Introduction to Monte Carlo Methods

 The Monte Carlo simulation

method was introduced in the late
1940s
 This simulation method was first
used in nuclear physics during
World War II
 Now Monte Carlo simulations are
used in
 Mathematics
 Physical sciences
 Engineering
 Climate change and
radiation
 Biological sciences
 Graphics and statistics
 Artificial Intelligence for
games, etc.
 Designs and visials
 Finance and business
 Law
1
Introduction to Monte Carlo Methods
For example:

2
Introduction to Monte Carlo Methods
What is Monte Carlo Simulation

imitate the Something that is

appearance or always changing
character of and is not steady

• It is a method where we simulate the behavior of a variable to enable us to

model the future impact of the variable
Having an effect
or influence on
something

• The essential idea is just to use repeated random sampling in order to obtain a
reliable result.
happening without
method or
conscious decision

• It is very useful when other approaches in mathematics are difficult or

impossible to be applied

3
Introduction to Monte Carlo Methods
What is Monte Carlo Simulation An example

Predicting the number emergency calls :

in order to optimize emergency medical services and reduce response time for patients

We can analyze past Days of the

week
Morning Afternoon Evening & Night

historical pattern of data 𝝁= 𝝁 = 25 𝝁 = 64

𝝈 = 1.3 𝝈 = 4.7
Monday

𝝈=
52.9
Tuesday
• Based on the last 1 or 2 years (> Wednesday 4.86
600 results) Thursday
• Get the mean (average) no of calls Friday
for each period and the standard Saturday
deviation (how far from the mean) Sunday

OR
we can simulate the future likelihood
of calls based on the historical pattern
of data – using Monte Carlo
simulation
4
 Introduction to Monte Carlo Methods
What is Monte Carlo Simulation An example

Question: Why do we need to simulate the variable, rather than just looking at its
historical behavior?

Historical discrete data over 775 days

𝝁 = 52.9 • Only 1 path of history (1 trial) is
250 𝝈 = 4.86 taken

200 • Assumes consistency over time

(history will repeat; the next
No of days

150 775 days will repeat in the same

manner)
100

50

0 - 20 20 - 40 40 - 60 60 - 80 80 - 100 100 - 120

No of calls in on Monday mornings 5
 Introduction to Monte Carlo Methods
What is Monte Carlo Simulation An example

Now, if we simulate the variable (no of

Historical discrete data over 775 days calls)

𝝁 = 52.9
𝝈 = 4.86
250 - we will randomly simulate the no. of
calls in the 775 days
200 - And repeat this many, many times
to represent different paths (trials)
No of days

150 - But within the possible probability

space
100 - And get the average calls

50

0 - 20 20 - 40 40 - 60 60 - 80 80 - 100 100 - 120

No of calls in on Monday mornings 6
Introduction to Monte Carlo Methods
What is Monte Carlo Simulation An example

Random numbers within the

probability space

Day
No of emergency calls on Monday mornings 1 2 3 4 ….775 Average
Trial 1 20 40 76 67 51
Trial 2 94 58 75 120 87
59
Trial 3 15 30 58 104 52 simulated
Trial 4 6 118 31 25 45 average
Trial 5 51 26 119 101 74
Trial 6 23 25 77 19 36
calls on
Trial 7 56 104 110 3 68 Monday
Trial 8 8 25 69 109 53 mornings
Trial 9 95 69 97 102 91
Trial 10 34 109 69 16 57
Trial 11 12 2 27 25 17
Trial 12 114 90 36 49 72
.
.
.. Trial 10,000
7
Introduction to Monte Carlo Methods
Understanding probability spaces

Poisson Distribution Normal Distribution

Uniform Distribution Bimodal Distribution

8
Introduction to Monte Carlo Methods
Applying Monte Carlo to Business Risk and Finance

Monte Carlo Simulations can be used to:

 Predict the possible future cost of a project

 Predict the possible profit of a project / business
 Estimate the Value at Risk in a particular project or Profit of a company
 Estimate the worst case scenario in operational risk
 Predict the scheduling risk in a project
 Predict the financial risk or technical risk in a project

 …anything can be predicted, as long as we have enough data to define the

probability space of the variable in a reliable manner

9
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Company Galaxy is developing a new smartphone named HomTom. The company wants to
predict the first year profits of this new phone by considering the following factors:

• Selling price per unit (p). The price for the new phone is RM 249 each.

• First-year administrative and advertising costs (ca). Currently, this cost is RM 1,000,000.

• direct labor cost per unit (ci).

• parts cost per unit (cp). Variable factors

• first-year demand (d).

fixed
10
Derive a simple formulae for profit Source: Bos Class: Data analysis modeling and analytical software (R, Python)
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Scenario based analysis:

Base Case Profit = (249 - 45 - 90)*(15,000) - 1,000,000 = 710,000

Best Case Profit = (249 - 43 - 80)*(30,000) - 1,000,000 = 2,780,000

Worst Case Profit = (249 - 47 - 100)*(0) - 1,000,000 = -1,000,000

• This analysis only estimates 3 possible events with no mention of probability of

each event happening
• Having only 3 predictions could have missed out various other possibilities

11
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 1 – Find out each variable’s (unknown factor) probability distributions

Direct Labour cost has a discrete distribution (not continuous)

Direct Labor Probability

Cost

$43 0.1
$44 0.2
$45 0.4
$46 0.2
$47 0.1

12
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 1 – Find out each variable (unknown factor)’s probability distributions

Part cost has a uniform distribution

13
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 1 – Find out each variable (unknown factor)’s probability distributions

Demand is not a discrete distribution, but a continuous distribution

Total of the
Demand units has a normal distribution probability
area under
Probability the curve is
ONE
0.4

0.3

0.2

0.1

14
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 2 – Generate a random number for every variable from their respective probability
distributions

c(i) – direct labour cost

Direct Labor Probability

Cost
$43 0.1
$44 0.2
$45 0.4
$46 0.2
$47 0.1

15
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 2 – Generate a random number for every variable from their respective probability
distributions

c(p) – part cost per unit Simply the

Lowest cost + prob (cost range)

80 + prob (100 – 80)

Random number
between 0 and 1

=rand() ..say is 0.3

Part cost will be

80 + 0.3 ( 100- 80) = 86

16
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 2 – Generate a random number for every variable from their respective probability
distributions

Demand To simulate the number of units sold, we

• Randomly choose a probability level
between 0 and 1
• With that probability level, find the
corresponding number of units sold
Mean or Standard
probability deviation
average

= NORM.INV(rand(), 15000, 4500)

17
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 3 – Compute the simulated profit from the 1st trial

Factor Random Number
Direct Labor Cost per Unit 0.4
43
Part Cost per Unit 0.3
Demand 0.6

Direct Labour cost per unit = RM43

Part cost per unit = 80 + 0.3 ( 100- 80) = RM86

Demand : (Use =NORM.INV(0.6,15000,4500) function) = 16,140

Simulated Profit = (249 – 43 – 86) x 16,140 – 1,000,000 = 936,800

18
Introduction to Monte Carlo Methods
Application example – determining profitability of a business

Applying Monte Carlo Simulation Technique:

Step 4 – Repeat the same steps 2 & 3 and get thousands of trials.

With the simulated profits, you can do several analysis, for example:

 Find the average profit and its standard deviation

 Find the best case and worst case profit
 Find the distribution pattern of the profit
 Find the top and bottom 10% best/ worst profit scenarios
 Find the 80% most likely of profits to be
 etc
19
Introduction to Monte Carlo Methods
Advantages and Disadvantages

Advantages

 We can simulate realities that have different scenarios with different probabilities.
 We can simulate realities in which real experiments are impossible (e.g. climate change)
 Addresses some analysts’ narrow subjective opinions
 A good substitute when (i) data is scarce ; (ii) using deterministic mathematical formulas are
cumbersome

Disadvantages

 Can take a lot of computing power – but now with the “cloud” it gets easier
 Simulated results can go horribly wrong if the underlying distributions of the simulated
variables (unknown) are not truly represented

20