowski tried to get it
while being swarmed with Giants. So the Giants won again 21-17. My gain
was the original 125 + 275 from the arbitrage or 400. This was lower than if
New England won but again, the mean reversion risk arbitrage turned a
potential loss by being on the losing team into a gain.
Some observations:
(1) The elite quarterbacks certainly include Eli Manning. He was 10 of 14
for 118 yards in the 4th quarter with his seventh game winning drive
of the season. Eli now has two Super Bowl wins, one more than his
illustrious brother Peyton who has one. Of course Peyton has four
MVPs during the regular season and Eli has none of these. Eli's two
Super Bowl MVPs are the same as Brady's two. He also is one of only
five players in the NFL history with multiple Super Bowl MVP
awards. Besides Brady and Eli, Terry Bradshaw and Bart Starr have
two with Joe Montana leading with three.
(2) It is surprising how many trivial 12 men on fixed penalties there are
and how much they hurt their teams.
(3) The Giants looked very solid not like a 7-7 team. They have really
improved and peaked at the right time. I am glad I saw it coming and
bet at 28-1.
(4) Brady played well - with a healthy Gronkowski they likely would
have won.
(5) Elo did not get the game right like it did the last two years. The
trending of Elo as well as its level and the serious tail issue for games
with a big favorite are things to study in the future. Elos performance
this year was so-so with 6 wins and 4 losses. Combining with mean
reversion and risk arbitrage makes it easier to win.
(6) The Las Vegas future book on next year's Super Bowl favors New
England at 5-1 with the Giants at 8-1 so they seem over bet!
Recap of the Playoffs

Figure 38.2 shows the evolution of the Elo ratings through the 16 game
season for the twelve playoff teams. Table 38.1 shows the results of the Elo
forecasts.
They started on Wildcard weekend, Saturday and Sunday January 7 and 8
for the wildcard 1 and 2 and division winners ranked 3rd and 4th in each
conference. That's:
in the AFC
Pittsburgh (12-4) at Denver (8-8)
and
Fig. 38.2 Evolution of Elo ratings for the twelve playoff teams in 2011/2012
Table 38.1: How did the Elo rating system do in the playoffs in 2012?
Cincinnati (9-7) at Houston (10-6)
in the NFC
Detroit (10-6) at New Orleans (13-3)

and
Atlanta (10-6) at the New York Giants (9-7)
By making bets early on the six best teams and later on Houston at 46-1
and the New York Giants at 28-1 (just as they seemed to be peaking). I
created a no lose arbitrage to win no matter who would win the Super Bowl.
The largest bet was on New England with substantial amounts on New
Orleans and Green Bay. The latter two were eliminated but New England
remains. To balance my New England Super Bowl bet I have hedged the
Giants long at with the main bet New England but a no lose situation and
a possible even bigger gain should New England win by 1-3 points.
Figure 38.3 shows the progression for the January 14 and 15 divisional
playoffs and January 22 Conference championship games and the February
5 Superbowl, including the results going into the Super Bowl of the
playoffs, followed by an analysis of some of the games and the record of
the Elo ratings.
Fig. 38.3 Results of the NFL Playoffs, 2012
The rules for assigning future games are in Figure 37.1; basically the
higher rated teams get to play the lower rated teams still alive in the
playoffs.
The analysis of some interesting games follows after a review of
arbitrage and risk arbitrage and the favorite long-shot bias which is greatly
used in our bets.
New Orleans at San Francisco: a game for the ages full of risk
arbitrage opportunities
New Orleans (14-3) had an explosive offense plus a defense that causes
many turnovers. At home, in a domed stadium, they have been dominant
but on the road in outdoor stadiums in the north, they have done well but
were not as dominant, scoring two touchdowns less per game.
San Francisco (13-3) had a top defense and a great runner, Frank Gore,
and an improving but not quite top notch quarterback, Alex Smith.
The Elo slightly favored New Orleans with 1075 versus 1068 or 52%
versus 48% chance of winning or a 1-point spread. The odds spread was +3
for New Orleans meaning a 59% to 41% edge for New Orleans. The Betfair
odds favored New Orleans and I bet them 10 at 1.56 and 190 at 1.57.
As we see below, this game was a good example of risk arbitrage - that
is, bet an A then later bet a B, in an A versus B game, in such a way that
you cannot lose when OAOB > 1; see the section below on risk arbitrage. It
is called risk arbitrage because this match might not occur. The usual
situation is that you bet initially on the favorite which will be at short odds
on the underdog rise so that a bet can be made to create a real arbitrage
where you cannot lose. But the game did not evolve that way. However as
shown below it is another typical risk arbitrage situation. That is the case
where the favorite gets behind and you bet more on them but now at better
odds assuming that the score will eventually mean revert so you can cover
the longer odds bets and create the arbitrage. The game was of this type not
once but multiple times. New Orleans had some turnovers and San
Francisco got to a 14-0 lead, so 100 more was bet at 3.3. Then after a San
Francisco field goal and a 17-0 lead 50 more was bet at 4.8. New Orleans
rallied to 17-14 at half time. This led to covering bets that is short New
Orleans/long San Francisco of 50 at 1.86, 100 at 1.88, 50 at 1.89, 100 at
1.90 and 50 at 1.81. The arbitrage was complete with a gain if either team
won +125.40 for New Orleans and +50 for San Francisco.
Early in the second half New Orleans had another turnover their fifth a
fumble. San Francisco has a top punter and field goal kicker so the score
went to 20-14 and later to 23-17. The last playoff team to win with 5
turnovers was the 1982 Jets. With 4 turnovers, the record is 9 wins 121
losses in the playoffs and 10% wins in the regular season. And with a -4
turnover deficit as here =1-5, the record is 1-71 with the one exception, the
1977 Raiders. Twenty-five more was hedged short SF at 1.61 to yield +135
New Orleans versus +35 for San Francisco for the arbitrage.
Darren Sproles then scored a 44 yard New Orleans touchdown from
Brees to give them a 24-23 lead. Now New Orleans is favored at 1.56 so 25
was shorted on New Orleans to give +136.4 New Orleans and +60.25 San
Francisco for the arbitrage.
The last 3.53 minutes of this game were terrifically exciting with four
lead changes and mean-reversion and risk arbitrage trades every minute or
so. Here's more of the recap: With 2.18 remaining Smith hit a 31-yard pass
to get San Francisco into field goal range. One option was to then run the
clock down, kick the field goal to gain the lead with so little time left for
New Orleans that they cannot score so San Francisco would win. But Smith
ran 28 yards to get a touchdown for a 29-24 San Francisco lead. The odds
were then San Francisco 1.52-1.56 and New Orleans 2.84-2.88.
A two point conversion attempt, that would have yielded a 7-point lead,
failed and the Betfair odds widened to 1.33-1.42 San Francisco 3.4-4 New
Orleans.
Then with 1.37 left Brees threw a 66 yard touchdown pass to Graham to
make the score 30-29 and with a two-point conversion 32-30, Brees to
Sproles.
So it looked like New Orleans had pulled the game out - but not quite.
The odds then became 1.23-1.25 New Orleans and 5-5.2 San Francisco.
A bet of 25 short on New Orleans yielded +128.90 New Orleans +85 San
Francisco for the arbitrage.
With 14 seconds left, on the 13-yard line, San Francisco was threatening
and the odds were 15-21 New Orleans 1.05-1.05 San Francisco. And, with
9 seconds, the odds were 26-36 New Orleans 1.03-1.04. Then Smith hit
another touchdown pass for 14 yards to Vernon Smith and the odds became
100-330 New Orleans no bid-1.01 San Francisco. I bet 3 at 100-1 on New
Orleans to yield +425.90 New Orleans +82.25 San Francisco.
San Francisco won 36-32 and I won +82.25 even though my original bet
was on New Orleans. It is a good example to the strategy to work the
position until you win using the mean reversion and arbitrage ideas. I
frequently have to work the position in the options markets when trouble
arises. And it is important to think through these corrective actions in
advance.
Figure 38.4 shows the evolution of the Elos as the playoffs progressed.
Fig. 38.4 Evolution of Elo ratings as the playoffs progressed
New York Giants at Green Bay

The Giants limped into the playoffs with a 9-7 record, but two impressive
wins indicated that they were a threat. In the wild card, they crushed Atlanta
24-2. Then it was on to Lambeau field and the favored Green Bay Packers.
Green Bay won last year's Super Bowl and had the best record in this
season, 15-1. They were at home and favored with Betfair odds of 1.35-
1.36. As the game progressed, it was clear that GreenBay was not as sharp
as usual with many dropped passes, fumbles and other misses. Meanwhile,
the Giant's quarterback, Eli Manning, was very sharp and in the end won
37-20. The Giants were on a roll and a force to be reckoned with. The
Giants went on to beat San Francisco 20-17 and were in the Super Bowl.
Pittsburgh at Denver: which Tebow will show up?

The highly favored veteran Pittsburgh Steelers came into the mile high
stadium in Denver quite banged up. Their top runner Mendenhall was
injured and their two time winning Super Bowl quarterback Big Ben
Roethlisberger had been hobbling around for several weeks.
Denver (8-8) won their division so had home field advantage and that is
one of the most home advantage stadiums with thin air at 5200 feet. Denver
started the season poorly then Tim Tebow, a strong runner for a
quarterback, started winning games later in the fourth quarter and in
overtime. His kneeling and waving his hands to God created a new
dictionary word: Tebowing. The Tebow-mania reached a high pitch and they
won seven straight games before being blown out in Denver by the Patriots.
The game was tight all the way and went into overtime. It ended
spectacularly with an 80 yard Tim Tebow pass as the first play in overtime.
It was a dramatic and sudden end to the year for Pittsburgh and a great
victory for Denver 29-23. Tebow was back.
Denver at New England: a learning experience for Tebow to

watch the great Brady
After the spectular win in Denver against a wounded Pittsburgh, the scene
moved to Foxborough and the mighty Patriots. Could Tebow pull off yet
another miracle? The odds greatly favored New England and bets of 95 at
1.15 , and 1.05 at 1.16 were made - so the odds were very short.
New England scored first with a pass from Brady to Wes Welker, the top
receiver in the NFL. The odds then became 1.09-1.11. Tebow had a good
run but then there was a fumble and the odds fell to 1.07-1.08. So half the
position was hedged at 1.09 to make the bet -100 Denver +23 New
England. Then Brady threw another touchdown to tight end Rob
Gronkowski who set the NFL touchdown record at that position. So the
score was 14-0 so the last 100 was hedged at 1.06 to give an arbitrage of +0
Denver +18.05 New England. This moved the position from win 31.05 but
with loss risk to this no risk arbitrage. Then Brady was intercepted to yield
a 14-7 score and odds of 1.09.
A Brady to Gronkowski touchdown made it 21-7 with odds at 1.04-1.05.
A fourth Brady touchdown, for 61 yards to Deion Branch made it 28-7 with
odds of 1.02-1.03. Yet another Brady touchdown, the fifth to Bronkowski
made it 35-7 with odds of 100-190 for Denver and no bid-1.01 for New
England.
Yet another, the record tying sixth touchdown (Steve Young and Darrell
La- monica did it before) made the score 42-7. Rather than let Brady set the
record, coach Bill Belichik went for a final goal and the game ended 45-10.
It was a powerful performance and Tim Tebow acknowledged that he
learned from this experience. Now Denver was out and New England
hosted Baltimore at home.
1Edited from Wilmott, March 2012.
The One That Got Away: The Hitable $2 Million
Pick 6 at the Breeders' Cup1
How a $2 million pick six at the Breeders' Cup should have been won
The Breeders' Cup included 14 major races over two days and was held
again at Santa Anita on Saturday, November 6 and 7, 2010. I went in 2009
and it is fun to see it live. This year on wide screen high definition TV it
was wonderful to watch. Being at home, the handicapping and betting is a
lot easier. There are many opinions. That's what makes a horse race. The
spreads on Betfair are fairly tight and it is easy to bet from Canada and you
frequently get better odds there than at the track. We don't actually bet at the
track but through rebate shops that give back part of the track take. That's
easy to do on the phone or by email. The rebaters take their cut and the
track gets more easy, low expense business to up their revenues.
The big race was as usual the $5 million classic. It is no longer the
world's richest race. The $6 million Dubai World Cup has that honor. For
March 2012 this race had a $10 million purse with the races that day worth
$25 million rivaling the purses at the Breeders' Cup. However the Classic is
the most important race in the world and frequently determines the horse of
the year. The two candidates for horse of the year are both female. The 3-
year-old Rachel Alexandra won all her 2009 races. She beat the top females
in the top female races by 20 lengths. I was at Churchill Downs to see this
in the Kentucky Oaks. In four races against males, she beat them handily.
So she would normally be an almost sure bet for horse of the year. But
Zenyatta, a five year old mare had won all 13 of her races but always
against females. She has a dynamite kick and just cruises by the other
horses near the buzzer to win easily. Some of her races were in slow times
(76 area on the Equiform scale I follow) and some in fast times (81 area).
To put this scale in perspective, the highest I ever saw was four 84's by
Ghostzapper. One of my most treasured but small bets was on Ghostzap-
per's 2004 Breeders' Cup win. There was a top filly in that race, Azeri - I
had watched her at Saratoga getting beat by females in a race so the fact
that she had numerous wins at short distances I was pretty sure she would
not be in the top 4 in this male dominated race. So a $20 superfecta bet
boxing for $5 Ghostzapper (lor 2) with Roses in May (1 or 2) with the two
next leading horses (3 or 4) and (3 or 4)came in to provide a $5000 payoff.
The big mistake was not betting more! I visited Ghostsapper at Frank
Stonach's farm, Adena Springs near Versailles, Ken-tucky and he looks
almost identical to his sire, Awesome Again, who also won the Breeders'
Cup Classic.
The 13/13 of Zenyatta is historic since only Personal Ensign (13/13) in

1988 and Colin in 1907 had undefeated records in the US among horses
with at least ten races at major racetracks. Tesio, the great Italian
Ghostzapper the fastest horse since trainer in the 1930s had the other three
Secretariat, winner of the Breeders' Cup of the five undefeated horses since
1900. Classic 2004 and provider of the Tesio was a great anomaly person
looking at many many generations to bread cheap to cheap to get great
champions. He did this without computers with a lot of help from his wife.
Currently the way to check such matings is to contact Steve Roman who
can run more then ten generations back. See Steve's website,
www.chefderace.com, for much valuable information. Free videos of many
great races like the 1988 Breeders' Cup Distaff match up of Personal Ensign
and Winning Colors (the 1988 Kentucky Derby winner), the amazing 1973
Belmont race of Secretariat, and Zenyatta's 2009 and 2010 Classic
Breeders' Cup races.
The Europeans and many other handicappers were pushing for Rip Van
Winkle, trained by Irish legend Aidan O“ Brien was the pick since he had
won his two races on 2009 when he did not face the superstar Sea the Stars.
The three times Sea the Stars beat Rip Van Winkle he was close behind and
well ahead of the competition. Unfortunately, the Sea the Stars connections
preferred to hedge and cash in on this horse reputed to be the best in Europe
in ten years by retiring him to stud duties. He had won the Epson Derby, the
2000 Guineas and the ARC. Perhaps they recalled the last “best horse of the
decade”, Dancing Brave, who could not lose and finished fourth in the
classic. The polytrack at Santa Anita not dirt or the European's grass could
have been a factor too.
Getting back to Zenyatta and Rip Van Winkle. The Betfair and track odds
showed the local biases. You could get better odds on Zenyatta in Europe
on Betfair and Rip Van Winkle in the US at the track. It was possible to do
an arbitrage here. I just concentrated on better odds on Zenyatta on Betfair.
My assumption was despite the fact that her running times were not super
outstanding she had the will to win. And indeed she did with Rip Van
Winkle finishing out of the money. She can now be retired as the only
undefeated mare who beat males in the toughest race in the world.
But let's discuss the Pick 6. In the Pick 6, to win you must have all six
winners who share 75% of the net pool with the 5/6 sharing the remaining
25%. The Pick 6 was races 4-9. There were three standouts but they were
definitely not certain winners. The other three races seemed wide open. So
you could play the Pick 6 in the following way: I thought about doing this
but did not - it was a $2 million mistake. You have a single ticket with about
10 horses in the three wide open races and single the three standouts. That
would cost about 10*10*10*1*1*1*$2=$2000, not a large Pick 6 ticket.
You only win if all three standouts win and they did.
The payoffs for $2 win tickets were as follows
Goldikova was the winner of the mile grass race last year and arguably
better this year. Her connections were the same as those of Miesque also a
two time winner of this race. The second standout, Conduit, was also the
winner last year of the mile turf.
The Pick 6 paid $1,838,305.20 for one winning 6/6 ticket and the 3*9=27
Pick 5/6 tickets (of the 10 losers in the three wide open races) paid
27*$4822.40 for a total of $130,204.80 plus the 6% rebate on the $2000 of
$120 for a grand total of $1,968,630.00. It is not quite $2 million but as
Johnnie Hooker played by Robert Redford in the 1973 movie Sting said:
“it's not enought but it's close”. Of course, taxes would take 25% at the
track and be sorted out later when filing and my winning would depress the
Pick 6 and Pick 5/6 prizes.
The big question is would I have gotten these three winners from my 10
picks in these races? Of course, more than 10 was possible. So let's look at
these three races. I use about 5-6 handicapping services plus my own
analysis of the daily racing form and one handicapper who just supplies
pace numbers. So the idea is to handicap the handicappers. In many bets
this is computerized.
Race 4 The Sprint, 6 furlongs Handicapper #1 had 3-1-5-6 (the winner,
Dancing in Silks) -8-4 Handicapper #2 had 5-3-8-1-6 So at 12-1 that's one
of the 10 for sure especially when you observe that its last race Beyer speed
rating at 106 was the highest of any horse in the race and it was right there
at Santa Anita. Three of the horses ran higher Beyers than #6 but not in
their last race.. Actually there were only nine horses in this race so likely I
would have taken them all including the 20-1 shot #2 and the 30-1 shot #7,
neither of which listed above. #2 had a Beyer of 110 in a Grade I at Santa
Anita so must be used. #7 looks greatly outclassed.
Race 5 Juvenile 2-year-olds on polytrack with 13 horses The winner
Vale of York (#7) was racing in the UK and in Italy. He always had short
odds and had two wins in five races and was close in the other three races.
The morning line odds were 20-1. Handicapper #1 had 5-13-4-9-10-8 (so
no 7 but he was a foreign shipper so likely not rated by this US service).
Handicapper #2 had 13-4-9-5-6. Only handicapper James Quinn with 13-8-
11-5-7-6 had 7 anywhere. These are two- year olds so there is a lot of noise.
There were four other longshot horses one at 20-1, one at 15-1, one at 30-1
and one at 50-1. So going with all 13 horses called all is suggested.
Race 7 Dirt Mile #2 Furthest Land won with 10 in the field. #1 had 4-3-
7-1¬9-2 (the winner). Handicapper #2 had 3-7-2-8. The pace numbers are
competitive. So 2 at 20-1 must be used.
Summary: all three of the wide open races had winners that were
competitive horses. So they would be on our ticket. But even if we bet on
all horses, these three races, the ticket only costs 9*12*10*1*1*1=$2340.
This ticket made a lot of sense so I should have played it.2 It would have
had 8+11+9=28 Pick 5/6s. Oh well, there is always next year.
I want to end this column with a short story of another big Pick 6 that got
away.
In 1991, a racing colleague Cary Fotias and I were hired by the SCA
Sports insurance company from Dallas, Texas, to help insure the Breeders'
Cup which was at Belmont near New York City. We insured the $2 to $3
million part. So the insurance company would guarantee a pool of at least
$3 million. For example, if $2.15 million was bet, they would be liable for
$850,000. We studied and proposed to bet a random amount if needed. The
idea being to get to $3 million and return the insurance company's money
by winning some Pick 6s and Pick 5/6s. It was risky as September 11 had
just occurred and all the Arab owners such as Sheik Mohammed of Dubai
were not in attendance. Their horses and trainers were though. It turned out
to be a glorious day so the crowd sent the Pick 6 pool well over $4 million.
Our client, the world's most famous bridge player, Robert Hammond, said
you two can just play about $25-30,000 of the tickets. So we had a $2000
ticket twice and what we call a gorilla ticket for $28,000. We had some 5/6's
and got most of the money back. The Pick 6 paid $250,000. The race we
lost was the sprint. Squirtle Squirt, which my handicapping colleague did
not like at 9-1 beat the front running filly, Extra Heat at 14-1, who we had
and had led all the way until the finish. So if she had won we would have
had three about $450,000 Pick 6s plus more 5/6s. Squirtle Squirt had run at
Belmont and had the very top jockey Jerry Bailey and was trained by the
recently deceased legendary trainer Broadway Bobby Frankel. Too bad. The
mistakes were Fotias forgetting that Bailey was the jockey and Ziemba not
overruling Fotias and including Squirtle Squirt on slightly more expensive
tickets. Such operational risk errors are common on Pick6 tickets. It is hard
to get all the bugs out. But it was fun! But the next week we won in a
similar situation at Santa Anita, while guaranteeing a $1 million Pick 6,
collecting $240,000 for the client and a nice bonus for us.
Exotic Racetrack Betting Book, 2013

Do look soon for my new book Exotic Betting at the Racetrack to be
published by World Scientific which analyzes Pick 3, 4, 5 and 6, trifecta,
superfecta and other exotic racetrack bets.
Update on the January column

Oh my god! They are not retiring Zenyatta, the third US undefeated horse
with at least ten races at major tracks. since 1900. She won all her races in
California and on synthetic surfaces. Rachel Alexandra was awarded the
Eclipse horse of the year honors. Of course, I thought a joint award was
warranted. But a dirt race between them seem likely in 2010. I sure hope it
will not be a repeat of the 1990 Breeders' Cup Distaff where the 3-year-old
Go for Wand broke a leg and had to be put down after going head to head
with the 5-year-old mare Bayokoa during the whole of the long Belmont
stretch. Postscript: neither was retired and they never met; see Chapter 40.

2Another way to play this is to have three sets of tickets in which you assume that at least two of the
three standouts will win. So you have combinations, i = 1; 2; 3 all at $2 each. So
depending on the you likely have a larger ticket than the three singles approach. You might win
more than one Pick 6 and more Pick 5/6s, but you might miss the Pick 6 as well unless the tickets are
well spread.
Two Super Horses1
Exploring the ramifications of being undefeated
The two greatest racehorses in the US, and among the very best in the
world, are Rachel Alexandra and Zenyatta, both females or mares as they
are called.2 In 2009, Rachel, at three, was beating the best 3-year-olds by 20
lengths. I saw one race, the Kentucky Oaks, the derby for fillies in May She
was sired by Medagiia d'Oro, a very top runner who was a lot faster than
any males running these days. He could run 82s on the scale of Equiform,
see Figures 40.1 and 40.7. Much of his ability shows up in Rachel. So with
female competition too easy, they ran Rachel against first 3-year-old males,
then against older males. And she won all four of these races against males
including the Preakness, the first female winner in 50 plus years. So it was a
terrific year with all her races on the east coast and on dirt. Her past
performances are in Figure 40.4.
Meanwhile, in California, a 5-year-old mare named Zenyatta was
winning all her races against the best females. They then entered Zenyatta,
who at 17 hands is bigger than any of the males, into the Breeders' Cup
Classic, the worlds's most difficult and important race. Zenyatta moved
from the back of the pack to win easily. Zenyatta has a style that stays
behind, then, when the race finish comes closer, zooms like a rocket ship to
beat the other horses. Her final of a mile is as fast or faster than
Secretariat's was. She has amazing agility. Her pace numbers in Figure 40.1
show this behavior. The way you read these numbers is as follows: For her
November 7 race, she ran 61 in the early part of the race, 54 (slower) in the
mid section, the zoomed to an 801 at the finish. The w means that she won
the race. For comparison, 80 has been good enough to wn most of the
recent Kentucky Derbies.
Fig. 40.1 Zenyatta's pace numbers (dates to be added)

Fig. 40.2 Pace Numbers. Source: Equiform
Fig. 40.3 Daily Racing Form Data : Zenyatta
The highest I have ever seen is the four 84s that Ghostzapper ran. The
that Zenyatta ran to win the 2009 Breeders' Cup Classic is not especially
fast.
There are 17 races here and 17 wins. Actually, she has won two more
races so is 19/19. If she remains undefeated, and likely she will have her
final race in the November 6 Breeders' Cup Classic at Churchill Downs in
Louisville, Kentucky, she will be the third such US horse since 1900!3
Colin (15/15) in 1907 and Personal Ensign (13//13) in 1988 were the others.
I consider Personal Ensign's final race, the 1988 Breaders'Cup distaff as
good as any I have ever seen.4 The great Italian trainer and breeder
Frederico Tesio (1869-1954) had three undefeated horses in the 1930s and
later, including Ribot (16/16), Nearco (14/14), Cavaliere d'Arpino (6/6)and
Donatello II who lost only one of nine races. And that's it! So this is a big
deal. Tesio, relying on his wife's help, had inexpensive horses on his farm
on Lake Maggiore in northern Italy, but he was a true anomaly person.
Looking back many generations, he bred cheap to cheap and got
champions! Others breed the best to the best and hope for the best.
I had expected that Rachel and Zenyatta would be joint Horse of the Year
winners but they gave the award to Rachel. She did run at more racetracks,
in more states, against males more times on dirt and Zenyatta won all but
one of her races on California synthetic surfaces and ran only against
females. So the stage was set for a 2010 showdown: who was the better
horse? Oaklawn Park offered a $5 million bonus if both of them appeared.
This got scuttled because Rachel actually lost two races by narrow margins
finishing second both times. Since one race was against the Zenyatta barn's
second string mare, Zardana, it was clear that the real Rachel was not
running. But Rachel has won her next two races convincingly, so she is
back. It remains to be determined if the two will meet before or during the
Breeders' Cup. Figures 40.1 and 40.7 compare their pace speed numbers
which are close. Their styles are different: Rachel starts the race on or near
the lead and powers a head. Zenyatta is slow, way behind then is a rocket-
ship to win the race.
Update: That's what I wrote in the first draft of this column dated August
9, 2010. But this did not happen. I was at Saratoga in late August/early
September and Rachel was running for the first time in a mile race. She
was in a beautiful speed dual with Life at Ten who had won her last six
races with speed ratings similar to but not quite as good as Rachel's on
August 29 in the Personal Ensign Grade I $300,000 race. With the two of
them close together and separated from the rest of the field. But such duels
often wear out the horses and, indeed, Persistently at 20-1 nipped Rachel at
the finish, see the chart of the race in Figure 40.5 and the photos I took of
the race in Figure 40.6. I give credit to the legendary trainer of mares
running races, Shug McGaughey for the upset. Prior to the race, it was
clear from the previous race and final speed numbers that Rachel as well as
Life at Ten were 20+ lengths better than Persistently in the pace figures; see
Figure 40.1. But once again we are reminded that they run the races on the
racetrack not in the computer! The finish was Persistently over Rachel by a
length over Life at Ten by 104 lengths in a slow time of 2:04:45.
Experience and calculations in my racing books such as Ziemba and
Hausch (1987) suggest that it is wise to make large place bets on such 2-5
favorites. I won that. So after a hard decision, Rachel was retired. To me,
this was not a correct decision as the real Rachel seemed back and deserved
a chance to try to get back to the top but owner Jess Jackson of Kendell-
Jackson win fame decided to give her a rest before breeding her to
champion Curlin which he also owns.
Figures 40.4, 40.7 and 40.5 have the past performances, Equiform pace
figures and the chart of the race, respectively. Figure 40.6ab show Rachel
taking the lead, then being run down by Persistently near the finish. These
two were like one reminiscent of the Affirmed-Alydar triple crown races -
again see Steve Roman's website chef-de-race.com to see these
confrontations. But they wore each other out and Persistently, the 20-1
longshot, beat Rachel Alexandra and Life at Ten was third. It was another
great training job by legendary trainer Shug McGaughey. If you look at the
pace figures of Rachel Alexandra and Persistently, it seems impossible for
Rachel Alexandra to lose. This is a reminder that races are run on the race
track not in the handicapping sheets. The ratings were Rachel Alexandra
116, Life at Ten 112, Miss Singhsix 103 and Persistently at 93. See also the
Equiform pace figures with Rachel Alexandra running 77's to 80 with
Persistently's best being 73 on that scale (Figure 40.7). This was a classic
Dr Z place bet which I made on Rachel Alexandra. But with a 95 Beyer
speed figure and now three seconds and two wins in 2010, Rachel
Alexandra was retired to rest and go at 4 to be bred to Curlin, who is also
owned by Jess Jackson of Kendall-Jackson wine fame.
Fig. 40.4 Daily Racing Form Past Performance Data: Rachel Alexandra and four others in the
Personal Ensign Race
Fig. 40.5 Rachel Alexandra Chart

Fig. 40.6 The Rachel Alexandra-Persistently face off
Fig. 40.7 Pace numbers of Rachel Alexandra, Life at Ten and Persistently

2We will see at the Breeders' Cup but the mare, Goldikova, a 10 time Grade I winner and Breeders'
Cup mile grass winner in 2008 and 2009 is in the running too. Goldikova won in 2010 also to be the
first three-peat but she lost in 2011.
3There are many undefeated or once beaten horses - a list is on Wikipedia. But if one restricts
consideration to top quality races, at least ten races, and post 1900, there are just these three in the
US.
4You can watch this racing masterpiece as well as Zenyatta's Breeders' Cup win and many other great
races for free on my colleague Steve Roman's website chef-de-race.com. Just go to videos. There is a
lot more on this excellent website which is updated regularly.
Farewell to the Queen and to the Princess of US
Thoroughbred Racing1
Honoring two female horses
Fig. 41.1 Two great mares: paintings by Nick Martinez
The outstanding thoroughbred horses do not have long running lives largely
for financial reasons. It is financially optimal to retire the top males as soon
as they have reached a high projected stud fee. This is usually after their
third or possibly fourth year. Geldings who cannot breed race longer as they
have no alternative - indeed the great John Henry was running in his 9th
year. For female horses who can at most breed one new foal each year, the
retirement decision is less biased towards early retirement but few race into
their 5th or 6th year.2
There are few events in sports greater than a top horse winning a top race
in spectacular fashion, especially if something impressive is achieved as
well.
In the US there were two great female horses the past two years with
fame and interest far exceeding any males. The princess, Rachel Alexandra,
won all her eight races as a three year old. I was at Churchill Downs when
she won the Kentucky Oaks by lengths. The Oaks is held the day before
the Kentucky Derby and features the best three year old fillies in the US
and possibly some shippers from overseas. So greater challenges were
needed and this was to run her against 3 year old males and older horses.
Her next race was a great challenge, namely the Preakness, the second of
the triple crown races. Rachel Alexandra won and was the only female in
the last 72 years to win this race. She followed the Oaks and Preakness wins
with an impressive performance with another 20 length victory against the
to 3 year old fillies in the Mother Goose Grade I stakes. Rachel Alexandra
finished 2009 with impressive wins against the best 3 year olds in the
Haskell and older horses in the Woodward. These five grade I wins plus
three other wins early in the year made if 8/8. She was not undefeated
having had six wins, two seconds and an out of the money finish in 2008 as
a two year old.
Zenyatta for horse of the year honors in 2010

Like Rachel Alexandra, the five year old mare Zenyatta won all her 2009
races finishing the year by winning in impressive fashion the Breeders' Cup
Classic. She had also won all her previous races so was undefeated.
Zenyatta has a racing style that's wonderful to watch. She starts slow and is
way behind the other horses, then, at a crucial point, turns on the rocket ship
and wins the race. See chef-de-race.com for this classic race which you can
see without cost. This race is one of the best ever. My all time favorite is the
1988 distaff where Personal Ensign won to finish 13/13, which is on this
website too.
In the Eclipse award voting for horse of the year, Rachel Alexandra won
with the advantage to her being races in more tracks with Zenyatta running
all but one of her races in California and mostly Santa Anita on synthetic
surfaces.
Of course, I was convinced that, as good as Rachel Alexandra was, the
regal Zenyatta, who is bigger and more agile than any of the males, was the
better horse. So a joint award seemed the best situation but the voters went
for Rachel Alexandra. Who was the better horse was agreed to be settled at
Oaklawn Park, where Zenyatta had run, a dirt track in the Apple Blossom in
April. The track put up a $5 million bonus assuming both horses ran But in
a prior race Zardana, the second filly in the barn of Zenyatta, actually beat
Rachel Alexandra. Then Rachel Alexandra lost another race, coming
second again. She regained her 2009 form and won her two next races
impressively but the match race was off. Rachel Alexandra's next race was
on August 29 at Saratoga in the $300,000 Personal Ensign race at miles,
a distance longer than she had ever raced. I was there and watched the race
which was a match between Rachel Alexandra and Life at Ten, who had
won six straight races.
Meanwhile, Zenyatta continued winning all her races in California on
synthetic surfaces except for the April Apple Blossom win on the dirt in
Arkansas. She came into the Breeders' Cup classic 19/19 and a huge
sentimental favorite to beat the best males again. She was even money at
Santa Anita but 5-2 (US odds) on Betfair with good place odds as well.
Place in the UK for this field is show in the US.
Earlier in the day Goldikova, the third great female horse shipped from
France and won again for the 3rd time the Mile grass race. No other horse
has won three Breeders' Cup races so this is a remarkable achievement. The
trainer, Freddy Head of Chantilly and Deauville fame, is a true veteran here.
He trained Miesque who won back to back Breeders' Cup Mile grass races
in 1987 and 1988.
Zenyatta faced another tough field but still looked the best. Several other
horses were upset possibilities. Quality Road had won a number of races in
impressive form and had several speed figures higher than Zenyatta's best.
See the past performances in Figure 41.2 and the Equiform pace numbers in
Figure 41.3. Zenyatta does not go for speed, she just goes to win. So her
times do not reflect her true brilliance. Blame had won two major races.
Andy Beyer also noted that synthetic track races are usually won with late
acceleration not early speed. But this classic field was populated with high
quality horses with a speed of sprinters who were running in high gear as
soon as the gate opened.The race started and Zenyatta was incredibly far
behind. The claim by handicapper Randy Moss was that this was her usual
start but the other horses started faster than usual. Maybe he was right but it
sure looked like Zenyatta had lost the race right at the start. The chart of the
race appears in Figure 41.4. Notice how Zenyatta was 11th after one mile
and only made her run in the last quarter mile.
Mike Smith, the hall of fame jockey who was the rider on all 19 wins,
pushed her along. She got boxed in and had to go wide twice and somehow
nearly won the race finishing a head behind Blame. What a tragedy! My
pick 6 ticket singling Goldikova and Zenyatta and going wide elsewhere
was now lost. Smith was heartbroken, blaming himself for losing the race.
Zenyatta was the best horse in the world and finished her career 19/20 with
one second. She was then retired at six. She did win the Eclipse award for
horse of the year with Blame second and Goldikova third. One could argue
for Goldikova with those three wins and eight grade I wins against males.
But they were on grass and in a less important race. History has shown that
the great female runners typically have great offspring. Personal Ensign
being a good example. We look forward to their offspring. Meanwhile its
sayonara to these two greats.
My colleague, the former jockey Nick Martinez has become a painter of
horses and the two prints that follow in Figure 41.1 are his work. Look for
him at Saratoga for your own copies of his work. Nick explained a theory to
me as to why Zenyatta lost. She started slow and was quite far behind the
pack - recall this is a dirt track and she mostly ran on synthetic. Then
around the turn, the pack broke into two groups with one group mostly
quite far away from the group Zenyatta was following. The claim is that
Smith could not see the first group and did not push Zenyatta soon enough.
So she was too far behind when she made her move and did not quite catch
Blame. I sat next to Smith in the Del Mar restaurant Pampelmouse in
September 2011and asked him about this. I sort of think Nick might be right
but Smith said it is a long story to discuss later. Since Smith has won 15
Breeders' Cup races, the most of any jockey and Martinez and I none, we
defee to Smith. He tried to win but did not quite make it.
Update February 7, 2011
Jerry and Ann Moss, Zenyatta's owners, have announced that she will be
bred to the 8-year old stallion Bernardini, who was the leading freshman
sire of 2010. He produced two grade I winners A Z Warrior and Biondetti.
Bernardini won the Preakness by lengths, the Jockey Club Gold Cup by
6+ lengths and the Travers by lengths and was second in the Breeders'
Cup Classic to Invasor. Bernardini was named the 2006 Eclipse Award for
best three-year old colt. Bernar- dini's fee of $75,000 is down from
$100,000 in his first breeding year. He is by super star 1992 horse of the
year stallion A P Indy who himself was sired by triple crown winner Seattle
Slew, who was bred to the triple crown winner Secretariat-sired mare
Weekend Surprise. Bernardini's dam is Grade I winner Cara Rafaela. This
impec¬cable breeding balances speed and stamina at the classic distances.
The foal which has now arrived will be well sought after and be ready for
running in 2014 at the earliest.3
Fig. 41.2 Daily Racing Form Past Performances
Fig. 41.3 Equiform Numbers
Fig. 41.4 Chart of Zenyatta's race

2Edited from Wilmott, March 2011.
3Source for photo of Berardini: http://sidfernando.wordpress.com/2010/10/13/seattle-slew-
secretariata-p-indy-the-sire-of-bernardini
The Dr. Z Place and Show Racetrack Betting
Systems Past and Present1
In which the author reviews the Dr. Z place and show racetrack betting systems past and present
This chapter recalls a memorable day on November 10, 1984 when Ed

Thorp and I attended, with my Vancouver colleague Bruce Fauman, the first
Breeders' Cup day at Hollywood Park. The purpose of the day in addition to
fun, was to test my Dr. Z system co- developed with Donald Hausch with
some early help from Mark Rubinstein. The idea of the system is simple:
use the data from a simple market, in this case the win probabilities to fairly
price bets in the more complex markets, such as place and show. For
example, with ten horses, there are 720 possible finishes for shower. Then
one searches for mispriced place and show opportunities. This is a weak
form violation of the efficient market hypothesis based solely on prices.
How much to bet depends on how much the wager is out of whack and it is
a good application of the Kelly betting system. The formulation below
shows such an optimization. There is a lot of data here on all the horses and
not much time at the track. So a simplified approach is suggested. Don and
I solved thousands of such models with real data and estimated
approximation regression equations that only involve four numbers,
namely, the amounts bet to win in the total pool and the horse under
consideration for a bet. Plus the total place or show pool and the place or
show bet on the horse under consideration.
These equations appear below. In our books Ziemba and Hausch (1984,
1986, 1987) we study this in various ways, including different track takes,
multiple bets for place and show on the same horse and how many can plan
the system before the edge is gone. See the discussion in Fauman's
accompanying paper with more technical detail in Ziemba and Hausch
(1986). This system revolutionized the way racetrack betting was perceived
viewing it as a financial market not just a race. This led to pricing of wagers
and the explosion of successful betting by syndicates in the US, Hong Kong
and elsewhere; see, for example, my joint books referenced here.
Transactions costs
The effect of transactions costs which is called slippage in commodity
trading is illustrated with the following place/show horseracing
formulation; see Hausch, Ziemba and Rubinstein (1981). Here qi is the
probability that i wins, and the Harville probability of an ij finish is
etc. That is qj/1 – qj is the probability that j wins a race that does not contain
i, that is, comes second. Q, the track payback, is about 0.82 (but is about
0.90 with professional rebates). The players' bets are to place pj and show sk
for each of the about ten horses in the race out of the players' wealth w0.
The bets by the crowd are Pi with and Sk with
The payoffs are computed so that for place, the first two finishers, say i and
j, in either order share the net pool profits once each Pi and pi bets cost of
say $1 is returned. The show payoffs are computed similarly. The model is
While the Harville formulas make sense, the data indicate that they are
biased. To correct for this, professional bettors adjust the Harville formulas,
using, for example, discounted Harville formulas, 2 to lower the place and
show probabilities for favorites and raise them for the longshots; see papers
in Hausch, Lo and Ziemba (1994, 2008) and Hausch and Ziemba (2008).
This is a non-concave program but it seems to converge when nonlinear
programming algorithms are used to solve such problems. But a simpler
way is via expected value regression approximation equations using 1000s
of sample calculations of the NLP model. These are
The expected value (and optimal wager) are functions of only four
numbers - the totals to win and place for the horse in question and the totals
bet. These equations approximate the full optimized optimal growth model.
See Hausch and Ziemba (1985). This is used in Dr. Z calculators. See the
discussion in Fauman's accompanying paper for a description of a typical
day's betting and, for more, technical detail see Ziemba and Hausch (1986).
An example is the 1983 Kentucky Derby.
Here, Sunny's Halo has about 1/6 of the show pool versus 1/4 of the win
pool so the expected value is 1.14 and the optimal Kelly bet is 5.2% of one's
wealth.
A race by race analysis of that Breeders' Cup day, plus the previous day
when I went by myself, with racing charts, optimal wagers, etc is in Ziemba
and Hausch (1986). Fauman wrote his own version which is attached, and
was written in 1995 but never published. So I have edited it slightly and
updated it in a few places.
A few comments appear as a postscript after his paper from Ziemba and
Hausch (1986). I did not alter his comments about me which on the whole
are correct. It is a colorful account so I bring it to the attention of Wilmott
readers. Before that begins, you might ask: does the system still work in
2011 and what is changed?
The main new features are:
(1) these days we bet at rebate shops by phone or electronically. The
rebate is a sharing of the track take by the track, the rebater and the
bettor. The effect is to take all bets from a track take of 13-30% for
various bets to about 10%;
(2) betting exchanges in the UK and elsewhere allow for short as well as
long wagers; and
(3) there is a lot of cross track and last minute betting and this takes time
to be sent to the pools at the racetrack. Hence, about 50% of the
wagers don't actually appear in the pools until after the horses are
running. So one must estimate the final odds (probabilities).
Syndicates exist that break even on their wagers yet make millions on the
rebate. Regarding the Dr. Z system, John Swetye works with me and we
wager with rebate searching for bets at 80 racetrack. Basically the system
still works but the task is not easy. One successful six month period with a
$5000 bankroll, the system lost 7%, received a 9% rebate. The total wagers
were $1.5 million giving a 2% or $30,000 profit.
Three to beat the Breeders' Cup by Bruce C Fauman

This is the description of our day at the 1984 Breeders' Cup by my former
colleague Bruce Fauman. Bruce died shortly after the internet stock market
crash on December 12, 2002 at age 59. To publish this I have slightly edited
and updated his original copy. My own description of that day with more
technical discussion, racing charts, etc appears in the 1986 book Betting at
the Racetrack.
A decade ago, at the first Breeders' Cup in 1984, three ordinary jamokes
in their forties, among whom one could count a total of three wives, six
kids, nine degrees, about 500 IQ points, and a system to beat the track, set
out for Hollywood Park to test their theories and equations on horseracing's
biggest day ever.
First Race: The Juvenile, 1 mile, for 2 year old colts and
geldings, purse $1,00,000.
The ground growls as I stand at the rail beside the sixteenth pole. The
vibrations travel faster through earth than through air, so I feel the horses
before I hear them. The front runners reach the head of the stretch, three
wide, a length behind one another. Their hooves sound not a distinct clip
clop beat., but a series of overlapping thuds upon the fast track surface.
Number 5, a bay with white stockinged forelegs, has gained steadily on
the early leader throughout the back stretch. As they pass me, he pulls even
and tries to fend off another bay's late charge between horses. At 35 miles
an hour, the horses spew fragments of track soil with each hoofs raising.
The warm November morning accentuates the aromas of rich soil, fresh
manure, damp straw, saddle leather and horse sweat, which in a more subtle
form are often invoked as barnyard bouquet when complimenting a well
balanced mature Burgundy.
Watching a horse race from the stands is as different from watching at
track side as it is from watching on television. I left our box to see the first
race from the rail and became so engrossed in the pre race ritual that I
forgot to make the first system bet of the day. The possible opportunity
appeared early on the tote, but by the time I remembered to ignore the
horses and recheck the odds, the horses were loading, and I was too late to
get to the betting window. Some say the first things to go in an ex athlete
are the knees. Not true; it's the short term memory. The grandson of both
Northern Dancer and Secretariat holds on to win by nearly a length.
Remembering our day's real purpose, I will no longer be so cavalier as to
watch a race for pleasure. We're not here to enjoy the races, but to beat the
track on what promises to be horseracing's biggest day in history
Second Race: The Juvenile Fillies, 1 mile, for 2 year old fillies,
purse $1,000,000.
Dr. Z introduces Ed, Jeff and me to the occupants of the box to our right,
Lindsay a local newspaper reporter and an English author of handicapping
books who mumbles his name, Foofraw or Frew faw, in upper class Brit
speak.
Bill Ziemba, is my colleague, sometime coauthor, and seminal mind
behind the system. I am the creator of his Dr. Z nom de plume, the reality
checker of his mathematical manipulations, and one of the few people
willing to tolerate his bustling, blustering and occasional boorishness in
return for access to the treasures of a polymath's mind.
Our third adventurer is Ed Thorp, who developed the original card
counting system for beating blackjack in the 1960s, and since no casino will
let him near a 21 table anymore, he has gone on to other things. But Ed is
always interested in any system that can actually beat the house, and is here
to see for himself whether the system will beat the racetrack.
Ed wears a cocoa heather herringbone jacket and beige slacks. I'm in my
I can go anywhere uniform of a navy blazer, gray flannels, button down
shirt and sincere necktie. Ed and I remove our ties, which we believed were
required in the clubhouse. Jeff's is unchanged from half mast. Bill is in a
conservative glen plaid suit adorned with a most un conservative floral tie.
Foofraw asks me which filly I like in the race, but before I can answer, he
tells me, “The 3 if she runs true to form, with the 9 right after that. Unless
the 6 filly has been laying in the weeds and I'm quite taken with number 8's
works, so she could upset.”
Bill and Ed made. the system bets on Chief's Crown in the first. Dr. Z
tells me at $2.40 to show his $100 wager puts him $20 ahead. Ed doesn't
say how much he bet. In addition to disappointment in my forgetfulness,
I'm also a little disappointed in the small payoff, while Dr. Z is pleased that
the horses are running true to form, which means the system should run true
to form as well.
We check the toteboard with each flash. The favored filly, number 4, has
offered a likely system bet since early in the wagering. The show bet
underlay fluctuates around 35%, which ought to give me an expected
payback of 1.17, or $1.17 for every dollar bet. My usual cutoff is 1.15,
which means I'll make a bet only when I have at least a 15% advantage.
However, for an event like the Breeders' Cup, with great horses and ideal
conditions, I reduce the cutoff to 1.10. Throwing darts at the racing program
gives an expectation of about 85 cents per dollar wagered, since the track
take is 15 cents of each dollar wagered. A fair bet is like tossing a coin,
where the expectation is 1.00. Unless a bet is for small sums between
friends, fair is for fools.
Just like Wall Street, the racetrack is a financial market, in which people
invest money in ventures with uncertain outcomes. The stock market is said
to be “'efficient,” because neither knowing how a stock has performed in
the past, nor having public information about the company's activities,
provides any prediction about the stock's future price. The current price
reflects all available information and is the best predictor of the future. The
racetrack is also a financial market, a turf market, and efficient as well. The
odds offered are the best predictor of any given horse's probability of
winning. The toteboard odds reflect all available information about each
horse's relative speed, stamina, breeding or other factors that predict
performance. Some handicappers rely directly on such historical
information about a horse to make betting decisions, while others eyeball
the animals being saddled or parading to the post. However, the tote odds
already reflect the handicappers' varying opinions. Forty years of research
confirm that horses whose post time odds predict a 25% probability of
winning do win 25%, of the time. There is a tendency for the turf market
investors to underbet heavy favorites and significantly overbet longshots.3
While the win betting market may be efficient, often the market for place
or show is not, giving a profitable opportunity to those who can recognize
the inefficiency and take a risk arbitrage position in the turf market.
I am not a gambler. However, I have been known to invest money in
events with uncertain outcomes, but only if I can expect to take out
somewhat more than I put in. When the efficient win pool indicates a horse
is two to one, but the show pool offers me a payoff equivalent to that of
three to one, I invest.
The system is based on similar reasoning and much more exactitude.4 Dr.
Z developed the nonlinear estimation and optimization routines to calibrate
the equations which tell us whether and how much to bet. Dr. Z and Donald
Hausch wrote a book on the system titled Beat the Racetrack. Originally
they used the terms “Hausch Ziemba algorithm” and “H Z method” to
describe the system. When I reviewed the draft of the 1981 working paper 5
and suggested Dr. Z as a better name for title system and a nom de plume
for Bill's non academic writing. Their manuscript now refers to it as the
“Dr. Z System.” Bill wrote a column for Gambling Times on lotteries and
horseracing for some time under the Dr. Z byline before noticing another
sports writer for Sports Illustrated, who I don't think is a real doctor of
anything, use the same Dr. Z moniker. I guess that's what happens when you
choose, such a common name; there are 273 John Smiths or J. Smiths in my
local white pages, but not even one Pocahontas. Later Bill told me a lawyer
checked and he has the rights to the name.
The system has been tested a number of times at our local racetrack. It's a
small track with a. small crowd, in a climate that makes. for an iffy track
surface, which offers only two or three system bet opportunities in a, typical
day's. ten race, card.. Furthermore, a wager of $200 or so is big enough to
influence the odds in the parimutuel betting pool in which the bets of the
losers are divided among the winners. We needed better controlled
conditions to validate the system. Today Hollywood Park has become our
laboratory. The Breeders' Cup is the biggest day, in racing history, offering
$10,000,000 in purses for the horse owners. The dirt track is fast' so each
horse should run true to form; the purses are huge, so no trainer or jockey
will hold back; and another ten million should pass through the betting
windows, making the pools big enough in which we will always be small
fish whose wagers won't alter the payoff odds.
The toteboard flash at one minute to post indicates an acceptable
25%,underlay. Dr. Z and I each bet $100 on number 4. Again Ed doesn't say
how much he bet. The horses are off to a start of thumping and bumping.
Our filly breaks stride early and finishes well back.
I feel my chest drop into my stomach as the race ends. I know you can't
win them all; nevertheless, I'd like to. The system delivers cashable tickets
only 75% of the time. I sure hope that Lindsay and Foofraw hadn't bet the
22 1 or 75 1 fillies that finished one two. They hadn't. A loser loves
company, even if they are I time losers. Foofraw tells Dr. Z that an exacta
bet on the longshots would have paid $8,000, ignoring the fact that this race
has no exacta betting. Bill and Don developed an exacta variant of the
system, which would almost never consider a bet that includes even one
such long odds horse, in their 1986 book Betting at the Racetrack, since it
would not price out to have an advantage.
Third Race: the Sprint, 6 furlongs, for 3 year olds and up, purse
$1,000,000.
En route to the track I picked up Ed Thorp, who lives in a gated community
on a Newport Beach hilltop. Aftcr passing muster with the gatehouse guard,
I drove up to Ed's brand new, old California mansion. He greeted me and
offered coffee, pointing to a bigger than a breadbox, brass plated, Italian
gizmo, and said he'd be ready to go in five. Ed is about as average looking
as a rocket scientist can be pushing fifty, gray shot walnut hair, but all still
there, tortoise shell glasses over blue eyes, an inch or so less than six feet,
medium build, probably within five pounds of his graduation weight
because of the marathons he still runs. I too am within five pounds of my
graduation avoirdupois of 230, but some of it must have migrated South
from my chest and shoulders, since my waistline has grown an inch or four.
I no longer run the long distances of my high school and college days,
which were seldom more than a furlong on a track and forty yards on a
football or rugby field. I've also lost half an inch of my six three, which my
daughter attributes to my hair having been blown away from driving too
fast in my convertible. My counter reminds her that until I had children I
had a full head of thick black hair; post hoc ergo propter hoc.
Mug of world class coffee in hand, I roamed the main floor and could
understand how there was several million in the place at least 10,000 square
feet of house, twelve foot ceilings, Architectural Digest kitchen, a view out
to China a, sunken tennis court, indoor outdoor pool. The kind of house fit
for one of the greatest hedge fund traders of all time.
As I salivated at the stereo and video components in the den, a face
Killroyed over the eight foot oatmeal leather sofa. Once standing, he
introduced himself as Ed's son Jeff, and said he was coming to Hollywood
Park with us. Jeff was 19 or 20, five ten, with rust brown hair and a few
freckles. He wore a dress shirt, a neck tie pulled halfway down, pressed
khakis and polished Weejuns.
Ed descended and sent Jeff upstairs to fetch something. When I asked
about the curious slot in the kitchen ceiling, Ed said that his wife didn't like
to carry packages in from her car. The slot was an industrial conveyor track,
leading from the garage, through double hinged doors and into the kitchen.
He demonstrated, hanging one of a dozen yellow plastic baskets from a
concealed hook and pressed buttons that smoothly carried the basket around
the loop. I guess an ex nerd with imagination and money can indulge
himself in creative gadgetry, for Ed exuded an inventor's pride in the
device. Judging from the house, the gadgets, and the cars, Ed spilled more
before breakfast than Dr. Z or I earned in a year.
In my Hertz hippopotamus on wheels, I told Ed, “I became a decent skier
thanks to you. Your book paid for four winters of skiing at Tahoe.” Ed is the
math professor who developed the card counting technique for beating
blackjack. After the casinos banned him from play he wrote Beat the
Dealer, a book describing the technique. I mastered his method, and before
the casinos changed their rules for everyone, I could play blackjack for an
hour or two before, dinner at a North Shore casino and pick up the $100 or
so that would pay for a week end of skiing, at Squaw Valley 15 years ago.
While a grad student, I skied 20 days a winter courtesy of Ed Thorp, and
now might have a chance to repay him.
I think Ed's accustomed to such occasional acknowledgements, for he
just shrugged. He's now into bigger things. After being banned from
blackjack, he worked out the techniques for stock warrant hedging, the
precursor of options theory, portfolio insurance and methods of valuing
various derivative securities. Ed now runs an investment pool of seven
figure amounts from each of a few dozen investors. His fund searches for
small discrepancies among the prices of equivalent securities, such as
convertibles, warrants, options and the underlying stock, then arbitrages
that discrepancy.
Ed played navigator, reaching into the satchel at his feet for an inch thick
road atlas of the L.A. area, and directed me through the back streets of
Inglewood, avoiding the heavy track bound traffic.
As the horses begin the parade to the post for the third, with ten minutes
to go, I scan the toteboard and see possible place and show system bets on
the 3 horse Ellio. Every minute or so Dr. Z keys data from the toteboard
into his gozinto. The calculator Bill holds in the palm of his hand contains a
custom chip on which he programmed the system. It has more computing
power than the M.I.T. mainframe Ed used for his original blackjack analysis
in 1959. At three minutes to post time I estimate underlays of about 50% to
place and 55% to show. Dr. Z confirms my approximation and indicates the
optimal bet size. He bets $110 to place and $215 to show; I do $100 and
$25 0, because I'm lazy. Ed buys tickets on both, but still isn't saying how
much he bet.
Our choice leads wire to wire, with just enough stamina to hold off a late
closing 35-1 bay, and wins by a nose.
The average margin of victory in a Class I stakes race is only one length,
or 20 feet in a mile and a quarter race, which makes the second best horse
99.7% as fast as the winner. There aren't many second place finishers in
business, sports, school, or the arts, who the public perceives to be 99.7% as
good as the person who finishes first. The winner is remembered but not the
runner up. How many of us can name the world's second best cellist or high
jumper?
I go to the window to cash my tickets. Ed just hands his to Jeff. While
I'm delighted at the $3.80 place and $2.80 show payoffs on a 6-5 horse,
when I return I see Dr. Z is unhappy. With both place and show bets on the
same horse, the system should adjust the optimal wager amount to reflect
the joint probabilities. Bill hadn't had time to complete the additional
calculations with only one minute to post. Afterward he computed that the
optimal bets should have been $84 to place and $351 to show, which would
have netted him another $31.
Fourth Race: The Mile, 1 mile on the grass, for 3 year olds and
up, purse $1,000,000.
I scan the crowd through Jeff's field glasses. Inside the glass walled V.I.P.
dining pavilion to my right, where neckties definitely are required, I see
Cary Grant at a table directly in line with the finishing pole. He's on the
Hollywood Park board of directors and deserves a prime table. I focus the
10x50s on him and think I'd like to look that good when I get to be his age.
Actually, I expect that by the time I get to be Cary Grant's age, I'll have
been dead for ten years.
Dr. Z hasn't eaten in more than, two hours, probably a daylight personal
best. He has only two paces fast and even faster and needs to refuel every
hundred ideas or so. Nevertheless, Bill is finicky about his diet and won't
ever eat junk food like most of us. Before the tote odds firm up, he canters
to the V.I.P. dining pavilion, where he's able to procure a take out order a
club sandwich on whole wheat toast, with no butter, no mayo and no bacon
for $18.00, plus tax and tip.
Dr. Z is a fortyish dervish with curly graying red hair and a beard to
match. Even while wearing a suit that a banker might buy, he frequently
sports the bright blue tam o'shanter knit by his wife. He's written a dozen
books in as many years on everything from stochastic optimization in
corporate finance, to Turkish tapestries, the mathematics of lotteries, and
now his system for beating the racetrack. Bill flits around the world, often
towing his wife and a beardless five year old minature of himself, to give
invited talks at universities and conferences before audiences who bob their
heads in understanding. Back home, most of his UBC colleagues are of two
types the majority, who cannot understand the depth and range of his work,
and therefore resent him, and the few who can understand but not match it,
and with quiet envy, resent him even more.
A conversation with Dr. Z is like taking a drink of water from a fire
hydrant. If I could bottle and sell injections of that energy and intellect, I'd
have even more money than Ed and a conveyor track system with a spur
line to my wine cellar.
Bill once told me that I was probably his best friend, and appeared
somewhat miffed when I didn't reciprocate. But throughout life I've had
only one very good friend at a time, and in the 14 years since she was naive
enough to marry me, my wife has been that friend. Besides, having Dr. Z
for a friend isn't all that easy. You've got to accept his occasional grating
idiosyncrasies with his brilliant insights, as indivisible as a quark. He talks
nearly full time at full speed, even more so than 1, and frequently while his
mouth is filled with one of the six meals a day it takes to fuel his mega
metabolism. He changes topics in mid sentence, because even at 200 words
a minute his mouth is three lengths behind his mind. Yet he's unnecessarily
generous with his coauthors, listing each alphabetically. But I think that Bill
Ziemba is hoping to find a new protege named Zollen or Zufiuyden.
Having me for a friend or colleague isn't that easy either. I don't suffer
fools gladly. I know a bad idea when I see one, and am outspoken enough to
say so, believing that keeping silent does no one a service. My
outspokenness extends to carrying on my part of an insulting dialogue with
the TV news anchor, shouting expletives about his half truths and omissions
at the man in the $50 haircut on Channel Two, who doesn't seem to respond
to my compelling debating points.
Back in our box, Dr. Z alternately punches toteboard data into the gozinto
and chomps the portion of his $18.00 sandwich that doesn't slop on his tie.
“Elizabeth Taylor is eating lunch over there,” he mumbles through the
turkey and toast in his teeth, and gestures at the dining pavilion, flinging
half a tomato slice on my shoe. I'm skeptical, since Dr. Z probably hasn't
had the patience to sit through a two hour movie since Liz was on her third
husband.
Dr. Z and I each handle the tote information differently. I do
approximations in my head. He keys in the data about every third flash and
gets precise results. The system's algorithm calculates how much to bet
,based upon the advantage offered and the bettor's risk capital. He shouts
the exact amount to bet, based on a bankroll updated for the day's income or
outgo from earlier races, and even the $18.00 club sandwich. Bill.gives
numbers like $366.47, which means I'll. bet either $350 or $400. Given Ed's
bankroll, he ought to bet ten times as much, but, doesn't say. The earlier
computer simulations used exact whole dollar amounts, a technical nicety
that isn't practical in the real world. Imagine the patience of a typical bettor
behind you in the queue should you request exactly 183 $2.00 show tickets
with half a minute to post time.
At each flash the tote's been bouncing on either side of the cutoff point
for a system bet on number 1, the only filly in the field. I want to make a
bet, but only if I have a healthy edge. Remember, fair is for fools.
We move from the box and stand at our key vantage point, as close to the
$100 betting window as we can yet still see the toteboard. With one minute
to post the show bet crosses the 1.10 threshold and Dr. Z and I each bet
$100. Ed is taciturn as usual, but I see that he keeps a half inch thick stack
of $100 bills in the inside left breast pocket of his jacket, fastened shut with
a two inch safety pin.
On the backstretch our filly is steadily overtaking horses. She wins the
race by more than a length, sets a new American record for a mile on turf,
and pays $2.80 to show. A veteran winner now, I eschew walking up to the
payoff window and nonchalantly hand my ticket to Jeff to cash.
Fifth Race: The Distaff, 11/4 miles, for fillies and mares, 3 year
olds and up, purse $1,000,000.
I'm getting the bettor's blues. When you lose, you regret losing the money
you bet. But when you win, you regret not making a bigger bet. Damon
Runyon said that all life was eight to five against. Runyon was optimist.
After returning with our pelf, Jeff goes to find lunch for the two of us,
foraging passable corned beef sandwiches, packets of regular and hot
mustard, a bag of Fritos and a couple of Pepsis. Ed pulls a container of
yogurt from his satchel.
I open my wallet to give Jeff money for the sandwiches. “Ed, look at
this,” I say, and pull the Thorp card from between my birth certificate and
medical insurance card. It's a chart of Ed's high low blackjack counting
system. I've carried it in my wallet since 1964, in case I happen to stumble
upon a casino, like on a cruise ship, a Caribbean island or Anytown
Nevada.
As I said, I don't gamble, and don't go places just to gamble, but I will
invest in opportunities with uncertain outcomes, so long as I am in the
neighborhood anyway, and most importantly, have advantage. When we go
to Palm Springs each winter, I play in a modest stakes poker game that
gives me such an edge. Leo Durocher is a semi regular in the game. The
winter before last, when I told him that Elston Howard had died earlier in
the day, Leo reminisced fondly about the Yankee catcher and said, “Ellie
was a winner in a loser's game.” “You too, kid,” he added “that's how you
stay at this table.” Since he's twice my age with half my hair, Leo can call
me kid; and he's right about how I play. At our level, poker is a loser's
game, in which money is lost by those players who make big mistakes, and
then divided up among the other players who don't make the mistakes.
Conversely, big money poker is a winner's game, in which the money is
made by the players who make brilliant decisions and collected from the
other players who don't make such plays. The racetrack is a major loser's
game, since the house cuts 15% or more off the top. To win at the races you
have to identify enough big mistakes by the crowd to offset that track take.
The system can do so for those who have the knowledge, patience and
discipline.
Ed focuses thoroughly on one major endeavor at a tune, gets seriously
rich by practicing it. Dr. Z has several projects underway at a time, each
getting his best for several hours or days at a time, but few earning him
anything but professional accolades. I am merely a dilettante, a dabbler in
many things, few in depth or with passion. I like academic research, and
keep up with the literature in several fields, but have little interest in doing
much myself. My few publications in big league academic journals aren't
sufficient for tenure, and my published recipes and satirical columns on
business mid economics don't count. Because I'm an omnivorous reader,
and the more meaningless the information, the more likely I retain it, I've
amassed at least twice the amount of useless knowledge about more
subjects than the two of them combined. I ought to make a good Jeopardy!
contestant since I can't get anyone to play Trivial Pursuit against me
anymore.
An odds on favorite offers the only possible system bet in the fifth. I
check each flash of the toteboard. In the last five minutes Foofraw has
touted four different horses as possible winners in a seven horse field.
Should one of them prevail, he'll say, “See, I gave you that winner,” or the
Brit mumble equivalent. I suspect that Dr. Z may be a closet handicapper,
because he says he thinks highly of the four year old favorite. Foofraw
jumps on the bandwagon and cites dosage index numbers, workout times
and the filly's recent races. Ed doesn't care about horse lingo, for he knows
that the value of dosage is already imputed in the tote odds. Didactically, he
explains to Foofraw that since none of us owns a horse in the race, we
cannot win any part of the $1,000,000 purse, only a return on our bets.
“Buying a hundred shares of General Motors stock isn't the same as buying
a Buick.”
With three minutes to post time, the possible system bet on number I still
hasn't materialized. We're here to test the system, but without a horse whose
odds meet our criterion we won't wager. Dr. Z and I check the toteboard at
each flash, as if our encouragement will cause the numbers to change and
give us the opportunity to bet. I can understand a little of how a compulsive
gambler must feel, ever eager to place a bet. At one minute to go I
guesstimate that the expected payout has across our threshold. Dr. Z's
computer confirms it. We buy our tickets as the horses are loading at the
gate. Ed and I walk down the stairs as the horses break from the gate.
Hearing the track announcer calling the race over the P.A., Ed asks, “What's
the horse we bet on?”
“What's on second,” I respond, on automatic shtick, in a pretty fair Bud
Abbott voice. “I think it's Princess something?”
“Bruce, I only hope she has four legs.”
Princess Rooney, all four legs intact, romps to win by half a dozen
lengths. She pays a disappointing $2.20, which is why this odds on favorite
was nevertheless only borderline as a system bet. Foofraw and Lindsay hold
an exacta, wheel with her and each of the other six entrants. So they win
their exacta but the bet actually loses money doing so, spending $30.00 for
each six ticket wheel on which they'll collect only $28.00.
Sixth Race: The Turf, 1 1/2 miles on the grass, for 3 year olds
and up, purse 1,000,000.
The Turf attracts several European horses. Foofraw is holding forth beside
us, rating the imports against the locals, although all are American bred. He
and the newspaper guy are serious handicappers. They pore over past
performance charts in the Daily Racing Form, and talk speed ratings and
bloodlines. Foofraw mumbles overmuch. Were a Henry Higgins present, I'd
offer two to one he'd pronounce Foofraw a non U fraud.
Lindsay and Foofraw favor the low probability, high payoff exotic bets.
Foofraw calls ours, “ladies' bets,” saying that wagering on a favorite to
show for a $2.60 payoff, isn't really wagering at all. Except for the pyhrric
exacta in the fifth, those two haven't cashed a ticket yet, while after each
race Jeff returns and deals out $100 bills to Ed, Dr. Z and me like it was a
card game. It takes a lot more real testosterone to maintain the self
discipline to bet our way than his. I don't mind if Foofraw insults my
manhood; keep those hundreds coming.
The starting gate is now at the head of the grass turf course. There are
eleven horses in the field, but only All Along, the crowd's second choice at
7-2, is a decent Dr. Z system bet. She's a 30% underlay.
In addition to keying data into the computer, Dr. Z also scrawls
information about the win, place and show pools at five, three and one
minutes before post time. A month earlier I'd hypothesized that a late drop
in the win pool odds might be so called smart money, a predictor that a
horse would win. If true, the anomaly would violate our assumptions of an
efficient turf market, yet provide us new higher payoff betting
opportunities. He was collecting information to test my conjecture. Dr. Z
isn't really sure the, stock market is efficient either, citing anomalies. Ed,
knows it isn't, and has made his fortune arbitraging many small
inefficiencies.
The zoom lens on Jeff's camera can't shoot the toteboard in a single
frame. Ed says we could capture the data quite easily by mounting a camera
with a 20mm wide angle lens and motor winder, focussing on the toteboard
and snapping the shutter every minute. I one up him suggesting a video
camera cabled to a digitizer which would not only capture the image but
also convert it to numerical data and directly input it to the computer. Ed
tops me with a scheme to transmit the computer's betting instructions by
radio signal to one of us at the $100 window. That's not too fanciful, for five
years ago a group of Ed's fans in Silicon Valley constructed a toe operated
computer built into a sneaker and programmed to at roulette.6
Ed has never been to Hollywood Park before, and I'm not sure he's even
been to any racetrack in twenty years. He's here to see the system in action,
in the real world, in real time, with real money. When Bill first explained
the reasoning behind the system and the basis of the complex optimization
calculations, Ed took all of five minutes to concur, probably performing the
pages of calculus equations in his head as Dr. Z spouted them. Ed's
experience in the financial markets has made him question any theory's
applicability in practice. He's seen what happens to the bid ask spread when
he tries to take a big position in a stock. He thinks that in a paramutuel
betting system, the payoff offered when we make a decision won't be the
same as when we go to cash our tickets if we get cash them at all. Ed also
doesn't care for the idea of putting money on humans or animals, because
there's too much random error in a one performance of one jockey on one
horse in one race. Furthermore, he's discomfited knowing the turf market
offers only ten races a day, and a top horse will typically run in a dozen
races a year for two or three years. A small number of investment
opportunities, high transaction costs and a fat tailed distribution are an
anathema to a man who makes hundreds of individual trades a day in the
Wall Street and Chicago financial markets. The few thousand dollars he'll
bet today are merely to test whether Dr. Z system is valid in practice.7
Hearing I too would be in Los Angeles that weekend, Dr. Z invited me to
join the expedition. The box he'd borrowed had empty seats and, since I was
in the neighborhood anyway, why not. When I was Jeff's age, my uncle
entered one of his horses in the Kentucky Derby. Since then I've enjoyed
horseracing, but in moderation. At home my trips to the track with Dr. Z are
purely for research purposes, although on a sunny afternoon we'll each
bring a daughter along to enjoy the event and to log a twofer of betting and
father bonding. Once or twice a year I go to the track with friends for
dinner, and as long as I'm there and I have an advantage, I'll make a few
system bets to pay for the wine, and sometimes for the dinner too.
Dr. Z is here for all of it. He wants to validate his system on horseracing's
biggest day in history. He also wants Ed Thorp to watch him do it, in real
time for real money. But Bill really does like horses and racing. He's trying
to get a release to use a picture of himself with Secretariat on the dust jacket
of Beat the Racetrack , the book he and Don Hausch have in press. If he
makes any money on the securities market anomalies project he is
beginning, I think he'd likc to squander it on owning a racehorse. I do
believe Dr. Z is a closet handicapper.
The field breaks cleanly from the gate and. Willy Shoemaker rides his
roan out to an early lead along the rail. The field is tightly bunched as they
cross the patch of dirt track and back onto the grass. On the backstretch All
Along steadily gains on the leaders and pulls into the lead before the stretch
run. She's nipped by a neck and finishes second to a 50-1 longshot. Since a
12-1 horse finishes third, this should be a big pavoff on our system bet. At
$4.40 to show, it's huge. Hooray for Hollywood Park!
Seventh Race: The Classic, 1 1/4 miles, for 3 year olds and up,
purse $3,000,000.
This is a tough race for most other bettors to get excited about. The
standout, a son of Seattle Slew, should go off odds on. While the payoff will
likely be modest, there'll almost certainly be a system bet available, because
the crowd does agree with Foofraw. Back at the office on Monday, there
will be no bragging rights in telling how you cashed a $2.20 ticket, so the
crowd won't wager heavily on the favorite in the place or show pools.
Super favorites often provide outstanding returns if you bet them with
enough conviction. While the system is new and quantitatively
sophisticated, some people had been doing the no brainer equivalent for
years. There's a likely apocryphal bettor in Kentucky called the Bridge
Jumper. Whenever offered a one to five or shorter odds on favorite, the
Jumper would bet $20,000 to show, buying several different tickets so he
needn't fill out IRS forms when he cashes them in separately. While
California requires a payout of $2.10 on a $2.00 bet, the Kentucky
minimum is $2.20. At one to five, assuming an efficient turf market and
independent sub races, the probability of the favorite finishing out of the
money is only one-half of one per cent but the track has to pay off as if it
were nine per cent, a huge underlay. Looming above is the one half of one
per cent risk he's a Joe Bltsftk that day, and the $20K goes kaput; hence the
bridge.
If you owned a thousand inch color TV, it would be the Diamond Vision
screen in Hollywood Park's infield that displays close ups and replays of
each race. Following each winning system bet, I watch the Diamond Vision
replay of the finish line through Jeff's field glasses and zoom in on the
efficient market equine benefactor who favored us by running true to form.
Dr. Z returns from fetching another $18.00 plus club sandwich, without
mentioning anyone famous. There are system bet possibilities for both place
and show on Slew O'Gold. Dr. Z likes the horse as well as the tote, citing
data from the Daily Racing Form. I Groucho my eyebrows and flick an
imaginary cigar, letting him know my doubts about his piety toward the
system.
The 30% underlay is enough for me to make both bets three minutes
before post time. Returning to the box I see the toteboard flash a big
change. In the last minute someone has bet another $50,000 to win, making
the respective place and show underlays 40% and 50%. Dr. Z punches in
the new numbers. The optimization genie in the gozinto still says our $250
place bets are in line, but the show bets should have been $927, not the
$500 we'd each bet. I push Jeff aside,mount the stairs in threes and sprint
toward the $100 window. I shout to the clerk from 20 feet, $100 Show,
Number One, four times.” Midway through my plea, the bell rings. I'm shut
out at the window.
Even though the race action isn't supposed to matter, I thrill in watching
this outstanding finish. The horses bump and shove on the long Hollywood
Park stretch run. Three horses are in contention. Slew O'Gold with Angel
Cordero, Jr., up is getting sandwiched between number 2, Wild Again, on
the rail and number 5, Gate Dancer with his muffed ear covers, veering in
from the outside. The horses finish 2, 5 and 1, no more than an armspan
separating first from third,, some five lengths ahead of the also rans. Wild
Again won at 30-1. He was supplemented at considerable cost by his
owners and the win was one of the starting points for the career of the great
jockey Pat Day.
The inquiry sign, on the toteboard lights, indicating either the stewards'
or a jockey's claim, of foul. Dr. Z keys in the final betting pool data, and
tells us his last pre race estimates were within three per cent of the
algorithm's optimum bet. Right now I don't care about the algorithm. Will
the race result hold up? Will the stewards, take down Slew O'Gold? As the
horses finished, we lose our place bets, and don't make enough on the show
bets to offset that loss, but if I'd gotten the additional $400 down, it would
have. However if the stewards completely disqualify Slew O'Gold my loss
will be that much greater. I ruminate on all the woulda coulda shoulda
combinations. The stewards take four or five minutes, which is geological
time for the tenterhooked ticket holders of the maybe in the money horses.
The Diamond Vision screen replays the stretch run at full speed, then in
slow motion. The inquiry light turns off and the tote displays the official
results. The finish order is now 2, 1 and 5; the stewards took Gate Dancer
down. Our place and show system bets both pay off, at $3.00 and $2.20.
Dr. Z too likes to invest in other uncertain situations, so long has he has
an advantage, and we have a few followers who will join us when we spot
one. A year ago we each independently found a one shot., one day lottery
opportunity which offered an expected payout of $2.38 for each $1.00
ticket. We rounded up the usual suspects from the department. The six of us
each descended on a different local retailer at 7:00 am on a Friday, politely
asked to commandeer the lottery terminal, and then spent the morning hours
making it spit out 1,000 or 2,000 tickets. All day Sunday we sorted through
grocery bags led with lottery tickets gleaning our winners, and collected our
predicted payoff.
While my confidence in Dr. Z is high, it's less than perfect. He sometimes
needs my reality-checking skepticism. Last winter Bill calculated a way to
regularly play the lottery that had a payback with an expected value of
almost $11,00 for each $1.00 ticket bought. We have each consulted for
government lotteries, he on the mathematics of game design and I on
marketing and strategy, so once we discussed the logic behind it, I trusted
his estimate. Our departmental syndicate bought 100 tickets on each lotto
draw. When after five months we were so far behind, I asked Dr. Z to
recheck his calculation of the expectation, and also to look at the higher
moments of the probability distribution. The next day we lunched at the
Faculty Club, and when he treated me to a Heineken, I knew there was bad
news coming. Yes, at 10.7267 the long run expected payout was as he'd
previously calculated. Since such a huge chunk of the lottery prize pool
goes to a winner who hits all six numbers of the 6/49, a 13.9 million to one
shot, it might take some time to achieve that expected value. Even if we
bought 1,000 tickets a week, it would be more than 15 years before we
could be 90% certain of being ahead of buying government bonds instead.
John Maynard Keynes said, “In the long run, we're all dead.”
Eighth Race: Fleet Nashrullah Stakes, 6 furlongs, for 2 year olds,

purse $60,000 added.
Much of the crowd was leaving. Cary Grant's table was empty. Elizabeth
Taylor had left, if she'd even been there. The seven Breeders' Cup races
were at the start of the day to reach East coast TV audiences, so Hollywood
Park slated two more races to fill out the program. The toteboard isn't
leaving with Cary and the illusory Liz, even though only about one third as
much is being bet on this race as on the Classic. None of the entrants are
important to Dr. Z and Ed still doesn't care to know the horses' names.,
There are five contenders and two long shots, in the seven horse field. The
tote has been pretty steady, offering a show bet on only the fourth favorite,
number 3. We make the bet.
Naturally, who was the catcher on Bud Abbotts baseball team, our pick
Teddy Naturally, leads the field throughout and wins by nearly ten lengths,
to pay $3.40 to show. While Ed still isn't talkiing about how much he's bet, I
see his other inside. breast pocket now holds a second stack of hundreds as
thick as the first, without a protective safety pin.
Ninth Race: Seabiscuit Claiming Stakes, 1 1/2 miles on the grass,

for 3 year olds and up, purse $100,000 added.
Even though this is a claimer, with claiming prices of $500,000 and
§1,000,000, there are no ringers in the race.
An early scan of the toteboard gives me possible place and show bets on
the favorite, number 10. Five minutes before post time I make the underlays
35% to place and 50% to show. Maybe he's another double dipper? Bill's
tracking the tote on his gozinto. As the horses are milling before the starting
gate, we walk up to bet both to place and to show. I'm approaching the
window, when Dr. Z shouts to me to come back. Starting at the vantage
point, he tells me the system bet to place has disappeared, and the optimum
show bet is $219. I put the extra bills back in my pocket and bet $200 to
show. This is the day's final race, so Dr. Z bets exactly $219.
Our horse starts from the outside post position, runs at the back of the
pack for the first half of the race, and then begins to chip away at the field
to finish second by a length. He pays $3.00 to show, and $4.00 to place, had
we made the place bet.
We cash the last tickets on our way to the parking lot. Dr. Z is going
North. Ed, Jeff and I are headed South. Ed and his street atlas route us
through a new set of one way Inglewood alleys, getting us to the freeway at
least two times faster than by driving the rhumb line.
Two days later, before going to the airport, I stopped in to be tested for
Jeopardy! and got a passing grade. When the contestant coordinator said
they received a late cancellation for the next day's taping, I postponed my
flight and stayed over. I postponed it again each of the following days. Just
to show that there are some payoffs for being a dilettante, I became an
undefeated five-day champion. For me it was that once in a lifetime
experience, winning a year's salary ($43,398) on a TV game show.
Assuming he had a typical ho hum time at the office, Ed made more money
those few days than I.
Bruce Fauman August, 1995.
Postscript by Ziemba
My bankroll was $1500 and the track payback at Hollywood Park was
0.8467. I used my calculator and a 1.10 cutoff with the proviso that for bets
with expected values above 1.02 (pretty well the breakeven point) but
below 1.10 I would bet $100 for the fun of it, if I liked the horse.
The last race of the day often has a Dr. Z system bet, since by then most
bettors are looking for a good way to get even, not for a favorite to show.
Lady's Secret was such a bet on November 9 and the pools evolved as
follows: With three minutes to go the pools were
The place bet completely fizzled in the last minute. A reminder not to bet
too early! I bet the $129 to show. On November 9, Lady's Secret won the
race and paid $3.00 to show so I made $64.50 profit. My final bankroll was
$1939.50 for a profit on the day of $439.50.
At post time the toteboard was
The Breeders' Cup was conceived by John Gaines of Gainesway farm on

the Paris Pike near Lexington, Kentucky. Its purpose is to promote racing at
the highest level both through bonus additions to purses at various race
tracks across North America and through a major culmination day of
racing. Breeders' Cup Day features seven races of which five have purses of
$1 million and the others have purses of $2 and $3 million. The day is
meant to bring together a tremendous collection of the top horses. With a
four hour TV special it's like a World Series of racing. The Breeders' Cup
program is paid for by stallion season donations. For a horse to be eligible
for Breeders' Cup races, the owner of the horse's sire must donate the
equivalent of one breeding season that year, be it worth $500,000 or $1,000.
Thus to have a chance to collect the large purses and fame, breeders have to
donate. It's a brilliant idea for breeding and racing and so far it has been
quite successful. In 1984 the purses were $1 million for each of the first
races with the 11 mile grass race for $2 million and $3 million for the
classic in 1984 a total of $10 million. In 2011 there are now 15 races over
two days with the classic at $5 million and some $2 and $1 million purses.
The inaugural Breeders' Cup Day was held on November 10, 1984 at
Hollywood Park. Ed Thorp and Bruce Fauman joined me. I pocketed my
earlier winnings and again started with a bankroll of $1500. Again I would
bet $100 on horses with expected values above 1.02 but below 1.10 and use
1.10 as my cutoff for Dr. Z system bets. The first race was the Breeders'
Cup Juvenile and featured the top two year old colts and geldings running
over a mile. The favorite was Chief's Crown. Near post time the toteboard
was as follows:
I bet $100 to show on Chief's Crown. He won the race and paid $2.40 so
I made $20 profit and my bankroll was $1520.
The third race was the Breeders' Cup sprint over 6 furlongs. The feeling
was that Eillo would dominate if he did not break down. This son of
Mr.Prospector had bandages on all four legs and ran in a very dangerous
style. He was a Dr. Z system bet both to place and show.
With three minutes to go
With one minute to go
The optimal bets of $223 to place and $423 to show were each based on
making only one bet. Since both wagers were Dr. Z system bets.it would be
too risky and overbetting to make both these wagers at these levels. There
are formulas in Ziemba and Hausch (1987) that allow you to compute the
optimal full Kelly bets which were $351 to show and $84 to place.8
However, I did not have time to make these calculations, keep track of the
mutuel pools for the Ziemba and Hausch (1986) book and still make the
bet. So as a conservative approximation I simply halved the suggested
wagers and bet $110 to place and $215 to show.
Eillo won the race leading wire to wire just nipping the charging
Commemorate at the finish. He paid $4.60 to win, an excellent $3.80 to
place and a respectable $2.80 to show. I made $99 on my place bet and $86
on my show bet for a profit of $185 on the race. My bankroll was now
$1605. The final toteboard was as follws:
The final Breeders' Cup race was for a purse of $3 million. The classic
was for three year olds and upwards over 1 1/4 miles. The choice was Slew
O'Gold. A win here would probably have sewed up horse of the year honors
over John Henry. Slew O'Gold was part of an entry with Mugatea. With one
minute to go the toteboard was:
Clearly it was time to load up! The edge for place and show is good but
the safety of a 3-5 place horse and even more so a 3-5 show horse plus
having an entry leads to a very big Kelly bet. With the huge pools you do
not influence the odds much at all. Betting for place by itself indicated a bet
of $613 and for show $951. When you consider the effect of both bets using
the formulas, the optimal bets become $367 for place and $780 for show.
These are gigantic bets with my fortune of $1785; but recall that is a Kelly
property when the chance of losing is small. I bet $250 to place and $500 to
show - roughly the 1/2, 1/2 idea I used before.
The race was a classic with Wild Again at the rail, Gate Dancer charging
on the outside and Slew O'Gold sandwiched in the middle with Angel
Cordero Jr. attempting to bring him in. There was a tremendous amount of
bumping among these three horses. In the end Wild Again won the race
followed by Gate Dancer and Slew O'Gold. Fortunately for us the stewards
took Gate Dancer down and awarded Slew O'Gold second place. Mugatea
finished last. Slew O'Gold paid $3.00 to place and only $2.20 to show
(breakage cut deeply into this payoff.) I made $125 on my place bet and
$50 on my show bet so my bankroll was now $1960. The final toteboard
and chart were as follows - so the Dr. Z bets on Slew O'Gold to place and
show were even better at post time than when I bet.
The eighth race was the Fleet Nasrullah Stakes, a $60,000 added event. A
bit of a comedown after the Breeders' Cup races but still a high class race.
The feature and late races are prime candidates for Dr. Z system bets. Late
in the day most bettors are behind and do not want to consider low paying
bets to place or show. These races also feature excellent horses. The 9th
race ws the Seabiscuit Claiming stakes, at $500,000 and up, this was no
ordinary claiming race. I bet $35 on Teddy Naturally who won the 8th race
(for a profit of $29.50) and I bet $219 to show on Late Act in the 9th. Late
Act finished second and paid $3.00. My profit was $109.50. That gave me a
final bankroll of $2094 for a nice profit of $594 on the day.
My colleagues Ed Thorp and Bruce Fauman did well also. Ed was betting
heavier than I was, using a $10,000 initial fortune that led to bets in the
range $500-750. He made $1851 on the day. He did not feel that he needed
to look at the Daily Racing Form or a program. This made a great
impression on the handicappers in the next box. Here a talented student of
betting was able to win big without knowing much about the horses while
they, the experts on handicapping, were having a rough go of it. The odds
board told the story. Bruce who made bets similar to mine made a tidy
$345.
We all had a fun time at the Breeders' Cup and it was very profitable.
Except for the mishap on Bessarabian, we had excellent luck that helped us.
On average Dr. Z system bets win about 60% of the time. The bets with
expected values between 1.02 and 1.09 on favorites have similar outcomes.
However, these bets will not on average have payoffs as good as the Dr. Z
system bets. To gain our edge of 10-20% we will have our ups and downs,
winning and losing streaks. However, if played properly, the Dr. Z system
should provide you with an upward drift in your bankroll at a rate of about
10% of the value of your wagers. I wound up losing $143 the day after
Breeders' Cup.
2The discounted probabilities come from
for α about 0.81 then one uses the in the second place position. For third one uses α2 about 0.64.
These empirical numbers vary over time and by track. This is more important for exacta pricing than
place and show because for the latter the win bias from the favorite-longshot and the second and third
biases tends to cancel. The favorite-longshot bias is the empirical observation that favorites are
underbet and longshots overbet; see graphs in Hausch and Ziemba (2008). I use these ideas in futures
options trading where there is a similar bias.
3See the graphs in Chapter 1.
4The 1984 book Beat the Racetrack, which was revised in 1987 into the book with the title Dr Z's
Beat the Racetrack, has simulated results from Exhibition Park, Aqueduct and Santa Anita plus
calculations on how the Dr. Z bets affect the odds, how many can play the system, etc.
5This was the paper Bill and Don along with Mark Rubinstein of portfolio insurance infamy
published in Management Science in 1981.
6Ed Thorp wrote a column about this in Gambling Times. A power function is estimated based on
what numbers cross the start and the time of one revolution. One then forecasts where the ball might
land. This leads to the winning system.
7Which he confirmed in the preface to Beat the Racetrack.
8The place bet is minimum (p*, 1.59p*-0.639s*) = minimum [223, 1.59(223)-0.639(423)] =min-
imum (223,84.20) = $84.20. Here p* = 223 and s* =423. The show bet is 0.907s* -0.134p*
=0.907(423)-0.134(223) =$351. So the optimal bets at one minute to post were $84 to place and $351
to show.
Bibliography
Abelson, A. (2009). Shadow over housing. Barron's, September 28.

Ackermann, C., R. McEnally, and D. Ravenscraft (1999). The performance
of hedge funds: Risk, return and incentives. Journal of Finance 54,
833–874.
Adler, L. (2010). Nearly one in four second quarter home sales a
foreclosure. Reuters, September 30.
Agarwal, V., N. D. Daniel, and N. Y. Naik (2004). Flows, performance, and
managerial incentives in hedge funds. Working paper, London Business
School.
Aliber, R. Z. (2008). Monetary turbulence and the Icelandic economy.
Working Paper.
Allianz Global Investors (2008). Funding unfunded pensions: Governance
and investments of Asian reserve funds.
Andrew, J. P., E. Stover-DeRocco, and A. Taylor (2009). The innovation
imperative in manufacturing. Boston Consulting Group.
Andrews, E. L. (2009). Bernanke worries about weak dollar. New York
Times, November 16.
Attends, B. (2010). You should have timed the market. Wall Street Journal,
September 29.
Aversa, J. (2010a). Economy loses speed in spring; more weakness ahead.
Associated Press, September 29.
Aversa, J. (2010b). Fed officials clash over efforts to aid economy.
Batista, J. (2009). Patriots' Belichick overplayed his hand. New York Times,
November 14.
Batson, A. (2009). China inc. looks homeward as us shoppers turn frugal.
The Wall Street Journal, September 20.
Becker, J. and G. Morgenson (2009). Geithner, member and overseer of
finance club. New York Times, April 27.
Benac, N. (2011). Housing danger: Slump, a liability for Obama.
Benartzi, S. and R. H. Thaler (2001). Naive diversification in defined
contribution savings plans. American Economic Review 91(1), 79–98.
Berge, K., G. Consigli, and W. T. Ziemba (2008). The predictive ability of
the bond stock earnings yield differential model. The Journal of
Portfolio Management, 34(3), 63–80.
Berkelaar, A. B., R. Kouwenberg, and G. T. Post (2004). Optimal portfolio
choice under loss aversion. Review of Economics and Statistics 86,
973–987.
Bertocchi, M., S. L. Schwartz, and W. T. Ziemba (2010). Optimizing the
Aging, Retirement and Pensions Dilemma. Wiley.
Berzon, A. (2010). Whether bonds or touchdowns, they're still a gamble:
firm engineers Vegas wedding between Wall Street and sports betting.
Wall Street Journal, February 10.
Birgeneau, R. J. (2011). Letter to UC Berkeley graduates asking for
financial contributions. September 13.
Blinder, A. and G. Hubbard (2011). Taxes and debt: Left and right dare to
agree. Yahoo Finance, September 19.
Booth, D. G. and D. Keim (2000). Is there still a January effect? In D. B.
Keim and W. T. Ziemba (eds.), Security Market Imperfections in World
Wide Equity Markets, pp. 169–178. Cambridge University Press.
Bouman, S. and B. Jacobsen (2002). The Halloween indicator, sell in May
and go away: Another puzzle. American Economic Review 92(5), 618–
1635.
Bowley, G. (2010). Lone $4.1 billion sale led to ‘flash crash’ in May. New
York Times, October 1.
Bremmer, I. (2007). Top political risks for 2007. Eurasia Group.
Brown, S. J., W. N. Goetzmann, and J. Park (2001). Careers and survival:
Competition and risk in the hedge fund and CTA industry. Journal of
Finance 56, 1869–1886.
Burton, J. (2007). Mapping the next China syndrome. Available at:
www.marketwatch.com.
Chopra, V. K. and W. T. Ziemba (1993). The effect of errors in mean,
variance and co-variance estimates on optimal portfolio choice.
Journal of Portfolio Management 19, 6–11.
Clark, R. and W. T. Ziemba (1988). Playing the turn-of-the-year effect with
index futures. Operations Research 35, 799–813.
CME (2010). CME Group statement on the joint CFTC/SEC report
regarding the events of May 6, 2010. October 1.
Cole, C. R. (2010). Unified risk theory — correlation, volatility M3 and
pineapples. Artemis Capital Management, Santa Monica, September
30.
Consigli, G., L. MacLean, Y.Zhao, and W.T.Ziemba (2009). The bond-stock
yield differential as a risk indicator in the financial markets. Journal of
Risk 11(3), 3–24.
Cowan, L. and R. Smith (2011). That IPO pop? Majority of 2011 US
listings are underwater. Wall Street Journal, September 15.
Craft, M. (2011). Stocks rally on support for European banks. Associated
Press, September 15.
Crimmins, C. (2010). Ireland faces ‘horrendous’ bank bill, Spain
downgraded. Reuters, September 30.
Dimson, E., P. Marsh, and M. Staunton (2009). Investment Returns
Yearbook. Credit Suisse Global.
Doeswijk, R. (2005). The optimism cycle: Sell in May. The Economist
156(2), 175200.
Donlan, T. G. (2010). More trillions owed. Barron's, October 9.
Douglass, J., O.Wu, and W. T. Ziemba (2004). Stock ownership decisions in
defined contribution pension plans. Journal of Portfolio Management
(Summer), 92100.
Duan, M. (2012). How the financial crisis in 2008 influenced China and
what was China's response to it. Presentation at ICMA Centre,
University of Reading.
Durden, T. (2010). 21st sequential weekly outflow confirms investors refuse
to be suckered into stock market. Yahoo Finance, September 29.
Durre, A. and P. Giot (2005). An international analysis of earnings, stock
prices and bond yields. Working paper, European Central Bank,
Frankfurt.
Dzhabarov, C. and W. T. Ziemba (2011). Do seasonal anomalies still work?
Journal of Portfolio Management 36(3), 93–104.
The Economist (2010). The world economy: The quest for growth. October
7.
El-Erian, M. (2008). When Markets Collide. McGraw-Hill.
Evanoff, D. D., G. G. Kaufman, and A. F. Malliaris (eds.) (2012). New
Perspectives on Asset Price Bubbles. Oxford University Press.
Fishback, D. (2008a). Index implied volatility is based on correlation and
time - it's just not magnitude! Don Fishback's Market Update, July 9.
Fishback, D. (2008b). VXO and the 20% decline in the Dow Industrials.
Don Fishback's Market Update, July.
Forsyth (2007). It's the liquidity, Larry, and it's starting to run out. Wall
Street Journal, July 24.
Forsyth, R. W. (2009). A foolish view of America's debt. Barron's,
November 18.
Forsyth, R. W. (2011). Why investors are fed up with the fed. Barron's,
September 19.
Fung, W. and D. A. Hsieh (1997). Empirical characteristics of dynamic
trading strategies: The case of hedge funds. Review of Financial
Studies 10, 275–302.
Fung, W. and D. A. Hsieh (2001). The risk in hedge fund strategies: Theory
and evidence from trend followers. Review of Financial Studies 14,
313–341.
Gergaud, O. and W. T. Ziemba (2012). Great investors: Their methods,
results and evaluation. Journal of Portfolio Management 38(4), 128–
147.
Geyer, A. and W. T. Ziemba (2008). The Innovest Austrian pension fund
planning model InnoALM. Operations Research 56(4), 797–810.
Ghosh, S., G. Bhalla, and W. T. Ziemba (2007). The January barometer,
19262005. Technical report, Sauder School of Business.
Gjerstad, S. and V. L. Smith (2009). From bubble to depression? Wall Street
Journal, April 6.
Goetzmann, W. N., J. E. Ingersoll, and S. A. Ross (2003). High-water
marks and hedge fund management contracts. Journal of Finance
58(4), 1685–1717.
Gorenstein, P. (2011). Jeffrey Miron: States are underestimating budget
crisis by trillions. Yahoo Finance, September 19.
Grant, D. (2009). Bargains available. Barron's, September 28.
Gultekin, M. N. and N. B. Gultekin (1983). Stock market seasonality:
International evidence. Journal of Financial Economics 12, 469–482.
Hale, G. (2007). Prospects for China's corporate bond market. Federal
Reserve Bank of San Francisco, March 16.
Hausch, D. B., V. Lo, and W. T. Ziemba (eds.) (2008). Efficiency of
Racetrack Betting Markets (2nd ed.). World Scientific, Singapore.
Hausch, D. B. and W. T. Ziemba (1985). Transactions costs, entries and
extent of inefficiencies in a racetrack betting model. Management
Science 31, 381–394.
Hausch, D. B. and W. T. Ziemba (eds.) (2008). Handbook of Sports and
Lottery Investments. North Holland.
Hausch, D. B., W. T. Ziemba, and M. E. Rubinstein (1981). Efficiency of
the market for racetrack betting. Management Science 27, 1435–1452.
Hensel, C. and W. T. Ziemba (1995). The January barometer. Journal of
Investing 4(Winter), 67–70.
Hensel, C. R. and W. T. Ziemba (2000a). Anticipation in the January effect
in the U.S. futures markets. In D. B. Keim and W. T. Ziemba (eds.),
Security Market Imperfections in World Wide Equity Markets, pp. 179–
202. Cambridge University Press.
Hensel, C. R. and W. T. Ziemba (2000b). How did Clinton stand up to
history? US stock market returns and presidential party affiliations. In
D. B. Keim and W. T. Ziemba (eds.), Security Market Imperfections in
World Wide Equity Markets, pp. 203–217. Cambridge University Press.
Herbst-Bayliss, S. (2007). Failed Amaranth trader woos investors for new
fund. Reuters, March 23.
Hirsch, J. A. and Y. Hirsch (2012). Stock Traders Almanac. Wiley.
Hirsch, Y. (1986). Don't Sell Stocks on Monday. Facts on File Publications,
NY.
Jen, S. (2007a). CHF mortgage loans ex-Switzerland are small. Morgan
Stanley Global Economic Forum, March 30, available at:
http://www.morganstanley.com/views/gef/archive/2007/20070330-
Fri.html\#anchor4655.
Jen, S. (2007b). How big could Sovereign Wealth Funds be by 2015?
Morgan Stanley. London, May 4.
Jen, S. (2008). Currencies: How much assets could SWFs farm out?
Morgan Stanley. London, January 10.
Johnson, S. (2009). The quite coup. The Atlantic, May.
Jordans, F. and G. Katz (2011). Rogue trader suspected in $2 billion loss at
UBS. Associated Press, September 15.
Jovanovic, M. and T. Zimmermann (2010). Stock market uncertainty and
monetary policy reaction functions of the Federal Reserve Bank. The
B.E. Journal of Macroeconomics 10(1), Article 21.
Kahn, M. (2010). Strong stocks or debilitated dollar? Barron's, September
29.
Kahneman, D. (2012). Thinking Fast and Slow. Allen Lane.
Kahneman, D. and A. Tversky (1979). Prospect theory: An analysis of
decisions under risk. Econometrica 47(2), 263–92.
Kallberg, J. G. and W. T. Ziemba (1983). Comparison of alternative utility
functions in portfolio selection problems. Management Science 29(11),
1257–1276.
Kamrany, N. and M. D. Intriligator (2011). A new approach to US
economic policy. Available at: www.huffingtonpost.com/nake-m-
kamrany/economic- policy b 955597, September 9.
Kasapis, M., S. M. Ayas, S. Matoli, S. Tsaggaridou, and M. Lamprianidou
(2012). Unemployment in Cyprus: Causes and factors. Presentation at
University of Cyprus.
Keilman, J. (2010). Secret millionaire donates fortune to Lake Forest
College. chicagotribune.com, March 5.
Keim, D. (1983). Size related anomalies and stock return seasonality:
Further empirical evidence. Journal of Financial Economics 12, 3–32.
Keim, D. B. and W. T. Ziemba (eds.). Security Market Imperfections in
World Wide Equity Markets. Cambridge University Press.
King, P. (2009). No matter which way you dissect it, Bill Belichick made
the wrong decision. SI.com, November 16.
Koivu, M., T. Pennanen, and W. T. Ziemba (2005). Cointegration analysis
of the Fed model. Finance Research Letters 2, 248–259.
Kouwenberg, R. and W. T. Ziemba (2007). Incentives and risk taking in
hedge funds. Journal of Banking and Finance 31, 3291–3310.
Kruger, D. (2011). Treasuries gain on Greece crisis; two-year yields drop to
record. Bloomberg, September 19.
Krugman, P. (2009). The big spender. New York Times, November 20.
Laing, J. R. (2009). Is the sun setting on Japan? Barron's, September 28.
Lane, D. and W. Ziemba (2008). Arbitrage and risk arbitrage in team Jai
Alai. In D. B. Hausch and W. T. Ziemba (eds.), Handbook of Sports
and Lottery Markets, pp. 253–271. North Holland.
Lewis, M. (2003). The Art of Winning an Unfair Game. Norton.
Liberto, J. (2011). Boehner: No tax hikes for super committee. CNN
Money, September 15.
Linden., M. (2011). The myth of lower marginal tax rates. Centre for
American Progress.
Linden, M. (2011). Rich people's taxes have little to do with job creation.
Centre for American Progress.
Lleo, S. and W. T. Ziemba (2012). Stock market crashes in 2007–2009:
Were we able to predict them? Quantitative Finance 12(8), 1161–1187.
Lynch, D. J. (2009). G-20 takes first steps to change. USA Today,
September 27.
Lyons, G. (2007). State capitalism: The rise of sovereign wealth funds.
Testimony before the US Senate Banking Committee on November 13.
MacLean, L., G. Consigli, Y.Zhao, and W.T.Ziemba (2008). Risk indicators
in equity markets. Working paper, Dalhousie University, July 3.
MacLean, L., Y. Zhao, G.Consigli, and W.T.Ziemba (2008). Estimating
parameters in a pricing model with state dependent shocks. In P.
Pardalos, M.Doumpos, and C. Zopounidis (eds.), The Handbook of
Financial Engineering. Springer- Verlag.
MacLean, L., Y. Zhao, and W. T. Ziemba (2009). Capital growth with
security and convex downside risk penalties. Working paper, Dalhousie
University.
MacLean, L., Y. Zhao, and W. T. Ziemba (2012). Optimal capital growth
with convex loss penalties. Working paper, Dalhousie University.
MacLean, L. C., E. O. Thorp, Zhao, and W. T. Ziemba (2011). How does
the Fortune's Formula-Kelly capital growth model perform? Journal of
Portfolio Management 37(4), 96–111.
MacLean, L. C., E. O. Thorp, and W. T. Ziemba (2010). The Kelly Capital
Gorwth Criterion: Theory and Practice. World Scientific, Singapore,
Hardback; 2012 revised paperback.
Mamudi, S. (2010). Beaten by Buffett. Market Watch, March 5.
Mauldin, J. (2008). $1.6 trillion and counting. e-letter, Frontline Weekly
Newsletter, July 11.
Mauldin, J. (2009a). If this is recovery? Frontline Thoughts, November 13.
Mauldin, J. (2009b). This time it's different. Thoughts from the Frontline
Weekly Newsletter, June 19.
Mauldin, J. (2009c). Where the wild things are. Frontline Thoughts,
November 20.
McCormick, L. C. (2010). Gross says Druckenmiller's exit marks 'old
normal' end in capital markets. Bloomberg, September 28.
Mehta, S. et al (2012). Opportunities and threats of investing in South
Korea, India and Russia. Presentation at ICMA Centre, University of
Reading.
Menegatti, C., M. Pineda, and R. Ziemba (2008). Global stag-deflation in
sight. RGE Analysts' Economonitor, (October 30),
www.rgemonitor.com.
Meulbroek, L. (2002). Company stock in pension plans: How costly is it?
Working Paper, Harvard Business School.
Mitchell, M. and T. Pulvino (2001). Characteristics of risk and return in risk
arbitrage. Journal of Finance 56, 2135–2175.
Mitchell, O. S. and S. P. Utkus (2002). Company stock and retirement plan
diversification. Working Paper, Wharton School.
Moffatt, C. (2006). Wallstreet Week “hall of fame” inductee introduced to
the George investment program class. Available at:
thegeorgeinvestmentsview.
Moulson, G. and N. Paphitis (2011). Greece, Europe struggle to contain
debt crisis. Associated Press, September 13.
Munnell, A. H. and M. Soto (2007). Why are companies freezing their
pensions? CRR Working Paper No. 2007-22.
Newman., R. (2011). Why big companies are axing jobs. US News,
September 16.
Patelis, A. (2007). The overflowing bathtub, the running tap and SWFs.
Merrill Lynch Economic Analysis, October 6. New York.
Pendergraft, P. (2008). What's wrong with the VIX? volatility index
behaving oddly. Investors' Daily Edge, available at: HoweStreet.com,
July 11.
Piovano, C. and D. Wagner (2011). Dollar access no long-term fix for
Europe's crisis. Associated Press, September 15.
Posnanski, J. (2009). Why Belichick made the right decision. SI.com,
November 16.
Poundstone, W. (2005). Fortune's Formula: The Untold Story of the
Scientific System that Beat the Casinos and Wall Street. Hill and Wang,
New York.
Quinn, B. (2010). Iceland financial crisis: voters reject debt repayment plan.
Christian Science Monitor, March 7.
Rajan, R. G. and A. Seru (2008). The failure of models that predict failure.
Graduate School of Business, University of Chicago working paper.
Reeves, J. (2012). Harvard proves (again) that active management sucks.
Investor Place.
Reinhart, C. M. and K. Rogoff (2010). This Time is Different. Princeton
University Press.
Rendon, J. and W. T. Ziemba (2007). Is the January effect still alive in the
futures markets? Finanzmarket and Portfolio Management 21, 381–
396.
Ritter, J. R. (1996). How I helped make Fischer Black wealthier. Financial
Man-agement 25(4), 104–107.
Rogers, J. (2003). Adventure Capitalist. Wiley.
Romer, D. (2006). Do firms maximize? Evidence from professional
football. Journal of Political Economy 114(2), 340–365.
Rosen, K. (2008). Misreading the VIX. Barrons.com, July 8.
Rosenberg, B., K. Reid, and R. Lanstein (1985). Persuasive evidence of
market inefficiency. The Journal of Portfolio Management 11(3), 9–16.
Roubini, N. (2006). Bob Shiller is sharply shrill...and the risks of a housing-
led systemic financial crisis. Available at:
http://www.rgemonitor.com/blog/ roubini/archive/2006–08/.
Roubini, N. (2008). Insolvency of the Fannie and Freddie predicted here
two years ago. what happens next? or how to avoid the Mother of all
bailouts. RGE- Moniter.com, July 11.
Roubini, N. (2009). A tale of two American economies. Globe and Mail,
November 17.
Rozanov, A. (2007). Sovereign Wealth Funds: Defining liabilities. State
Street Global Advisors.
Rozeff, M. and W. Kinney (1976). Capital market seasonality: The case of
stock returns. Journal of Financial Economics 3, 379–402.
Rugaber, C. S. (2011). Inflation squeezing consumers in weak economy.
Rugaber, C. S. and D. Carpenter. (2011). Household wealth dipped in the
spring. Associated Press, September 16.
Ruoen, R. (2007). Private correspondence.
Samuel, H. (2010). Rogue trader ordered to repay five billion euros. Dialy
Telegraph, October 6.
Sears, S. M. (2010). The good sense in hedging gold. Barron's, October 6.
Setser, B. (2008). What to do with over a half a trillion a year?
Understanding the changes in the management of China's foreign
assets. RGE Monitor. New York, January.
Setser, B. and C. Rosenblatt (2006). China reserve watch: Q3 2006. Roubini
Global Economics, available at: www.rgemonitor.com.
Setser, B. and R. Ziemba (2008). Understanding the new financial
superpower - the management of GCC official foreign assets. RGE
Monitor. New York, January.
Setser, B. and R. Ziemba (2009). GCC Sovereign Funds: A reversal of
fortune. Working Paper, January 2009. Council on Foreign Relations,
available at www.cfr.org.
Shanghai Ranking Consultancy (2011). Academic ranking of world
universities. Available at:
www.shanghairanking.com/ARWU2011.html.
Shiller, R. (2005). Irrational Exuberance. Princeton University Press.
Siegel, J. (2002). Stocks for the Long Run: The Definitive Guide to
Financial Market Returns and Long-Term Investment Strategies.
McGraw-Hill.
Siegel, J. (2008). Stocks for the Long Run (4th ed.). McGraw-Hill.
Siegmann, A. and A. Lucas (2002). Explaining hedge fund investment
styles by loss aversion. Working paper, Vrije Universiteit Amsterdam.
Simon, R. and H. R. Hagerty (2009). House price drops leave more
underwater. Wall Street Journal, May 6.
Smith, Y. (2009). China lambastes dollar carry trade, diverting attention
from its currency manipulation. RGE Asia EconoMonitor, November
16.
Snowberg, E. and J. Wolfers (2008). Examining explanations of a market
anomaly: preferences or perceptions? In D. B. Hausch and W. T.
Ziemba (eds.), Handbook of Sports and Lottery Markets, pp. 103–136.
North Holland.
Stone, D. and W. T. Ziemba (1993). Land and stock prices in Japan. Journal
of Economic Perspectives, 7(3), 149–165.
Stovall, S. (2005). As goes January, so goes the year? Business Week
Online. Avail-able at: www.businessweek.com
Swensen, D. F. (2000). Unconventional Success: A Fundamental Approach
to Personal Investment. Free Press.
Task, A. (2010a). Bernake pushing on a string- Fed can't reflate the 'Real
Economy', Harvard professor says. Yahoo, October 6.
Task, A. (2010b). Great recession pushes gap between rich and poor to
record levels. Yahoo, October 1.
Thaler, R. H. and W. T. Ziemba (1988). Anomalies: Parimutuel betting
markets: Racetracks and lotteries. Journal of Economic Perspectives 2,
161–174.
Thomas, Jr., L. (2011). Greece near the precipice, raising fear. New York
Times, September 19.
Tompkins, R., W. T. Ziemba, and S. Hodges (2008). The favorite-longshot
bias in the S&P500 and FTSE100 index futures options: the return to
bets and the cost of insurance. In D. B. Hausch and W. T. Ziemba
(eds.), Handbook of Sports and Lottery Markets, pp. 161–180. North
Holland.
Vincent, A. (2007). China's stocks tumble. BNP Paribas, March 2.
Walcoff, M. (2009). Stocks soaring for games sponsors. The Vancouver
Sun.
Wang, W. (2012). Real estate mania in China. Presentation at ICMA Centre,
Uni-versity of Reading.
Ward, A. (2010). Iceland rejects Icesave repayment deal. Financial Times,
March 6.
Yale Endowment Committee (2012). Annual report.
Young, A. (2007). Volatility: Too low for too long? TD Ameritrade, January
10.
Ziemba, R. (2007a). Responses to Sovereign Wealth Funds: Are 'draconian'
measures on the way? RGE Monitor. New York, November.
iemba, R. (2009a). Is a draconian response on the way? Roubini Global
Economics.
Ziemba, R. (2009b). Tapping the sovereign wealth cookie jar. Roubini
Global Eco-nomics.
Ziemba, R. (2008a). The price of oil. Available at:
http://video.aol.com/video- detail/oil-marches-back-to-
100/698563074/?icid=VIDURVBUS11.
Ziemba, R. and W. T. Ziemba (2007). Scenarios for risk management and
global investment strategies. Wiley.
Ziemba, W. T. (1994a). Investing in the turn of the year effect in the futures
mar-kets. Interfaces 24(3), 46–61.
Ziemba, W. T. (1994b). Worldwide security market regularities. European
Journal of Operational Research 74, 198–229.
Ziemba, W. T. (2003). The Stochastic Programming Approach to Asset
Liability and Wealth Management. AIMR, Charlottesville, VA.
Ziemba, W. T. (2005). The symmetric downside risk Sharpe ratio and the
evaluation of great investors and speculators. Journal of Portfolio
Management, 108–122.
Ziemba, W. T. (2007b). The Russell-Yasuda Kasai InnoALM and related
models for pensions, insurance companies and high net worth
individuals. In S. A. Zenios and W. T. Ziemba (eds.), Handbook of
Asset and Liability Management, Volume 2, Handbooks in Finance, pp.
861–962. North Holland.
Ziemba, W. T. (2008b). Efficiency of racetrack, sports and lottery betting
markets? In D. B. Hausch and W. T. Ziemba (eds.), Handbook of Sports
and Lottery Markets, pp. 183–222. North Holland.
Ziemba, W. T. (2010). Ideas in asset liability management in the tradition of
H.M. Markowitz. In J. Guerard (ed.), Essays in Honor of H.M.
Markowitz, pp. 213–258. Springer.
Ziemba, W. T. (2011). Investing in the turn-of-the-year. Finanzmarket and
Portfolio Management 25(4), 455–472.
Ziemba, W. T. (2012a). Calendar Anomalies and Arbitrage. World
Scientific, Singapore.
Ziemba, W. T. (2012b). Response to Paul A. Samuelson letters and papers
on the Kelly capital growth investment strategy. Working Paper.
Ziemba, W. T. (2013). Portfolio optimization. In A. Bell, C. Brooks, and M.
Protopczuk (eds.), Handbook of Research Methods and Applications in
Empirical Finance. Edward Elgar.
Ziemba, W. T. and D. B. Hausch (1984). Beat the Racetrack. Harcourt,
Brace and Jovanovich.
Ziemba, W. T. and D. B. Hausch (1986). Betting at the Racetrack. Dr Z
Investments, San Luis Obispo, CA.
Ziemba, W. T. and D. B. Hausch (1987). Dr Z's Beat the Racetrack.
William Morrow.
Ziemba, W. T. and S. L. Schwartz (1991). Invest Japan. Probus, Chicago.
Ziemba, W. T. and S. L. Schwartz (1992). Power Japan. Probus, Chicago.
About the Authors
Rachel E. S. Ziemba is Director of Emerging

Europe, Middle East and Africa Global
Macroeconomics at Roubini Global Economics
where she leads much of the organization's global
scenario work from its London office. She holds a
BA in History (Hounors) from the University of
Chicago and an MPhil in International Relations
from St. Antony's College, Oxford University
where her thesis concerned on the politics of
dollarization in developing economies. She has
traveled widely and has worked for the Canadian
International Development Agency (CIDA) in
Egypt, and the International Development Research Centre (IDRC) in
Ottawa, Canada before joining RGE. Her current work focuses on global
trade and investment flows with a focus on the management of sovereign
assets, as well as mapping how the transitions in the global political scene
affect the global economy through the resource market. She has written and
given talks on a wide range of macroeconomic, political and developmental
themes at a range of academic, financial sector and policy (government and
think tanks). She is frequently represented in financial media outlets such as
Bloomberg, CNBC, Financial Times.
William T. Ziemba is the Alumni Professor of Financial Modeling and

Stochastic Optimization (Emeritus) at the Sauder School of Business,
University of British Columbia, Vancouver, Canada and Professor of
Finance at the ICMA Centre of the University of Reading. He is a well
published and known academic around the world with books, research
articles and talks on a wide range of topics related to investments and other
areas. He trades and consults through William T
Ziemba Investment Management Inc which
manages private accounts. He has held visiting
professorships at MIT, Chicago, Berkeley, UCLA
and Cambridge, LSE, and Oxford in the UK. He
has or currently consults for Buchanan Partners
and RAB Capital in London, the Frank Russell
Company, Morgan Stanley, Canyon Capital
Advisors, Credit Swiss First Boston,Ketchum
Trading and Edward O Thorp and Associates in
the US, Yamaichi Research Institute in Japan, Siemens Innovest and
Gruppo Uni Credit in Europe and Market Research in Nassau. A list of his
books and other activities and publications is on the website
www.williamtziemba.com.
Index
A
absolute return 113-16, 118
Abu Dhabi Investment Authority see ADIA
accuracy 155, 191, 193-5, 231
negative monthly barometer 193-4
ADIA (Abu Dhabi Investment Authority) 85-7, 89-93
allocation 32, 95-6, 105-6, 110, 113-14, 118, 156
Amaranth 28, 296, 353-65
anomalies 169, 171-2, 186-7, 192-3, 215, 222, 324, 327, 329, 348, 425, 528
anticipation 176-80, 182-3, 187, 363, 385
arbitrage 3-5, 7, 9, 11, 125, 213, 390-1, 425, 485-6, 488-90, 492, 494, 499,
523
Asia 86, 99, 91, 95-7, 109, 270, 293, 301 379, 381, 393, 407-9, 411-15,
418, 451
asset allocation 79, 87, 95-6, 101-5, 107, 110, 114-16, 144, 258
and governance issues 103-111
asset classes 14, 77-8, 90-2, 95, 99, 103-4, 157, 248, 302-3, 309, 313, 391,
428, 430
various 85-6, 303, 313
asset-liability management 104, 118, 166, 168
asset weights 154, 159, 161
assets 64-7, 75, 79-80, 89-92, 95-6, 103-10, 151-3, 155-8, 163, 165-6, 246-
8, 301-2, 311, 316, 350-1
alternative 95-6, 99, 101, 103, 107-8, 110
foreign 88, 93, 104, 391
real 106, 115-16, 118
risk free 301
toxic 303, 313-14, 379-80
Atlanta Falcons 474, 479-81, 487-8
average hedge funds 35, 37-9, 41, 43, 45
Average Monthly Small- and Large-Cap
Stock 219-20
average return 36, 47-8, 50-1, 53, 155, 164, 220
average turnover 31-2
B
Baltimore 213, 474, 478-81, 483, 487-8
Bank of China 388-9
banks 85, 235, 239-40, 243-4, 250-2, 266, 320-2, 333-4, 342, 369-73, 382-
3, 389, 398-9, 426-7, 435-6
largest 273, 275, 442
barometer 188, 191-203, 373, 377, 471
barometer signal 192-3, 384
bearish 193-4, 369-70, 379-80, 434
Berkshire Hathaway 25, 61-4, 245, 304-6, 330, 361-2, 375-7
Betfair 4-5, 8, 12, 213, 254, 375, 378, 443-4, 465, 471, 473, 478-9, 483,
489-91, 493-4
bets 5-8, 127-30, 132, 138, 141-3, 360-5, 465-6, 471-3, 477-80, 484-6, 488-
91, 495-6, 515-16, 520-3, 525-37
optimal 128, 524, 535-6
Bhalla and Ziemba 192-5
blackjack 127, 129-31, 134, 422, 519, 523
Bond Stock Earnings Yield Difference crash model (see BSEYD)
Bond Yield 276-81
bonds 15, 32, 85-7, 95-6, 114-16, 144-7, 157-9, 162-4, 167, 213, 220-2,
258-60, 330-1, 338, 379-80
long 264, 274, 289, 298-9, 313-14, 336
borrowers 42, 252, 311, 319, 369, 371, 383
BP Statistical Review 413-14
Breeders' Cup 4, 323-4, 379, 493, 495-6, 499, 501, 509-10, 518, 520-1, 531,
533-4, 536-7
Breeders' Cup Classic 4, 142, 494, 499, 501, 508
BRIC 455-6
BSEYD 13-23, 191, 235-6, 242, 260-8, 270, 272,5, 280, 282-4, 285-7, 289-
91, 299, 313, 349, 352, 372-5, 440
bubbles 98, 208, 237, 241-2, 299, 311-12, 316, 319, 334, 336, 367, 379,
381-3, 385, 398
bullish 193-5, 299, 369-70, 385
C
CAD (current account deficit) 252, 332, 397
California 4, 8, 10, 108, 121, 309, 317, 382, 384, 427, 431, 433, 446, 499,
508-9
Canada 14-15, 17-19, 21, 89, 102-4, 108, 110, 121, 123-4,, 126, 132-7, 239-
40, 289-90, 319-20, 359, 433, 451, 478
Canadian banks 42, 240, 322, 372
Cantelli's inequality, using 275-81
capital 76, 88, 93, 96, 99, 101-3, 106-8, 110, 145, 321-2, 334-5, 339, 361,
411, 442
human 78-9, 81, 449-50
capital growth theory 127, 171, 175
CAR (cumulative abnormal returns) 329
cash 14-15, 76-7, 95-6, 114-16, 128, 141, 144, 146-7, 172, 175, 177, 179,
181, 185-6, 200, 219-22, 224-5, 379-80
CBOE volatility index see VIX CC (Commodity Corporation) 62-3
central banks 86, 90, 96, 251, 299, 301, 321, 373, 389, 392, 396, 398-9, 442
CFTC (Commodity Futures Trading
Commission) 66, 365
chance 5, 99, 118, 127-9, 132-4, 136, 138, 142-3, 167, 176, 191, 377, 433,
466-8, 485
Chicago 36, 243, 439, 474, 479-81
China 11, 87-90, 96-9, 257, 266-7, 286-7, 292-3, 295-9, 301-2, 332-5, 372-
5, 387-9, 391-4, 449-51, 457
Chinese banks 388-9
Chinese equities 299, 333
Chinese holdings of US assets 388-9
Chinese Investment Company see CIC
Chinese investment markets 37, 293-4, 296-9, 301, 313, 439
Chopra-Ziemba 30-1
CIC (Chinese Investment Company) 88-9, 93
Cleveland 240, 474, 481
Clinton 12, 213, 216, 222, 224, 375, 378, 443
CMOs 252, 316-18, 371-2
commissions 3, 5, 26, 35-6, 93, 171-2, 176, 425
Commodity Corporation see CC Commodity Futures Trading Commission
see CFTC commodity pool operators see CPO
Commodity Trading Advisors see CTA
company stock 75-8, 80-2
complacency 229, 231-2
complex markets 129, 244, 428, 515
confidence 143, 194-5, 264-8, 276-81, 285, 287, 290, 303, 313-14, 330,
342, 439, 445, 531
high 134-6, 138
level 49, 207, 260-1, 265-6, 285, 287
Confidence Interval Stock Price 276-8
Congress 221-3, 235, 250, 375, 377, 411-12, 436, 443-4, 446
Consigli and Ziemba 14-15, 259, 263, 289-90
constraints 80-1, 128, 146-7, 151, 153, 156-7, 159-62, 165-7, 363, 454
consumption 98, 319-20, 393-4, 396, 402, 416, 438
contracts 117, 175, 177, 180, 182-3, 339, 356, 358-9, 370-1, 424, 435
control 42, 47-50, 96-7, 147, 230-1, 313-15, 324, 331, 371, 381, 390
corrections 15, 18-19, 21, 23, 45, 259, 263, 294, 420, 429
correlations 144, 152, 157-60, 193-5, 213, 227, 250, 325, 339-40, 354, 369,
371, 430, 446
mixing 158-9, 161-2
CPO (Commodity Pool Operators) 65, 67-8
crash indicators 279, 285, 287
crash signal 247, 260-1, 264-5, 267-9, 270, 272-3, 275-81, 285-7
faint crash signal 278, 281
first crash signal 267, 269, 276-81
crashes 13-14, 235-6, 246-7, 257-8, 264-5, 270-2, 274-5, 277, 279-80, 285,
287, 289-92, 311-15, 346-7, 373
flash 423-4, 430
market crashes 91, 229, 276, 279, 292
credit 98-9, 243, 272, 303, 312-15, 319, 321, 332-3, 369, 383-4, 389, 398,
442, 478, 502
credit markets 85, 388
CTA 65, 68
cumulative abnormal returns (CAR) 329
currencies 60, 71, 96-8, 124-5, 250-2, 269-71, 297, 338-9, 344-5, 369, 374-
5, 382, 393, 396-7, 434
current account deficit see CAD Cyprus 251, 416-17, 449, 460-2
D
danger zone signal 13-15, 235-6, 246-7, 249, 258-9, 262-5, 268-9, 272,
274-5, 280, 283, 289-91, 299-300, 313-15, 345-7, 372-4
database 17, 45-50, 53
day's range 230-1, 295, 314-15, 324, 331, 381, 434
debt 98, 121, 132, 238, 247, 303-4, 312, 316, 319-20, 326-7, 382, 421-2,
430, 435-6, 440-2
demand growth 408-9, 414-15
Democratic administrations 214-16, 218, 220-1, 223-5
Democratic and Republican
administrations 215-17, 220-1, 224
Democrats 9, 213, 215-25, 250, 252, 375, 377, 435, 442-3, 445
Denver 213, 240, 474, 481, 483, 486-8, 491-2
derivatives 87, 128, 243-5, 313-14, 337-9, 341, 343, 345, 347, 349, 351,
353, 355, 359, 375-6
how to lose money in 339-365
Dimson 31, 76, 145, 157, 159, 162, 307-8, 310, 384-5
dollar share 90, 392
dollars 8, 57-8, 64, 75, 90, 121, 134-6, 242-3, 250-1, 271, 301-2, 314-15,
382-3, 391-2, 520
Douglass, Wu and Ziemba (DWZ) 76-8, 81
Dow Compare 230-1, 292, 295, 314-15, 324, 331, 381
Dow Jones 229, 233-4, 312, 423, 428
DSSR (Downside Symmetric Sharpe Ratio) 31, 38, 47-53, 55-60, 62-9,
119, 245, 283, 346, 352, 362, 423, 428
DWZ see Douglass, Wu and Ziemba
Dzhabarov and Ziemba 174, 183, 185, 188-90, 200, 205-6, 210-11, 327-8,
425, 426
E
earnings 13-14, 243, 246, 249, 257, 262-3, 272, 281, 283-4, 291, 296, 313-
15, 318, 347, 440
bond stock 235, 257, 262, 273-4, 287, 292, 303, 313, 377, 440
huge 45, 421
ECB 251, 373, 396, 435-6, 442
economists 321-2, 391-2, 395-7, 405, 433, 442
economy, weak 114, 247-8, 280, 375-6, 422, 433, 438
edge 3, 25, 29, 56-7, 121, 128, 130-1, 133, 135-6, 140, 142-3, 338, 349,
515-16, 536-7
education 99, 117, 393, 433, 446, 450, 453, 462
election 9, 12, 214, 216, 220-1, 253-4, 442-3
Elo Ranking System 473-8, 486-8, 490
EM (emerging markets) 74, 87, 96-8, 101-2, 108, 113, 147-8, 291, 352,
395, 401, 408, 430, 452-3, 457
emerging markets see EM employees 58-60, 75-6, 78-82, 110, 149-50, 342,
440
endowments 60, 80, 99, 101, 104, 113-19, 303
equities 31-2, 95-6, 101, 104, 107-8, 116, 118, 151, 157-64, 262-3, 301-3,
307, 309, 369, 390
domestic 109, 114-16, 428
wide range of 85-7
equity market correction 258, 390-1
equity markets 15-17, 45, 91, 110, 158, 214, 245, 259, 263, 289-90, 303-4,
313, 333, 345-6, 369, 389-91
errors 29-31, 82, 140, 142, 155, 474-5
ETFs (exchange traded funds) 36, 99, 298, 311, 428
euro 36, 242-3, 247-8, 250-3, 270-1, 330, 374-6, 382, 385-6, 392, 419, 421-
2, 434, 436, 441
Europe 26, 45, 96-8, 104, 110, 159, 163, 252, 385, 393, 408-9, 411, 414-17,
419, 441-2, 494
European banks 373, 436, 441-3
European bonds 151, 157-9, 164, 441
European equities 106, 157, 159, 162, 164
evaluating superior hedge funds 55-73
excess return 18-23, 171-2, 207, 267
monthly 19, 21, 23
exchange rate volatility 124-5
exchange traded funds see ETFs expected return 10, 77-9, 81, 114, 147, 163
expected value 8, 133, 136, 517, 531, 533-5, 537
exports 96-9, 302, 392-3, 401, 411-14, 416-17, 438, 450
exposure 86-7, 90, 104, 107, 252, 338, 357, 370
F
fair value 25, 124-5, 179, 236, 263, 379, 381
favorite-longshot bias 3, 5, 7-11, 499, 516
favorites 4-5, 7-9, 129-30, 213, 252, 360, 377, 443, 478-9, 483-4, 488-9,
516-17, 527, 529-30, 532-4
extreme 7-9
fees 15, 28, 35-6, 38-9, 45-7, 49, 51-2, 57, 60, 63-4, 341-2, 356, 359, 363,
425
high 43, 45, 49
field goal 213, 465-6, 468, 473-5, 485, 490
final wealth 15, 55, 128-9, 143, 206, 213, 235, 264, 323
Finance Search 230-1, 292, 314-15, 324, 331, 381, 434
financial institutions 85-6, 90, 238-40, 282, 322, 369, 372-3, 388, 441, 445
financial markets 3, 26, 42, 45, 134-6, 191, 217, 239, 369, 379, 419, 471,
515-16, 520, 528-9
financials 85-6, 90, 92, 229-30
five days, first 196, 199, 201, 303, 313, 374, 384
fixed income 87, 103, 110, 114-16, 220-1, 349, 369
forecasts 193-4, 199, 234, 263, 283-4, 313, 315, 528
foreign exchange reserves 86, 301
foreigners 242, 268, 296-7, 299, 335
fractional Kelly strategies 27, 130, 132, 136, 143-4
France 93, 102, 104, 134, 159, 203, 310, 320, 385, 416, 422, 442, 451, 457,
509
function 39-43, 47, 52, 77, 80-1, 128, 144, 154, 158, 199, 517
fund managers 14, 35-6, 38-40, 42, 46, 51, 347, 362
fund value 39-42, 47, 51
function of 39-41
initial 39, 42
funds 14-15, 25, 35-43, 45-53, 56-60, 62-8, 85-97, 101-11, 296-8, 303-5,
341-5, 347-54, 356, 359-65, 373
fund of 38, 45-6, 49, 51-3, 68 359, 365
index 26, 35-6, 266
individual 38, 66, 68
misuse of 66, 68
national wealth 96-7, 107
new 87, 89, 105, 361, 363
worst 66, 68
futures contracts 27, 125, 132, 174, 176-7, 179-81, 185-90, 201, 210-1, 233,
235, 338, 345-6, 352, 357-8, 369-71
futures exchanges 370-1
futures markets 106, 171, 176-9, 182, 184, 187, 290, 356-8, 423, 428, 485
G
GCC (Gulf Cooperation Council) 88, 95-6, 400
GDP 109, 251, 268, 296, 304, 315, 326-7, 332-5, 374-5, 429-31, 436, 441-
2, 449-51, 454, 457
GDP growth 98-9, 122, 129, 131-2, 141-2, 146-7, 176-7, 301, 315, 335,
383, 395, 429, 449-50, 457
geopolitical risks, environment of 411-12
Gergaud and Ziemba 59, 64-8, 71-4
Germany 14-15, 17-20, 93, 159, 203, 235, 257, 263, 289-90, 320, 409, 416,
422, 431, 451
Geyer and Ziemba 32, 79, 118, 149-50, 152, 157-8, 160, 162-6, 430
Giants 213, 474, 481, 483-8, 491
GIC (Govermment Investment Corporation) 85-7, 89-90, 92, 107
GLD (gold ETF) 331, 429
Glitnir 261, 266, 272-6, 279
global economy 244, 294, 298, 385, 387-8, 391-2, 396-7, 403, 409-10, 449
go-to capital source 85, 94
gold 36, 142, 242, 248-9, 252, 310, 331, 374, 377, 428-9, 434-5, 437-8, 513
Goldikova 495, 499, 509
Goldman Sachs 45, 330-1, 355, 420-2
Govermment Investment Corporation see GIC
government bonds 18, 95, 105, 219-21, 259, 301, 303
government debt 15, 319-20, 430
government spending 96-7, 103, 405-6
governments 85, 87, 95, 98-9, 101-11, 132-4, 240, 302-4, 335, 383, 390,
397-400, 405-9, 442, 449-50
Greece 251, 384-5, 399, 419, 421-2, 440-1, 460-2
Green Bay Packers 474, 478-81, 487-8, 491
Greenspan 91, 244, 294, 318-19, 371
GSPC 231, 314, 381, 434
Gulf Cooperation Council see GCC
H
half Kelly 29, 128-9, 136, 138, 141, 143, 176-7
handicappers 494-6, 520, 527, 536
HARA (hyperbolic absolute risk aversion) 38, 41
Hausch and Ziemba 3, 8, 130, 465, 516-17
hedge fund managers 12, 38-9, 43, 52, 58-9, 62, 64, 66, 356, 362
hedge funds 15, 28, 33, 38-9, 42-3, 45-53, 59, 85-6, 104, 282-3, 341-4, 346-
8, 352-3, 356-61, 363-5
Hensel and Ziemba 177, 180, 182-4, 186, 191-2, 203, 213-17, 219, 221-4
Hirsch 191-2, 194, 197, 199, 201
hitable 493, 495, 497
holidays 303, 313, 326
Hollywood Park 515, 518, 521-2, 524, 528, 530-1, 533-4
home field advantage 468, 475-6, 478-9, 491
horses 4-6, 8, 10, 129, 493-6, 499, 502, 508-10, 515-23, 527-30, 532-6
households 299, 319-20, 332, 397, 411-12, 449-50
housing 82-3, 95, 99, 105, 130, 191, 239-41, 311-12, 314-20, 332, 334, 349,
372-3, 384, 398, 400, 414-16, 426-8, 430, 434-5, 439, 454, 475, 491-3,
519, 522, 526
housing market 282, 316, 426-7, 439
housing prices 297, 311-12, 315, 317, 372, 439, 459-60
Hunter 296, 355, 359-61, 363-4
hyperbolic absolute risk aversion see HARA
hypothesis 206-7, 214, 216-17, 224
I
Iceland 14, 235, 242, 251, 257, 260-1, 264, 266- 7, 269-72, 274, 276, 286-7,
313-15, 419, 451
ICI (Investment Company Institute) 428
incentive fees 38-43, 45-9, 51-2, 347, 363
effect of 38-9, 47, 49
higher 41, 51-2
incentives 37-9, 42, 46-7, 52, 56-7, 107, 347, 362-3, 397, 402, 414, 427
incentives and risk taking in hedge funds 45, 47, 49, 51, 53
increasing market share 398, 413
Index Futures 171-89, 370-1
index level 230-1, 240, 246, 253, 264-6, 267- 9, 273, 284-6, 292, 306, 314-
15, 324, 374-5, 381, 434
Indianapolis 465, 468, 473-4, 478-81
indicator 276-81, 292, 303, 313, 383
individual hedge funds 46, 52
inflation 76, 117, 219, 249, 303-4, 311-12, 323-4, 331, 375-6, 379-80, 393,
398-9, 422-3, 429-31, 454
information 3, 42, 94, 129, 149-50, 156, 163, 166, 191, 231, 296-7, 315,
363, 390-1, 494
initial investment 65-7, 223
initial wealth 29, 57-8, 128-30, 132, 136, 138, 145, 157, 161
InnoALM model 149-53, 155, 157, 159, 161, 163-167, 430
Innovest Austrian Pension Fund Financial Planning Model see InnoALM
model
interest rates 91, 122, 154-5, 251, 263-4, 270-2, 301-2, 313-16, 324-5, 330-
2, 338-41, 371-6, 383- 4, 430-1, 440-1
Investment Company Institute see
ICI
investment in own-company stock 75, 77, 79, 81, 83
investment strategies 39-42, 95, 105, 143, 223
good 85, 296
minimum 65-7
optimal 18, 40
investment style 42, 47-8, 50, 52
investor camps 25, 27, 29, 31
investors 27-9, 39, 45-7, 58-60, 95, 98-9, 105-7, 134-6, 202-3, 232, 295-8,
339-40, 351-4, 362-5, 390-1
foreign 294, 296, 298-9
great 3, 25-6, 28, 31, 55, 61, 119, 245, 304
Ireland 6, 102, 104, 107-8, 251, 266-7, 384- 5, 419, 421-2, 440, 451
Israel 414, 416-18, 442, 450-1
Italy 159, 241, 251, 320, 327, 341, 344, 384-5, 409, 419, 421-2, 440-2, 451,
460-1, 496
J
Japan 17-19, 235, 262-3, 289-91, 293-4, 301, 320, 335-6, 344-7, 369, 371-
5, 393-4, 405, 409, 449-51
Japanese stock market 14-15, 121-2, 126, 347, 369
job losses 309, 375-6, 383-4, 386, 439, 445, 461
K
KAIST (Korea Advanced Institute for
Science and Technology) 449-50
Kaupthing 261, 266, 273-4, 277, 279-80
Kelly Capital Growth Investing 3, 25, 27, 29-30, 127-31, 133, 135, 137-9,
139, 141, 143, 145, 147
Kelly criterion 27, 56-7, 82, 132, 138-9, 141, 343
Kelly bet, optimal 533-5
Kentucky Derby 8, 10, 129-30, 323-5, 494, 508, 517, 529
KIA (Kuwait Investment Authority) 85-7, 93, 95-6
KIC (Korean Investment Company) 89, 93
Korea Advanced Institute for Science and
Technology see KAIST Korean Investment Company see KIC
Kouwenberg and Ziemba 38-42, 45, 47-9, 51, 347, 362
kurtosis 47-8, 50-1, 53, 158
Kuwait Investment Authority see KIA
L
LAIS 273, 278, 281
Landsbanki 266, 271-5, 278-9, 281, 419
large cap stocks 171, 175, 187, 213-17, 223-5, 241
largest decile US stocks 171-2
last reported date 65-7
liabilities 10, 25, 86, 104-5, 166-7, 316
liquid assets 95-7, 99, 110, 305
Lleo and Ziemba 14, 235, 257, 260, 262, 264-70, 273-6, 279, 282-3, 285-7,
292, 314, 362, 440
Long Term Capital Management see LTCM
losers 125, 355, 363-4, 495, 521
loss aversion, implicit level of 39-40, 47, 51
losses
large 43, 147, 234-5, 337-8, 340, 342-3, 352-4, 428
yearly 57, 201
lottery games 132-4, 136, 138
lottery numbers 132-6, 138, 154-6
popular 133-6
unpopular numbers 133-6, 138
LTCM (Long Term Capital Management) 28, 338-43, 346, 354, 357, 361,
363, 383, 456
M
MacLean 15, 27, 29, 61, 127, 141, 143-4, 236, 263
MacLean, Sanegre, Zhao and Ziemba see MSZZ
MacLean and Ziemba 28, 131, 133, 135-7, 141
management fee 38-9, 45-6, 49-51, 53, 349, 360, 363
managers 14, 25-6, 35, 38-43, 45, 51, 60, 91, 113, 347-8, 363
active 26, 35-6
external 87, 90-1
Manchester Trading 350, 353
Manning 213, 465-6, 478, 483-5
March futures 179, 184-6, 187
market actors 294, 365, 395-7
market capitalization 172, 266, 298-9, 390-1, 451, 471
market conditions 92, 119, 158, 191-2, 231
market corrections 263, 289-90
market impact 26, 35-6, 171-2
market prices 25, 40, 83, 125, 416, 459
market share 411-13
market timing 35, 119, 140, 207
market value 116, 221, 239, 292, 452
current 262-3, 316
markets
efficient 26-7, 126, 471
complete 40, 273, 275
global 91, 94, 229, 231, 296, 302, 390
international 101-2, 203, 293, 451
move 85-6, 362
simple 129, 515
spot 411, 414-16
turf 520, 528
Maudlin 235-6, 283-4, 315, 384
maximum drawdown 42, 47-51, 53
mean-variance model 77-9, 167
measures, bond-stock 236, 245, 249, 251, 279, 289-90
MENA (Middle East and North Africa)
region 400, 405-7
millionaires 136, 138, 442, 446
mini crashes in US and world equity
markets 389- 301
Mitchell 42, 47, 75-6, 80, 82
mixing-correlation cases 159, 161, 164
model 13-15, 31, 149-51, 162-3, 166, 168, 191, 235-7, 257-8, 262-4, 286-7,
313-15, 340-1, 347-8, 515-16
simple 191, 354
monthly losses 27, 57, 61, 64, 66, 350, 362
Morgan Stanley 87-9, 183-4, 294, 365, 373
mortgages 42-3, 239-40, 252, 282, 303-4, 311-12, 316-20, 332, 340, 371-3,
383-4, 388-9, 427
MSCI World Index 379, 381
multistrategy fund 72-4
MSZZ (MacLean, Sanegre, Zhao and Ziemba) 143-7
mutual funds 35, 64, 66, 81, 297, 428, 430
N
Nasdaq 65, 230-1, 241-2, 292, 295, 314-15, 324, 331, 381, 428, 434
natural gas markets 296, 353-60, 365, 409, 411-12, 415, 418, 435
futures prices 355-6
net asset value 25, 49, 65, 291, 352
New England 213, 379, 465-6, 468, 474, 478-81, 483-8, 491-2
New Orleans 335, 473-6, 478-81, 487-90
New York Mercantile Exchange (NYMEX) 355, 358
New York Stock Exchange see NYSE
NFL 477, 483, 485, 487, 489, 491
Niederhoffer 28, 340, 345-6, 348-9, 352-3
Nikkei stock average see NSA
normal distribution 29-30, 79, 140, 158-9, 161-2, 193-5, 260-1
Norway 86-7, 89-90, 92-3, 102-4, 108, 159, 203, 316, 451
NSA (Nikkei stock average) 13, 121-6, 229, 262, 293, 380, 458, 471
NSA volatility 124-5
NYMEX (New York Mercantile Exchange) 355, 358
NYSE (New York Stock Exchange) 172, 221, 299, 315, 471
O
October 15-16, 194-5, 205-7, 210-11, 218, 220, 236, 247, 258-9, 282-3,
323-4, 327, 358-9, 379-80, 425-6
OECD 102-4, 109
OEX (S&P100 index) 233-4
oil exporters 88, 96-7, 393, 406-9
oil funds 88-9
oil markets 339, 360, 411
oil prices 95-7, 107, 109, 247-8, 295-6, 313-14, 354, 361, 405, 408, 414,
454, 456-7
Olympic risks 387, 389, 391, 393
OMX Iceland All-Share PR 315
optimal weight of stocks 40-1
option prices 289-90, 303, 313, 379-80, 423, 430
outperform 64, 66, 168, 171, 202-3
own-company stock 75, 77, 79, 81, 83, 419
P
p-value 47-8, 50-1, 53
pace numbers 496, 499-500, 505
PE ratios 235-6, 259, 264, 270, 272-3, 276, 280-1, 283, 285, 298, 301, 379,
440, 471
peak 4, 85, 124-5, 229, 233-4, 240, 264-5, 267, 269, 273-81, 285, 318, 333,
371-2, 429-30
peak winter demand 357-8
penalties 43, 342-3, 354, 360, 411-12, 424
pension funds 75, 91-2, 99, 101-7, 109-10, 149, 151, 213, 341, 348, 365,
437
sovereign 101, 103, 110
public pension funds 101, 103-6, 108, 110
pension reserve funds 103-4, 107
pensions 75-6, 91, 101, 103-5, 107-9, 111, 239-40, 303, 335, 444
PIIGS 385, 419, 421, 440-1
PIMCO 14-15
Pittsburgh 213, 474, 479-81, 483, 486-8, 491
place bet 6, 91, 129-30, 142, 191, 214, 341, 388-9, 391-2, 401, 406-7, 450,
502, 515-17, 519-25, 527-37
players 130, 132, 138, 413, 466, 469, 473, 477, 481, 515-16, 526
policies, economic 217, 224-5, 333, 406-7, 446
population 12, 108, 327, 393, 402, 406-7, 418, 436-7, 441-2, 453, 455, 461
portfolio managers 35-6, 119, 379, 381
portfolio weights 77-8, 81, 147, 153, 164-5
portfolios 30, 61-2, 79-80, 85-7, 92-3, 113-14, 162-3, 191-2, 224, 309, 340,
356, 360, 375-6, 434
optimal 77-9, 81, 158
Portugal 102, 104, 251, 384-5, 419, 421-2, 431, 440
positions 61, 64, 80, 96, 119, 125, 141, 175, 177, 338-9, 341-7, 354-6, 359-
60, 369-71, 490-2
similar 338-9, 342-3
predicting 191-2, 234, 272, 289-91, 303, 313-14, 433, 440
prediction signals, various 303, 313
predictive value 202, 303, 313
premiums 179-80, 232, 259, 291, 306, 342, 352, 375-7
presidential party 213, 215, 217-19, 221, 223-5
presidential terms 215, 217, 219, 221, 225
price index 240, 273, 300
prices 25-7, 64, 122-3, 171-2, 175, 194-5, 229, 239-41, 246-8, 250, 289,
354-60, 398-9, 411-12, 458-9
fair 124-5
house 43, 316, 318-19, 371
low 125, 348, 411-12
predicted 236-7
private equity 37, 85-7, 90-1, 94, 104, 113-16, 118
probability 4, 29, 79, 81, 128-9, 131-4, 136-8, 142, 145, 156, 176-7, 192-6,
259-60, 466-8, 515-16
conditional 191-4
problems 42, 76, 79-80, 82, 252-3, 338, 340-3, 361, 386, 398, 413-14, 419,
427, 437-40, 449-50
production 357, 359, 391-2, 397, 408, 411, 413-18, 429, 440
profits 3-4, 8, 42, 132-3, 179-80, 192-3, 202, 270, 321, 358, 373, 388-90,
408-9, 421, 533-6
corporate 321, 390, 445
small 7, 9, 338
purses 6, 493, 513, 518-19, 521-2, 524, 526-7, 529, 531-5
Q
Qatar Investment Authority see
QIA
QIA 87, 89, 93
Quantum Funds 305, 342-3, 356
R
races 4-10, 93, 129, 142, 323-4, 413-14, 443, 485, 493-7, 499-502, 508-9,
513, 515-17, 519-21, 525-36
classic 323-4, 508
final 501, 532
first 518-19, 534
racetrack 7-8, 436, 497, 501-2, 518-22, 524, 526, 528-9
racetrack betting systems 515, 517, 519, 521, 523, 525, 527, 529, 531, 533,
535, 537
Rachel Alexandra 497, 499, 502-5, 507-9
rally 6, 45, 125, 232, 235, 254, 282, 285, 291, 294, 298, 323, 379-81, 471,
513
random numbers 133-6
rate of return 9, 11, 113-14
rates 25, 141, 146, 157, 240, 243, 251, 269, 272, 299, 306, 373, 398-9, 445-
6, 459-60
exchange 121, 123, 301, 315, 396-7
free 42, 47, 141-2, 147
real estate 113, 115, 208, 252, 270-1, 282, 303, 313-14, 329, 334, 336, 369,
371, 379-80, 384
real price 240, 242, 248, 407
rebalancing 15, 165-6, 259, 396-7
recession 250, 291, 294, 304, 369-70, 374-6, 398, 429-30, 433-4, 446
regions 105, 317, 321, 344, 385, 397, 400-3, 405-7, 411, 413-15, 417, 419,
453, 462
regulators 85, 93-4, 99, 239-40, 365, 396-7
Rendon and Ziemba 174, 177-9, 181, 184-7
repeated investments 171-2, 175
Republican administrations 214-15, 217, 223-4
Republicans 9, 213-20, 222-5, 435, 437, 442-6
reserve funds 89, 103
resources 42, 86, 106-7, 144, 360, 383, 409, 418, 449-50
returns 9, 11, 31-2, 47-8, 51-2, 59, 64-8, 96-9, 113-16, 162, 191-5, 200-2,
290, 305-6, 357
year-end 194-5
return distributions 63, 80, 143-4, 176, 221
return measures 47-50, 53
Roubini Global Economics (RGE) 89, 333
risk 26, 38-43, 46-53, 55-8, 75-6, 124-5, 138-9, 141-2, 243-5, 319-21, 337-
41, 353-4, 375, 421-3, 473
currency 123, 297
increased 41, 47, 51, 329
war 452-3
risk-adjusted performance measures 47-9
risk arbitrage 3, 5, 7, 9, 11, 125, 213, 483-9, 491-2
risk arbitrage convergence trade 121, 123, 125
risk assets, higher 87, 99, 101
risk aversion 30, 77-80, 206, 293, 309
function of 77-8
aversion parameter 78, 81
risk control 27, 340, 349, 352-4, 361, 363, 365, 439
risk factors 26, 236-7
risk measures 42, 47, 79, 118, 346
linear convex 153-4
risk premium see RP risk profiles 86-7, 351
risk taking 45, 47, 49, 51, 53, 309
RMB currency 88, 294, 391-2, 459-60
RP (Risk premium) 3, 26-8, 206-7, 258, 263, 513
RTS Index 456-7
Russell 174-5, 178-85, 187-8, 190, 201, 205-6, 211, 224, 323, 327-8, 430,
434
Russia 86, 88-9, 96-7, 102, 107, 272, 295, 342-4, 363, 392, 413-14, 416-17,
441, 449, 455-7
Russian Government 315, 414
Russian New Europe Fund (RNE) 291, 352
S
San Francisco 36, 168, 213, 240, 317, 474, 481, 483, 487-90
sample 48, 157, 164, 166, 182, 258, 265
Santa Anita 4-6, 493-5, 497, 508-9, 515, 521
savings 90, 95-7, 101-2, 104, 106, 108-9, 271, 397, 402, 429, 441, 454
scenarios 3, 25, 79, 145-7, 150, 152, 156-7, 162, 166-8, 323, 339-40, 342-3,
352-4, 362, 372
bad 28-9, 75, 141, 167, 245, 337-8, 340-1, 344, 348, 353, 360-1, 365,
372, 474
seconds 56-7, 60, 168, 213, 466, 484-6, 490, 502, 508
Sell in May see SIM Shanghai stock exchange 264, 299-300, 332
Shanghai Stock Index 264-70, 298, 315
share price 17, 275-8, 441
shares 25-6, 37-8, 62, 75-6, 82-3, 91-2, 107, 109-10, 245, 275-8, 299, 305,
375-6, 390, 446-7
Sharpe ratio 18, 31, 47-53, 55, 57-8, 64-8, 139-40, 235, 245, 350
short term interest rates in Japan 374-5
signal threshold 268-9, 279-81
signals 7, 191-2, 197, 235-6, 260-2, 264-6, 268-9, 273-9, 285-7, 291-2, 303,
313-15, 321-3, 329-31, 383-4
SIM (Sell in May) 205-7, 209-11
Futures Sell in May 205-6
simulations 27, 29, 61, 123, 128-9, 157, 167, 194-5, 209, 339
Singapore 86-7, 95-6, 107-9, 125, 297, 332, 345, 353, 439, 450-1
skewness 42, 47-51, 53, 158
Slew O'Gold 530, 535-6
slowdown 297, 374-5, 387, 396-8, 411-12
small cap stocks 171, 173, 213, 215-17, 221, 223-4, 241
small stock advantage 171-2, 217
South Korea 102, 108, 449-53
sovereign wealth funds (SWFs) 33, 85-96, 93-5, 99, 101, 103-4, 106, 108-9,
235, 372-3, 418
S&P500 181-9, 191-8, 200-1, 216-21, 229-38, 245-7, 258-9, 280-7, 289-96,
305-6, 313-15, 323-32, 352-3, 369-77, 379-81
S&P return 194-5
Spain 102, 110, 203, 251, 384-5, 419, 421-2, 431, 438, 440-1, 451, 460-2
SSE Composite Index see Shanghai Stock Index
Standard Chartered 85-6, 94
Staunton 157, 159, 307-8, 310, 384-5
stochastic programming 76, 79-80, 167-8, 340
stock indices 76, 272, 289, 375, 430
stock market 18-23, 167, 199, 213-14, 232, 236, 252-3, 262-4, 268, 272,
374-5, 291-2, 299, 313, 422-3, 428
stock market anomalies 134-6
stock market corrections 16, 18-22
stock market crashes 14, 114, 257-281, 313-14
stock prices 27, 75, 214, 224, 237, 257, 264, 267, 282, 292, 312, 344, 347-9
stock return 270, 272, 281
stock return volatility 158, 160, 163
stocks 40-1, 75-9, 81-3, 144-7, 157-62, 172, 213-16, 229-32, 257-9, 270-1,
309, 375-7, 379-80, 420-3, 429-30
large cap 171, 175, 202, 218, 221
small cap 171-2, 175-6, 216,218, 292, 471
strategies 14-23, 35, 61, 65-7, 113, 128-30, 138, 146-7, 163, 165-6, 175,
202-3, 205-7, 223-4, 323
strategy signals 19, 21, 23
students 12, 83, 117, 134-5, 175, 330, 341, 435, 443, 449-50
summer 68, 207, 232, 282, 296, 313-14, 357-9, 386, 426
Super Bowl 213, 465, 468, 471-9, 481, 483-9, 491
super horses 499, 501, 503, 505
supply 137, 398, 406, 408, 411-12, 417-18
SWFs see sovereign wealth funds system 10, 62, 99, 129-30, 354, 515-16,
518-21, 523-4, 526-30, 537
system bets 520, 523, 525, 527-9, 531-4, 536-7
T
taxes 17, 36, 78, 82, 133-5, 167, 319, 342, 375, 377, 382, 385, 436-8, 441-
2, 444-7
teams 131, 191, 213, 233, 341, 363, 465, 467-8, 473-9, 481, 483, 486, 489,
532
Temasek 85-7, 89, 91, 94
test 16, 46-7, 50, 53, 163, 168, 174, 207, 216, 221-2, 329, 476, 515, 518,
527- 9
Thorp 15, 25-9, 61, 64, 66, 68, 82, 121-3, 127, 141-3, 147-8, 331, 515, 522-
3, 528- 9
tickets 135-8, 494-7, 509, 523, 525, 527-9, 531
total wealth 59, 154
toteboard 520, 523, 525, 527-8, 531-5
touchdown 213, 465-8, 474, 481, 483-5, 489, 491-2
TOY see turn-of-the-year track 4-5, 7-8, 10, 45, 129, 318, 396, 435, 468,
493-5, 508, 515-16, 518-20, 522, 529- 30
trade time 230-1, 292, 295, 314-15, 324, 331, 381, 434
traders 27, 56-7, 119, 179-80, 282, 311, 338-9, 353-4, 358, 360-4, 370-1,
396-7, 400, 423
rogue 354, 359, 361-2, 365, 369-70, 439
trades 124-6, 171, 175-7, 179-80, 182-6, 202, 251, 294, 341-2, 352-3, 357-
9, 361-5, 382, 400-2, 424-5
commodity 171-2, 175
convergence 121-2, 124
trading 57, 63-4, 124-5, 171-2, 188-90, 205-6, 210-11, 231-2, 289-90, 328,
351-3, 355, 357-8, 370-1, 423-4
trading constraints 77-9, 81
trading volume 173, 451-2
Turkey 96-7, 295, 329, 352, 395-402, 417, 442, 456, 461
turn-of-the-year 127, 171, 173, 175-81, 182-7, 198, 215, 485
U
UK 3, 7-8, 14-15, 17-19, 22, 132, 134-7, 239, 249, 263, 270-1, 289-90, 313-
14, 320-1, 421-2
UMASS DHF Database 65, 68
unemployment 329, 383-4, 386, 419, 426, 429, 436-8, 441, 446, 460-2
United States 90, 240, 327, 357-8, 370-1, 411-12, 451
university 13, 117-18, 359, 439, 449-50
upside 42, 47-51, 53, 405
US assets 90, 235, 392
US dollars 13, 36, 122-3, 229, 231, 241-2, 247-8, 289-90, 330, 344-5, 374-
5, 382, 429, 434-5, 459
US mortgages 239, 319, 369, 371-2
US stock market 249, 275, 379, 386, 419, 440, 471
US Treasuries 90, 95, 98, 312, 372, 438, 441
V
value, theoretical 124, 126, 262-3
value function 39-40
Value Line see VL variables 49, 152-3, 191, 332, 372, 461-2
variances 29-31, 79, 118-19, 129, 140, 143, 147, 155
variation, daily mark-to-the-market
account 171, 175
VIX (volatility index) 37, 45, 229-34, 236-7, 253-4, 290-4, 303, 313, 324,
379-81, 384, 386, 419-20, 422-3, 426, 434, 471
VIX and Violent Market 239, 241, 243, 245, 247, 249, 251, 253
VIX Fear Index 229, 231, 233, 235, 237
VL (Value Line) 172, 174-87
volatility 42, 47-53, 76, 157, 176, 179, 192-5, 209, 221-2, 227, 229, 231-2,
352-4, 357-8, 423
implied 124, 126, 192-3, 232-3
volatility regimes, current 163, 165
volume 26, 126, 174-5, 184, 187, 233, 331, 424, 428
W
wagers 10, 128, 132-3, 136, 138, 142-3, 172, 176, 473, 515-16, 518, 520-1,
527, 529, 535
wealth 25, 27, 47, 59-60, 64-7, 78-9, 81, 127-9, 131, 140-4, 151-2, 154-5,
162-3, 362-3, 515-17
wealth levels 27, 127, 129, 136, 154, 163, 176, 305
weights 32, 155, 159, 163-4, 236-7, 266, 272, 475
optimal 31, 39, 161-4, 166
winners 6-7, 27, 134, 177, 182, 305, 323-4, 335, 363, 476-7, 494-6, 513,
523, 526-7, 531
winning 3-4, 25, 28, 128, 133-6, 138, 140, 142, 184, 361, 440, 477, 495-6,
499, 508-9
chance of 8, 129, 133, 143, 467, 489
probability of 213, 483, 520
winter 194, 207, 296, 354, 357-9, 523, 526
world economy 239, 248, 297, 365, 436, 439
Y
Yale endowment fund 115-18
Z
Zenyatta 4-6, 493-5, 499, 501-2, 507-10, 513
Ziemba, William (WTZ) 9, 11, 57, 174, 184, 188-90, 200, 205-6, 210-11,
294, 328, 343-4, 346-7, 426, 494
Zhao and Ziemba 27, 29, 61, 143-4, 263
Ziemba and Hausch 7-10, 128, 142-3, 502, 515-18, 535
Ziemba and Schwartz 13, 37, 122, 235, 246, 257-9, 294, 301-2, 347
Ziemba and Ziemba 45, 57, 59-60, 113, 191, 235-6, 242, 251, 268, 272,
275, 281, 313-14, 372, 375

Investing in The Modern Age - Rachel E S Ziemba

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Investing in The Modern Age - Rachel E S Ziemba

Uploaded by

Copyright:

Available Formats

Investing

Library of Congress Cataloging-in-Publication Data

British Library Cataloguing-in-Publication Data

Copyright © 2013 by World Scientific Publishing Co. Pte. Ltd.

In-house Editor: Chye Shu Wen

1. Arbitrage, Risk Arbitrage and the Favorite-longshot Bias

II Hedge Funds, Sovereign Wealth Funds and Other Investment

4. Average Hedge Funds and their Evaluation

III Seasonal Effects and Other Anomalies

15. Investing in the January Turn-of-the-Year Effect with Index Futures

IV Volatility, Correlation and Liquidity

19. Thoughts on the VIX Fear Index

V Can We Predict Stock Market Crashes?

21. Stock Market Crashes in 2006-2009: Were We Able to Predict Them?

VI Bubbles and Debt

VII Investing and Arbitrage in NFL Football and Horse Racing

36. Blunder or Correct Decision? The Belichick Decision to go for it on

About the Authors

Pure arbitrage exist when

where Q = 0.975 is the payback percent or 21.67%. So the other horses B

So (Oah – 1)(Obj – 1) = 3.5(0.229) < 1 so it is not an arbitrage. To get an

The favorite-longshot bias

Risk arbitrage in the US presidential election

One of Romney's main challengers was Newt Gingrich, who is a tough

1 Edited from Wilmott, November 2010.

I was a visiting professor at Tsukuba University and consultant to the

Predictive Ability of the Bond-Stock Earnings Yield Difference

Table 2.2: Strategies Used to Evaluate the Robustness (Predictive Ability)

respectively, where d is the length of the interval considered. The exit

and the entry threshold level is

where k = d · α and x(k) is the k-smallest value of the d observations starting

Table 2.4: Evaluation of the Performance of the Strategies for the US

Table 2.5: Stock Market Corrections (10% plus) in Germany, 1975–2005

Table 2.6: Evaluation of the Overall Performance of the Strategies for

Table 2.7: Stock Market Corrections (10% plus) in Canada, 1975-2005

Table 2.8: Evaluation of the Overall Performance of the Strategies for

Table 2.12: Evaluation of the Overall Performance of the Strategies for

This chapter begins with a categorization of the efficient market camps

How do investors and consultants do in all these cases?

The importance of getting the mean right.

Fig. 3.1 The main theorem of mean-variance analysis

The certainty equivalent, CE, of a portfolio with utility function u equals

Assuming exponential utility and normal distributions yields exact

Observe that the mean-variance problem is

Conclusion: spend your money getting good mean estimates; use

1 Edited from Wilmott, March 2006.

Figure 4.3 shows the effect on the optimal investment strategy of

1 Edited from Wilmott, September 2007.

An update on the financial markets and why ownership is so very important

There are a number of studies in the literature concerning hedge fund

Table 5.1: Characteristics of the Hedge Fund Data*

Kouwenberg and Ziemba do find significant differences in average

Table 5.3: Cross-sectional Regressions of Hedge Funds Risk and Returns*

Table 5.5 contains the estimation results of the cross-sectional

Table 5.5: Cross-sectional Regressions of Fund of Funds Risk and Returns*

1 Edited from Wilmott, July 2010.

The downside symmetric Sharpe ratio

So we replace σp with σx− where means take only minus values.

The Renaissance Medallion Fund

We calculated the annual standard deviation for the DSSR by multiplying

Renaissance Medallion continues to win but three new Renaissaince funds

Evaluating superior funds in the UMASS hedge fund data base

Outstanding funds in the UMASS DHF universe