You are on page 1of 16

Risk

Taking
The devil is in the details
Standard deviation
• Average distance from the mean
• Generally considered as a measure of variation.

Standard deviation (σ) =

• Mean absolute deviation (MAD)

MAD =

2
SD vs. MAD

• The effect of squaring


• Squaring the difference leads to inflated dispersion
• Sample values of SD and MAD
• Excel exercise with calculation of SD and MAD of a set of numbers
• Is there a relationship between the difference in SD and MAD?

• Why is SD so important then?

First project task

3
Probability distributions
• What are probability distributions
• Consider the data of student marks out of 20
[12 14 14 17 18]

• Different ways to analyze this information


• Max/min, center, dispersion, greater/less than a certain value

• Probability distributions allow us to analyze data


• Determining the shape of the distribution is critical

9/3/20XX 4
Normal (gaussian) distribution
• Exercise (excel): coin toss with 20 vs. 30 samples
• Observe the shape with 20 tosses vs. 30 tosses
• Is 30 the magic number?

9/3/20XX 5
Generating a distribution

• Simulate using excel and R


Excel: use the NORMINV(rand(), mean, sd) function
R: rnorm(n, mean, sd)

• Predicting probability P(x > k)


• You can manually count the number of observations above the desired point

9/3/20XX Presentation Title 6


Estimating parameters

• Sample size
• Estimating population mean and standard deviation

• Greater the sample size, better the estimation

9/3/20XX Presentation Title 7


Moments of a distribution

• Moments describe the shape of a distribution


• We can have multiple moments, but the first 4 moments are often sufficient

• 1st moment: Average distance from 0

• 2nd moment: Average squared distance from the mean

• 3rd moment: Average cubed distance from the mean

• 4th moment: Average fourth-powered distance from the mean

9/3/20XX Presentation Title 8


Moments
• Example: Alternate sets of data about the time it takes from a point to the airport.

9/3/20XX Presentation Title 9


Other distributions
• A number of other types of probability distributions
• Poisson, lognormal, power law, beta, gamma.

• We look at the Poisson distribution (only)

9/3/20XX Presentation Title 10


Poisson distribution
• Distribution of the number of occurrences of discrete events within a fixed interval of time.

• It has a single parameter, lambda (λ)


• λ = expected value or the average rate of events

• Probability of an event is independent of past events


• No memory

• Often used to model rare events


• Like number of customer arrivals per minute

11
Example
P(x) =

An investor is known for making an average of 2 trades per day. Find the
probability that he makes
(1) 3 trades in a day
(2) No trades in a day
(3) More than 5 trades in a day

12
Example
P(x) =

λ=2
P (x = 3) = (2^3)/3! * e^(-2) = 0.18

13
Example
P(x) =

λ=2
P (x = 0) = (2^0)/0! * e^(-2) = 0.135

14
Example
P(x) =

P(x > 5) = 1 – P(x<=5)

λ=2
P (x = 0) = 0.135
P (x = 1) = 0.271
P (x = 2) = 0.271
P (x = 3) = 0.180
P (x = 4) = 0.090
P (x = 5) = 0.036

P (x > 5) = 1 – cumulative probability P(0 to 5 trades)


P (x > 5) = 1 – 0.983 = 0.017
15
Thank you
Mohsin Khawaja
mkhawaja@iba.edu.pk

You might also like