Problem Set 4 for Stochastic Processes II - M375T, Assignments of Mathematics

Problem set 4 for the course stochastic processes ii - m375t, taught by gerard brunick. The problems cover various topics in stochastic processes, including exponentiation of martingales, choosing constants for martingales, and strategies for gambling on coin flips.

Typology: Assignments

Pre 2010

Uploaded on 08/30/2009

koofers-user-quf
koofers-user-quf 🇺🇸

10 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
Course: Stochastic Processes II - M375T Instructor: Gerard Brunick
Problem Set 4
We will go over this problem set on Monday 4/6.
We are working in the coin-flip setup with Mn,Pn
i=1
1
Hi
1
Tifor all of the following
problems.
1. In this problem we look at what happens when you exponentiation a martingale.
(a) Set Xn=eMn. Is (Xn)nN0a martingale, submartingale, or supermartingale?
Hint: write Xn=eMnXn1and use the convexity of the exponential function.
(b) Let Nn=eMnnC for some constant C. Choose Cso that (Nn)nN0is a martingale.
Hint: write Nn=eMneCNn1.
2. Let Nn=M2
nnC for some constant C. Choose Cso that (Nn)nN0is a martingale.
Hint: write M2
n= (∆Mn+Mn1)2and expand to get Nnwritten in the form Nn=
? + Nn1.
3. Set Θ0= 0, Θn=Mn1for nN, and In=Pn
i=1 ΘnMn.
(a) Write Inas a function of Mnand n. Hint: notice that Mn1(∆Mn) = 1
2M2
n
M2
n1(∆Mn)2, and the expression on the right is essentially a telescoping sum.
(b) Is n)nN0adapted? Is it predictable? Is Ina martingale?
4. Let X= 2 P3
i=1
1
Hidenote twice the number of heads that occur in the first three flips
(so Xis a random variable not a process).
(a) Set Nn=E[X|Fn]. Compute the process (Nn)nN0explicitly and check that it is
a martingale. Hint: it may help to consider the cases n3 and n > 3 separately.
For the n3 case it may help to draw a tree of all possibilities.
(b) Find a process n)nNsuch that Nn=N0+Pn
i=1 ΘiMi. You may help to look
at the tree from the previous part and try to use backward induction.
(c) Suppose that in addition to the regular bets on heads and tails, you can also
buy a “magic double triple header ticket” for some cost $C, where Cis a non-
random constant. When you buy such a ticket, you get $2 for each heads that
appears in the next three flips. If C < 3, explain how you can take advantage
of this mispricing to make money with no risk. Such an opportunity is called an
arbitrage.
Hint: The “hedging strategy” that you computed in the previous part of this
problem gives you one way to produce a payoff equal to X, and the buying the
ticket gives you another. How can you play these two possibilities off against one
another?
Page: 1/2
pf2

Partial preview of the text

Download Problem Set 4 for Stochastic Processes II - M375T and more Assignments Mathematics in PDF only on Docsity!

Course: Stochastic Processes II - M375T Instructor: Gerard Brunick

Problem Set 4

We will go over this problem set on Monday 4/6.

We are working in the coin-flip setup with Mn ,

∑n i=1 1 Hi −^1 Ti for all of the following problems.

  1. In this problem we look at what happens when you exponentiation a martingale.

(a) Set Xn = eMn^. Is (Xn)n∈N 0 a martingale, submartingale, or supermartingale? Hint: write Xn = e∆Mn^ Xn− 1 and use the convexity of the exponential function. (b) Let Nn = eMn−nC^ for some constant C. Choose C so that (Nn)n∈N 0 is a martingale. Hint: write Nn = e∆Mn^ e−C^ Nn− 1.

  1. Let Nn = M (^) n^2 − nC for some constant C. Choose C so that (Nn)n∈N 0 is a martingale. Hint: write M (^) n^2 = (∆Mn + Mn− 1 )^2 and expand to get Nn written in the form Nn = ? + Nn− 1.
  2. Set Θ 0 = 0, Θn = Mn− 1 for n ∈ N, and In =

∑n i=1 Θn∆Mn. (a) Write In as a function of Mn and n. Hint: notice that Mn− 1 (∆Mn) = (^12)

M (^) n^2 − M (^) n^2 − 1 − (∆Mn)^2

, and the expression on the right is essentially a telescoping sum. (b) Is (Θn)n∈N 0 adapted? Is it predictable? Is In a martingale?

  1. Let X = 2

i=1 1 Hi denote twice the number of heads that occur in the first three flips (so X is a random variable not a process).

(a) Set Nn = E[X | Fn]. Compute the process (Nn)n∈N 0 explicitly and check that it is a martingale. Hint: it may help to consider the cases n ≤ 3 and n > 3 separately. For the n ≤ 3 case it may help to draw a tree of all possibilities. (b) Find a process (Θn)n∈N such that Nn = N 0 +

∑n i=1 Θi∆Mi. You may help to look at the tree from the previous part and try to use backward induction. (c) Suppose that in addition to the regular bets on heads and tails, you can also buy a “magic double triple header ticket” for some cost $C, where C is a non- random constant. When you buy such a ticket, you get $2 for each heads that appears in the next three flips. If C < 3, explain how you can take advantage of this mispricing to make money with no risk. Such an opportunity is called an arbitrage. Hint: The “hedging strategy” that you computed in the previous part of this problem gives you one way to produce a payoff equal to X, and the buying the ticket gives you another. How can you play these two possibilities off against one another?

Page: 1/

Course: Stochastic Processes II - M375T Instructor: Gerard Brunick

In the following problems, we describe a strategy that one could follow when gambling on coin flips. You are asked to give the integrand process (Θn)n∈N (in particular, Θ 0 doesn’t matter) such that the stochastic integral In =

∑n i=1 Θi∆Mi^ corresponds to the amount of money that you would have at time n if you followed the given strategy.

  1. Suppose that you choose to bet one dollar on heads for the first two flips, and then you quit gambling. What process (Θn)n∈N corresponds to this strategy? Is this process predictable? Is (In)n∈N 0 a martingale?
  2. Suppose that you choose to bet one dollar on heads the first flip, the third flip, and for each odd numbered flip after that, but you choose to bet on tails for the even numbered flips. What process (Θn)n∈N corresponds to this strategy? Is this process predictable? Is (In)n∈N 0 a martingale?
  3. Suppose that you bet two dollars on heads when the flip is going to be heads, by you don’t bet when the flip will and tails, so you never lose money. What process (Θn)n∈N corresponds to this strategy? Is this process adapted? Is this process predictable? Is (In)n∈N 0 a martingale?
  4. Suppose that you choose to continue betting one dollar on tails until the first time that you lose a bet. What process (Θn)n∈N corresponds to this strategy? Is this process predictable? Is (In)n∈N 0 a martingale?
  5. Suppose that you choose to continue betting one dollar on heads until you have won two dollars or lost two dollars. What process (Θn)n∈N corresponds to this strategy? Is this process predictable? Is (In)n∈N 0 a martingale? If it in not clear how to write the general case, it is enough to describe Θi for i ∈ { 1 , 2 , 3 , 4 , 5 , 6 } explicitly.

Page: 2/