MScFE xxx [Course Name] - Module X: Collaborative Review Task

© 2019 - WorldQuant University – All rights reserved.

Revised: 06/06/2019
1
MScFE 620 Discrete-time Stochastic Processes – Table of Contents

Module 4: Trading in Discrete Time ................................................................ 3

Unit 1: Portfolios and Strategies ............................................................................................... 4

Unit 2: A Worked Example of Arbitrage .................................................................................6

Unit 3: Equivalent Martingale Measures ................................................................................9

Unit 4: Completeness................................................................................................................... 13

Problem Set ..................................................................................................................................... 15

Bibliography ................................................................................................................................... 20

Collaborative Review Task ........................................................................................................ 21

© 2019 - WorldQuant University – All rights reserved.

2
MScFE 620 Discrete-time Stochastic Processes - Module 4: Summary

Module 4: Trading in Discrete Time

By viewing the values of variables that occur at distinct, individual points in time, we have a useful
lens through which the frequently erratic practice of trading can be studied. This module
introduces the concept of a trading strategy in discrete time and how it can be used to define no-
arbitrage and completeness of a financial market. It also discusses the two fundamental theorems
of asset pricing in discrete time.

© 2019 - WorldQuant University – All rights reserved.

3
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 1

Unit 1: Portfolios and Strategies

We consider a financial market with a finite time horizon and trading in discrete time. Specifically,
we will assume that the market consists of 𝑑 + 1 assets whose prices are discrete-time stochastic
processes. We let 𝑇 ∈ ℕ+ be the time horizon and assume that the trading dates are 𝑡 = 0,1,2, … , 𝑇.
We let (𝐵, 𝑆) = (𝐵, 𝑆 1 , … , 𝑆 𝑑 ) be a stochastic process that represents that prices of all the 𝑑 + 1
assets. This means that for each 𝑖 = 1, … , 𝑑, 𝑆 𝑖 = {𝑆𝑡𝑖 : 𝑡 = 0,1, … , 𝑇} is a discrete-time stochastic
process.

To properly talk about randomness and the flow of information, we will work on a filtered
probability space (Ω, ℱ, 𝔽, ℙ) and assume that (𝐵, 𝑆) is adapted to 𝔽, where 𝔽 = {ℱ𝑡 : 𝑡 = 0,1, … , 𝑇}.
We will also assume that Ω has finitely many elements.

We call the tuple ((Ω, ℱ, 𝔽, ℙ), (𝐵, 𝑆)) a market.

We will assume that the asset 𝐵 is a riskless bank account with a predictable interest rate process
𝑟 = {𝑟𝑡 : 𝑡 = 1,2, … , 𝑇}. That is,

𝐵𝑡 = 𝐵𝑡−1 (1 + 𝑟𝑡 ), 𝑡 = 1,2, … , 𝑇,

where 𝑟𝑡 ≻ −1 for all 𝑡.

To simplify the presentation – and without loss of generality – we will assume that 𝐵 has a strictly
positive price process (i.e. 𝐵 > 0) and express the prices of all the assets in units of 𝐵. Asset 𝐵 is
said to be a numeraire. To this end, we define the discounted prices 𝑋 = (𝑋1 , … , 𝑋 𝑑 ) by

𝑆𝑖
𝑋𝑖 ≔ , 𝑖 = 1, … , 𝑑.
B

Thus, the discounted prices are (1, 𝑋) = (1, 𝑋1 , … , 𝑋 𝑑 ). From now onwards, we will work with the
discounted prices.

A trading strategy is a predictable process (𝜂, 𝜑) = (𝜂, 𝜑1 , … , 𝜑𝑑 ), where 𝜑𝑡𝑖 is the number of units
of asset 𝑖 held from time 𝑡 − 1 to time 𝑡 and 𝜂𝑡 is the amount invested in the riskless asset over the
same period. The value of a trading strategy (𝜂, 𝜑) is the stochastic process 𝑉((𝜂, 𝜑)) =
{𝑉𝑡 ((𝜂, 𝜑)): 𝑡 = 0,1, … , 𝑇} defined by

𝑉0 ((𝜂, 𝜑)) = (𝜂1 , 𝜑1 ) ⋅ (1, 𝑋0 ) = 𝜂1 + ∑ 𝜑1𝑖 𝑋0𝑖

𝑖=1

© 2019 - WorldQuant University – All rights reserved.

4
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 1

And

𝑉𝑡 ((𝜂, 𝜑)) ≔ (𝜂𝑡 , 𝜑𝑡 ) ⋅ (1, 𝑋𝑡 ) = 𝜂𝑡 + ∑ 𝜑𝑡𝑖 𝑋𝑡𝑖 𝑡 ≥ 1.

𝑖=1

The gains process associated with (𝜂, 𝜑) is the stochastic process 𝐺(𝜑) = {𝐺𝑡 (𝜑): 𝑡 = 0,1, … , 𝑇}
defined by

𝐺0 (𝜑) ≔ 0, 𝐺𝑡 (𝜑) ≔ ∑ 𝜑𝑘 ⋅ (𝑋𝑘 − 𝑋𝑘−1 ) for 𝑡 = 1,2, … , 𝑇.

𝑘=1

(Note that there are no gains from trading the riskless asset.)

A trading strategy 𝜑 is self-financing if

(𝜂𝑡 , 𝜑𝑡 ) ⋅ (1, 𝑋𝑡 ) = (𝜂𝑡+1 , 𝜑𝑡+1 ) ⋅ (1, 𝑋𝑡 ) for all 𝑡 = 1, … , 𝑇.

The interpretation of this condition is that the value of the strategy only changes through the
gains from trading – there is no influx or outgo of funds from investment at any point in time.

Equivalently, 𝜑 is self-financing if and only if

𝑉𝑡 ((𝜂, 𝜑)) = 𝑉0 ((𝜂, 𝜑)) + 𝐺𝑡 (𝜑) for 𝑡 = 0,1, … , 𝑇.

Since there are no gains from trading asset 0, we see that for a self-financing strategy knowledge
of the initial value of the strategy and 𝜑1 , … , 𝜑𝑑 allows us to recover uniquely the amount invested
in the riskless asset via the formula:

𝑉0 ((𝜂, 𝜑)) + 𝐺𝑡 (𝜑) = 𝜂𝑡 + ∑ 𝜑𝑡𝑖 𝑋𝑡𝑖 .

𝑖=1

Thus, we can (and will) represent a trading strategy by specifying an initial value 𝑣0 and
predictable holdings in the other assets 𝜑 = (𝜑1 , … , 𝜑𝑑 ). We will write this as (𝑣0 , 𝜑).

© 2019 - WorldQuant University – All rights reserved.

5
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 2

Unit 2: A Worked Example of Arbitrage

From now until the end of the course, we assume that all trading strategies are self-financing. Also,
to simplify notation, we will sometimes write 𝜑 0 ≔ 𝜂 and talk about a strategy 𝜑 =
(𝜑0 , 𝜑1 , … , 𝜑𝑑 ). We will also denote the discounted riskless asset by 𝑋 0 , i.e. 𝑋 0 ≡ 1.

A trading strategy 𝜑 is an arbitrage strategy if

𝑉0 (𝜑) = 0, 𝑉𝑇 (𝜑) ≥ 0 ℙ − a.s. and ℙ(𝑉𝑇 (𝜑) > 0) > 0.

In words, an arbitrage strategy is a strategy that requires no investment to enter into (𝑉0 = 0), yet
the strategy poses no risk of loss (𝑉𝑇 ≥ 0) and a possibility of making a positive gain (ℙ(𝑉𝑇 (𝜑) >
0) > 0). A market model is arbitrage-free if it does not allow for any arbitrage
strategies/opportunities.

Let 𝑇 = 1, 𝑑 = 2 and Ω = {𝑎, 𝑏}, ℱ0 = {∅, Ω}, ℱ1 = ℱ = 2Ω , ℙ = 0.5𝛿𝑎 + 0.5𝛿𝑏 . Define (𝑋 0 , 𝑋1 , 𝑋 2 ) =

(1, 𝑋1 , 𝑋 2 ) as follows:

Define the strategy 𝜑 by

𝜑1 = (𝜑10 , 𝜑11 , 𝜑12 ) = (10, 1, −3).

Then

𝑉0 (𝜑) = (10, 1, −3) ⋅ (1, 2, 4) = 10 + 2 − 12 = 0,

and
10 + 4 − 9 = 5 𝜔= 𝑎
𝑉1 (𝜑) = 𝜑1 ⋅ 𝑋1 = {
10 + 1 − 6 = 5 𝜔 = 𝑏.

Hence 𝜑 is an arbitrage strategy.

In general, it is much more difficult to show that a market has no arbitrage opportunities. We will
now find conditions on the market that are equivalent to the no arbitrage condition.

© 2019 - WorldQuant University – All rights reserved.

6
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 2

Let ℙ∗ be a probability measure on (Ω, ℱ). We call ℙ∗ an equivalent martingale measure (EMM) for
𝑋 if ℙ∗ is equivalent to ℙ (i.e. ℙ(𝐴) = 0 ⟺ ℙ∗ (𝐴) = 0 ∀𝐴 ∈ ℱ) and 𝑋 is an (𝔽, ℙ∗ )-martingale. We
will denote by 𝒫 the (possibly empty) set of EMM’s for 𝑋. We will also write 𝔼∗ to denote the
expectation with respect to ℙ∗.

1
Let 𝑇 = 2, 𝑑 = 1 and Ω = {𝑎, 𝑏, 𝑐, 𝑑}, ℙ = 4 (𝛿𝑎 + 𝛿𝑏 + 𝛿𝑐 + 𝛿𝑑 ) and

Then ℙ∗ is an EMM for (1, 𝑋). (1 is trivially a martingale, so we will only concentrate on 𝑋.) We
choose the filtration to be the natural filtration of 𝑋 given by

ℱ0𝑋 = {∅, Ω}, ℱ1𝑋 = {∅, Ω, {𝑎, 𝑏}, {𝑐, 𝑑}} and ℱ2 = 𝜎({𝑋0 , 𝑋1 , 𝑋2 }) = 2Ω .

The equivalence of ℙ and ℙ∗ is clear since the only null set is the empty set for both of them. Now
we show that 𝑋 is a ℙ∗-martingale.

First note that 𝔼(|𝑋𝑛 |) ≤ 36 < ∞ for all 𝑛 ∈ 𝕀.

𝑋𝑛 ∈ 𝑚ℱ𝑛𝑋 ∀𝑛 ∈ 𝕀 = {0,1,2} is trivial since we are using the natural filtration of 𝑋.

1 2 1 1
𝔼∗ (𝑋1 |ℱ0𝒳 ) = 𝔼∗ (𝑋1 |{∅, Ω}) = 𝔼∗ (𝑋1 ) = × 18 + × 18 + × 9 + × 9 = 12 = 𝑋0
9 9 6 2

1 2 1 1
9 𝑋2 (𝑎) + 9 𝑋2 (𝑏) 6 𝑋2 (𝑐) + 2 𝑋2 (𝑑)
𝔼 ∗
(𝑋2 |ℱ1𝒳 ) = 𝛼𝐼{𝑎,𝑏} + 𝛽𝐼{𝑐,𝑑} = 𝐼{𝑎,𝑏} + 𝐼{𝑐,𝑑}
1 2 1 1
+ +
9 9 6 2

= 18𝐼{𝑎,𝑏} + 9𝐼{𝑐,𝑑} .
Clearly, 𝔼∗ (𝑋2 |ℱ1𝒳 ) = 𝑋1 , so 𝑋 is a martingale.

© 2019 - WorldQuant University – All rights reserved.

7
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 2

Now suppose that ℙ∗ is an EMM for 𝑋 and let 𝜑 be a strategy with 𝑉0 (𝜑) = 0 and 𝑉𝑇 (𝜑) ≥ 0. Since
𝑉𝑡 (𝜑) = 𝑉0 (𝜑) + ∑𝑡𝑘=1 𝜑𝑘 (𝑋𝑘 − 𝑋𝑘−1 ) is a martingale transform with 𝑉𝑇 (𝜑) ≥ 0, it follows that
𝑉(𝜑) is a martingale too. (See Shiryaev's “Essentials of Stochastic Finance”.) Hence 𝔼∗ (𝑉𝑇 (𝜑)) = 0,
which implies that ℙ(𝑉𝑇 (𝜑) > 0) = 0. So, there are no arbitrage opportunities.

The converse of the previous also holds, i.e. absence of arbitrage opportunities also implies the
existence of an EMM. This remarkable result is called The Fundamental Theorem of Asset Pricing I
(FTAP I).

Theorem

[FTAP I]

For a financial market (𝛺, ℱ, 𝔽, ℙ), the following are equivalent:

1 There are no arbitrage opportunities.
2 𝒫 ≠ 0.

This theorem gives us an alternative way of deciding whether or not a market admits arbitrage
opportunities.

© 2019 - WorldQuant University – All rights reserved.

8
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 3

Unit 3: Equivalent Martingale Measures

Pricing and Replication

We will now assume that ℱ𝑇 = ℱ.

A contingent claim 𝐻 is an ℱ𝑇 -measurable random variable. We call a contingent claim 𝐻 a

derivative of the underlying assets 𝑋 if 𝐻 is measurable with respect to ℱ𝑇𝑋 .

We think of 𝐻 as a liability that expires at time 𝑇. Such a liability is called a European contingent
claim. In a later module we will meet contingent claims whose expiry date can be at any time
before time 𝑇. Such contingent claims are called American contingent claims.

Here are some common examples of European derivatives 𝐻 (with 𝑑 = 1 and 𝑋 = 𝑋1 ):

o Call option: 𝐻 = (𝑋𝑇 − 𝐾)+ , where 𝐾 > 0 is called the strike price.

o Put option: 𝐻 = (𝐾 − 𝑋𝑇 )+ , where 𝐾 > 0 is called the strike price.

1 +
o Asian call option: 𝐻 = ( ∑𝑇 𝑋 − 𝐾} , where 𝐾 is the strike price.
𝑇+1 𝑡=0 𝑡

o Digital option: 𝐻 = 𝐼{ max 𝑋𝑡 ≥𝐾}

, where 𝐾 > 0 is the barrier.
0≤𝑡≤𝑇

Given a contingent claim 𝐻, we want to answer the following questions:

1 What is the price of 𝐻 at time 0? That is, how much should one pay at time 0 in return for
the random payment 𝐻 at time 𝑇?
2 Having sold the contingent claim at time 0, how can the seller protect himself against the
random outflow 𝐻 by investing in the primary assets 𝑋?

The first question has to do with pricing, while the second one is to do with replication.

A strategy/portfolio 𝜑 such that 𝑉𝑇 (𝜑) = 𝐻 ℙ-a.s. is called a replicating portfolio or replicating

strategy. If there exists a replicating strategy for 𝐻, we say that 𝐻 is attainable.

It turns out the two questions (of pricing and replication) are related. Indeed, if 𝒫 ≠ ∅ and 𝐻 can be
replicated by a portfolio 𝜑 (i.e. 𝑉𝑇 (𝜑) = 𝐻), then for any ℙ∗ ∈ 𝒫,

𝑉0 (𝜑) = 𝔼∗ (𝑉𝑇 (𝜑)) = 𝔼∗ (𝐻).

© 2019 - WorldQuant University – All rights reserved.

9
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 3

That is, the initial capital required to replicate 𝐻 is always 𝔼∗ (𝐻), and this quantity is independent
of the EMM ℙ∗. It makes sense then to call this the no-arbitrage price of 𝐻 and we will denote it by
π(𝐻).

Thus, an attainable contingent claim 𝐻 has a unique no-arbitrage price π(𝐻), and that price is
equal to 𝔼∗ (𝐻) for any EMM ℙ∗.

1
Let us consider an example. Let 𝑇 = 1, 𝑑 = 1 and Ω = {𝑎, 𝑏, 𝑐}, ℙ = 3 (𝛿𝑎 + 𝛿𝑏 + 𝛿𝑐 ) and

An equivalent martingale measure ℙ∗ is characterized by numbers ℙ∗ (𝑎) =: 𝛼, ℙ∗ (𝑏) =: 𝛽 and

ℙ∗ (𝑐) =: 𝛾, such that

𝛼, 𝛽, 𝛾 > 0, 𝛼+𝛽+𝛾 =1

and the martingale property on 𝑋1 :

𝛼 + 2𝛽 + 4𝛾 = 2.

Solving this system of equations gives

1
𝒫 = {ℙ∗𝑝 = (2𝑝, 1 − 3𝑝, 𝑝) ∶ 0 < 𝑝 < } .
3
So, the market admits no arbitrage opportunities.

Continuing with the previous example, consider the contingent claim 𝐻 defined as follows:

© 2019 - WorldQuant University – All rights reserved.

10
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 3

Then 𝐻 is attained by the portfolio 𝜑1 = (𝜑10 , 𝜑11 ) = (−1,2) with 𝑉0 (𝜑) = (−1,2) ⋅ (1,2) = 3. Also, if
ℙ∗𝑝 is an EMM, then

𝜋(𝐻) = 𝔼∗ (𝐻) = 2𝑝 + 3(1 − 3𝑝) + 7𝑝 = 3 = 𝑉0 (𝜑),

as expected.

What happens if 𝐻 is not attainable? Does it still have a unique no-arbitrage price? Consider the
following contingent claim 𝐻 ′ :

To find a replicating strategy 𝜑 = (𝜑0 , 𝜑1 ), we need to solve the following system of equations:

𝜑10 + 𝜑11 = 2

𝜑10 + 2𝜑11 = 2

𝜑10 + 4𝜑11 = 6,

which clearly does not have a solution. So 𝐻 ′ is not attainable.

1
Also, if ℙ∗𝑝 = (2𝑝, 1 − 3𝑝, 𝑝) ∶ 0 < 𝑝 < 3 is an EMM, then

𝔼∗𝑝 (𝐻 ′ ) = 2(2𝑝) + 2(1 − 3𝑝) + 6𝑝 = 4𝑝 + 2,

which depends on 𝑝. So 𝐻 ′ has no unique no-arbitrage price.

So how do we price a non-attainable claim 𝐻? It turns out that the (infinite) set of no-arbitrage
prices for 𝐻 is given by

{𝔼∗ (𝐻): ℙ∗ ∈ 𝒫}

© 2019 - WorldQuant University – All rights reserved.

11
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 3

In the previous example, this is

1 4
{4𝑝 + 2 ∶ 0 < 𝑝 < } = (2,2 + ).
3 3

4
One can check that if the price of 𝐻 ′ is greater than 2 + 3 or less than 2, then arbitrage
opportunities will exist.

© 2019 - WorldQuant University – All rights reserved.

12
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 4

Unit 4: Completeness

A market is complete if every contingent claim is attainable.

Consider the market from the previous section. Let 𝑇 = 1, 𝑑 = 1 and Ω = {𝑎, 𝑏, 𝑐}, ℙ =
1
(𝛿
3 𝑎
+ 𝛿𝑏 + 𝛿𝑐 ) and

The contingent claim 𝐻 ′ defined by

is not attainable, so the market is not complete.

The Fundamental Theorem of Asset Pricing II (FTAP II) characterizes completeness of a market.

Theorem

[FTAP II]

For a financial market ((𝛺, ℱ, 𝔽, ℙ), 𝑋) with no arbitrage opportunities, the following are
equivalent:
1 The market is complete.
2 𝒫 has exactly one element.

© 2019 - WorldQuant University – All rights reserved.

13
MScFE 620 Discrete-time Stochastic Processes - Module 4: Unit 4

3 𝑋 has the predictable representation property (PRP) with respect to every ℙ∗ ∈ 𝒫 : every
(𝔽, ℙ∗ )-martingale 𝑀 with 𝑀0 = 0 can be written as a martingale transform with respect to
𝑋.

The third part of the theorem gives us a replicating strategy for each claim 𝐻 in a complete market.
Indeed, let 𝐻 be a contingent claim with 𝔼∗ (|𝐻|) < ∞. Then the process 𝑀 = {𝑀𝑡 : 𝑡 = 0,1, … , 𝑇}
defined by 𝑀𝑡 ≔ 𝔼∗ (𝐻|ℱ𝑡 ) is a martingale with 𝑀𝑇 = 𝐻. By PRP, there exists a predictable process
𝜑𝐻 such that

𝑀𝑡 = 𝔼 ∗ (𝐻)
+ ∑ 𝜑𝑘𝐻 ⋅ (𝑋𝑘 − 𝑋𝑘−1 ), 𝑡 = 1,2, … , 𝑇.
𝑘=1

Substituting 𝑡 = 𝑇, we see that the process 𝜑𝐻 is a replicating strategy for 𝐻. Consider

Ω = {𝜔 = (𝜔1 , … , 𝜔 𝑇 ): 𝜔𝑖 ∈ {−1,1}, 𝑖 = 1,2, … , 𝑇}

and let 𝑌𝑡 (𝜔) ≔ 𝜔𝑡 for 𝜔 = (𝜔1 , … , 𝜔 𝑇 ). Define ℱ0 = {∅, Ω} and ℱ𝑡 = 𝜎({𝑌1 , … , 𝑌𝑡 }) for 𝑡 ≥ 1.
Consider two assets 𝑋 0 and 𝑋1 whose discounted assets are 𝑋 0 ≡ 1, 𝑋01 = constant, and

1
𝑋𝑡1 = 𝑋𝑡−1 𝑒𝑥𝑝(𝜎𝑡 𝑌𝑡 + 𝜇𝑡 ) , 𝑡 ≥ 1,
for some predictable processes 𝜇 and 𝜎 with 0 ≤ |𝜇𝑡 | < 𝜎𝑡 for every 𝑡. Assume that ℙ is such that
ℙ({𝜔}) > 0 for every 𝜔 ∈ Ω. Then there is a unique EMM ℙ∗ for 𝑋. Furthermore, if 𝑀 is a ℙ∗-
martingale, then

𝑀𝑡 = 𝑀0 + ∑ 𝜑𝑘𝑀 (𝑋𝑘 − 𝑋𝑘−1 ),

𝑘=1

where

𝑀𝑘 − 𝑀𝑘−1
𝜑𝑘𝑀 = .
𝑋𝑘 − 𝑋𝑘−1

This also gives a replicating portfolio for any contingent claim 𝐻 as

𝔼∗ (𝐻|ℱ𝑡 ) − 𝔼∗ (𝐻|ℱ𝑡−1 )
𝜑𝑡𝑀 = .
𝑋𝑡 − 𝑋𝑡−1

© 2019 - WorldQuant University – All rights reserved.

14
MScFE 620 Discrete-time Stochastic Processes - Module 4: Problem Set

Problem Set

Problem 1

Consider a market ((Ω, ℱ, 𝔽, ℙ), 𝑋) with one risky asset 𝑋 = {𝑋0 , 𝑋1 , 𝑋2 } defined as follows:

Ω = {𝑎, 𝑏, 𝑐, 𝑑}, 𝔽 = 𝔽𝑋 , ℱ = 2Ω .

Consider the self-financing trading strategy (𝑣0 , 𝜑) defined by

𝑣0 = 10, 𝜑1 = 2, 𝜑2 = 𝐼{𝑋0 >𝑋1 } − 𝐼{𝑋1 ≥ 𝑋0 } .

Here 𝑣0 is the initial capital and 𝜑 is the investment in 𝑋. Find 𝐺1 (𝜑)(𝑎), the gains from trading at
time 1.

Solution:

We know that the gains process associated with (𝜂, 𝜑) is the stochastic process 𝐺(𝜑) =
{𝐺𝑡 (𝜑): 𝑡 = 0,1, … , 𝑇} defined by

𝐺0 (𝜑) ≔ 0, 𝐺𝑡 (𝜑) ≔ ∑ 𝜑𝑘 ⋅ (𝑋𝑘 − 𝑋𝑘−1 ) for 𝑡 = 1,2, … , 𝑇.

𝑘=1

(Note that there are no gains from trading the riskless asset.)

In this case, we have to focus on 𝐺1 (𝜑)(𝑎), thus

𝐺1 (𝜑)(𝑎) ≔ ∑ 𝜑𝑘 ⋅ (𝑋𝑘 − 𝑋𝑘−1 ) = 𝜑1 ⋅ (𝑋1 (𝑎) − 𝑋0 (𝑎)) = 2 ∗ (26 − 15) = 22.

𝑘=1

The solution is 22.

© 2019 - WorldQuant University – All rights reserved.

15
MScFE 620 Discrete-time Stochastic Processes - Module 4: Problem Set

Problem 2

Consider a market ((Ω, ℱ, 𝔽, ℙ), 𝑋) with one risky asset 𝑋 = {𝑋0 , 𝑋1 } defined as follows:

Ω = 𝑎, 𝑏, 𝑐, 𝑑, 𝔽 = 𝔽𝑥 , ℱ = 2Ω

Which one of these measures is an EMM for 𝑋?

1 1 3
1 ℙ∗ = 5 𝛿𝑎 + 2 𝛿𝑏 + 10 𝛿𝑐
1 1 5
2 ℙ∗ = 4 𝛿𝑎 + 3 𝛿𝑏 + 12 𝛿𝑐
1 1 7
3 ℙ∗ = 4 𝛿𝑎 + 6 𝛿𝑏 + 12 𝛿𝑐 *

4 ℙ∗ = 𝛿𝑎 + 𝛿𝑏 + 𝛿𝑐

Solution:

Let ℙ∗ be a probability measure on (Ω, ℱ). We call ℙ∗ an equivalent martingale measure (EMM) for
𝑋 if ℙ∗ is equivalent to ℙ (i.e. ℙ(𝐴) = 0 ⟺ ℙ∗ (𝐴) = 0 ∀ A ∈ ℱ) and 𝑋 is an (𝔽, ℙ∗ )-martingale.

The only option that holds the above conditions is

1 1 7
ℙ∗ = 𝛿𝑎 + 𝛿𝑏 + 𝛿𝑐 .
4 6 12

The equivalence of ℙ and ℙ∗ is clear since the only null set is the empty set for both of them. Now
we need also to show that 𝑋 is a ℙ∗-martingale.

First note that 𝔼(|𝑋𝑛 |) ≤ 8 < ∞ for all 𝑛 ∈ 𝕀. 𝑋𝑛 ∈ 𝑚ℱ𝑛𝑋 ∀𝑛 ∈ 𝕀 = {0,1} is trivial since we are using
the natural filtration of 𝑋.

1 1 7
𝔼∗ (𝑋1 |ℱ0𝑋 ) = 𝔼∗ (𝑋1 |{∅, Ω}) = 𝔼∗ (𝑋1 ) = ×8+ ×5+ × 3 = 4 = 𝑋0 .
4 6 12

© 2019 - WorldQuant University – All rights reserved.

16
MScFE 620 Discrete-time Stochastic Processes - Module 4: Problem Set

Problem 3

Consider a financial market with the following components:

Ω = {𝑎, 𝑏, 𝑐}, ℱ = 2Ω , ℱ0 = {∅, Ω}, and ℱ1 = ℱ.

Let ℙ, 𝑋 = {𝑋0 , 𝑋1 }, and 𝐻 be defined as follows:

Find the price of 𝐻.

Solution:

We need to find a replicating strategy 𝜑 = (𝜑0 , 𝜑1 ), thus we need to solve the following system of
equations:

1𝜑0 + 8𝜑1 = 4

1𝜑0 + 5𝜑1 = 10

1𝜑0 + 2𝜑1 = 16.

The solution is 𝜑0 = 20 and 𝜑1 = −2. The initial price of 𝑋 is 4, and we assume that the risk free
asset cost is 1, thus the strategy cost is 20 ∗ 1 − 2 ∗ 4 = 12, which should also be the price of 𝐻.

Problem 4

Consider a financial market with the following components:

Ω = {𝑎, 𝑏, 𝑐}, ℱ = 2Ω , ℱ0 = {∅, Ω}, and ℱ1 = ℱ.

Let ℙ, 𝑋 = {𝑋0 , 𝑋1 }, and 𝐻 be defined as follows:

© 2019 - WorldQuant University – All rights reserved.

17
MScFE 620 Discrete-time Stochastic Processes - Module 4: Problem Set

If 𝐻 is attainable and has a unique no arbitrage price of 0, then what is 𝛾?

Solution:

We need to find a replicating strategy 𝜑 = (𝜑0 , 𝜑1 ), thus we need to solve the following system of
equations:

1𝜑0 + 120𝜑1 = −20

1𝜑0 + 112𝜑1 = 𝛾

1𝜑0 + 90𝜑1 = 10.

From the first and third equation we get that 𝜑1 = −1. If we substitute the value of 𝜑1 in the
second equation, we get

𝛾 + 112
𝜑0 = .
100

As the statement says that the value of 𝐻 is zero, we can get 𝛾 from the following condition
(𝐻 = 0),

1 ∗ 𝜑0 + 100 ∗ 𝜑1 = 0, 𝛾 = −12.

Thus, the solution is 𝛾 = −12.

Problem 5

Consider a market ((Ω, ℱ, 𝔽, ℙ), 𝑋) with one risky asset 𝑋 = {𝑋0 , 𝑋1 , 𝑋2 } and a contingent claim 𝐻
defined as follows:

Ω = {𝑎, 𝑏, 𝑐, 𝑑}, 𝔽 = 𝔽𝑋 , ℱ = 2Ω .

© 2019 - WorldQuant University – All rights reserved.

18
MScFE 620 Discrete-time Stochastic Processes - Module 4: Problem Set

Compute the price of 𝐻 at time 1.

Solution:

It is important to remark that the problem is asking about the price at 𝑡 = 1. Thus, we need to
focus on the filtration on 𝑡 = 1 or, in other words, we will have to compute two different prices at
𝑡 = 1, 𝐻1 {𝑎, 𝑏} and 𝐻1 {𝑐, 𝑑}.

Let’s start for the path {𝑎, 𝑏}. In this case, we have to solve the following system of equations:

24𝜑1 + 30𝜑2 = 12

24𝜑1 + 20𝜑2 = 0.

12
The solution of the above system is 𝜑1 = −1 and 𝜑2 = . Thus the price 𝐻1 {𝑎, 𝑏} is equal to
10

12 24
𝐻1 {𝑎, 𝑏} = −24 ∗ 1 + 24 ∗ = .
10 5

If we follow the same steps for {𝑐, 𝑑}, we first solve the system:

12𝜑1 + 14𝜑2 = 10

12𝜑1 + 10𝜑2 = 0.

−25 10
The solution is 𝜑1 = − 12
and 𝜑2 = 4
. The price, 𝐻1 {𝑐, 𝑑}, will be

25 10
𝐻1 {𝑐, 𝑑} = −12 ∗ + 12 ∗ = 5.
12 4

Finally, we just can express the solution as

24
𝐻1 = 𝐼{𝑎,𝑏} + 5 𝐼{𝑐,𝑑} .
5

© 2019 - WorldQuant University – All rights reserved.

19
MScFE 620 Discrete-time Stochastic Processes - Module 4: Bibliography

Bibliography

Bingham, N. and Kiesel, R. (2004). Risk-Neutral Valuation: Pricing and Hedging of Financial
Derivatives. London: Springer.

Föllmer, H. and Schied, A. (2004). Stochastic Finance. Berlin: Walter de Gruyter.

Shiryaev, A. (2001). Essentials of Stochastic Finance: Facts, Models, Theory. Translated by N.

Kruzhilin. Singapore: World Scientific.

Shreve, S. (2005). Stochastic Calculus for Finance I: The Binomial Asset Pricing Model. New York:
Springer.

© 2019 - WorldQuant University – All rights reserved.

20
MScFE 620 Discrete-time Stochastic Processes - Module 4: Collaborative Review Task

Collaborative Review Task

In this module, you are required to complete a collaborative review task, which is designed to test
your ability to apply and analyze the knowledge you have learned during the week.

Brief
1
Let 𝑇 = 2, 𝑑 = 1 and Ω = {𝑎, 𝑏, 𝑐, 𝑑}, ℙ = 4 (𝛿𝑎 + 𝛿𝑏 + 𝛿𝑐 + 𝛿𝑑 ) and

Assume that 𝔽 is the natural filtration of 𝑋 given by

𝐹0𝑋 = {∅, Ω}, 𝐹1𝑋 = {∅, Ω, {𝑎, 𝑏}, {𝑐, 𝑑}}, and 𝐹2 = 𝜎({𝑋0 , 𝑋1 , 𝑋2 }) = 2Ω .
1 Show that the market is complete.
2 Find the unique no-arbitrage price of 𝐻.
3 Construct a replicating strategy for 𝐻.

Remember to show all your calculations, otherwise marks will be deducted.

© 2019 - WorldQuant University – All rights reserved.

21