Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
MMF1941H: Stochastic Analysis - Assignment # 1
25 Oct 2021
1 Instructions
1. Assignments are due on Monday Nov 8th, 7:30pm EST. Submissions are accepted late for 1
day with a 10% penalty, submissions after Nov 2nd will not be accepted.
2. Please have your final report typeset using LATEX and submit your report individually. Pro-
vide code separate in a Python script file that you attach in your submission.
3. You may discuss these questions with your fellow students, however the write-up must be
yours and yours alone, sharing of the write-up before the deadline is not allowed.
2 Problem: Bachelier Call Option Pricing
Let X be a standard normal random variable and let µX ∈ R and variance σ2X > 0 parameters.
We are looking at the value of a call option in the Bachelier Model ie
v = E
(
(µX + σXX− K)+
)
for a given strike K.
1. (10 pts) Show that for K ∈ R
E((X− K)+) = ϕ(K)− K(1−Φ(K))
holds where ϕ is the pdf of the standard normal distribution and Φ the respective cdf.
(Hint: you can exploit that ϕ′(x) = −xϕ(x) holds).
2. (5 pts) Use the previous result to show that analytically
v = σXϕ(
K− µX
σX
)− (K− µX)
(
1−Φ(K− µX
σX
)
)
holds.
1
MMF1941H: Stochastic Analysis
3. (10 pts) For a parameter a ∈ R we can define a measure Qa via the definition
Qa(A) = E
(
1A
exp(aX)
E(exp(aX))
)
.
For x ∈ R calculate the Qa (X ≤ x) and conclude that under Qa X again follows a normal
distribution and determine its parameters.
4. (5 pts) Write Python code to simulate the option value 1000 times under the measure P
with a sample size of 5000 simulations each for µX = 4, σ2X = 4 and K = 8. Share the code
and provide a histogram of the results. Also calculate the exact value analytically per the
above formula.
5. (5 pts) Denoting vj a single MC estimate (based on 5000 simulations) for j = 1, . . . , M with
M = 1000 we can define the sample variance as
s =
1
M− 1
M
∑
j=1
(
vj − 1M
M
∑
k=1
vk
)2
.
Calculate the sample variance for your previous experiment.
6. (10 pts) Note that for any a the option value can be re-written as
v = EQa
(
dP
dQa
(µX + σXX− K)+
)
which mathematically will yield the same answer for any a. Define
g(a) = EQa
((
dP
dQa
(µX + σXX− K)+
)2)
which can be evaluated through MC simulation given the fact the distribution of X under
the measure Qa is known. Write Python code plot the function g(a) for the above selection
of parameters and a ∈ [0, 5]. Note that equivalently,
g(a) = E
(
dP
dQa
(
(µX + σXX− K)+
)2)
holds which you could use alternatively for implementation purposes.