Professional Documents
Culture Documents
Monte Carlo Simulation
Monte Carlo Simulation
Monte Carlo Simulation
Simulation
Natalia A. Humphreys
April 6, 2012
University of Texas at Dallas
Aknowledgement
Wayne L. Winston, Microsoft Excel Data
Overview
Part I
Questions answered with the help of MCS
History
Typical simulations
analysis
Challenges
We are constantly faced with uncertainty,
Questions answered
with the help of MCS
How should a greeting card company
portfolio?
individuals
MCS Use
General Motors (GM), Procter and Gamble
MCS Use: GM
Forecast net income for the corporation
Predict structural costs and purchasing costs
investment portfolios.
By definition, Value at Risk at security level p
Pr(X<VaR_p(X))=p
In practice, p is selected to be close to 1: 95%,
99%, 99.5%
risk
postpone a project
MCS Applications
Physical Sciences
Engineering
Computational Biology
Applied Statistics
Games
Design and visuals
Finance and business (Actuarial Science)
Telecommunications
Mathematics
Part II
Well now discuss how Monte Carlo simulation
=RAND() function
When you enter the formula =RAND() in a
the time
Example 1: Discrete
Random Variable
Simulation
Demand for a calendar is governed by the
DEMAND
PROBABILITY
10,000
0.10
20,000
0.35
40,000
0.30
60,000
.25
Discrete r.v.
Simulation(cont.)
How can we have Excel play out, or simulate,
20,000
40,000
60,000
CUTOFF DEMAND
0
10,000
0.1
20,000
0.45
40,000
0.75
60,000
1
2
3
4
SIM
DEMAND
0.82309742 60,000
2
0.07607429 10,000
8
0.36420163 20,000
4
0.69811636 40,000
5
col_index_num, range_lookup )
1)=60,000
TRUE=1, FALSE=0
If VLOOKUP can't find lookup value, and
FRACTION
OF TIME
DEMAND
PROBABIL
ITY
10,000
0.10
10,000
0.10250
20,000
0.35500
20,000
0.35
40,000
0.29250
40,000
0.30
60,000
0.25000
60,000
0.25
Example 2: Normal
Random Variable
Simulation
Suppose we want to simulate 400 trials or
variable.
Standard Normal
Random Variable
Its distribution has a form of a bell curve
and =1
Connection between
any Normal r.v. and a
Standard Normal r.v.
If Z is N(0, 1) and is Y is N(, ^2), then
Y=Z+
Normal Random
Variable Simulation
Suppose we want to simulate 400 trials or
TRIAL
RAND
NORMAL RV
0.258433031
33,518.16
0.344835199
36,006.98
0.927522163
54,575.82
0.248403053
33,204.76
PROBABILIT
Y
10,000
0.10
20,000
0.35
40,000
0.30
60,000
.25
$1.50
Leftover cards will be disposed at $0.20 per
card
produced
2
3
4
5
6
7
8
rand
demandcard
unit prod cost
unit price
unit disp cost
revenue
total var cost
9
10
10,000
0.40092709
1
20,000
$1.50
$4.00
$0.20
$40,000.00
$15,000.00
VLOOKUP(rand, lookup, 2)
The number of unites sold is
price
Total production cost in 8: produced*unit
production cost
If we produce more cards than are
Cards to Produce?
(cont.)
Disposal cost in 9:
Cards to Produce?
(cont.)
24,985
45,984
57,311
44,218
Cards to Produce?
(cont.)
Enter 1-1000 on the left corresponding to our
1,000 trials
Cards to Produce?
(cont.)
Cards to Produce?
Conclusion
Producing 40,000 cards always yields the
Confidence Interval
for Mean Profit
Into what interval are we 95% sure the true mean
will fall?
This interval is called the 95% confidence interval
Problems
1
Problems (cont.)
2
PROBABILITY
15
0.10
20
0.20
25
0.30
30
0.25
35
0.15
Problems (cont.)
2
Advantages of MCS
MCS provides a number of advantages over
deterministic, or single-point estimate
analysis:
Probabilistic Results
Graphical Results
Sensitivity Analysis
Scenario Analysis
Correlation of Inputs
Probabilistic Results
Results show not only what could happen, but
Graphical Results
Because of the data a Monte Carlo simulation
to other stakeholders.
Sensitivity Analysis
With just a few cases, deterministic analysis
Scenario Analysis
In deterministic models, its very difficult to
analysis.
Correlation of Inputs
In Monte Carlo simulation, its possible to
References
Wayne L. Winston, Microsoft Excel Data
http://www.palisade.com/risk/monte_carlo_si
mulation.asp