Adventures in Financial Data Science: The Empirical Properties of Financial and Economic Data, 2nd Edition Graham L. Giller

Adventures in Financial Data Science:
The Empirical Properties of Financial

and Economic Data, 2nd Edition
Graham L. Giller
Visit to download the full and correct content document:
https://ebookmeta.com/product/adventures-in-financial-data-science-the-empirical-pro
perties-of-financial-and-economic-data-2nd-edition-graham-l-giller/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Analyzing Financial Data and Implementing Financial

Models Using R 2nd Edition Clifford S. Ang
https://ebookmeta.com/product/analyzing-financial-data-and-
implementing-financial-models-using-r-2nd-edition-clifford-s-ang/
A Solution Manual for Statistics and Data Analysis for

Financial Engineering 2nd Edition John L. Weatherwax
https://ebookmeta.com/product/a-solution-manual-for-statistics-
and-data-analysis-for-financial-engineering-2nd-edition-john-l-
weatherwax/
Data Quality Engineering in Financial Services (Early

Release) Brian Buzzelli
https://ebookmeta.com/product/data-quality-engineering-in-
financial-services-early-release-brian-buzzelli/
Statistics of Earth Science Data Their Distribution in

Time Space and Orientation 2nd Edition Graham J
Borradaile
https://ebookmeta.com/product/statistics-of-earth-science-data-
their-distribution-in-time-space-and-orientation-2nd-edition-
graham-j-borradaile/
Regulations in the Energy Industry Financial Economic
and Legal Implications Dorsman A.
https://ebookmeta.com/product/regulations-in-the-energy-industry-
financial-economic-and-legal-implications-dorsman-a/
Decision Making with Quantitative Financial Market Data

1st Edition Alain Ruttiens
https://ebookmeta.com/product/decision-making-with-quantitative-
financial-market-data-1st-edition-alain-ruttiens/
Data-Driven Science and Engineering Steven L. Brunton
https://ebookmeta.com/product/data-driven-science-and-
engineering-steven-l-brunton/
Beginning Data Science in R 4: Data Analysis,

Visualization, and Modelling for the Data Scientist 2nd
Edition Thomas Mailund
https://ebookmeta.com/product/beginning-data-science-in-r-4-data-
analysis-visualization-and-modelling-for-the-data-scientist-2nd-
edition-thomas-mailund/
Risk Analytics: From Concept to Deployment (World

Scientific Financial Data Analytics), 2nd Edition
Edward H K Ng
https://ebookmeta.com/product/risk-analytics-from-concept-to-
deployment-world-scientific-financial-data-analytics-2nd-edition-
edward-h-k-ng/
Downloaded from www.worldscientific.com
12678_9789811250644_tp.indd 1 6/5/22 7:38 AM

World Scientific Series in Finance
(ISSN: 2010-1082)
Series Editor: William T. Ziemba (University of British Columbia (Emeritus) and

London School of Economics, UK)
Advisory Editors:
Greg Connor (National University of Ireland, Maynooth, Ireland)
George Constantinides (University of Chicago, USA)
Espen Eckbo (Dartmouth College, USA)
Hans Foellmer (Humboldt University, Germany)
Christian Gollier (Toulouse School of Economics, France)
Thorsten Hens (University of Zurich, Switzerland)
Robert Jarrow (Cornell University, USA)
Hayne Leland (University of California, Berkeley, USA)
Haim Levy (The Hebrew University of Jerusalem, Israel)
John Mulvey (Princeton University, USA)
Marti Subrahmanyam (New York University, USA)
Published*:
Vol. 19 Adventures in Financial Data Science: The Empirical Properties of

Financial and Economic Data
Second Edition
by Graham L. Giller (Giller Investments, USA)
Vol. 18 Sports Analytics

by Leonard C. Maclean (Dalhousie University, Canada) &
William T. Ziemba (University of British Columbia, Canada)
Vol. 17 Investment in Startups and Small Business Financing

edited by Farhad Taghizadeh-Hesary (Tokai University, Japan),
Naoyuki Yoshino (Keio University, Japan),
Chul Ju Kim (Asian Development Bank Institute, Japan),
Peter J. Morgan (Asian Development Bank Institute, Japan) &
Daehee Yoon (Korea Credit Guarantee Fund, South Korea)
Vol. 16 Cultural Finance: A World Map of Risk, Time and Money

by Thorsten Hens (University of Zurich, Switzerland),
Marc Oliver Rieger (University of Trier, Germany) &
Mei Wang (WHU – Otto Beisheim School of Management, Germany)
Vol. 15 Exotic Betting at the Racetrack

by William T. Ziemba (University of British Columbia, Canada)
Vol. 14 Dr Z’s NFL Guidebook

by William T. Ziemba (University of British Columbia, Canada) &
Leonard C. MacLean (Dalhousie University, Canada)
*To view the complete list of the published volumes in the series, please visit:
www.worldscientific.com/series/wssf
Soundararajan - 12678 - Adventures in Financial Data Science.indd 1 5/2/2022 2:23:15 pm

12678_9789811250644_tp.indd 2 6/5/22 7:38 AM

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

Names: Giller, Graham L., author.
Title: Adventures in financial data science : the empirical properties of financial and
economic data / Graham L Giller, Giller Investments, New Jersey.
Description: Second Edition. | Hackensack, NJ : World Scientific, 2022. |
Series: World scientific series in finance, 2010-1082 ; Vol. 19 | Revised edition. |
Includes bibliographical references and index.
Identifiers: LCCN 2021058959 | ISBN 9789811250644 (hardcover) |
ISBN 9789811251818 (ebook) | ISBN 9789811251825 (ebook other)
Subjects: LCSH: Financial services industry--Data processing. | Investments--Data processing.
Classification: LCC HG4515.5 .G55 2022 | DDC 332.640285--dc23/eng/20220131

LC record available at https://lccn.loc.gov/2021058959
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.
Copyright © 2022 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.
For any available supplementary material, please visit

https://www.worldscientific.com/worldscibooks/10.1142/12678#t=suppl
Desk Editors: Soundararajan Raghuraman/Pui Yee Lum
Typeset by Stallion Press

Email: enquiries@stallionpress.com
Printed in Singapore
Soundararajan - 12678 - Adventures in Financial Data Science.indd 2 5/2/2022 2:23:15 pm

June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page v
For my wife, for her support,

our children, who put up with my “idiosyncracies,”
and for my parents, who inspired me to share.
In loving memory of
Jack and Lorna Baugh & Gordon and Alice Giller.
I am grateful to the many friends I have found over my years

working in the financial services industry, including those who
agreed to take a look at drafts of this work and provide feedback.
In particular: Winston Featherly-Bean, Pete Kyle, Claus Murmann,
Yuri Malitsky, Alex Ribeiro-Castro, Sutesh Sharma, Adrian
Schofield and Jaipal Tuttle.
v
B1948 Governing Asia
This page intentionally left blank

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page vii
Preface
This revised second edition of Adventures in Financial Data Science

is being written as over half of America has been vaccinated against

the Coronavirus, and life may be returning to normal in parts of the
United States, although the δ-variant is also continuing its deadly
spread, particularly in The South. As a long read with a potentially
narrow audience, the book has surprised me with its success and
led to me finding new friends throughout the world. As a personal
experience, it has been immensely enriching to write it.
I decided to take the opportunity to refresh some of the charts,
particularly those relevant to financial and economic data, from the
viewpoint now possible after the onset of the Coronavirus recession.
For the chapter on the Coronavirus itself, I have chosen to analyze
the accuracy of the extrapolations made from the data, freezing the
model in the Autumn of 2020 but grading it’s performance one year
later.
This book features mostly my empirical work, although I did
include some theory at the end. It will now be followed by a sec-
ond, smaller, volume Essays on Trading Strategy, that outlines my
view of the analytic framework that a trader ought to use to make
decisions under uncertainty. That book will be entirely devoted trad-
ing strategy theory and will have only slight overlap with this one.
vii
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page viii
viii Adventures in Financial Data Science
Many parts of this work can be viewed as a plea against the use
of the Normal distribution in places where it has no business being
used. Finance is one of them.
Graham L. Giller
Holmdel, NJ, 2021
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page ix
About the Author
Graham Giller is one of Wall Street’s original

data scientists. Starting his career at Morgan

Stanley in the UK, he was an early member
of Peter Muller’s famous PDT group and went
on to run his own investment firm. He was
Bloomberg LP’s original data science hire and
set up the data science team in the Global Data
division there. He then moved to J.P. Morgan to
take the role of Chief Data Scientist, New Prod-
uct Development, and was subsequently Head
of Data Science Research at J.P. Morgan and Head of Primary
Research at Deutsche Bank. He is currently CEO of Giller Invest-
ments (New Jersey), LLC, a private research firm.
ix

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xi
Contents
Preface vii
About the Author ix

List of Figures xv
List of Tables xxxi
Chapter 1. Biography and Beginnings 1

1.1 About this Book . . . . . . . . . . . . . . . . . . . . 1
1.2 Family . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Oxford, Physics and Bond Trading . . . . . . . . . . 10
1.4 Morgan Stanley and P.D.T. . . . . . . . . . . . . . . 13
1.5 Self Employed . . . . . . . . . . . . . . . . . . . . . . 23
1.6 Professional Data Science . . . . . . . . . . . . . . . 23
Chapter 2. Financial Data 31

2.1 Modeling Asset Prices as Stochastic Processes . . . . 31
2.2 Abnormality of Financial Distributions . . . . . . . . 35
2.3 The US Stock Market Through Time . . . . . . . . . 48
2.4 Interest Rates . . . . . . . . . . . . . . . . . . . . . . 56
2.5 LIBOR and Eurodollar Futures . . . . . . . . . . . . 71
2.6 Asymmetric Response . . . . . . . . . . . . . . . . . 88
2.7 Equity Index Options . . . . . . . . . . . . . . . . . 113
2.8 The VIX Index . . . . . . . . . . . . . . . . . . . . . 124
xi
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xii
xii Adventures in Financial Data Science
2.9 Microwave Latency Arbitrage . . . . . . . . . . . . . 131

2.10 What I’ve Learned about Financial Data . . . . . . 138
Chapter 3. Economic Data and Other

Time-Series Analysis 141
3.1 Non-Farm Payrolls . . . . . . . . . . . . . . . . . . . 142
3.2 Initial Claims . . . . . . . . . . . . . . . . . . . . . . 156
3.3 Twitter . . . . . . . . . . . . . . . . . . . . . . . . . 165
3.4 Analysis of Climate Data . . . . . . . . . . . . . . . 184
3.5 Sunspots . . . . . . . . . . . . . . . . . . . . . . . . . 214
Chapter 4. Politics, Schools, Public Health,

and Language 227
4.1 Presidential Elections . . . . . . . . . . . . . . . . . 227
4.2 School Board Elections . . . . . . . . . . . . . . . . . 240

4.3 Analysis of Public Health Data . . . . . . . . . . . . 249
4.4 Statistical Analysis of Language . . . . . . . . . . . . 271
4.5 Learning from a Mixed Bag of Studies . . . . . . . . 284
Chapter 5. Demographics and Survey Research 285

5.1 Machine Learning Models for Gender
Assignment . . . . . . . . . . . . . . . . . . . . . . . 285
5.2 Bayesian Estimation of Demographics . . . . . . . . 293
5.3 Working with Patreon . . . . . . . . . . . . . . . . . 296
5.4 Survey and Opinion Research . . . . . . . . . . . . . 305
5.5 Working with China Beige Book . . . . . . . . . . . 321
5.6 Generalized Autoregressive Dirichlet Multinomial
Models . . . . . . . . . . . . . . . . . . . . . . . . . . 326
5.7 Presidential Approval Ratings . . . . . . . . . . . . . 339
Chapter 6. Coronavirus 347

6.1 Discrete Stochastic Compartment Models . . . . . . 348
6.2 Fitting Coronavirus in New Jersey . . . . . . . . . . 352
6.3 Independent Models by State . . . . . . . . . . . . . 358
6.4 Geospatial and Topological Models . . . . . . . . . . 365
6.5 Looking Back at this Work . . . . . . . . . . . . . . 384
6.6 COVID Partisanship in the United States . . . . . . 388
6.7 Final Conclusions . . . . . . . . . . . . . . . . . . . . 393
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xiii
Contents xiii
Chapter 7. Theory 395

7.1 Some Remarks on the PDT Trading Algorithm . . . 395
7.2 Cosine Similarity . . . . . . . . . . . . . . . . . . . . 396
7.3 The Construction and Properties of Ellipsoidal
Probability Density Functions . . . . . . . . . . . . . 402
7.4 The Generalized Error Distribution . . . . . . . . . . 429
7.5 Frictionless Asset Allocation with Ellipsoidal
Distributions . . . . . . . . . . . . . . . . . . . . . . 438
7.6 Asset Allocation with Realistic Distributions
of Returns . . . . . . . . . . . . . . . . . . . . . . . . 444
Epilogue 449
E.1 The Nature of Business . . . . . . . . . . . . . . . . 449
E.2 The Analysis of Data . . . . . . . . . . . . . . . . . . 450
E.3 Summing Things Up . . . . . . . . . . . . . . . . . . 451
Appendix A. How I Store and Process Data 453

A.1 Databases . . . . . . . . . . . . . . . . . . . . . . . . 453
A.2 Programming and Analytical Languages . . . . . . . 454
A.3 Analytical Workflows . . . . . . . . . . . . . . . . . . 454
A.4 Hardware Choices . . . . . . . . . . . . . . . . . . . 455
Appendix B. Some of the Data Sources

I’ve Used for This Book 457
B.1 Financial Data . . . . . . . . . . . . . . . . . . . . . 457
B.2 Economic Data . . . . . . . . . . . . . . . . . . . . . 457
B.3 Social Media and Internet Activity . . . . . . . . . . 458
B.4 Physical Data . . . . . . . . . . . . . . . . . . . . . . 458
B.5 Health and Demographics Data . . . . . . . . . . . . 458
B.6 Political Data . . . . . . . . . . . . . . . . . . . . . . 458
Bibliography 461
Index 469

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xv
List of Figures
1.1 The author at The Blue Coat School, Coventry, 1987. . 3

1.2 Lorna Baugh with her Bletchley Park Service Medal. . 6

1.3 Gordon Pryce Giller (center), and his family, taken in
London. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 On a punt in Oxford with my good friend Eu Jin. . . . 12
1.5 Sketch from Jim Simons’ office, circa 1999, reproduced
from memory. . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1 Daily closing values of the S&P 500 Index since
inception. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2 Daily closing values of the S&P 500 Index since inception
with a logarithmic vertical axis. . . . . . . . . . . . . . . 39
2.3 Histogram of the daily returns of the S&P 500 Index
since inception. . . . . . . . . . . . . . . . . . . . . . . . . 40
2.4 Histogram of the daily returns of the S&P 500 Index
since inception. . . . . . . . . . . . . . . . . . . . . . . . . 41
2.5 Results of a set of successive F tests for normal distri-
butions of S&P 500 returns with constant variance for
1957 to 2021 inclusive. . . . . . . . . . . . . . . . . . . . . 43
2.6 Time series of daily returns of the S&P 500 Index since
inception in 1967 to date. . . . . . . . . . . . . . . . . . . 44
xv
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xvi
xvi Adventures in Financial Data Science
2.7 Scatter plot of squared daily return of the S&P 500 Index
(in percent) versus that quantity for the prior day, with
a fitted linear regression line. . . . . . . . . . . . . . . . . 44
2.8 Distributions of innovations for a GARCH (1, 1) model
for the daily returns of the S&P 500 for 1957 to 2020
inclusive. A Normal distribution for the innovations is
used in the modeling. . . . . . . . . . . . . . . . . . . . . . 46
2.9 Distributions of innovations for a GARCH (1, 1) model
for the daily returns of the S&P 500 for 1957 to 2021
inclusive. A Generalized Error Distribution for the inno-
vations is used in the modeling. . . . . . . . . . . . . . . 48
2.10 Time series of the estimated average daily return of the
S&P 500 Index by year from independent GARCH (1, 1)
regressions with innovations from the Generalized Error

Distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.11 Time series of the estimated first lag autocorrelation
coefficient of the daily returns of the S&P 500 Index by
year from independent GARCH (1, 1) regressions with
innovations from the Generalized Error Distribution. . . 51
2.12 Time series of the estimated kurtosis parameter, κ,
from daily returns of the S&P 500 Index by year from
independent GARCH (1, 1) regressions with innovations
from the Generalized Error Distribution. . . . . . . . . . 52
2.13 The empirical distribution function for standardized
estimates of the average daily return of the S&P 500
Index by year, and the cumulative distribution function
of the Normal distribution to which it should converge. 53
2.14 The empirical distribution function for standardized
estimates of the correlation of daily returns of the
S&P 500 Index by year, and the cumulative distribution
function of the Normal distribution to which it should
converge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.15 Time series of 3-month (91 day) US Treasury Bill
discount rates. . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.16 Distribution of daily changes in the discount rate of
3-month (91 day) US Treasury Bills. . . . . . . . . . . . 59
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xvii
List of Figures xvii
2.17 Distribution of standardized residuals from fitting the

daily changes in the discount rate of 3-month (91 day)
US Treasury Bills to a Vasicek-GARCH (1, 1) model. . . 60
2.18 Distribution of standardized residuals from fitting the
daily changes in the discount rate of 3-month (91 day)
US Treasury Bills to a censored Vasicek-GARCH (1, 1)
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.19 Imputed daily interest rate volatility for daily changes
in the discount rate of 3-month (91 day) US Treasury
Bills estimated from data in which no-change days have
been removed. . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.20 Estimated first lag autocorrelation of the daily changes
in the discount rate of 3-month (91 day) US Treasury
Bills from data in which no-change days have been

removed, by year. . . . . . . . . . . . . . . . . . . . . . . . 62
2.21 Time series of p values from performing Whittle’s test
for state dependence on the direction of daily changes in
the discount rate of 3-month US Treasury Bills. . . . . . 65
2.22 Distributions of scale of the innovations imputed when
estimating the model of Equation (2.24) from the daily
changes in the discount yield of US 3-month Treasury
Bills. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.23 Distribution of standardized daily rate change scales for
the model specified in Section 2.4.1.5. The blue bars rep-
resents the counts of data and the red line is the best
fitting Gamma distribution. . . . . . . . . . . . . . . . . . 70
2.24 Time series of both 3 month LIBOR rates and 3-month
(91 day) US Treasury Bill discount rates for the period
over which data coincides. . . . . . . . . . . . . . . . . . . 74
2.25 Cross-sectional regression between the 3-month (90 day)
change in LIBOR and the expected change based on the
Eurodollar Futures price 90 days prior. . . . . . . . . . . 79
2.26 Plots of estimated coefficients for the regression model
of Equation (2.48) with the final dates, T , taken to the
be settlement dates of the Eurodollar Futures. . . . . . . 80
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xviii
xviii Adventures in Financial Data Science
2.27 Scatter plots and regression lines for the model of Equa-
tion (2.48) with the final dates, T , taken to be the
settlement dates of the Eurodollar Futures. . . . . . . . 81
2.28 The test statistic, χ21 ,
for testing the hypothesis that
daily price changes for Eurodollar Futures are Normally
distributed. . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2.29 The estimated value of the kurtosis parameter, κ, when
a GARCH (1, 1) model for the daily price change of
Eurodollar Futures is estimated contract by contract. . 84
2.30 The test statistic, χ22 , for testing the hypothesis
that daily price changes for Eurodollar Futures are
homoskedastic. . . . . . . . . . . . . . . . . . . . . . . . . . 84
2.31 The estimated values of the GARCH (1, 1) parameters,

AˆT and B̂T when a model for Eurodollar Futures is esti-
mated contract by contract. . . . . . . . . . . . . . . . . . 85
2.32 Estimated average daily change in the price of Eurodol-
lar Futures for consecutive 90-day wide maturity buckets. 87
2.33 The level of the S&P 500 Index around three major
market disruptions and the associated daily volatility
of returns obtained by fitting a GARCH (1, 1) model
with innovations drawn from the Generalized Error
Distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.34 The implied variance response functions for
GARCH (1, 1) and PQ GARCH (1, 1) models estimated
from the daily returns of the S&P 500 Index using data
from 2000 to 2009. . . . . . . . . . . . . . . . . . . . . . . 95
2.35 Scatter plot of the estimated upside and downside
response variables D̂y and Êy for daily return of the
S&P 500 Index for the years 1928 to 2021. . . . . . . . . 96
2.36 Empirical distribution functions for the estimates D̂y
and Êy from a fit of the PQ GARCH (1, 1) model to the
daily returns of the S&P 500 Index year by year. . . . . 97
response variables D̂y and Êy for daily changes in
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xix
List of Figures xix
3-month (91) US Treasury Bill rates for the years 1954

to 2020. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
2.38 Empirical distribution functions for the estimates D̂y
and Êy from a fit of the PQ GARCH (1, 1) model to
the daily changes in 3-month (91 day) US Treasury Bill
Rates year by year. . . . . . . . . . . . . . . . . . . . . . . 99
response variables D̂y and Êy for daily returns of the US
dollar–British pound exchange rate for the years 1971 to
2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
2.40 Empirical Distribution Function for the p value of a test
of D = E for the US dollar–British pound exchange rate,
and the expected PDF from a Uniform Distribution. . . 100
2.41 Distribution of the stock specific kurtosis parameter, κ̂i ,

for fitting a PQ GARCH (1, 1) model to the daily returns
of S&P 500 Index member stocks. . . . . . . . . . . . . . 103
2.42 Distribution of the stock specific heteroskedasticity
parameters B̂i , D̂i , and Êi , for fitting a PQ GARCH (1, 1)
model to the daily returns of S&P 500 Index member
stocks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
2.43 Distribution of the stock specific autocorrelation param-
eter, ϕ̂i , for fitting a PQ GARCH (1, 1) model to the daily
returns of S&P 500 Index member stocks where the con-
ditional mean includes this term and a market return. . 105
2.44 Distribution of the stock specific market covariance
parameter, β̂i , for fitting a PQ GARCH (1, 1) model to
the daily returns of S&P 500 Index and distribution of
the variance explained (R2 ) by that model. . . . . . . . 107
2.45 Distributions of monthly returns for equal variance
investment in three basic option strategies on the
S&P 500 Index and buy-and-hold investing in the index
itself. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
2.46 Time series of total returns for equal variance investment
in three basic option strategies on the S&P 500 Index
and buy-and-hold investing in the index itself. . . . . . . 118
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xx
xx Adventures in Financial Data Science
2.47 Distribution of the test statistic, the mean difference

in monthly returns between a call buying strategy and
a buy-and-hold strategy for the S&P 500 Index, from
50,000 bootstrap simulations. . . . . . . . . . . . . . . . . 119
2.48 Average value of the discount of the forward price of the
S&P 500 Index, S(t, T ), implied by the option market to
the spot price, St , as a function of the forward interval,
T − t. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
2.49 Median value of the rate at which the S&P 500 Index
Options discount the forward price, S(t, T ), relative to
the spot rate, St , from September 2012, to date. . . . . 123
2.50 Time series of the VIX index and the daily volatility
computed by fitting a PQ GARCH (1, 1) model to the
daily returns of the S&P 500 Index. . . . . . . . . . . . . 126

2.51 Data used to examine the variance linearity of the VIX
relative to the lead 1 variance forecast obtained by fitting
a PQ GARCH (1, 1) model to the daily returns of the
S&P 500 Index. . . . . . . . . . . . . . . . . . . . . . . . . 128
2.52 Data used to examine the covariance linearity of the VIX
relative to the daily returns of the S&P 500 Index. . . . 130
2.53 Topographic map illustrating the location of Aurora, IL,
and Carteret, NJ. . . . . . . . . . . . . . . . . . . . . . . . 132
2.54 Conditioned and unconditioned average price moves at
the NASDAQ around the times of price changing ticks
on ES futures at the CME. . . . . . . . . . . . . . . . . . 136
2.55 Conditioned and unconditioned average price moves at
the NASDAQ around the times of price changing ticks
on ES futures at the CME when the order book imbal-
ance is accounted for. . . . . . . . . . . . . . . . . . . . . . 137
3.1 Results of fitting a GJRGARCH (1, 1) model to the
monthly “return” of US NFP. . . . . . . . . . . . . . . . . 143
3.2 Block bootstrap distribution of the change in AIC (c) for
fitting a GJR-GARCH (1, 1) model to monthly “returns”
of NFP (seasonally adjusted). . . . . . . . . . . . . . . . . 149
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxi
List of Figures xxi
3.3 Scan of the change in AIC(c) for fitting an AR(n)-GJR-

GARCH (1, 1) model to monthly relative changes of NFP
(seasonally adjusted). . . . . . . . . . . . . . . . . . . . . . 150
3.4 Out of sample standardized forward prediction errors for
an AR(4)-GJR-GARCH (1, 1) model for the monthly rel-
ative changes in NFP. . . . . . . . . . . . . . . . . . . . . 152
3.5 Out of sample regression of forward predictions for an
AR(4)-GJR-GARCH (1, 1) model onto the monthly rel-
ative changes in NFP. . . . . . . . . . . . . . . . . . . . . 153
3.6 Screenshot of the Google Trends user interface taken in
August, 2020. . . . . . . . . . . . . . . . . . . . . . . . . . 158
3.7 Time-series of Google Trends activity index for the key-
word “unemployment insurance” and Initial Claims (not

seasonally adjusted). . . . . . . . . . . . . . . . . . . . . . 159
3.8 Linear regression of Initial Claims (not seasonally
adjusted) onto Google Trends activity index for the key-
word “unemployment insurance”. . . . . . . . . . . . . . 160
3.9 Comparison of the performance of a Time Varying Coef-
ficients and Ordinary Least Squares model for the regres-
sion of Initial Claims (not seasonally adjusted) onto a
Google Trends index for the keyword “unemployment
insurance”. . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
3.10 Comparison of the performance of a Time Varying Coef-
ficients model for Initial Claims (seasonally adjusted) to
the released data. . . . . . . . . . . . . . . . . . . . . . . . 164
3.11 Histogram of the standardized forward prediction errors
for a TVC model of Initial Claims (seasonally adjusted)
from Google search trends. . . . . . . . . . . . . . . . . . 164
3.12 Tweet from known activist investor Carl Icahn announc-
ing that he had bought Apple, Inc. shares. . . . . . . . 165
3.13 Regression analysis of Bloomberg’s BTWTNF Index data
and the release for Monthly Change in NFP (seasonally
adjusted). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxii
xxii Adventures in Financial Data Science
3.14 Histograms of the account age for Twitter users who

were collected when listening to the streaming API for
tweets mentioning stock tickers in the S&P 400 Index. . 173
3.15 The autocorrelation function of the graded sentiments
of captured tweets and the distribution of times between
tweets that were captured. . . . . . . . . . . . . . . . . . 175
3.16 Analysis of the cross-correlation between graded senti-
ments of captured tweets and the returns of the S&P 500
Index for May and June, 2012. . . . . . . . . . . . . . . . 177
3.17 Population pyramid for Twitter users based on inferred
demographics. . . . . . . . . . . . . . . . . . . . . . . . . . 180
3.18 Time series of average organic Twitter sentiment for
tweets geolocated to the USA and the University of

Michigan’s Index of Consumer Sentiment based on tra-
ditional consumer surveys. . . . . . . . . . . . . . . . . . . 180
3.19 Time-series of average organic Twitter sentiment for
tweets geolocated to the USA and weekly Initial Claims
(for unemployment insurance, seasonally adjusted). . . . 181
3.20 Regression between organic Twitter sentiment and the
decimal log of Initial Claims (seasonally adjusted). . . . 182
3.21 Time-series of the Central England Temperature during
the author’s lifetime. . . . . . . . . . . . . . . . . . . . . . 186
3.22 Smoothed time-series of the Central England Tempera-
ture since 1659. . . . . . . . . . . . . . . . . . . . . . . . . 187
3.23 Smoothed time-series of the Central England Temper-
ature since 1659 and five bootstrap simulations of that
curve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
3.24 Smoothed time-series of the Central England Tempera-
ture since 1659 and the mean of 5,000 bootstrap simu-
lations of that curve. . . . . . . . . . . . . . . . . . . . . . 189
3.25 Contour plot of the ΔA.I.C.(c) surface for fitting a lep-
tokurtotic AR(m) × SAR(y) model to the Central Eng-
land Temperature series for data from December 1700,
to December 1849. . . . . . . . . . . . . . . . . . . . . . . 193
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxiii
List of Figures xxiii
3.26 Estimated and implied weights for an AR(7) × SAR(34)

model for the Central England Temperature estimated
from the Pre-Industrial Period data (January 1659 to
December 1849). . . . . . . . . . . . . . . . . . . . . . . . 194
3.27 Distribution of standardized residuals from fitting a
leptokurtotic AR(7) × SAR(34) model to the Central
England Temperature series for data until December
1849. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
3.28 Autocorrelation function of the residuals from fitting a
leptokurtotic AR(7) × SAR(34) model to the Central
England Temperature series for data until December
1849. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
3.29 Autocorrelation function of the squared residuals from

fitting a leptokurtotic AR(7) × SAR(34) model to
the Central England Temperature series for data from
December 1700 to December 1849. . . . . . . . . . . . . . 198
3.30 Estimated idiosyncratic standard deviations of the aver-
age monthly temperature in the Central England Tem-
perature series when fitted to an AR(7) × SAR(34)
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
3.31 Autocorrelation function of the squared residuals from
fitting a AR(7) × SAR(34) model with structural het-
eroskedasticity to the Central England Temperature
series for data until December 1849. . . . . . . . . . . . . 201
3.32 Contour plot of the ΔA.I.C.(c) surface for fitting a lep-
tokurtotic AR(m) × SAR(y) model to the Central Eng-
land Temperature series for data from December 1700,
to December 1849, with Structural Heteroskedasticity. . 203
3.33 Linear regression of the autoregression and seasonal
autoregressive coefficients for the Pre-Industrial period
onto those of the Industrial Period. . . . . . . . . . . . . 206
3.34 Central England Temperature in the 2010s and predicted
temperature from AR(7)× SAR(29) models fitted in the
Pre-Industrial and Industrial periods. . . . . . . . . . . . 208
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxiv
xxiv Adventures in Financial Data Science
3.35 Time series of the Mean Monthly Sunspot Num-

ber (the Wolf Series) and the Central England
Temperature. . . . . . . . . . . . . . . . . . . . . . . . . . 210
3.36 Distribution of the Monthly Mean Sunspot Number from
1749 to date. . . . . . . . . . . . . . . . . . . . . . . . . . . 211
3.37 Distribution of the Transformed Monthly Mean Sunspot
Number from 1749 to date. . . . . . . . . . . . . . . . . . 212
3.38 The Sunspot Number for the last three Solar cycles. . . 215
3.39 Results of scanning the model order, m, in an
Markov(2) × AR(m) model of the incremental “excita-
tion” of the Sunspot generating system. . . . . . . . . . 219
Estimated lag coefficients for a Markov(2) × AR(26)
3.40
model of the incremental “excitation” of the Sunspot
generating system. . . . . . . . . . . . . . . . . . . . . . . 222
3.41 In sample regression of the model onto observations for
a Markov(2) × AR(26) model of the incremental “exci-
tation” of the Sunspot generating system. . . . . . . . . 223
3.42 Out-of-sample predictions of a Markov(2) × AR(26)
model of the incremental “excitation” of the Sunspot
generating system. . . . . . . . . . . . . . . . . . . . . . . 224
4.1 Predictions and outcomes of a logistic regression model
of Presidential elections using height difference and
desire for change as independent variables. . . . . . . . . 234
4.2 Variation of precision, recall and F -score with the deci-
sion threshold, β, for predicting a Republican President
based on logistic regression. . . . . . . . . . . . . . . . . . 235
4.3 Distribution of optimal decision threshold for 10,000
bootstraps of the use of a logistic regression model
to predict a Republican President based on candidate
heights and desire for change. . . . . . . . . . . . . . . . . 237
4.4 Variation of precision, recall and F -score with the deci-
sion threshold, β, for predicting a Republican President
based on a Naı̈ve Bayes classifier. . . . . . . . . . . . . . 239
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxv
List of Figures xxv
4.5 Side-by-side comparison of the probabilities for a Repub-

lican Presidential win based on a logistic regression
model and a naı̈ve Bayes classifier. . . . . . . . . . . . . . 239
4.6 Heights of male Presidential candidates from 1896 to
2020. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
4.7 Screenshot of the official election results that would be
certified by the Monmouth County Clerk for the Holmdel
Board of Education Election of 2019. . . . . . . . . . . . 249
4.8 The actual relationship between human body weight and
height from the BRFSS sample data for 2019. . . . . . . 252
4.9 Distribution of the residuals to a fitted logistic model
for human body weight as a function of height for the
BRFSS sample data for 2019. . . . . . . . . . . . . . . . . 254

4.10 Moments of the conditional distributions of the residuals
to the model for human body weight versus age. . . . . 257
4.11 Empirical distribution functions for a fitted logistic
model for human body weight as a function of height
for the BRFSS sample data for 2019. . . . . . . . . . . . 258
4.12 Distribution of consumption of alcoholic drinks per day
for the BRFSS sample data for 2019. . . . . . . . . . . . 261
4.13 Distribution of consumption of meals including fried
potatoes per day for the BRFSS sample data for 2019. 263
4.14 Time-series of parameter estimates for a logistic-
quadratic model of human body weight as a function
of height and age. . . . . . . . . . . . . . . . . . . . . . . . 266
4.15 Time-series of parameter estimates for linear adjust-
ments to the logistic-quadratic model for human body
weight based on exercise and drinking habits. . . . . . . 268
4.16 Time-series of weighted averages of inputs to model for
human body weight. . . . . . . . . . . . . . . . . . . . . . 269
4.17 Frequency-rank analysis for the Brown Corpus illustrat-
ing the best fitting models. . . . . . . . . . . . . . . . . . 276
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxvi
xxvi Adventures in Financial Data Science
4.18 Frequency-rank analysis for four bootstrap samples of

unordered pseudowords drawn according to the entire
Brown Corpus character frequencies of the discovered
alphabet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
4.19 Frequency-rank analysis for four bootstrap samples of
semi-ordered pseudowords drawn according to the entire
Brown Corpus character frequencies of the discovered
alphabet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
5.1 Distribution of precision, recall and F -score for a logistic
regression model of gender based on vowels within a first
name. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
5.2 Regression tree to predict assigned gender from first
names based upon a random sample of 20% of the SSA’s

Baby Names Database. . . . . . . . . . . . . . . . . . . . . 290
5.3 Distribution of precision, recall and F -score for a regres-
sion tree model of gender based on vowels within a first
name. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
5.4 The probable age of a woman called Veronica given
known age and gender, derived from the BRFSS and
ACS data. . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
5.5 The probability that Leslie is male given their age,
derived from the BRFSS and ACS data. . . . . . . . . . 296
5.6 Estimated population pyramid for Patreon users. . . . . 298
5.7 Comparison of the Patreon “Not Financial Instability”
index and the University of Michigan’s Index of Con-
sumer Sentiment. . . . . . . . . . . . . . . . . . . . . . . . 302
5.8 Comparison of the University of Michigan’s Index of
Consumer Sentiment and my own data. . . . . . . . . . . 307
5.9 Estimates of the time-varying coefficients in the regres-
sion of Equation (5.13). . . . . . . . . . . . . . . . . . . . 309
5.10 Estimated value of the Index of Consumer Sentiment
based on private survey data and the public series. . . . 310
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxvii
List of Figures xxvii
5.11 Comparison of consumer expectations of inflation from

private surveys and the University of Michigan’s data. . 312
5.12 Estimated value of expectations of short-term inflation
based on private survey data and the results of the Uni-
versity of Michigan’s survey. . . . . . . . . . . . . . . . . 313
5.13 Empirical distribution function for Z score variables to
University of Michigan data and the cumulative proba-
bility distribution of the Normal distribution. . . . . . . 314
5.14 Screenshots showing the delivery of online surveys and
microrewards to participants. . . . . . . . . . . . . . . . . 315
5.15 Scaling of question response time with number of
choices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
5.16 Analysis of the China Beige Book all sectors workforce

change index and quarterly revenues for Wynn Resorts,
Limited. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
5.17 Region of stationarity in expectation for the case α0 = n̄.
The region is open for positive ψ provided |ϕ| < 1. . . . 339
5.18 Presidential approval ratings data as processed by Nate
Silver’s 538. . . . . . . . . . . . . . . . . . . . . . . . . . . 341
5.19 Time-series of estimated approval ratings for Presi-
dent Trump during 2020 with confidence regions and
forecasts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
6.1 Early results of fitting a Stochastic Discrete Compart-
ment Model to the Coronavirus outbreak in Monmouth
County, NJ. . . . . . . . . . . . . . . . . . . . . . . . . . . 353
6.2 Estimated values for the piecewise constant Effective
Reproduction Rate of the Coronavirus outbreak in Mon-
mouth County, NJ. . . . . . . . . . . . . . . . . . . . . . . 355
6.3 Current results of fitting a Stochastic Discrete Compart-
ment Model, with piecewise constant βt , to the Coron-
avirus outbreak in Monmouth County, NJ. . . . . . . . . 355
6.4 Current results of fitting a Stochastic Discrete Compart-
ment Model in New Jersey. . . . . . . . . . . . . . . . . . 356
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxviii
xxviii Adventures in Financial Data Science
6.5 Current results of fitting a model for the Coronavirus

outbreak in the entire United States. . . . . . . . . . . . 357
6.6 Estimated values for the piecewise constant in the
United States. . . . . . . . . . . . . . . . . . . . . . . . . . 358
6.7 Observed covariance of the estimates R̂0 and R̂, from
Table 6.2 and Presidential Approval ratings by State for
Donald Trump. . . . . . . . . . . . . . . . . . . . . . . . . 361
6.8 Observed covariance of the estimates R̂0 and R̂, from
Table 6.2 and the decimal log of the population counts
of the States. . . . . . . . . . . . . . . . . . . . . . . . . . . 364
6.9 Observed covariance of estimated Case Fatality Rates
for the States and the decimal log of the population. . . 365
6.10 Computed “shortest time” route between Middletown,

NJ, and Midtown, Manhattan. . . . . . . . . . . . . . . . 367
6.11 Undirected graph for the connectivity of New Jersey
counties as output by Mathematica. . . . . . . . . . . . . 368
6.12 Undirected graph for the communities of New Jersey
counties as output by Mathematica, with labels assigned
by the author. . . . . . . . . . . . . . . . . . . . . . . . . . 369
6.13 Current results of fitting a topological model for
the Coronavirus outbreak in six counties in central
New Jersey. . . . . . . . . . . . . . . . . . . . . . . . . . . 374
6.14 Estimated values for the time-varying scale factor, λt ,
in topological model for the Coronavirus outbreak in six
counties in central New Jersey. . . . . . . . . . . . . . . . 375
6.15 Estimated time-series of the number of infected persons
in Monmouth County, New Jersey, based on the single
county and topological models. . . . . . . . . . . . . . . . 376
6.16 Undirected graph for the connectivity of the United
States as output by Mathematica. . . . . . . . . . . . . . 377
6.17 Undirected graph for the communities of the United
States as output by Mathematica, with labels assigned
by the author. . . . . . . . . . . . . . . . . . . . . . . . . . 378
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxix
List of Figures xxix
6.18 Results of fitting a topological model for the Coronavirus

outbreak in the Midwest. . . . . . . . . . . . . . . . . . . 379
6.19 Results of fitting a topological model for the Coronavirus
outbreak in the Midwest. . . . . . . . . . . . . . . . . . . 380
6.20 Effective reproduction rate scale-factor implied by a
topological model for the Coronavirus outbreak in the
Midwest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
6.21 Estimates of the number of people infected with COVID-
19 in Nebraska based on a single state model and a topo-
logical model for eight states in the midwest. . . . . . . 382
6.22 Time series of estimated effective reproduction rate, by
state, for the United States. . . . . . . . . . . . . . . . . . 389
6.23 Map of the effective reproduction number by state for

the COVID-19 epidemic in July 2021. . . . . . . . . . . . 390
7.1 The Generalized Error Distribution probability den-
sity function for various values of the kurtosis
parameter, κ. . . . . . . . . . . . . . . . . . . . . . . . . . 431
7.2 Excess kurtosis measure γ2 of the Generalized Error Dis-
tribution, for κ > 12 . . . . . . . . . . . . . . . . . . . . . . 432
7.3 Univariate standardized Generalized Error Distribution
for several values of the kurtosis parameter, κ. . . . . . 434
7.4 Variance scale factor for constructed multivariate
Generalized Error Distributions as a function of
the kurtosis parameter, κ, for various numbers of
dimensions, n. . . . . . . . . . . . . . . . . . . . . . . . . . 435
7.5 Excess kurtosis measure γ2,n multivariate General-
ized Error Distributions as a function of the kurto-
sis parameter, κ, with κ > 12 and various number of
dimensions, n. . . . . . . . . . . . . . . . . . . . . . . . . . 436
7.6 Behavior of the scaling function xΨ1/2 (x) for κ =
0.5, 0.8 and 1. . . . . . . . . . . . . . . . . . . . . . . . . . 442
7.7 Behavior of the Inverting Function Φ1/2 (x) as κ → 1. The
dotted diagonal line represents the Normal distribution
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxx
xxx Adventures in Financial Data Science
theory Φν (x) = 1 and the dotted

√ horizontal line shows
the upper bound Φ1/2 (x) < 2 for κ = 1. . . . . . . . . . 443
7.8 Portfolio scaling factors 1/Ψ1/2 {Φ 1 (x)} for a single asset
2
as κ → 1. The dotted line represents the Normal distri-
bution theory. . . . . . . . . . . . . . . . . . . . . . . . . . 443
7.9 Standardized portfolio expected return x2/Ψ 1 {Φ 1 (x)}
2 2
for a single asset as κ → 1. The dotted line represents
the Normal distribution theory. . . . . . . . . . . . . . . . 444
7.10 Illustration of the form of the Markowitz trading rule
and the simple barrier rule of Equation (7.167). . . . . . 446
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxxi
List of Tables
2.1 Single factor ANOVA table for estimated daily returns

of the S&P 500 Index by year. . . . . . . . . . . . . . . . 54

2.2 Single factor ANOVA table for estimated autocorrelation
of daily returns of the S&P 500 Index by year. . . . . . 54
2.3 Maximum likelihood regression results for the Markov
Chain model for the direction of the daily change in the
discount rate of 3-month US Treasury Bills. . . . . . . . 69
2.4 Maximum likelihood regression results for the scale of
daily changes 3-month US Treasury Bill Rates with a
GARCH (1, 1) structure and a Gamma distribution for
the innovations. . . . . . . . . . . . . . . . . . . . . . . . . 70
2.5 Maximum likelihood regression results for the Markov
Chain model for the direction of the daily change in the
3-month LIBOR rates. . . . . . . . . . . . . . . . . . . . . 75
2.6 Maximum likelihood regression results the fit of a GJR-
GARCH (1, 1) model to the daily returns of the S&P 500
Index from 1928 to date. . . . . . . . . . . . . . . . . . . . 91
2.7 Maximum likelihood regression results the fit of a PQ-
GARCH (1, 1) model to the daily returns of the S&P 500
Index from 1928 to date. . . . . . . . . . . . . . . . . . . . 94
xxxi
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxxii
xxxii Adventures in Financial Data Science
2.8 Table of values for the t test for Zero Mean applied
to the heteroskedasticity parameters obtained when fit-
ting a PQ GARCH (1, 1) model to current members of
the S&P 500 Index that have at least 3 years of data
history. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
2.9 Robust linear regression results for fit of the square
of the VIX to the lead 1 variance predictor from a
PQ GARCH (1, 1) model for the daily returns of the
S&P 500 Index for 1990 to date. . . . . . . . . . . . . . . 127
2.10 Robust linear regression results for fit of the daily change
in the VIX onto the daily return of the S&P 500 Index
for 1990 to date. . . . . . . . . . . . . . . . . . . . . . . . . 129
3.1 Maximum likelihood regression results the fit of a GJR-
GARCH (1, 1) model to the monthly “returns” of NFPs

(seasonally adjusted) from 1940 to 2015. . . . . . . . . . 145
3.2 Maximum likelihood regression results the fit of a PQ -
GARCH (1, 1) model to the monthly “returns” of NFPs
(seasonally adjusted) from 1940 to 2015. . . . . . . . . . 147
3.3 Maximum likelihood regression results the fit of an
AR(4)-GJR-GARCH (1, 1) model to the monthly rela-
tive changes of NFPs (seasonally adjusted) from 1940 to
2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
3.4 Predictions and reported data for the monthly relative
changes of NFPs (seasonally adjusted) for 2019–2020. . 155
3.5 Four sample formats of Twitter messages used the fol-
lowers experiment. . . . . . . . . . . . . . . . . . . . . . . 168
3.6 Single factor ANOVA table for change in Twitter follow-
ers grouped by indicated trading success. . . . . . . . . . 169
3.7 Results from fixed individual effects panel regression of
Initial Claims by State onto lagged values and contem-
poraneous organic Twitter sentiment. . . . . . . . . . . . 184
3.8 Maximum likelihood regression results for the fit of a
Markov(2) × AR(6) model to the power transformed
Sunspot Number. . . . . . . . . . . . . . . . . . . . . . . . 220
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxxiii
List of Tables xxxiii
4.1 Independent logistic regression results to determine the

predictor variables useful in explaining Presidential
elections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
4.2 Logistic regression results for a joint model for Presiden-

tial elections. . . . . . . . . . . . . . . . . . . . . . . . . . . 233
4.3 Independent wins logistic regression results to predict

the outcomes of School Board elections. . . . . . . . . . . 244
4.4 Independent wins logistic regression model predictions of

the outcomes of the 2019 Holmdel School Board election. 245
4.5 Independent vote counts regression results to predict the

outcomes of School Board elections. . . . . . . . . . . . . 247
4.6 Independent vote counts regression model predictions

of the vote share for the 2019 Holmdel School Board
election. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
4.7 Estimated parameters and consistency test for logistic

model to relationship between human body weight and
height. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
4.8 Estimated parameters and consistency test for logistic

model to relationship between human body weight and
height and age. . . . . . . . . . . . . . . . . . . . . . . . . 259
4.9 Basic statistics and regression results for the corpora

analyzed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
4.10 Symbol statistics for the alphabet discovered in the

Brown Corpus after processing. . . . . . . . . . . . . . . . 278
4.11 Sample statistics for the bootstrapped estimators of the

Frequency-Rank distribution parameters for unordered
pseudowords generated from the Brown Corpus. . . . . 282
4.12 Sample statistics for the bootstrapped estimators of

the Frequency-Rank distribution parameters for semi-
ordered pseudowords generated from the Brown Corpus. 282
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxxiv
xxxiv Adventures in Financial Data Science
5.1 Cross-tabulation of employment situation versus weekly

hours worked and employer type from a private con-
sumer survey. . . . . . . . . . . . . . . . . . . . . . . . . . 320
5.2 Maximum likelihood regression of the quarterly revenues

of Wynn Resorts, Limited on the national workforce
indicator computed from China Beige Book data. . . . . 326
5.3 Maximum likelihood estimates of the parameters for the

generalized autoregressive Dirichlet multinomial model
for Presidential approval ratings. . . . . . . . . . . . . . . 344
6.1 Estimated parameters for discrete stochastic compart-

ment model for Coronavirus outbreaks. . . . . . . . . . . 357
6.2 Estimated parameters for discrete stochastic compart-

ment model for Coronavirus outbreaks in the United
States by State. . . . . . . . . . . . . . . . . . . . . . . . . 359
6.3 Single factor ANOVA table for the estimated values

of R̂0 dependent on the party of the Governor of the
States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
6.4 Single factor ANOVA table for the estimated final val-
ues of R̂ dependent on the party of the Governor of the
States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
6.5 Estimated characteristic rates for COVID-19 in the

Midwest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
6.6 Single factor ANOVA table for the variation of estimated

values of β̂ij dependent on the level of contact between
the States. . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
6.7 Table of predicted total counts of cases and deaths

for the Coronavirus ourbreak in various regions of the
United States. . . . . . . . . . . . . . . . . . . . . . . . . . 388
6.8 Single factor ANOVA table for the variation of estimated

values of R̂it dependent on the level of partisanship of
the States. . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
June 8, 2022 10:44 Adventures in Financial Data Science. . . 9in x 6in b4549-fm page xxxv
List of Tables xxxv
6.9 Estimated value of the effective reproduction rate, R̂it

of COVID-19 for July 2021, by State, and the value cor-
. . . . . . . . . . . . . . . . . . 393
rected for partisanship, R̂it
7.1 Results of numerical simulations of the cosine similarity
between independent random bitmaps of length 10 000,
100 and 10 bits. . . . . . . . . . . . . . . . . . . . . . . . . 401
7.2 Cartesian and polar coordinates for 1 to 5 dimensions. . 406

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

June 8, 2022 10:42 Adventures in Financial Data Science. . . 9in x 6in b4549-ch01 page 1
Chapter 1
Biography and Beginnings
1.1. About this Book

1.1.1. What this Book is?

This book is not a biography, as I am not a famous person, and it
is not a textbook either. It is a narrative of analytic work that I
have done, structured around both my professional career and some
of the things that just interested me. I am hoping to share with
you what Richard Feynman described as “The Pleasure of Finding
Things Out” [40]. For roughly half of my career I’ve run my own
businesses, strongly focused on analytical work, and when I needed
a metric to judge myself I fell back on a simple question: Did I learn
something today? That is my personal goal, to learn things about the
Universe, and what I hope to do here is share some of the things I’ve
learned in my pursuit of practical data science. Nevertheless, I will
start by giving you some insights into my background, so you can
better understand who I am and how I think about things.
1.1.2. What this Book is Not?

There are many books on data science that are essentially compendi-
ums of code snippets that exhibit various elementary analyzes. This
book does not follow that format. There is value to pedagogic work,
but I believe strongly that understanding comes from figuring out the
quirks of each individual data set you work with, and no reader will
gain that by merely copying my work. I will show you some results
1
2 Adventures in Financial Data Science
I have found and discuss what I think they mean, and whether sci-
ence tells us they are believable. I will provide some links to the
public data sets I have accessed, some of which require small fees to
access. Financial data has never been free because people use it to
make a lot of money, and the data generators are well aware of that.
1.2. Family
1.2.1. About Me
I am English and grew up in the West Midlands. First in Coventry,
and later in a small village called Church Lawford. I now live in New
Jersey on my in-law’s family farm, just outside New York City on the
Jersey Shore. For the majority of my career I’ve been professionally
involved in what is now called “data science,” but for me that started
much, much, earlier. I have a doctorate in Experimental Particle
Physics, from Oxford University, where I also did my undergraduate
work. My work featured very large scale (for the time) computer
based data analytics in the field of Cosmic Ray Astronomy. For some
reason both time-series analysis and statistics “clicked” for me, and
I’ve spent a lot of my professional career organically applying those
tools to data from more “social” sciences. I’ve worked in proprietary
trading at Morgan Stanley in London, as a “quant,” and in New York,
as a strategy developer and portfolio manager in the now famous and
very successful Process Driven Trading unit. I ran my own investment
fund for about a decade, a small effort for “friends and family,” not
some giant world shaping enterprise, and subsequently spent time
as a consultant data scientist before joining Bloomberg LP’s Global
Data unit as their first data science hire. I set up the data science
group there before joining J.P. Morgan as Chief Data Scientist in a
unit called New Product Development (N.P.D.). From there I moved
to take the role as Head of Primary Research at Deutsche Bank before
that firm decided it no longer wanted to be in the equity business. At
the time of writing, I am sitting at home, as the COVID-19 outbreak
appears to be entering a third wave.
1.2.2. Grandad and the Oil Tanker

One of the questions I’m often asked is “why did you decide to enter
finance?” I usually respond with the story about Grandad and the
Biography and Beginnings 3

Figure 1.1: The author at The Blue Coat School, Coventry, 1987.
Photo Credit Phillip Shipley.
oil tanker. My grandfather was Jack Baugh and he had a great effect
on me. I figured out that he didn’t always see eye-to-eye with my
Father, which I guess is not unusual, but generally we all got on and
spent many Christmases and Summers together. When I was a child
I knew that it was expected that one aspire to some kind of career,
and my choices followed the typical paths of children. The first thing
I announced I wanted to be was a professional cricketer. I am not very
good at sport, as everybody who knows me well can testify to, but
my Grandad was very keen on cricket, having played for his work’s
team, and that rubbed off on me. When I found out that only native
born Yorkshiremen were permitted to play for Geoff Boycott’s team
my aspirations crashed to the ground, and I was puzzled as to why
a professional sports team would have such a rule.
My second announced choice was “Chartered Accountant.” This,
again, was influenced by my Grandad. His official title was “Chief
Cashier of the South Yorkshire Passenger Transport Executive,” but
what he was was a management accountant and what he did was,
with his buddy, run the state owned bus company that provided
public transport to the city of Sheffield. In modern times his title
would have been Chief Financial Officer. I guess he could see that
his grandson was good at maths, so he would try to interest me in his
work. Once he took me with him to the Midland Bank in Sheffield,

on some personal business, and I was pleasantly surprised to see the
manager of that institution come out from the backs of the banking
hall to personally greet him. For that was where the bus company
also did its banking business. That was when I realized that he was,
relatively, important.
When your grandfather wants to entertain you he puts you on
his knee and tells you stories, and most of Jack’s framing experi-
ences were those from his service in the Royal Air Force during the
Second World War. His role was to help airbases install and use a
groundbreaking technology called RADAR, which I found fascinat-
ing, especially as what he was describing was nothing like what I had
seen in films. He described two cathode ray tubes showing a straight
line with a “blip” on it. The first represented vertical distance and
the second horizontal distance and the blip was the reflected signal
from an enemy plane. I was expecting the sweeping lines with linger-
ing dots that we are used to from images of air traffic control and
ship radar, but this was before that time.
Once he told me about the bus company having “too much money
to keep in the bank over the weekend,” and so they needed to figure
out what to do with it. I guess they had good insights as to what
was happening to oil prices, which would arise from running a bus
company and, presumably, being in receipt of regular phone calls
from commodities brokers, but I was startled when his response to
my question “so what are you going to do?” was “we’re going to
buy an oil tanker off the coast of South Africa and sell it back on
Monday morning.” I think this is what you tell your grandchild when
you want them to be interested in “high finance” when they grow up.
I don’t know when this transaction took place, but it was prob-
ably in the late 1970s or early 1980s. Periods in which inflation was
running hot in the United Kingdom and a time when the prices of
physical commodities appreciated in real terms. I now know that
that’s a fairly unusual circumstance, when spot prices for crude oil
are expected to appreciate due to the contango of the forward curve,
and that “normal backwardation” would make this a money losing
trade. On the other hand, in the 1970s, there was probably a lot
more value to being on the receiving end of brokers’ phone calls than
now, so likely the information they were receiving was more useful
in a less efficient market. Of course, I didn’t know at the time that
I would become directly involved in plays on the structure of forward

curves, many years later.
1.2.3. Grandma, Bletchley Park, and Partial

Differential Equations
Jack was married to Lorna, and their hallway telephone table always
carried a black and white photo of them from their wedding day,
with him in his R.A.F. uniform. Unlike Grandad, Grandma didn’t
talk much about the war. She had strong political opinions and I
occassionally witnessed them arguing over policy issues of the day.
She was good at crosswords, and made cakes and pies, and the sort
of comfort food we all look to their grandparents for and remember
fondly after they are gone. When I stayed with them for a week or
so over the long summer holidays, she made the picnics that we took
to watch cricket.
Crosswords, played a much more significant role in her life than
that of many people. While Jack was serving overseas in the R.A.F.,
Lorna was doing something altogether different at home. She told
me that one day, when she was in a Mathematics class at university,
a “man from the ministry” (as the English used to say) came into
the lecture theatre and asked if anybody in the room was good at
crosswords. She raised her hand and a few days later was no longer a
student but a codebreaker at Bletchley Park. There is a scene in the
movie The Imitation Game that refers directly to this recruitment
strategy [133]. The movie, very early on, also includes a scene where
people were sliding sheets of celluloid over each other looking for
a coinciding light patch. This is exactly what she told me she did.
To me, at the time, this sounded nothing like what “codebreaking”
ought to be, but I was impressed nonetheless.
My mother tells me that earlier on she was less forthcoming about
her work. In fact, everybody at Bletchley Park had been sworn to
secrecy and “signed the Official Secrets Act.” And, like many of her
peers, she kept those secrets. Later, when people finally started talk-
ing about what is probably Great Britain’s biggest contribution to
winning that war, and it was covered in the press and on televi-
sion, she apparently turned to her family after a feature on the
evening news and said “well now you all know what I did in the
war.”

Figure 1.2: Lorna Baugh with her Bletchley Park Service Medal.
Photo Credit Carole Baugh.
I have another memory about Grandma Baugh and maths, one

that I’ll never forget. In my first year at Oxford, I struggled to keep
up with the work not having had the grounding in more advanced
mathematics in secondary school that many of my peers had had,
and that the Oxford Physics programme assumed. In Hilary Term,
I think, we were studying partial differential equations, the key tool
of classical physics. I just didn’t get it. I understood the general
principal of separation of variables as a method of solution, that
seemed clear, but the wave equation was bothering me. The wave
equation is
∂ 2f 2
2∂ f
= c (1.1)
∂t2 ∂x2
and the solutions are f (x + ct) and f (x − ct). To me that didn’t
seem to actually “solve” anything! When solving ordinary differ-
ential equations one finds actual solutions, not generic expressions.
The solutions to the Harmonic Oscillator equation, for example,
d2f
= −ω 2 f (1.2)
dt2
are sin ωt and cos ωt. Those are actually solutions, they tell you what
the form of the unknown function, f , actually is. At the time, I was
struggling to process what to do with such much more general solu-
tion, and then in walks Grandma with my parents, come to take me
to tea on a Saturday afternoon in Oxford.
“What are you doing?” ask Grandma, so I showed her my work.
“Oh, P.D.E.’s. I love them. I used to make them up and do them
for fun.” This succeeded in making me feel even more out of place.
My mother, who was a school teacher and history enthusiast and no
form of scientist whatsoever, always told me her mother was “not the
best teacher,” due to her tendency to declare “Oh that’s easy, I don’t
understand why you don’t understand this!”
1.2.4. Grandpa, Dad, and Caring about People

My father’s family were somewhat different. My father was first a
minister and later a school teacher, starting with Religious Education
and moving on to “high school” mathematics. His father was also a
minister, but that was not his first career, which was Organ Building.
That is the construction of musical instruments for churches, and
he was a talented pianist and lover of music. This was a “reserved
profession,” meaning exempt from the draft, so he wasn’t called up
into military service. However Grandpa, and we always called him
“Grandpa” to distinguish him from “Grandad,” did not sit aside. He
lived and worked in the East End of London (my father was born
within the sound of Bow Bells making him an genuine Cockney),
so he volunteered to join the London Fire Service and fought fires
during the nights in the London Docklands. An area that I would
later return to, in my financial career.
Growing up with a parent who is a practicing minister is an expe-
rience. One trait that the entire family seems to share is liking the
sound of their own voices, and I have always enjoyed public speak-
ing. This seems to be something that started early as, one Sunday,
my mother was surprised to find me standing in the pulpit of the

Figure 1.3: Gordon Pryce Giller (center), and his family, taken in London. It was
only after seeing this photo that I realized the origin of my somewhat unruly hair.
Photo via Susan Giller.
church, just reading the numbers of the hymns into the microphone
and being broadcast on the P.A. system.
We all have complicated relationships with our parents and I feel
that I get on much better with my father now, when I am 52 and he
is almost 81, than I did when I was a child. However, my most impor-
tant memory of him is only tangentially about me, and it dates from
when I was at home from college as a student. I woke, late at night,
to hear my father saying into the telephone “. . . and why do you
want to talk to Graham Giller?” It turned out that a schoolfriend’s
father had found a note, with my name and home phone number, in
his wife’s purse and was calling to challenge a suspected boyfriend.
That note must have been there for several years, for it dated from
when our schoolteachers were on strike and we had to leave school
at lunchtimes because the custodial staff were on strike. This was a
common occurrence in the 1980s. Every day, on the way from school
Another random document with
no related content on Scribd:
“I know it, I know it,” she answered, in a tone of such sweet
humility, as made Esmond repent that he should ever have
dared to reproach her. “I know how wicked my heart has been;
and I have suffered too, my dear. I confessed to Mr. Atterbury—I
must not tell any more. He—I said I would not write to you or go
to you—and it was better even that, having parted, we should
part. But I knew you would come back—I own that. That is no
one’s fault. And to-day, Henry, in the anthem, when they sang it,
‘When the Lord turned the captivity of Zion, we were like them
that dream,’ I thought yes, like them that dream—them that
dream. And then it went, ‘They that sow in tears shall reap in joy;
and he that goeth forth and weepeth, shall doubtless come again
with rejoicing, bringing his sheaves with him;’ I looked up from
the book and saw you. I was not surprised when I saw you. I
knew you would come, my dear, and saw the gold sunshine
round your head.”
She smiled an almost wild smile as she looked up at him.
The moon was up by this time, glittering keen in the frosty sky.
He could see for the first time now clearly, her sweet careworn
face.
“Do you know what day it is?” she continued. “It is the 29th
of December—it is your birthday! But last year we did not drink it
—no, no. My lord was cold, and my Harry was likely to die: and
my brain was in a fever; and we had no wine. But now—now you
are come again, bringing your sheaves with you, my dear.” She
burst into a wild flood of weeping as she spoke: she laughed and
sobbed on the young man’s heart, crying out wildly, “bringing
your sheaves with you—your sheaves with you!”
So they fare to the lit house, and to the tragedy which is the tragedy
of all womankind; of beauty fading while desire endures, the passion
to be loved persists; most tragic of all when a mother meets in a
daughter her careless conquering rival.
As they came up to the house at Walcote, the windows from

within were lighted up with friendly welcome; the supper-table
was spread in the oak parlour; it seemed as if forgiveness and
love were awaiting the returning prodigal. Two or three familiar
faces of domestics were on the look-out at the porch—the old
housekeeper was there, and young Lockwood from Castlewood,
in my lord’s livery of tawny and blue. His dear mistress pressed
his arm as they passed into the hall. Her eyes beamed out on
him with affection indescribable. “Welcome,” was all she said, as
she looked up, putting back her fair curls and black hood. A
sweet rosy smile blushed on her face; Harry thought he had
never seen her look so charming. Her face was lighted with a joy
that was brighter than beauty—she took a hand of her son, who
was in the hall waiting his mother—she did not quit Esmond’s
arm.
“Welcome, Harry!” my young lord echoed after her. “Here we
are all come to say so. Here’s old Pincot: hasn’t she grown
handsome?” and Pincot, who was older, and no handsomer than
usual, made a curtsey to the Captain, as she called Esmond,
and told my lord to “Have done, now.”
“And here’s Jack Lockwood. He’ll make a famous grenadier,
Jack; and so shall I; we’ll both ’list under you, Cousin. As soon
as I am seventeen, I go to the army—every gentleman goes to
the army. Look! who comes here—ho, ho!” he burst into a laugh.
“’Tis Mistress Trix, with a new ribbon; I knew she would put one
on as soon as she heard a captain was coming to supper.”
This laughing colloquy took place in the hall of Walcote
House, in the midst of which is a staircase that leads from an
open gallery, where are the doors of the sleeping chambers: and
from one of these, a wax candle in her hand, and illuminating
her, came Mistress Beatrix—the light falling indeed upon the
scarlet ribbon which she wore, and upon the most brilliant white
neck in the world.
Esmond had left a child and found a woman, grown beyond
the common height, and arrived at such a dazzling
completeness of beauty, that his eyes might well show surprise
and delight at beholding her. In hers there was a brightness so
lustrous and melting, that I have seen a whole assembly follow
her as if by an attraction irresistible: and that night the great
Duke was at the playhouse after Ramillies, every soul turned
and looked (she chanced to enter at the opposite side of the
theatre at the same moment) at her, and not at him. She was a
brown beauty: that is, her eyes, hair and eyebrows and eye-
lashes were dark: her hair curling with rich undulations, and
waving over her shoulders; but her complexion was as dazzling
white as snow in sunshine; except her cheeks, which were a
bright red, and her lips, which were of a still deeper crimson. Her
mouth and chin, they said, were too large and full, and so they
might be for a goddess in marble, but not for a woman whose
eyes were fire, whose look was love, whose voice was the
sweetest low song, whose shape was perfect symmetry, health,
decision, activity, whose foot as it planted itself on the ground
was firm but flexible, and whose motion, whether rapid or slow,
was always perfect grace—agile as a nymph, lofty as a queen—
now melting, now imperious, now sarcastic—there was no single
movement of hers but was beautiful. As he thinks of her, he who
writes feels young again, and remembers a paragon.
So she came holding her dress with one fair rounded arm,
and her taper before her, tripping down the stair to greet
Esmond.
“She hath put on her scarlet stockings and white shoes,”
says my lord, still laughing. “Oh, my fine mistress! is this the way
you set your cap at the Captain?” She approached, shining
smiles upon Esmond, who could look at nothing but her eyes.
She advanced holding forward her head, as if she would have
him kiss her as he used to do when she was a child.
“Stop,” she said, “I am grown too big! Welcome, Cousin
Harry,” and she made him an arch curtsey, sweeping down to
the ground almost, with the most gracious bend, looking up the
while with the brightest eyes and sweetest smile. Love seemed
to radiate from her. Harry eyed her with such a rapture as the
first lover is described as having by Milton.
“N’est-ce pas?” says my lady, in a low, sweet voice, still
hanging on his arm.
Esmond turned round with a start and a blush, as he met his
mistress’s clear eyes. He had forgotten her, rapt in admiration of
the filia pulcrior.
I have said some hard things, Gentlemen, upon Thackeray and

have indicated some dislike of him here and there, or, at least, some
impatience. But to the man who could at once so poignantly and so
reticently bring those two scenes into contrast—with all its meaning
—all meaning—modulated to so perfect a balance of heart and
intelligence wedded in human speech—well, to that man I conclude
by bowing the head, acknowledging a real master: a great
melancholy man with his genius running in streaks, often in thin
streaks about him but always, when uttered, uttered in liquid lovely
prose.
THE VICTORIAN BACKGROUND
I
I INTEND, in this and two following lectures, Gentlemen, taking my
illustrations in the main from Victorian times, to examine with you
how one and the same social question, urgent in our politics,
presented itself to several writers of imaginative genius, all of whom
found something intolerable in England and sought in their several
ways to amend it.
At the beginning of this enquiry let me disclaim any parti pris
about the duty of an imaginative writer towards the politics of his
age. Aristophanes has a political sense, Virgil a strong one even
when imitating Theocritus; Theocritus none: yet both are delightful:
Lucretius has no care for politics, Horace has any amount, and both
are delightful again: the evils of his time which oppress the author of
Piers Plowman, affect Chaucer not at all: Dante is intensely political,
Petrarch, far less sublime as a poet, disdains the business; Villon is
for life as it flies, Ronsard for verse and art (and the devil take the
rest); Spenser, with a sore enough political experience, casts it off
almost as absolutely as does Ariosto. Shakespeare has a strong
patriotic sense and a manly political sense: but he treats politics—let
us take King John and Coriolanus for examples—artistically, for their
dramatic value. He knows about
The oppressor’s wrong, the proud man’s contumely
and that they can be unendurable: but he does not use them for
propaganda (odious word!) whatever the minute of utterance. Milton
put all his religion into verse, his politics into prose; save for a
passage or two in Lycidas and Paradise Lost he excluded politics
from his high poetry. On the other hand Dryden had a high poetic
sense of politics, and it pervades the bulk of his original poetry, while
the opening of his famous Essay of Dramatic Poesy strikes an
introductory note as sure as Virgil’s, through whom a deep
undercurrent of politics runs from the first page of the Eclogues to
the last of the Æneid. Our poets of the eighteenth century were
social and political in the main: since if you once take Man for your
theme, you, or some one following you, must be drawn on irresistibly
to compare the position you assign him in the scheme of things with
his actual position in the body politic, to consider the “Rights of Man,”
“man’s inhumanity to man” and so forth. An Essay on Man (with the
philosophy Pope borrowed for it) leads on to The Deserted Village:
Ill fares the land, to hast’ning ills a prey,

Where wealth accumulates and men decay
—to Crabbe’s Poor House, Hall of Justice, Prison; to Blake’s lyrical

laments over small chimney-sweeps, blackamoors, foundlings and
all that are young and desolate and oppressed, and the vow to
sweep away “these dark Satanic mills” (of which I shall have more to
say by and by) “and build Jerusalem in England’s green and
pleasant land.” Turn now to Keats and you are returned upon mere
poetry, in the Latin sense of mere. Keats has no politics, no
philosophy of statecraft, little social feeling: he is a young apostle of
poetry for poetry’s sake.
Beauty is truth, truth beauty,—that is all

Ye know on earth, and all ye need to know.
But of course, to put it solidly, that is a vague observation—to

anyone whom life has taught to face facts and define his terms,
actually an uneducated conclusion, albeit most pardonable in one so
young and ardent. Let us, for a better, go on to the last and grandest
word of his last, unfinished, poem:
“High Prophetess,” said I, “purge off
Benign, if so it please thee, my mind’s film.”
“None can usurp this height,” returned the Shade,
“But those to whom the miseries of this world
Are misery, and will not let them rest.”
Such a spirit, preëminently, was Shelley; of whom, when the last

word of disparagement has been said, or the undeniable truth, put
into a phrase by Mr. Max Beerbohm, “a crystal crank,” the equally
undeniable fact remains that Shelley suffered tortures over the woes
of his fellow-creatures, while Byron (for a contrast) cares scarcely at
all for the general woe surrounding him, everything for his own
affliction in a world which had paid him tribute far above the earnings
of common men, and yet not only (as Shelley does) casts the blame
on tyrants and governments, but the cure for his egoistical troubles
on political machinery, revolutions. I go on, taking names and
illustrations almost at random. Contrast any Radical utterance of
Tennyson’s—his Lady Clara Vere de Vere, for example—with poor
Thomas Hood’s Song of the Shirt. Why, it fades away: Hood’s
passionate charity simply withers up the other’s personal self-
assertive inverted snobbery. If you have stuff in you, contrast the
note of
With fingers weary and worn,

With eyelids heavy and red,
A woman sat, in unwomanly rags,
Plying her needle and thread
with the whine of Lady Clara Vere de Vere—
The grand old gardener and his wife

Laugh at the claims of long descent
—which is just
When Adam delved, and Eve span,

Who was then the gentleman?
—on the pianola. Observe, pray, that I am not comparing the poetic
gift, in which (as in other gifts of the gods) Tennyson very greatly
outweighted Hood. I am merely setting some poets against others
and contrasting the degrees in which they exhibit social or political
sensitiveness. We should all allow, probably, that Robert Browning
was a greater poet and a stronger thinker than his wife: but probably
deny to him the acute indignation against human misery, social
wrong, political injustice, evinced by the authoress of The Cry of the
Children or Casa Guidi Windows. Of the two friends, Matthew Arnold
and Arthur Hugh Clough, we should as probably admit Arnold to be
the better poet as Clough to be the less occupied with his own soul,
the more in vain attempt to save other men. So again among the
Pre-Raphaelites Swinburne raves magnificently for the blood of
tyrants: but when it came to lifting the oppressed, to throwing himself
into the job, what a puff-ball was he beside William Morris who had
announced himself as no more than “the idle singer of an empty
day”!
One fishes in the night of deep sea pools:

For him the nets hang long and low,
Cork buoyed and strong: the silver gleaming schools
Come with the ebb and flow
Of universal tides, and all the channels glow.
Or holding with his hand the weighted line

He sounds the languors of the neaps,
Or feels what current of the springing brine
The cord divergent sweeps,
The throb of what great heart bestirs the middle deeps.
Thou also weavest meshes, fine and thin,
And leaguer’st all the forest ways:
But of that sea, and the great heart therein
Thou knowest nought: whole days
Thou toil’st, and hast thy end—good store of pies and jays.
II
So far we have spoken of poets—fairly selected, I trust—and
have found that there are poets and poets; and some are Olympian
in attitude, looking down deep below the surface from a great height
as a gannet spies his fish; but high aloof, concerned rather with
universal themes than with the woman of Canaan clamorous in the
street crying for her daughter, “Truth, Lord: yet the dogs eat of the
crumbs which fall from their masters’ table.”
Now if we turn to our novelists, from Defoe to Scott, we find that
the novel from its first virtual beginning in our country and for a
century or more, has for social diseases in the body politic little
concern and practically no sense at all. Defoe has strong political
sense, but keeps it for his tracts and pamphlets: in Robinson Crusoe
(and specially in the third volume, The Serious Reflections of
Robinson Crusoe), in Moll Flanders, in Roxana, he is always a
moralist, but a religious moralist. If—to twist a line of Hamlet—there’s
something rotten in the state of Denmark, it does not come within the
scope of the novelist whose office is to combine amusement with
general edification. So—leaving out the edification—it is in Tristram
Shandy, so in The Vicar of Wakefield. Richardson is all for the
human heart as he reads it, and female virtue. Fielding with his
genial manly morality—Fielding, magistrate of a London Police
Court, and a humane one—discloses little sense in his novels of any
vera causa in our system supplying the unfortunates for whom, in
daily life, he tempers justice with mercy. You will not, I think, cite
Jonathan Wild against me. Noble fellow, as he drops down the
Thames—stricken to death, and knowing it—on that hopeless
voyage to Lisbon, his thoughts are hopeful for England and the glory
of her merchant shipping: and (says he) it must be our own fault if it
doth not continue glorious:
for continue so it will, as long as the flourishing state of our trade

shall support it, and this support it can never want, till our
legislators, shall cease to give sufficient attention to the
protection of our trade, and our magistrates want sufficient
power, ability, and honesty to execute the laws: a circumstance
not to be apprehended, as it cannot happen till our senates and
our benches shall be filled with the blindest ignorance, or with
the blackest corruption.
Smollett’s recipe for a novel is just a rattling picaresque story

enlivened by jocular horse-play. Respect Fanny Burney and idolise
Jane Austen as we will, they move their plots on a narrow and
sheltered stage: while the romantics, working up from Horace
Walpole to Scott, call in the past to redress the poverty of the present
and the emptiness of a general theory of the arts which, deservedly
sovereign in its day, has passed by imitation into convention, and
through convention, as always, into mere inanition.
III
Now if you will take, as a convenient starting-point for your
enquiry, the year 1832—the year that saw the passing of the Great
Reform Bill and the death of Scott: if you will start (I say) with that
year beyond which, when I first made acquaintance, with the English
School here, our curiosity was forbidden to trespass—you will find
that then, or about then, certain terrible diseases in our
Commonwealth were brewing up to a head. As everyone now
recognises, we must seek the operating cause of these in what we
now agree to call the “Industrial Revolution”; that is in the process as
yet unrestricted by law, encouraged by economic theory, moving at
once too fast for the national conscience to overtake or even to
realise it and with a step of doom as rigidly inexorable as the
machinery, its agent and its symbol, converting England into a
manufacturing country, planting the Manchester of those days and
many Manchesters over England’s green and pleasant land, and
leaving them untended to grow as they pleased polluting her
streams, blackening her fields, and covering—here lies the
indictment—with a pall of smoke, infinite human misery: all this
controlled and elaborated by cotton-lords and mine-owners who
prospered on that misery.
The plight of rural, agricultural, England is another story. Here in
Lancashire, Derbyshire, Yorkshire was a monstrous revolution
gathering strength (as I say) beyond men’s power even to realise it.
And if they realised it, there was Political Economy assuring them
that it had to be. And it continued (as you will remember) long after
poor Wragg strangled and left her illegitimate infant on the dismal
Mapperly hills and the egregious Mr. Roebuck asked, if, the world
over or in past history, there was anything like it. “Nothing. I pray that
our unrivalled happiness may last.”
We all recognise it now, and the wicked folly of it—or at least I
hope we do. My purpose to-day, Gentlemen, is not to excite vain
emotions over a past which neither you nor I can remedy at all, but
simply to show that—as, after all, we are a kindly nation—the
spectacle of industrial England about and after 1832 became
intolerable to our grandfathers: how it operated upon two
extraordinarily different minds: and (if I can) how irresistible is the
wind of literature, through what mouthpiece soever it breathes with
conviction.
IV
But before examining how two of the most dissimilar minds
conceivable—one a man’s, the other a woman’s—reacted upon it, I
must indicate the enormity of the challenge.
France had passed through her Revolution and her Terror, with
graphic details of which our public speakers and writers had taken
pains to make our country familiar enough: and England had won out
of the struggle, having taken the side she chose, all oblivious (as we
are, maybe, to-day) that victory in arms is at best but the beginning
of true victory, and that she herself was in the throes of a revolution
not a whit the less murderous than that of France, and only less
clamant because its victims, instead of aristocrats and politicians and
eminent saviours of their country following one another by scores in
tumbrils to die scenically in the Place de la République, the Place of
the Guillotine, were serfs of the cotton-mill and the mine, wives,
small children, starved unscenically, withered up in foetid cellars or
done to death beside the machines of such a hell-upon-earth as
Manchester had grown to be out of towns in which an artificer,
however humble, had once been permitted to rejoice in that which
alone, beyond his hearth and family, heartens a man—the well-
executed work of hand and brain. The capitalists of that time simply
overwhelmed these towns, expanding, converting them into barracks
for workers. Who these workers were, let an advertisement in a
Macclesfield paper of 1825 attest—
To the Overseers of the Poor and to families desirous of

settling in Macclesfield. Wanted between 4,000 and 5,000
persons between the ages of 7 and 21 years.
Yes, let us pass the hideous towns with but one quotation, from
Nassau Senior—
As I passed through the dwellings of the mill-hands in Irish

Town, Ancoats and Little Ireland, I was only amazed that it was
possible to maintain a reasonable state of health in such homes.
The towns, for in extent and number of inhabitants they are
towns, have been erected with the utmost disregard of
everything except the immediate advantage of the speculative
builder.... In one place we found a whole street following the
course of a ditch, because in this way deeper cellars could be
secured without the cost of digging, cellars not for storing wares
or rubbish, but for dwellings of human beings. Not one house in
the street escaped the cholera.
“Such,” wrote Chadwick, that careful observer, “is the absence of

civic economy in some of our towns that their condition in respect of
cleanliness is almost as bad as that of an encamped horde or an
undisciplined soldiery.”
But from the poor men and women—who had sold themselves
into these slums and industrial slavery—let us turn to their hapless
children, who, after all, had never asked to be born. Your Malthus in
that age, and your Mr. Harold Cox in this, are positive (God forgive
them!) that a number of these brats never ought to be born. (I don’t
know the price of millstones, but they ought to be cheap and handy,
and properly labelled.) I shall lay stress on these children,
Gentlemen, because—as children do so often—they brought back
the gospel—or something of it. For these weaklings, as they were
the foundation of the manufacturer’s wealth, by their illimitable woe
enabling him to cut his wages, in the end brought about his
exposure. To us—for always to us in our day the past wears a haze
softening it into sentiment—Elizabeth Barrett Browning’s Cry of the
Children is nothing, or suspected as sentimental, to be classed
alongside with anything (say) by Mrs. Hemans or L. E. L. Listen to a
couple of stanzas or three—
“For oh,” say the children, “we are weary,

And we cannot run or leap;
If we cared for any meadows, it were merely
To drop down in them and sleep.
Our knees tremble sorely in the stooping,
We fall upon our faces, trying to go;
And underneath our heavy eyelids drooping
The reddest flower would look pale as snow.
For, all day, we drag our burden tiring
Through the coal-dark, underground;
Or, all day, we drive the wheels of iron
In the factories, round and round.
For all day the wheels are droning, turning;

Their wind comes in our faces,
Till our hearts turn, our head with pulses burning,
And the walls turn in their places:
Turns the sky in the high window, blank and reeling,
Turns the long light that drops adown the wall,
Turn the black flies that crawl along the ceiling:
All are turning, all the day, and we with all.
And all day the iron wheels are droning,
And sometimes we could pray,
‘O ye wheels’ (breaking out in a mad moaning),
‘Stop! be silent for to-day!’”
And well may the children weep before you!

They are weary ere they run;
They have never seen the sunshine, nor the glory
Which is brighter than the sun.
They know the grief of man, without its wisdom;
They sink in man’s despair, without its calm;
As slaves, without the liberty in Christdom,
As martyrs, by the pang without the palm....
Let them weep! let them weep!

They look up with their pale and sunken faces,
And their look is dread to see,
For they mind you of their angels in high places,
With eyes turned on Deity.
“How long,” they say, “how long, O cruel nation,
Will you stand, to move the world, on a child’s heart,—
Stifle down with a mail’d heel its palpitation,
And tread onward to your throne amid the mart?
Our blood splashes upward, O gold-heaper,
And your purple shows your path!”
But the child’s sob in the silence curses deeper
Than the strong man in his wrath.
V
Now, I dare say some of you, even while I read this, were
dismissing it in your minds as early-Victorian humanitarianism, faded
philanthropy, outworn sentiment. Yes, but even a sentiment, if it
works simultaneously upon a generation of great and very dissimilar
writers, is a fact in the story of our literature—a phenomenon, at
least, which made itself an event—to be studied by you scientifically.
One of the first rules of good criticism, and the sheet-anchor of the
historical method, is to put yourself (as near as may be) in the other
fellow’s place: and if you take but a very little pains to do so, you will
soon discover that Mrs. Browning was not writing “for the fun of the
thing,” exuding, or causing to be exuded, any cheap tears. We are
accustomed to Manchester to-day: we take it for granted as a great
community with a most honourable Press to represent its opinions.
But we only take it for granted because it has become tolerable, and
it only became tolerable, then dignified—it only became a city—
because our Victorian writers shamed its manufacturers out of their
villainies. In the twenties, thirties, and “hungry forties” of the last
century Manchester was merely a portent, and a hideous portent, the
growth of which at once fascinated our economists and frightened
our rulers. Think of the fisherman in the Arabian Nights who,
unstopping the bottle brought ashore in his net, beheld a column of
smoke escape and soar and spread, and anon and aloft, overlooking
it, the awful visage of a Genie. Even so our economists watched an
enormous smoke ascend from Manchester and said, “Here is
undreamed-of national prosperity”; while our ministers stared up into
the evil face of a monster they had no precedent to control. You
understand, of course, that I use “Manchester” as a symbolic name,
covering a Lancashire population which grew in the first twenty years
of the century from 672,000 to 1,052,000. But let a very different
person from Mrs. Browning—let Benjamin Disraeli, then a young
man, describe the portent.
From early morn to the late twilight our Coningsby for

several days devoted himself to the comprehension of
Manchester. It was to him a new world, pregnant with new trains
of thought and feeling. In this unprecedented partnership
between capital and science—
Mark you, not between capital and labour, but between capital and
science, still by machinery arming capital to vaster strength—
In this unprecedented partnership between capital and science,

working on a spot which Nature had indicated as the fitting
theatre for their exploits, he beheld a great source of the wealth
of nations which had been reserved for these times, and he
perceived that this wealth was rapidly developing classes whose
power was imperfectly recognised in the constitutional scheme,
and whose duties to the social system seemed altogether
omitted
—“and whose duties to the social system seemed altogether

omitted.” There, in Disraeli’s words, you have it. Every prolonged war
raises a new governing class of prosperous profiteers who turn their
country’s necessity to glorious gain. So it was a hundred years ago
at the conclusion of the long Napoleonic struggle: so it is to-day. So it
goes on ever. A profiteering class of speculators and (as Cobbett
would say) “loan-mongers” emerges at the top of any great war. Ex-
soldiers tramp the roads for work, for bread. Decent folk, bred in the
incurable belief that England, whoever suffers, must pay her debts,
sell out and suffer, breaking up old homes, cutting neighbourly ties,
disappearing, taxed out of endurance, electing to suffer, for honour’s
sake. Succeeds a generation or two which, at school or University,
are baptised into the old honourable cult. The gravity of an
Englishman, because they are English after all, revives and takes
possession of young hearts, made generous by education, forgetful
of old woes. And so in time—give it a couple of generations—the
descendants of the sponge and the parvenu will have shed the hair
from the hoof, will leap to the summons of noblesse oblige, and in
their turn make haste to die by Ypres or the Somme, transmitting
somehow the mettle of England into a future denied to them.
VI
But you will say that, although this revolt in the better minds of
England, a hundred years ago, may be a fact, I have as yet quoted
but the evidence of a poetess and a novelist. Very well, then: I go to
Blue Books and the reports of several commissions, reminding you
that I lay most stress on the children because it happened through
their almost inconceivable sufferings that, such as it was, victory
came.
In 1831 Michael Sadler (a great man, in spite of Macaulay, and
the ancestor of a great one—if I may insert this word of long
admiration for the first senior man who spoke to me at my first
undergraduate dinner in Hall, more than forty years ago)—in 1831
this Michael Sadler, member for Newark, introduced a Ten Hours Bill,
and moved its second reading in a speech that roundly exposed,
along with other woes of the poor, the sacrifice of child life in the
mills. The Bill was allowed a second reading on condition that the
whole subject should be referred to a Select Committee, over which
Sadler presided.
Now let me quote a page from Mr. and Mrs. Hammond’s recently
published study of Lord Shaftesbury, then Lord Ashley, who, though
so many have laughed at him, devoted his life that they should laugh
if they chose, but willy-nilly on the right side of their mouths, and not
with a grin unacceptable to any Divinity presumed as having created
Man in His image—
The Report of Sadler’s Committee is a classical document; it

is one of the main sources of our knowledge of the conditions of
factory life at the time. Its pages bring before the reader in the
vivid form of dialogue the kind of life that was led by the victims
of the new system. Men and women who were old at twenty,
from all the industrial districts, from Manchester, from Glasgow,
from Huddersfield, from Dundee, from Bradford, from Leeds,
passed before their rulers with their tale of weariness, misery,
and diseased and twisted limbs. A worsted spinner of
Huddersfield, Joseph Hebergram, aged seventeen, described
his day’s work at the age of seven. His hours were from five in
the morning to eight at night, with one solitary break of thirty
minutes at noon. All other meals had to be taken in snatches,
without any interruption of work. “Did you not become very
drowsy and sleepy towards the end of the day and feel much
fatigued?” “Yes; that began about three o’clock; and grew worse
and worse, and it came to be very bad towards six and seven.”
“What means were taken to keep you at your work so long?”
“There were three overlookers; there was one a head
overlooker, and there was one man kept to grease the
machines, and there was one kept on purpose to strap.” His
brother, who worked in the same mill, died at sixteen from spinal
affection, due to his work, and he himself began to grow
deformed after six months of it. “How far do you live from the
mill?” “A good mile.” “Was it very painful for you to move?” “Yes,
in the morning I could scarcely walk, and my brother and sister
used, out of kindness, to take me under each arm, and run with
me to the mill, and my legs dragged on the ground; in
consequence of the pain I could not walk.” Another witness, an
overseer in a flax spinning mill at Dundee, said that there were
nine workers in the room under his charge who had begun work
before they were nine years old, and that six of them were splay-
footed and the other three deformed in other ways. A tailor at
Stanningley, Samuel Coulson, who had three daughters in the
mill, described the life of his household when the mill was busy.
In the ordinary time the hours were from six in the morning to
half-past eight at night; in the brisk time, for six weeks in the
year, these girls, the youngest of them “going eight,” worked
from three in the morning to ten or half-past ten at night. “What
was the length of time they could be in bed during those long
hours?” “It was near eleven o’clock before we could get them
into bed after getting a little victuals, and then at morning my
mistress used to stop up all night, for fear that we could not get
them ready for the time; sometimes we have gone to bed and
one of us generally awoke.” “Were the children excessively
fatigued by this labour?” “Many times; we have cried often when
we have given them the little victualling we had to give them; we
had to shake them, and they have fallen asleep with the victuals
in their mouths many a time.”
Another witness, Gillett Sharpe, described how his boy, who
had been very active and a good runner, gradually lost the use
of his limbs at the mill. “I had three steps up into my house, and I
have seen that boy get hold of the sides of the door to assist his
getting up into the house; many a one advised me to take him
away; they said he would be ruined and made quite a cripple;
but I was a poor man, and could not afford to take him away,
having a large family, six children under my care.”
—and so on, and so on. Sadler forced the horrible tale upon
Parliament. Unhappily, being pitted against Macaulay at Leeds in the
General Election of 1832, he lost his seat, though Manchester sent
an appeal signed by 40,000 factory-workers: and he never returned
to the House of Commons. He died in 1835 at fifty-five, worn out by
his work on behalf of these poor children.
VII
His mantle descended to Lord Ashley: and Ashley, after bitter
defeats, won on the mine-children what had been lost in the cotton-
mills. For the mines took an even more hideous toll of childhood than
did the mills. Listen to this, extracted from the Report of the
Commission of 1840–1842, which shocked all England by its
disclosures—
In every district except North Staffordshire, where the

younger children were needed in the Potteries, the employment
of children of seven was common, in many pits children were
employed at six, in some at five, and in one case a child of three
was found to be employed. Even babies were sometimes taken
down into the pits to keep the rats from their fathers’ food. The
youngest children were employed as trappers; that is, they were
in charge of the doors in the galleries, on the opening and
closing of which the safety of the mine depended. For the
ventilation of the mine was contrived on a simple principle; there
were two shafts, one the downcast, the other the upcast. A fire
was lighted at the foot of the upcast to drive the air up the shaft,
and air was sucked down through the downcast to fill the
vacuum. This air was conducted by means of a series of doors
through all the workings of the mine on its passage to the
upcast, and these doors were in the charge of a little boy or girl,
who sat in a small hole, with a string in his or her hand, in
darkness and solitude for twelve hours or longer at a time.
“Although this employment,” reported the Commission, “scarcely
deserves the name of labour, yet as the children engaged in it
are commonly excluded from light, and are always without
companions, it would, were it not for the passing and re-passing

Adventures in Financial Data Science: The Empirical Properties of Financial and Economic Data, 2nd Edition Graham L. Giller

Uploaded by

Copyright:

Available Formats

You might also like

Adventures in Financial Data Science: The Empirical Properties of Financial and Economic Data, 2nd Edition Graham L. Giller

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

Adventures in Financial Data Science: The Empirical Properties of Financial and Economic Data, 2nd Edition Graham L. Giller

Uploaded by

Copyright:

Available Formats

Adventures in Financial Data Science:

The Empirical Properties of Financial

Analyzing Financial Data and Implementing Financial

A Solution Manual for Statistics and Data Analysis for

Data Quality Engineering in Financial Services (Early

Statistics of Earth Science Data Their Distribution in

Decision Making with Quantitative Financial Market Data

Data-Driven Science and Engineering Steven L. Brunton

Beginning Data Science in R 4: Data Analysis,

Risk Analytics: From Concept to Deployment (World

12678_9789811250644_tp.indd 1 6/5/22 7:38 AM

Series Editor: William T. Ziemba (University of British Columbia (Emeritus) and

Vol. 19 Adventures in Financial Data Science: The Empirical Properties of

Vol. 18 Sports Analytics

Vol. 17 Investment in Startups and Small Business Financing

Vol. 16 Cultural Finance: A World Map of Risk, Time and Money

Vol. 15 Exotic Betting at the Racetrack

Vol. 14 Dr Z’s NFL Guidebook

Soundararajan - 12678 - Adventures in Financial Data Science.indd 1 5/2/2022 2:23:15 pm

12678_9789811250644_tp.indd 2 6/5/22 7:38 AM

Library of Congress Cataloging-in-Publication Data

Classification: LCC HG4515.5 .G55 2022 | DDC 332.640285--dc23/eng/20220131

British Library Cataloguing-in-Publication Data

Copyright © 2022 by World Scientific Publishing Co. Pte. Ltd.

For any available supplementary material, please visit

Desk Editors: Soundararajan Raghuraman/Pui Yee Lum

Typeset by Stallion Press

Soundararajan - 12678 - Adventures in Financial Data Science.indd 2 5/2/2022 2:23:15 pm

For my wife, for her support,

I am grateful to the many friends I have found over my years

This page intentionally left blank

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

This revised second edition of Adventures in Financial Data Science

is being written as over half of America has been vaccinated against

viii Adventures in Financial Data Science

About the Author

Graham Giller is one of Wall Street’s original

data scientists. Starting his career at Morgan

This page intentionally left blank

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

About the Author ix

Chapter 1. Biography and Beginnings 1

Chapter 2. Financial Data 31

xii Adventures in Financial Data Science

2.9 Microwave Latency Arbitrage . . . . . . . . . . . . . 131

Chapter 3. Economic Data and Other

Chapter 4. Politics, Schools, Public Health,

4.2 School Board Elections . . . . . . . . . . . . . . . . . 240

Chapter 5. Demographics and Survey Research 285

Chapter 6. Coronavirus 347

Chapter 7. Theory 395

E.3 Summing Things Up . . . . . . . . . . . . . . . . . . 451

Appendix A. How I Store and Process Data 453

Appendix B. Some of the Data Sources

This page intentionally left blank

B1948_1-Aoki.indd 6 9/22/2014 4:24:57 PM

1.1 The author at The Blue Coat School, Coventry, 1987. . 3

1.2 Lorna Baugh with her Bletchley Park Service Medal. . 6