← All notes

Mathematics for Finance

Contents
  1. Financial Markets
  2. Contingent Claim
  3. Binomial Model
  4. Trading Strategies and Fair Pricing
  5. Call-Put Parity
  6. Optional Stopping Theorem
  7. Necessary Wealth
  8. Pricing an American Call
  9. Arbitrage
  10. American Call-Put Parity
  11. Martingale Measure
  12. First Fundamental Theorem of Asset Pricing
  13. Second Fundamental Theorem of Asset Pricing
  14. Optional Stopping
  15. Ito's Stochastic Integral
  16. Localization
  17. Ito's Process
  18. Martingale Representation Theorem
  19. Pricing a European Contingent Claim
  20. Black-Scholes Formula
  21. Geometric Asian Call

Introduction

We will discuss mathematics for options, pricing methods, and optimizing the performance of a portfolio. We will cover discrete (finite probability spaces, tree models) as well as continuous math (Brownian Motion, Black-Scholes formula, Martingale theory).

The theory in the course revolves around the “No Arbitrage theorem” which is closely related to one-price principle. Similarly, the idea of one-price principle theorizes that a set of investments with the same net outcome, albeit with different transactions, should have the same price.

Financial Markets

A financial market consists of tradable products such as stocks, bonds, currencies and indices. The market consists of two parties - the buyers and the sellers, who buy and sell these products respectively trying to make an arbitrage from the transactions. Formally, an arbitrage opportunity is a chance for a risk-free profit.

Odds and Arbitrage

Let us consider the example of a horse race, where P(S wins)=3/4P(\text {S wins}) = 3/4 and P(W wins)=1/4P(\text{W wins}) = 1/4. How do we calculate the odds against WW winning? It is given by P(W wins)/P(W loses)=3P(\text{W wins})/P(\text{W loses}) = 3. This situation is called 3-to-13\text{-to-}1 odds. That means, a dollar bet $1\$ 1 brings $3\$ 3 if WW wins. Similarly, for SS, it brings $0.3\$ 0.3. These bets are an example of a fair game, the net reward 14(3)+34(−1)\frac{1}{4} (3) + \frac{3}{4}(-1) is 00.

A bookmaker sets the odds for such scenarios. Consider an other example where the odds are 9-to-59\text{-to-}5 against WW and 2-to-52\text{-to-}5 against SS. We claim that this results in an arbitrage for the bookmaker. To illustrate this, consider the bets $10\$ 10 on SS and $5\$ 5 on WW. The payoff for the bookmaker when SS wins is 5−1025=15 - 10\frac{2}{5} = 1. When WW wins, the profit is 10−595=110 - 5\frac{9}{5} = 1. A bookmaker typically decides the odds after the bets are made so that arbitrage is made in any scenarios.

Ideally, one wants to make profit from such games. How do we do this? In the first example, we considered the probabilities and calculated the odds. To decide the odds for arbitrage, we do the reverse. In the above scenario, given the odds, the probabilities are P(S)≈71%P(S) \approx 71\% and P(W)≈36%P(W) \approx 36\%. The probabilities add up to slightly above 100%100\%, which makes up the arbitrage for the bookmaker.

Contingent Claim

A derivative or a contingent claim is a security whose value depends on the value of an underlying asset. Forward contracts, futures contracts and options are examples of such securities. Interestingly, the markets around the world also use derivatives whose underlying asset is also a derivative. Such securities are called as structured products.

With such contrived products, it is important to choose a fair price to ensure there are no arbitrage opportunities in the market. To do so, we introduce the required mathematical notation - XX is a real-valued random variable defined on a probability space (Ω,F,P)(\Omega, \mathcal F, P) and G\mathcal G is a sub-σ\sigma-algebra of F\mathcal F. For unfamiliar readers, in probability theory, a probability space consists of three elements -

  • A sample space Ω\Omega, which is the set of all possible outcomes. For example, the sample space for a dice roll is Ω={1,2,3,4,5,6}\Omega = \{1, 2, 3, 4, 5, 6\}.
  • An event space or a set of events is represented by F\mathcal F. For example, the events constituting an odd dice roll is F={1,3,5}\mathcal F = \{1, 3, 5\}.
  • A probability function PP, assigns a probability (a number between 00 and 11) to each event in the event space.

The term σ\sigma-algebra on a set XX refers to a nonempty collection Σ\Sigma of subsets of XX closed under complement, countable unions, and countable intersections. The pair (X,Σ)(X, \Sigma) is called a measurable space.

In general, we model the price of a contract with a random variable XX that evolves across time with X(w)≥0X(w) \geq 0 (all the values for the random variable are positive). For example, for the call option described in #options, we have

X(w)=(S(w)t−K)+X(w) = (S(w)_t - K)^+

Similarly, for a put option, where the holder has the right to sell a certain asset at a fixed price, we get

X(w)=(K−ST)+X(w) = (K - S_T)^+

There can be other kinds of structured products like

  • X=max⁡(S1,…,ST)X = \max(S_1, \dots, S_T)
  • X=(1T∑t=1TSt−K)+X = \left(\frac{1}{T} \sum_{t = 1}^T S_t - K\right)^+

In the above definitions, the variables w,Stw, S_t capture the evolution of price across time, and are described in the later section. The key takeaway is that, we are trying to price a contingent claim XX - find C0(X)C_0(X)

Forwards Contract

A forward constract is an agreement to buy or sell an asset at a certain future time for a certain price. Consider the following example,

  • AA agrees at time t=0t = 0 to sell one share at time TT for $F\$ F. We assume that AA buys the share at t=0t = 0 at price S0S_0.
  • BB agrees at time t=0t = 0 to buy one share at time TT for $F\$ F

We assume that A,BA, B don’t invest their own money, and borrow money from the bank to perform transactions. Given such a scenario, how do we decide a fair price FF? To simplify the calculations, we assume the following -

  • No transaction costs or dividends.
  • Market is liquid - every transaction has a buyer and seller available.
  • Investor is small compared to the market. That is, the action of the investor is not momentous enough to change the price of the stock in the market.
  • Short selling/borrowing of stock is allowed.

BB can invest the amount in the bank for a continuous compound interest rate rr during this term. The claim is that F=S0erTF = S_0 e^{rT} where S0S_0 is the price of the share at t=0t = 0. To analyze this, we consider the following cases -

  • F>S0erTF > S_0 e^{rT}

    AA sells the share for $F\$ F and repays the loan of $S0\$ S_0 with interest - $S0erT\$ S_0 e^{rT}. The profit AA gets is F−S0erTF - S_0 e^{rT}

  • F>S0erTF > S_0e^{rT}

    BB short sells one share for $S0\$ S_0 and invests $S0\$ S_0 in the bank. The money in the bank will grow to S0erTS_0 e^{rT} at t=Tt = T, and BB buys one share at FF. The profit for BB here is S0erT−F>0S_0 e^{rT} - F > 0

So the fair price is S0erTS_0 e^{rT} where neither AA nor BB can take advantage.

Note. A futures contract is similar to a forward contract but is traded on a financial exchange. They typically have a delivery month rather than a delivery date, and are followed by a settlement procedure called marking to market.

Options

An option is a contract which gives the holder of the option the right, but not the obligation, to buy other sell a given security at a given price (called the strike price) within a fixed time period [0,T][0, T]. A call option gives the option holder the right to buy at the given price, whereas the put option gives the option holder the right to sell at the given price.

European Call Options

A European option can only be exercised by the holder of the option at the expiration time TT (unlike an American Option, which will be discussed in the later sections). We have the following equation for profit of a call option holder, assuming the option is free -

P=(ST−K)+={ST−KST>K0ST≤KP = (S_T - K)^+ = \begin{cases} S_T - K & S_T > K \\ 0 & S_T \leq K \end{cases}

How do we decide the fair price at t=0t = 0 for the option? We need to use probability results to model the price movement to decide this. In discrete time settings, we use different models like Cox, Ross and Rubenstein. In continuous time frame, we use Brownian motion using the Black-Scholes model. In both cases, there exists a unique fair price. However, if the product is slightly more complicated, then there is no unique fair price.

Binomial Model

Also known as Cox-Ross-Rubenstein Model (CRR) is a discrete time model for a financial market defined for t=0,1,…,Tt = 0, 1, \dots, T. At each time instant the model considers the following two types of assets

Bonds

Considered as a risk-free asset, a bond yields at a constant rate of return r≥0r \geq 0 over each time period. The price is modeled as Bt=B0(1+r)tB_t = B_0 (1 + r)^t for t=0,1,…,Tt = 0, 1, \dots, T. B0B_0 is typically assumed as 11.

Stocks

Stocks are considered as risk-models where the price evolution is captured using binomial random variables. We consider the model St=S0ζ1…ζt=St−1ζtS_t = S_0 \zeta_1 \dots \zeta_t = S_{t - 1} \zeta_t where S0>0S_0 > 0 is the constant initial price. That is, the stock price process is modeled as an exponential random walk. In this model, we assume P(ζt=u)=pP(\zeta_t = u) = p and P(ζt=d)=1−pP(\zeta_t = d) = 1- p where 0<d<1+r<u0 < d < 1 + r < u where d,ud, u are the down and up movements respectively. That is, St+1=StuS_{t + 1} = S_t u represents the price of the stock if the price has moved upward at time tt.

Concretely, the probability space (Ω,F,P)(\Omega, \mathcal F, P) is such that Ω\Omega is the finite set of 2T2^T possible outcomes for the values of stock price (T+1)(T + 1)-tuple (for S0,S1,…,ST)S_0, S_1, \dots, S_T), F\mathcal F is the σ\sigma-algebra consisting of all possible subsets of Ω\Omega, and PP is the probability measure on (Ω,F)(\Omega, \mathcal F) associated with the Bernoulli probability pp.

Let the field F0={ϕ,Ω}\mathcal F_0 = \{ \phi, \Omega\} and

F1=σ(S1)={ϕ,Ω,{w:S1(w)=S0u},{w:S1(w)=S0d}}=σ({S1=S0u}),{S1=S0d})\begin{align*} \mathcal F_1 &= \sigma(S_1) \\ &= \{ \phi, \Omega, \{w: S_1(w) = S_0u \}, \{w: S_1(w) = S_0 d\}\} \\ &= \sigma(\{S_1 = S_0u\}), \{S_1 = S_0d\}) \end{align*}

This is a partition of Ω\Omega. Similarly, for t=2t = 2, we get

F1={ϕ,Ω,{uu,ud},{dd,du}}\mathcal F_1 = \{ \phi, \Omega, \{uu, ud\}, \{dd, du\}\}. Extrapolating this,

Ft=σ(S1,…,St)=σ(ζ2,…,ζt)\mathcal F_t = \sigma(S_1, \dots, S_t) = \sigma(\zeta_2, \dots, \zeta_t)

We get #(F)t=2t\#(\mathcal F)_t = 2^t. This evolution can be represented using a binomial tree of the following form -

duddduuduu… \begin{align*} &\boxed{} \\ \boxed{d} &\quad \boxed{u} \\ \boxed{dd} \quad \boxed {du} &\quad \boxed{ud} \quad \boxed{uu} \\ &\dots \end{align*}

Trading Strategies and Fair Pricing

A strategy ϕ\phi in stocks and bonds consists of sequence of random variable pairs ϕ={(αt,βt);t=1,…,T}\phi = \{(\alpha_t, \beta_t); t = 1, \dots, T\} where αt\alpha_t represents the holdings of a stock and βt\beta_t represents the holdings of a bond on day tt (The unit for time can be taken as minutes, hours, etc., but we consider days). We also have the following assumptions

  1. Predictability: αt,βt∈Ft−1,t=1,…,T\alpha_t, \beta_t \in \mathcal F_{t - 1}, t = 1, \dots, T
  2. Self-financing: αtSt+βtBt=αt+1St+βt+1Bt\alpha_t S_t + \beta_t B_t = \alpha_{t + 1} S_t + \beta_{t + 1}B_t. That is, there is no influx or outflux of money asides from the changes in the underlying assets. We have some initial investment and assume that α1,β1\alpha_1, \beta_1 are F0\mathcal F_0-measurable are constant.

The value of the portfolio at time tt is given by Vt(ϕ)=αtSt+βtBt,t=1,…,TV_t(\phi) = \alpha_t S_t + \beta_t B_t, t = 1, \dots, T and V0(ϕ)=α1S0+βB0=α1S1+β1V_0(\phi) = \alpha_1 S_0 + \beta B_0 = \alpha_1 S_1 + \beta_1 We can now define the concept of arbitrage formally.

Arbitrage

An arbitrage is a trading strategy ϕ\phi such that V0(ϕ)=0V_0(\phi) = 0 and Vt(ϕ)≥0V_t(\phi) \geq 0 . That is, we don't have any initial investment and get a profit at time t=Tt = T. However, note that Vt(ϕ)V_t(\phi) is a random variable, so the inequality is assumed to be a point-wise inequality wherein VT(ϕ)(w)≥0,∀w∈ΩV_T(\phi)(w) \geq 0, \forall w \in \Omega. This is a stronger assumption than P(VT(ϕ)>0)>0≡E(VT(ϕ))>0≡VT(ϕ)(w)>0P(V_T(\phi) > 0) > 0 \equiv E(V_T(\phi)) > 0 \equiv V_T(\phi)(w) > 0 for some w∈Ωw \in \Omega. Notice that the last equivalence does not have any probability term in the expression. This equivalence is important to note since it makes the definition invariant of the underlying probability assumptions.

Hedge

A hedging or a replicating strategy ϕ\phi has the property VT(ϕ)=XV_T(\phi) = X. That is, to price a contingent claim XX, if we find a trading strategy ϕ\phi with cost (initial investment) V0(ϕ)V_0(\phi), such that VT(ϕ)=XV_T(\phi) = X, then V0(ϕ)V_0(\phi) is a fair price of XX by the one-price principle. Essentially, with the price “V0(ϕ)V_0(\phi)” we bought an opportunity that resulted in a value of "VT(ϕ)V_T(\phi)" at time TT. If a contingent claim XX has the same value at time TT, then its fair price should be V0(ϕ)V_0(\phi) at t=0t = 0.

Using this as a motivation, let us price a European call option. Our approach is as follows - an investment in the underlying stock and a bond should result in the same outcome as an investment in the option itself. The movements in the option will be correlated with those of the underlying stock, allowing us to correctly determine a fair price at every time instant.

Hedging for T=1T = 1

Consider the following example, where we try to hedge XX at t=1t = 1. That is, we are trying to find a strategy ϕ=(α1,β1)\phi = (\alpha_1, \beta_1) such that V1(ϕ)=α1S1+β1B1≡XV_1(\phi) = \alpha_1 S_1 + \beta_1 B_1\equiv X.

We have two cases based on the stock movement -

  • α1S0u+β1(1+r)=Xu\alpha_1 S_0 u + \beta_1 (1 + r) = X^u
  • α1S0d+β1(1+r)=Xd\alpha_1 S_0 d + \beta_1 (1 + r) = X^d

Equivalently, α1S0(u−d)=Xu−Xd\alpha_1 S_0(u - d) = X^u - X^d

α1=Xu−Xdu−d1S0β1=uXd−dXuu−d1B1\begin{align*} \alpha_1 &= \frac{X^u - X^d}{u - d}\frac{1}{S_0} \\ \beta_1 &= \frac{uX^d - d X^u}{u - d}\frac{1}{B_1} \end{align*}

What do the above expressions mean? Investing α1\alpha_1 in stock and β1\beta_1 in bond at the end of day 00 will result in same price as XX on day 11. Manipulating the expressions further,

V0(ϕ)=α1S0+β1=Xu−Xdu−d+uXd−dXuu−d11+r=11+r{Xu(1+r−du−d)+Xd(u−1−ru−d)}=11+rE∗(X)\begin{align*} V_0 (\phi) &= \alpha_1 S_0 + \beta_1 \\ &= \frac{X^u - X^d}{u- d} + \frac{uX^d - d X^u}{u - d}\frac{1}{1 + r} \\ &= \frac{1}{1 + r}\left\{X^u \left(\frac{1 + r - d}{u - d}\right) + X^d \left(\frac{u - 1 - r}{u - d}\right)\right\} \\ &= \frac{1}{1 + r} \mathbb E^* (X) \end{align*}

The above expression is reminiscent of an expectation formulation where p∗=1+r−du−dp^* = \frac{1 + r - d}{u - d} and 0<p∗<10 < p^* < 1. Here, the set of discounted stock prices {S0,S1/(1+r)}\{S_0, S_1/(1 + r)\} form a p∗p^*-martingale.

What is a Martingale? A Martingale is a sequence of random variables representing a stochastic process for which, at a particular time, the conditional expectation of the next value in the sequence is equal to the present value, regardless of all prior values.

In the above case, we have

E∗(S1/(1+r)∣F0)=S0u⋅p∗+S0d⋅(1−p)∗=S0  ⟹  E∗(S1)=S0(1+r)\begin{align*} \mathbb E^* (S_1/(1 + r) \vert \mathcal F_0) &= S_0 u \cdot p^* + S_0 d \cdot (1 - p)^* \\ &= S_0 \implies \mathbb E^*(S_1) = S_0(1 + r) \end{align*}

It is important to realize that computing expectations under p∗p^* is a mathematical device. We are not assuming that the stock price actually moves according to this probability. That is, p∗p^* may be unrelated to the subjective probability pp that we associate with the binomial model for movements in the stock price.

Theorem 1 (T = 1) For a given XX, the value E∗(X)/(1+r)\mathbb E^*(X)/(1 + r) is the non-arbitrage price of XX at t=0t = 0. Proof. Let C0C_0 is the market price of the contingent claim XX and V0=E∗(X)/(1+r)=V0(ϕˉ)V_0 = \mathbb E^*(X)/(1 + r) = V_0(\bar \phi) where ϕˉ=(αˉ1,βˉ1)\bar \phi = (\bar \alpha_1, \bar \beta_1) (the values we calculated before). Consider the following cases -

  1. Say C0>V0C_0 > V_0 To make an arbitrage in this situation, we sell XX for C0C_0. We invest the amount V0V_0 in the stock and amount C0−V0>0C_0 - V_0 > 0 in bonds at t=0t = 0. At time t=1t = 1, we get V1(ϕˉ)=XV_1(\bar \phi) = X yielding (C0−V0)(1+r)>0(C_0 - V_0)(1 + r) > 0 as profit from the bond.

  2. Say C0<V0C_0 < V_0 We do the opposite from above, where we buy the asset XX. Short sell ϕˉ\bar \phi for V0V_0, and invest C0C_0 in XX and V0−C0V_0 - C_0 in the bond. We get the profit (V0−C0)(1+r)(V_0 - C_0)(1 + r) at t=1t = 1.

  3. Say C0=V0C_0 = V_0 Here, we can show that no one can make an arbitrage. Consider the strategy in stock, bond and contingent claim ψ=(α1,β1,γ1)\psi = (\alpha_1, \beta_1, \gamma_1). We need to argue that no arbitrage can be made with any such strategy. We have,

V0(ψ)=α1S0+β1+γ1C0V1(ψ)=α1S1+βB1+γ1X  ⟹  E∗(V1(ψ))=E∗(α1S1+β1B1+γ1X)=α1E∗(S1)+β1(1+r)+γ1E∗(X)\begin{align*} V_0(\psi) &= \alpha_1 S_0 + \beta_1 + \gamma_1 C_0 \\ V_1(\psi) &= \alpha_1 S_1 + \beta B_1 + \gamma_1 X \\ \implies \mathbb E^*(V_1(\psi)) &= \mathbb E^*(\alpha_1 S_1 + \beta_1 B1 + \gamma_1 X) \\ &= \alpha_1 \mathbb E^*(S_1) + \beta_1 (1 + r) + \gamma_1 \mathbb E^*(X) \end{align*}

We can distribute the expectation this way due to the linearity property. Recall that, we typically assume B0=1B_0 = 1 for brevity. Continuting the calculations,

E∗(V1(ψ))=α1S1(1+r)+β1(1+r)+γ1V0(ϕˉ)(1+r)=α1S1(1+r)+β1(1+r)+γ1C0(1+r)=(1+r)V0(ψ)\begin{align*} \mathbb E^*(V_1(\psi)) &= \alpha_1 S_1(1 + r) + \beta_1(1 + r) + \gamma_1 V_0 (\bar \phi)(1 + r) \\ &= \alpha_1 S_1(1 + r) + \beta_1(1 + r) + \gamma_1 C_0 (1 + r) \\ &= (1 + r) V_0 (\psi) \end{align*}

If V0(ψ)=0V_0(\psi) = 0, then E∗(V1(ψ))=0\mathbb E^*(V_1(\psi)) = 0, and if V1(ψ)≥0V_1(\psi) \geq 0 , then P∗(V1(ψ)=0)=1P^*(V_1(\psi) = 0) = 1. That is, V1(ψ)=0V_1(\psi) = 0 meaning no arbitrage can be made with any strategy ψ\psi.

General TT

This analysis can be extended to derive the hedging strategy a general t>1t > 1. Given an XX, find ϕ={(αt,βt):t=1,…,T}\phi = \{(\alpha_t, \beta_t): t = 1, \dots, T\} such that VT(ϕ)=XV_T(\phi) = X (resulting in C0=V0(ϕ)C_0 = V_0(\phi)). To derive such a strategy, we work backwards in time starting from VT(ϕ)=XV_T(\phi) = X. This way, we can also calculate the intermediate price - the price to enter the market at some intermediary time tt. It is helpful to think of the tree visualization for the price evolution.

VT(ϕ)=αTST+βTBT=X\begin{align*} V_T(\phi) = \alpha_T S_T + \beta_T B_T = X \end{align*}

conditional on FT−1\mathcal F_{T - 1}. Focusing on T−1T - 1 and TT, the analysis is similar to the previous case wherein XU=X(w1…wT−1u)X^U = X(w_1 \dots w_{T - 1} u) and Xd=X(w1…wT−1d)X^d = X(w_1 \dots w_{T - 1} d). Doing so we get

αT=Xu−Xdu−d1ST−1βT=uXd−dXuu−d1(1+r)T}∈FT−1\left. \begin{align*} \alpha_T = \frac{X^u - X^d}{u - d}\frac{1}{S_{T - 1}} \\ \beta_T = \frac{uX^d - dX^u}{u - d}\frac{1}{(1 + r)^T} \end{align*} \right\} \in \mathcal F_{T - 1}

Using the property of Martingales,

E∗(X∣FT−1)=E∗[VT(ϕ)∣FT−1]=E∗[αTST+βTBT∣FT−1]=αTE∗(ST∣FT−1)+βTBT=αT(p∗ST−1u+(1−p∗)ST−1d)+βTBT=αTST−1(1+r)+βTBT\begin{align*} \mathbb E^* (X \vert \mathcal F_{T - 1}) = \mathbb E^* [V_T(\phi) \vert \mathcal F_{T - 1}] &= \mathbb E^*[\alpha_T S_T + \beta_T B_T \vert \mathcal F_{T - 1}] \\ &= \alpha_T \mathbb E^*(S_T \vert \mathcal F_{T - 1}) + \beta_T B_T \\ &= \alpha_T \left(p^* S_{T - 1} u + (1 - p^*) S_{T - 1} d\right) + \beta_T B_T \\ &= \alpha_T S_{T - 1}(1 + r) + \beta_T B_T \end{align*}

Note. αT,βT\alpha_T, \beta_T are constants with respect to the condition FT−1\mathcal F_{T - 1}, and that is why the linearity property can be used.

Continuing the previous calculations

E∗(VT∣FT−1)=E∗(αTST+βTBT∣FT−1)=αTST−1(1+r)+βTBT−1(1+r)=(1+r)[αTST−1+βTBT−1]=(1+r)[αT−1ST−1+βT−1BT−1]∵ self-financing=(1+r)VT−1(ϕ)\begin{align*} \mathbb E^*(V_T \vert \mathcal F_{T - 1}) &= \mathbb E^*(\alpha_T S_T + \beta_T B_T \vert \mathcal F_{T - 1}) \\ &= \alpha_T S_{T - 1}(1 + r) + \beta_T B_{T - 1}(1 + r) \\ &= (1 + r) \left[\alpha_T S_{T - 1} + \beta_T B_{T - 1}\right] \\ &= (1 + r) \left[\alpha_{T - 1} S_{T - 1} + \beta_{T - 1} B_{T - 1}\right] &\because \text{ self-financing} \\ &= (1 + r) V_{T - 1}(\phi) \end{align*}

As a result, we have a recursive property

Vt−1(ϕ)=11+rE∗(Vt(ϕ)∣Ft−1)αt=Vtu−Vtdu−d1St−1βt=uVtd−dVtuu−d1Bt−1\begin{align*} V_{t - 1}(\phi) &= \frac{1}{1 + r} \mathbb E^*(V_t(\phi) \vert \mathcal F_{t - 1}) \\ \alpha_t &= \frac{V_t^u - V_t^d}{u - d} \frac{1}{S_{t - 1}} \\ \beta_t &= \frac{uV^d_t - dV^u_t}{u - d} \frac{1}{B_{t - 1}} \end{align*}

for t=T,…,1t = T, \dots, 1

The above relation can also be alternately derived using the Tower law to relate the value with Vt(ϕ)V_t(\phi) for a general tt.

Tower Law: If XX is a random variable whose expected value E⁡(X)E ⁡( X ) is defined, and YY is any random variable on the same probability space, then E(X)=E(E(X∣Y))E(X) = E (E (X \vert Y ) ), i.e., the expected value of the conditional expected value of XX given YY is the same as the expected value of XX.

Then, we get

Vt(ϕ)=11+rE∗(Vt+1(ϕ)∣Ft)=11+rE∗(11+rE∗(Vt+2(ϕ)∣Ft+1)∣Ft)=1(1+r)2E∗(E∗(Vt+2∣Ft+1)∣Ft)=1(1+r)2E∗(Vt+2∣Ft)\begin{align*} V_t(\phi) &= \frac{1}{1 + r} \mathbb E^*(V_{t + 1} (\phi) \vert \mathcal F_t) \\ &= \frac{1}{1 + r} \mathbb E^*\left(\frac{1}{1 + r} \mathbb E^*(V_{t + 2}(\phi) \vert \mathcal F_{t + 1} ) \vert \mathcal F_t\right) \\ &= \frac{1}{(1 + r)^2} \mathbb E^*\left(\mathbb E^*\left(V_{t + 2} \vert \mathcal F_{t + 1}\right) \vert \mathcal F_{t} \right) \\ &= \frac{1}{(1 + r)^2} \mathbb E^*(V_{t + 2} \vert \mathcal F_t) \end{align*}

Consequently,

VT(ϕ)=1(1+r)T−tE∗(VT(ϕ)∣Ft)=1(1+r)T−tE∗(X∣Ft)V0(ϕ)=1(1+r)TE∗(X)\begin{align*} V_T(\phi) &= \frac{1}{(1 + r)^{T - t}} \mathbb E^*(V_T (\phi) \vert \mathcal F_t) \\ &= \frac{1}{(1 + r)^{T - t}} \mathbb E^*(X \vert \mathcal F_t) \\ V_0 (\phi) &= \frac{1}{(1 + r)^T} \mathbb E^*(X) \end{align*}

Conditional Expectation of Martingales

Given (Ω,D,P)(\Omega, \mathcal D, P) with g⊂Fg \subset \mathcal F, X∈L′(F)X \in L'(\mathcal F) (the space of random variables that are integrable with respect to the σ\sigma-algebra F\mathcal F) then Y=E(X∣g)Y = \mathbb E(X \vert g) is the random variable such that

  1. Y∈L′(g)Y \in L'(g)

  2. E(YZ)=E(XZ)\mathbb E(YZ) = \mathbb E(XZ) for all ZZ that are gg-measurable.

where L′L' represents integrability. The above property is the general statement for the law of general statician.

Special case. When Ω\Omega is finite (#(Ω)<∞\#(\Omega) < \infty), then F={all subsets of Ω}\mathcal F = \{ \text{all subsets of } \Omega\} and G=σ(G1,…,Gn)\mathcal G = \sigma(G_1, \dots, G_n) form a partition (Gi∩Gj=ϕ,∪j=inGj=ΩG_i \cap G_j = \phi, \cup_{j = i}^n G_j = \Omega) of Ω\Omega with P(Gk)>0P(G_k) > 0.

A random variable ZZ is gg-measurable iff Z(w)=∑k=1n1Gk(w)⋅ck≡Z(w)=ck if w∈GkZ(w) = \sum_{k = 1}^n \mathbb 1_{G_k} (w) \cdot c_k \equiv Z(w) = c_k \text{ if } w \in G_k where ckc_k is the cardinality of GkG_k.

Then,

E[X∣G](w)=∑k=1n(1Gk(w)⋅E[X∣Gk])\mathbb E[X \vert \mathcal G](w) = \sum_{k = 1}^n \left(\mathbb 1_{G_k}(w) \cdot E[X \vert G_k]\right)

From Bayes' rule

E(X∣Gk)=E(X.1Gk)/P(Gk)=E(X;Gk)/P(Gk)\mathbb E(X \vert G_k) = \mathbb E(X . \mathbb 1_{G_k}) /P(G_k) = \mathbb E(X; G_k)/P(G_k)

Using the above results, we get

  1. X→E(X∣Y)X \to \mathbb E(X \vert Y) is linear & positive (X≥0  ⟹  E(X∣g)≥0 X \geq 0 \implies \mathbb E(X \vert g) \geq 0)

  2. E(E(X∣g))=E(X)\mathbb E(\mathbb E (X \vert g)) = \mathbb E(X)

  3. E(XZ∣g)≡E(X∣y)Z\mathbb E(XZ \vert g) \equiv E(X \vert y) Z

  4. If E∈g\mathbb E \in g, then E(X∣g)=XE(X \vert g) = X and X⊥ ⁣ ⁣ ⁣⊥gX \perp\!\!\!\perp g, then E(X∣Y)=E(X)\mathbb E(X \vert Y) = \mathbb E(X).

Exercise.

Show that E(⋅∣G)E( \cdot \vert \mathcal G) is a ⊥\perp projection where d(X,X~)=E(X−X~)2d(X, \tilde X) = \sqrt{\mathbb E(X - \tilde X)^2}

Tower Property

For any G⊂H(⊂F)\mathcal G \subset \mathcal H ( \subset \mathcal F), we have E[E[X∣H]∣G]=E[X∣G]\mathbb E[\mathbb E[X \vert \mathcal H] \vert \mathcal G] = \mathbb E[X \vert \mathcal G].

What does G⊂H\mathcal G \subset \mathcal H mean? In the discrete case where G=σ(G1,…,Gn)\mathcal G = \sigma(G_1, \dots, G_n) and H=σ(H1,…,Hm)\mathcal H = \sigma(H_1, \dots, H_m), we have Gk=∪˙j∈J(k)HjG_k = \dot \cup_{j \in J(k)} H_j where J(1),…,J(n)J(1),\dots, J(n) is a partition of {1,…,m}\{1, \dots, m\}.

Now, to show the tower property,

E[X∣Gk]=E[X;Gk]⏟P(Gk)=∑j∈J(k)E(X;Hj)⏟P(Hj)P(Hj∣Gk)=∑j∈J(k)E(X∣Hj)P(Hj∣Gk)\begin{align*} \mathbb E[X \vert G_k] &= \underbrace{\mathbb E[X; G_k]}_{P(G_k)} \\ &= \sum_{j \in J(k)} \underbrace{\mathbb E(X; H_j)}_{P(H_j)}P(H_j \vert G_k) \\ &= \sum_{j \in J(k)} \mathbb E(X \vert H_j)P(H_j \vert G_k) \end{align*}

In the trading strategy context, we have F0⊂F1⊂⋯⊂Fn\mathcal F_0 \subset \mathcal F_1 \subset \dots \subset \mathcal F_n.

Martingales

We now introduce Martingales formally - A Martingale is a sequence of random variables {M0,…,MT}\{M_0, \dots, M_T\} such that

  1. Adapted: Mt∈FtM_t \in \mathcal F_t, for all tt

  2. Integrable: E(∣Mt∣)<∞\mathbb E(\vert M_t\vert) < \infty for all tt

  3. E(Mt∣Ft−1)=Mt−1\mathbb E(M_t \vert \mathcal F_{t - 1}) = M_{t - 1}, 1≤t≤T1 \leq t \leq T. From Tower law, E[MT∣Fs]=Ms\mathbb E[M_T \vert \mathcal F_s] = M_s.

Let us now revert to our discussion on trading strategies with these formal definitions in mind.

As we have seen before in #General TT, ϕ\phi that satisfied Vt∗(ϕ):=Vt(ϕ)/(1+r)tV_t^*(\phi) := V_t(\phi)/(1 + r)^t is a p∗p^*-martingale, where p∗=1+r−du−dp^* = \frac{1 + r - d}{u - d}. A special case of this is St∗S^*_t is a p∗p^*- martingale. We have the following theorem,

Theorem 2. (General TT) V0V_0 is the arbitrage-free price of VV

Proof. We follow a similar structure as before,

  1. C0≠V0C_0 \neq V_0, then arbitrage is available

  2. C0=V0C_0 = V_0, we have for any strategy ψ={(αt,βT,γ1):t=1,…,T}\psi = \{(\alpha_t, \beta_T, \gamma_1): t = 1, \dots, T\}, V0(ψ)=0,VT(ψ)≥0V_0(\psi) = 0, V_T(\psi) \geq 0 due to self-financing. Let ϕˉ={(αˉt,βˉT):t=1,…,T}\bar \phi = \{(\bar \alpha_t, \bar \beta_T): t = 1, \dots, T\} be the hedge for XX. Then, VT(ϕˉ)=XV_T(\bar \phi) = X and V0(ϕˉ)=V0V_0(\bar \phi) = V_0.

    V0(ψ)=α1S0+β+γ1C0VT(ψ)=αTSt+βTBT+γ1X\begin{align*} V_0 (\psi) &= \alpha_1 S_0 + \beta + \gamma_1 C_0 \\ V_T(\psi) &= \alpha_T S_t + \beta_T B_T + \gamma_1 X \end{align*}

    Continuing the calculations,

    E∗(VT∗(ψ))=E∗(αTST∗+βT+γ1X∗)=E∗(VT(ϕ))+γ1E∗(X∗)=V0(ϕ)+γ1C0=V0(ψ)=0  ⟹  VT(ψ)=0\begin{align*} \mathbb E^*(V^*_T(\psi)) &= \mathbb E^*(\alpha_T S^*_T + \beta_T + \gamma_1 X^*) \\ &= \mathbb E^*(V_T(\phi)) + \gamma_1 \mathbb E^*(X^*) \\ &= V_0(\phi) + \gamma_1 C_0 \\ &= V_0 (\psi) = 0 \implies V_T(\psi) = 0 \end{align*}

    Therefore, no arbitrage is possible in this case.

Let us consider a numerical example to understand this better - Consider the call option X=(ST−K)+X = (S_T - K)^+, where T=4,u=1.2,d=0.8,r=0T = 4, u = 1.2, d = 0.8, r = 0 (p∗=0.5p^* = 0.5). Let S0=50S_0 = 50 and K=40K = 40. What is the fair price of XX?

Then, we have the following evolution for STS_T,

50406032487225.638.457.686.420.4830.7246.0869.12103.68\begin{align*} &50\\ 40&\quad 60\\ 32\quad &48\quad 72\\ 25.6\quad 38.4&\quad 57.6\quad 86.4\\ 20.48\quad 30.72\quad &46.08\quad 69.12\quad 103.68 \end{align*}

Consequently, the tree for V0V_0 is

13.54121.661.5210.323203.0417.646.4006.0829.1263.68\begin{align*} &13.54\\ 1&\quad 21.66\\ 1.52\quad &10.32\quad 32\\ 0\quad 3.04&\quad 17.6\quad 46.4\\ 0\quad\quad 0 \quad &6.08\quad 29.12 \quad 63.68 \end{align*}

We construct the tree recursively, starting from the bottom-most level. At time TT, we know VTV_T, and then we use Vtup∗+Vtd(1−p∗)V_t^u p^* + V_t^d(1 - p^*) to get Vt−1(1+r)V_{t - 1}(1 + r).

Call-Put Parity

A call option is represented as CT=(ST−K)+C_T = (S_T - K)^+ and a put option is represented as PT=(K−ST)+=(ST−K)−P_T = (K - S_T)^+ = (S_T - K)^-.

For any bb, we have b+−b−=bb^+ - b^- = b and b++b−=bb^+ + b^- = b. In other words, the payoffs in call and put options are related to each other as

CT−PT=ST−KCT+K=ST+PT\begin{align*} C_T - P_T = S_T - K \\ C_T + K = S_T + P_T \end{align*}

Claim. For t<Tt < T, we have

Ct+K(1+r)T−t=St+PtC_t + \frac{K}{(1 + r)^{T - t}} = S_t + P_t

This result can be proved using One price principle. Think of each side of the equation as separate investments

  1. First investment has a call option and bonds with KK.

  2. Second investment has a put option and one share of the stock

We can see that at time TT, the values become equal CT+K=ST+PTC_T + K = S_T + P_T

American Options

In American Options, we have Y0,Y1,…,YTY_0, Y_1, \dots, Y_T random payouts where each Yt∈FTY_t \in \mathcal F_T. The holder of the options can take the payment at any time t∈[0,T]t \in [0, T] of choice. That is , unlike European options, American options have the choice of payment at any intermediate time. For example, consider a call option Yt=(St−K)+Y_t = (S_t - K)^+ for t=0,…,Tt = 0, \dots, T.

Formally, we have the stopping time property represented as T:Ω→{0,1,…,T}\mathcal T: \Omega \to \{0, 1, \dots, T\} such that {T≤t}={w∈Ω:T(W)≤t}∈Ft,∀t=0,1,…,T\{\mathcal T \leq t\} = \{w \in \Omega: \mathcal T(W) \leq t\} \in \mathcal F_t, \forall t = 0, 1, \dots, T. This is equivalent to saying, {T=t}∈Ft,∀t\{T = t\} \in \mathcal F_t, \forall t.

For example, consider Z0,Z1,…,ZTZ_0, Z_1, \dots, Z_T, we have T:=min⁡{t:ZT∈B}∧T\mathcal T:= \min \{t: Z_T \in B\} \wedge T where A∧b=min⁡(a,b)A \wedge b = \min(a, b). Then, {T=t}={Z0)∉B,Z1∉B,…,Zt−1∉B,Zt∈B}∈Ft\{\mathcal T = t\} = \{Z_0) \not\in B, Z_1 \not\in B, \dots, Z_{t - 1} \not \in B, Z_t \in B\} \in \mathcal F_t.

Optional Stopping Theorem

Consider a stopping time τ\tau and a Martingale (Mt)(M_t), we have E(Mτ)=E(M0)\mathbb E(M_\tau) = \mathbb E(M_0). Essentially, in a fair game, the average payout from exiting the option at an intermediate time is the same as the average payout till time t=Tt= T.

Here MτM_\tau is a random variable of the form Mτ(w)(w)M_{\tau(w)}(w) which we have represented in a singular form.

Is the converse true? That is, a sequence of random variables have the optional stopping property where expectations are the same for the beginning and an intermediate time. Then do they constitute a Martingale? Yes, we have

MT=M0+∑t=1T(Mt−Mt−1)Mτ=M0+∑t=1T(Mt−Mt−1)1{t≤τ}\begin{align*} M_T &= M_0 + \sum_{t = 1}^T (M_t - M_{t - 1}) \\ M_\tau &= M_0 + \sum_{t = 1}^T (M_t - M_{t - 1}) \mathbb 1_{\{t \leq \tau\}} \end{align*}

The event {t≤τ}\{t \leq \tau\} is equivalent to {τ≤t−1}c\{\tau \leq t - 1\}^c. This allows us to get back the notion of Ft−1\mathcal F_{t - 1}

E(Mτ)=E(M0)+∑t=1TE[(Mt−Mt−1)1{t≤τ}]=E(M0)+∑t=1TE[E[(Mt−Mt−1)1{t≤τ}∣Ft−1]]=E(M0)+∑t=1TE[E[(Mt−Mt−1)∣Ft−1]1{t≤τ}]=E(M0)\begin{align*} \mathbb E(M_\tau) &= \mathbb E(M_0) + \sum_{t = 1}^T \mathbb E[(M_t - M_{t - 1}) \mathbb 1_{\{t \leq \tau\}}] \\ &= \mathbb E(M_0) + \sum_{t = 1}^T \mathbb E[\mathbb E[(M_t - M_{t - 1}) \mathbb 1_{\{t \leq \tau\}} \vert \mathcal F_{t - 1}]] \\ &= \mathbb E(M_0) + \sum_{t = 1}^T \mathbb E[\mathbb E[(M_t - M_{t - 1}) \vert \mathcal F_{t - 1}] \mathbb 1_{\{t \leq \tau\}}] \\ &= \mathbb E(M_0) \end{align*}

Let us consider an example of an American Put option to understand this better -

Yt=(K−St)+Y_t = (K - S_t)^+

where T=1,u=2,d=0.5,r=0.2,S0=100,K=150T = 1, u = 2, d = 0.5, r = 0.2, S_0 = 100, K = 150. The value for p∗p^* is 7/157/15. Then, we have the following tree evolution for StS_t

1005010025100100400\begin{align*} &100 \\ 50 &\quad 100 \\ 25 \quad 100 &\quad 100 \quad 400 \end{align*}

The payment YtY_t looks like

501001012550500\begin{align*} &50 \\ 100^{1} &\quad 0 \\ 125 \quad 50 &\quad 50 \quad 0 \end{align*}

How do we calculate a fair price in this case? We introduce the notion of a super-martingale. In comparison, in European call options, we get for VtV_t

41.9875122.2ˉ12550500\begin{align*} &41.98 \\ 75^{1} &\quad 22.\bar 2 \\ 125 \quad 50 &\quad 50 \quad 0 \end{align*}

From the seller's perspective, we have UtU_t, and we use something called as a necessary wealth process - calculating the amount of money the seller requires at every node to pay the option holder. To do so, we need to consider the maximum of all possible payouts at every node. In the node labeled 11 above, if the buyer exits the option at t=1t = 1, then the seller needs a payout of 100100. However, if the buyer exits at time t=2t = 2, the seller requires atleast 7575 at t=1t = 1 (They can invest the 7575 in the same stock using short sell to get the required payouts). Considering these scenarios, the seller requires 100100 at the 11 node. Doing such a process for every node, we get UtU_t

53.0910022.2ˉ12550500\begin{align*} &53.09 \\ 100 &\quad 22.\bar 2 \\ 125 \quad 50 &\quad 50 \quad 0 \end{align*}

We can see that the price of this option is slightly higher than European since the buyer has the choice of exiting the contract at any intermediate time.

To formally describe the process we did above, we consider the following recursion

Ut−1=max⁡(11+rE∗(Ut∣Ft−1),Yt−1)t=T,T−1,…,1Ut−1∗=max⁡(E∗(Ut∗∣Ft−1),Yt−1∗)≥E∗(Ut∗∣Ft−1)\begin{align*} U_{t - 1} &= \max \left(\frac{1}{1 + r} \mathbb E^*(U_t \vert \mathcal F_{t - 1}), Y_{t - 1}\right) \quad t = T, T - 1, \dots, 1 \\ U^*_{t - 1} &= \max\left(\mathbb E^*(U^*_t \vert \mathcal F_{t - 1}), Y^*_{t - 1}\right) \\ &\geq \mathbb E^*(U^*_t \vert \mathcal F_{t - 1}) \end{align*}

The variables Ut∗=Ut(1+r)U^*_{t} = U_t (1 + r) are called as super-martingales.

Let us now delve deeper into the necessary wealth argument.

Necessary Wealth

We have Y={Y0,…,YT}Y = \{Y_0, \dots, Y_T\} representing the random variable for the payouts at each time tt. We formulated that

UT:=YTUt−1:=max⁡(11+rE∗(Ut∣Ft−1),Yt−1)\begin{align*} U_T := Y_T \\ U_{t - 1} := \max\left(\frac{1}{1 + r} \mathbb E^*(U_t \vert \mathcal F_{t - 1}), Y_{t - 1}\right) \end{align*}

for t=T,T−1,…,1t = T, T - 1, \dots, 1. For each Ut−1U_{t - 1},

Ut−1≥11+rE∗(UT∣Ft−1)Ut−1∗≥E∗(Ut∗∣Ft−1)\begin{align*} U_{t - 1} &\geq \frac{1}{1 + r} \mathbb E^*(U_T \vert \mathcal F_{t - 1}) \\ U^*_{t - 1} & \geq \mathbb E^*(U_t^* \vert \mathcal F_{t - 1}) \end{align*}

{UT∗:t=0,1,…,T}\{U_T^*: t = 0, 1, \dots, T\} form a p∗p^*-supermartingale. Also note that, Ut∗≥Yt∗U^*_t \geq Y^*_t for t=0,1,…,Tt = 0, 1, \dots, T. It is called as a Snell Envelope of {Yt}\{Y_t\}.

Consider the probability space (Ω,F,P)(\Omega, \mathcal F, P) with the event spaces (Ft)t=0T(\mathcal F_t)_{t = 0}^T. Then, we have the following claim -

Claim. There does not exist a supermartingale (Zt)(Z_t) such that

  1. Zt≥YtZ_t \geq Y_t, for all tt

  2. Zt=max⁡t≤τ≤TE(Yτ∣Ft)Z_t = \max_{t \leq \tau \leq T} \mathbb E(Y_\tau \vert \mathcal F_t)

  3. The maximum in the above equation is attained at τ∗(t)=min⁡(s≥t:Zs=Ys)∧T\tau^*(t) = \min(s \geq t: Z_s = Y_s) \wedge T

  4. The random variables ZT∧τ∗(0)Z_{T \wedge \tau^*(0)} for t=0,1,…,Tt = 0, 1, \dots, T is a martingale.

Proof. Define Zt:=max⁡t≤τ≤TE(YT∣Ft)Z_t := \max_{t \leq \tau \leq T } \mathbb E(Y_T \vert \mathcal F_t). We can compute ZtZ_t since it is a finite optimization problem. Since {Yt}\{Y_t\} form a martingale, we get

Zt≥E[YT∣Ft]=YtZ_t \geq \mathbb E[Y_T \vert \mathcal F_t] = Y_t

We use this result to show that ZtZ_t form a supermartingale. Consider the stopping time σ\sigma between tt and TT. Then, for attaining maximum

Zt=E[Yσ∣Ft]E[Zt∣Ft−1]=E[E[Yσ∣Ft]∣Ft−1]=E[Yσ∣Ft−1]≤Zt−1\begin{align*} Z_t &= \mathbb E[Y_\sigma \vert F_t ]\\ \mathbb E[Z_t \vert \mathcal F_{t - 1}] &= \mathbb E[\mathbb E[Y_\sigma \vert \mathcal F_t ] \vert \mathcal F_{t - 1}] \\ &= \mathbb E[Y_\sigma \vert \mathcal F_{t - 1}] \leq Z_{t - 1} \end{align*}

How do we now show that this is minimal? ZtZ_t being minimal implies that if ZtZ_t is a supermarginale majorant of {Yt}\{Y_t\}, then Xt≥ZtX_t \geq Z_t for all tt.

Xt(w)≥Yt(w)∀t,∀w  ⟹  Xτ(w)(w)≥Yτ(w)(w),∀w\begin{align*} X_t(w) \geq Y_t(w) \forall t, \forall w \\ \implies X_{\tau(w)} (w) \geq Y_{\tau(w)}(w) , \forall w\end{align*}

Then, equivalently, Xτ≥YτX_\tau \geq Y_\tau for t≤τ≤Tt \leq \tau \leq T.

Xt≥E(Xt∣Ft)≥E[YT∣Ft]≥E[YT∣Ft],∀t\begin{align*} X_t &\geq \mathbb E(X_t \vert \mathcal F_t) \geq \mathbb E[Y_T \vert \mathcal F_t] \\ &\geq \mathbb E[Y_T \vert \mathcal F_t], \forall t \end{align*}

Resulting in Xt≥ZtX_t \geq Z_t. So far, we have shown (1) and (2) in the claim. Also, in the above proof, we use the following property. Let XtX_t be a supermartingale ifor t≤τ≤Tt \leq \tau \leq T. Then E(Xτ∣Ft)≤Xt\mathbb E(X_\tau \vert \mathcal F_t) \leq X_t. The proof is left as an exercise.

How do we show (3) and (4) in the claim. For (4), we simply need to check the martingale property

E[Z(t+1)∧τˉ−Zt∧τˉ∣Ft]=?0=E[1t<τˉ(Z(t+1)∧τˉ−Zt∧τˉ∣Ft]=E[1t<τ(Zt+1=Zt)∣Ft]=1t<τˉ(E(Zt+1∣Ft)−Zt)\begin{align*} \mathbb E[Z_{(t + 1) \wedge \bar \tau} - Z_{t \wedge \bar \tau} \vert \mathcal F_t] &\stackrel{?}{=} 0 \\ &= \mathbb E[\mathbb 1_{t < \bar \tau}(Z_{(t + 1) \wedge \bar \tau} - Z_{t \wedge \bar \tau} \vert \mathcal F_t] \\ &= \mathbb E[\mathbb 1_{t < \tau} (Z_{t + 1} = Z_t) \vert \mathcal F_t] \\ &= \mathbb 1_{t < \bar \tau} \left(\mathbb E(Z_{t + 1} \vert F_t) - Z_t\right) \\ \end{align*}

where τˉ(w)=τ∗(0)(w)\bar \tau(w) = \tau^*(0) (w). Now, rewriting the definition of ZtZ_t, we get

Zt=max⁡(E(ZT+1∣Ft,Yt)Z_t = \max \left(\mathbb E(Z_{T + 1} \vert \mathcal F_t, Y_t\right)

meaning for t<τˉ  ⟹  Zt>Ytt < \bar \tau \implies Z_t > Y_t (from definition of τˉ\bar \tau in (3)) implying Zt=E(Zt+1∣Ft)Z_t = \mathbb E(Z_{t + 1} \vert \mathcal F_t). This proves the required property.

The last argument is to show that, given a supermartingale Z~\tilde Z with Z~t≥Yt\tilde Z_t \geq Y_t for all tt, then Z~t=Zt\tilde Z_t = Z_t for all tt. This proves uniqueness.

We have

Z~t−1=max⁡(Yt−1,E(Z~∣Ft−1))≤max⁡(Yt−1,E(Zt∣Ft−1))≤max⁡(Yt−1,Zt−1)=Zt−1\begin{align*} \tilde Z_{t - 1} &= \max(Y_{t - 1}, \mathbb E(\tilde Z \vert \mathcal F_{t - 1})) \\ &\leq \max(Y_{t - 1}, \mathbb E(Z_t \vert \mathcal F_{t -1})) \\ &\leq \max(Y_{t - 1}, Z_{t - 1}) = Z_{t - 1} \end{align*}

We have already shown that (Zt)t−0T(Z_t)_{t - 0}^T is the least supermartingale majorant of (Yt)t=0T(Y_t)_{t = 0}^T. This proves uniqueness.

Pricing an American Call

Can we find a ϕ∗\phi^* such that Vt(ϕ∗)≥UtV_t(\phi^*) \geq U_t. We had derived the following before

α~t=Utu−Utdu−d1St−1β~t=uUtd−dUtuu−d1Bt−1\begin{align*} \tilde \alpha_t = \frac{U^u_t - U^d_t}{u - d} \frac{1}{S_{t - 1}} \\ \tilde \beta_t = \frac{uU^d_t - dU^u_t}{u - d} \frac{1}{B_{t - 1}} \end{align*}

An important to thing to note here is that these values are random variables and are measurable under Ft−1\mathcal F_{t - 1}. However, this strategy will not be self-financed due to the super-martingale property. The value of the portfolio is

α~tSt+β~tBt≡Ut\tilde \alpha_t S_t + \tilde \beta_t B_t \equiv U_t

Then we have,

Ut−1=Yt−1∨(11+rE∗(Ut∣Ft−1))δ~t:=Ut−1−11+rE∗(Ut∣FT−1)≥0∵δ~t≥0⇔Yt−1>11+rE∗(UT∣Ft−1)\begin{align*} U_{t - 1} &= Y_{t - 1} \vee \left(\frac{1}{1 + r} \mathbb E^*(U_t \vert \mathcal F_{t - 1})\right)\\ \tilde \delta_t &:= U_{t - 1} - \frac{1}{1 + r} \mathbb E^*(U_t \vert \mathcal F_{T - 1}) \geq 0 \\ &\because \tilde \delta_t \geq 0 \Leftrightarrow Y_{t - 1} > \frac{1}{1 + r} \mathbb E^*(U_T \vert \mathcal F_{t - 1}) \end{align*}

Now, consider the following strategy

αt∗=α~tβt∗=β~t+∑s=1tδ~s/Bs−1≥β~t\begin{align*} \alpha^*_t &= \tilde \alpha_t \\ \beta_t^* &= \tilde \beta_t + \sum_{s = 1}^t \tilde \delta_s/B_{s - 1} \geq \tilde \beta_t \end{align*}

This trading strategy ϕ∗={(α∗,β∗):t=1,…,T}\phi^* = \{(\alpha^*, \beta^*): t = 1, \dots, T\} is a super-hedging strategy with Vt(ϕ∗)≥UtV_t(\phi^*) \geq U_t. We introduced δ~t\tilde \delta_t to make it a self-financing strategy.

Define τˉ=τˉ(0)=min⁡(T:Ut=YT)∧T\bar \tau = \bar \tau(0) = \min(T: U_t = Y_T) \wedge T. Then, Ut−1>Yt−1U_{t - 1} > Y_{t - 1} and δ~t=0\tilde \delta_t = 0 for t=1,…,τˉ(w)t = 1, \dots, \bar \tau(w). Consequently, Vt∧τ(ϕ∗)=Ut∧τV_{t \wedge \tau}(\phi^*) = U_{t \wedge \tau}, and Ut∧τ∗=Ut∧τ(1+r)t∧τU^*_{t \wedge \tau} = \frac{U_{t \wedge \tau}}{(1 + r)^{t \wedge \tau}} forming a martingale.

Arbitrage

For a trading strategy ψ\psi, we have V0(ψ)=0V_0(\psi) = 0 and VT(ψ)≥0V_T(\psi) \geq 0. Alternately, P(VT(ψ)>0)>0P(V_T(\psi) > 0) > 0. In an American option, there is an asymmetry between the buyers and the sellers. The buyer has to choose only one time t≤Tt \leq T to exit the option whereas the seller has to safeguard against an arbitrage in all possibilities.

The seller's arbitrage for a strategy ϕs\phi^s (in stocks and bonds) is then defined as

V0(ϕs)=C0∀τVτ(ϕs)≥Yτ>sometimes\begin{align*} V_0(\phi^s) &= C_0 \\ \forall \tau V_\tau(\phi^s) &\geq Y_\tau \\ &> \quad \quad \text{sometimes} \end{align*}

The seller sells an option for C0C_0 and makes an investment of C0C_0 in ϕs\phi_s. Then an arbitrage for the seller is

VT=(Vτ(ϕs)−Yτ)(1+r)T−τV_T = (V_\tau(\phi^s) - Y_\tau)(1 + r)^{T - \tau}

where τ\tau is the time when the buyer exits the option. This value is always greater than or equal to 00 with a non-zero probability that it is strictly greater than 00.

On the other hand, the buyer's arbitrage for a strategy ϕb\phi_b is defined as

V0(ϕb)=−C0∃τVτ(ϕb)+Yτ≥0>sometimes\begin{align*} V_0(\phi^b) &= -C_0 \\ \exists \tau V_\tau(\phi^b) + Y_\tau &\geq 0 \\ &> \quad \quad \text{sometimes} \end{align*}

The arbitrage at time TT is (Vτ(ϕb)+Yτ)(1+r)T−τ(V_\tau(\phi^b) + Y_\tau)(1 + r)^{T - \tau}.

What would be the good time to exercise the option? In the following sections, we'll see that τ=τˉ\tau = \bar \tau is the optimal time to exit the option.

Example

Consider a put option YT=(K−St)+Y_T = (K - S_t)^+ where K=150,S0=100,u=2,d=0.5,r=0.2,T=2K = 150, S_0 = 100, u = 2, d = 0.5, r = 0.2, T = 2. We get U0=53.09U_0 = 53.09 and Y0=50Y_0 = 50. Then, the super-hedging strategy is given by

\begin{align*} \end{align*}

To summarise, for a payout random variable Y={Y0,…,YT}Y = \{Y_0, \dots, Y_T\}, we have UtU_t as the necessary wealth at each time tt. Then, the super-hedging strategy ϕ∗\phi^* has the following properties -

  • Vt∗(ϕ∗)V^*_t(\phi^*) is a martingale

  • V∗(ϕ∗)≥Ut∗≥Yt∗V^*(\phi^*) \geq U^*_t \geq Y^*_t for all tt

We define τˉ\bar \tau such that τˉ=min⁡{t∣Ut=Yt}∧T\bar \tau = \min\{t \vert U_t = Y_t\} \wedge T. With this definition, we have

Vt∗(ϕ∗)=Ut∗V_t^*(\phi^*) = U_t^*

for t=0,1,…,τˉt = 0, 1, \dots, \bar \tau and V0(ϕ∗)=U0V_0(\phi^*) = U_0.

Theorem. U0U_0 is the no-arbitrage price for American call option based on YY.

Proof.

  1. C0>U0C_0 > U_0, then there's arbitrage for the seller. The seller sells the claim for C0C_0 and buys ϕ∗\phi^* for U0U_0. The remaining amount, C0−U0C_0 - U_0 is invested in bonds. Then, for this strategy ϕs\phi^s, we have

    Vt(ϕs)=Vt(ϕ∗)+(C0−U0)BtVτ(ϕs)=Vτ(ϕ∗)+(C0−U0)Bτ⏟>0≥Uτ+…≥Yτ+…\begin{align*} V_t(\phi^s) &= V_t(\phi^*)+ (C_0 - U_0)B_t \\ V_\tau(\phi^s) &= V_\tau(\phi^*) + \underbrace{(C_0 - U_0)B_\tau}_{>0} \\ &\geq U_\tau + \dots \\ &\geq Y_\tau + \dots \end{align*}
  2. C0<U0C_0 < U_0, we have a buyer's arbitrage for the strategy ϕb\phi^b

    ϕb={−U0in −ϕ∗U0−C0in bond\phi^b = \begin{cases} -U_0 & \text{in } -\phi^* \\ U_0 - C_0 & \text{in bond} \end{cases}

    Then,

    V0(ϕb)=−C0Vτ(ϕb)=−Vτ(ϕ∗)+(U0−C0)Bτ\begin{align*} V_0(\phi^b) &= -C_0 \\ V_\tau(\phi^b) &= - V_\tau(\phi^*) + (U_0 - C_0)B_\tau \end{align*}

    Using τ=τˉ\tau = \bar \tau,

    Vτˉ(ϕb)+Yτˉ=−Vτˉ(ϕ∗)+(U0−C0)Bτˉ+Yτˉ=(U0−C0)Bτˉ>0\begin{align*} V_{\bar \tau}(\phi^b) + Y_{\bar \tau} &= -V_{\bar \tau}(\phi^*) + (U_0 - C_0)B_{\bar \tau} + Y_{\bar \tau} \\ &= (U_0 - C_0) B_{\bar \tau} > 0 \end{align*}
  3. C0=U0C_0 = U_0. We need to show separately that there's no seller's and buyer's arbitrage

    1. No seller's arbitrage - Suppose ϕs\phi^s is a seller's arbitrage strategy. We have

      V0(ϕs)=C0,Vτ(ϕs)≥YτV_0(\phi^s) = C_0, V_\tau(\phi^s) \geq Y_\tau

      Note that all elements need not be strictly greater for an arbitrage. Computing the expectation,

      E∗(Yτ∗)<E∗(Vτ∗(ϕs))=V0(ϕs)=U0=C0\mathbb E^*(Y^*_\tau) < \mathbb E^*(V^*_\tau(\phi^s)) = V_0(\phi^s) = U_0 = C_0

      for all stopping times τ\tau. However,

      why strictly less than in the above equation?

      E∗(Yτ∗)=E∗(Vτ∗(ϕs))=V0(ϕs)=C0\mathbb E^*(Y^*_\tau) = \mathbb E^*(V_\tau^*(\phi^s)) = V_0(\phi^s) = C_0

      This shows that no seller's arbitrage is possible.

    2. No buyer's arbitrage - Suppose ϕb\phi^b is such that

      V0(ϕb)=−C0∃τ0Vτ0(ϕb)+Yτ0≥0(>0 sometimes)\begin{align*} V_0(\phi^b) &= -C_0 \\ \exists \tau_0 V_{\tau_0} (\phi^b) + Y_{\tau_0} &\geq 0 \quad (> 0 \text{ sometimes}) \end{align*}

      Then, computing the expectation

      E∗(Vτ∗(ϕb))=V0(ϕb)=−C0E∗(Vτ∗(ϕb)+Yτ∗)=−C0+E∗(Yτ∗)≤−C0+E∗(Uτ∗)≤−C0+U0=0\begin{align*} \mathbb E^*(V^*_\tau(\phi^b)) &= V_0(\phi^b) = -C_0 \\ \mathbb E^*(V_\tau^*(\phi^b) + Y_\tau^*) &= -C_0 + \mathbb E^*(Y^*_\tau) \\ &\leq -C_0 + \mathbb E^*(U^*_\tau) \\ & \leq -C_0 + U_0 = 0 \end{align*}

      This shows that no buyer's arbitrage is possible. □\square

Example

Consider anMerican Call S0=4,r=0.25,u=2,d=0.5,K=4,T=3S_0 = 4, r = 0.25, u = 2, d = 0.5, K = 4, T = 3. This yields p∗=0.5p^* = 0.5. We have the following

St  ⟹  42814160.52832S_t \implies \begin{align*} &4 \\ 2 &\quad 8 \\ 1 \quad &4 \quad 16 \\ 0.5 \quad 2 &\quad 8 \quad 32 \end{align*} Yt  ⟹  004001200028Y_t \implies \begin{align*} &0 \\ 0 &\quad 4 \\ 0 \quad &0 \quad 12 \\ 0 \quad 0 &\quad 0 \quad 28 \end{align*} Ut=Vt  ⟹  64516251442508564500428U_t = V_t \implies \begin{align*} &\frac{64}{5} \\ \frac{16}{25} &\quad \frac{144}{25} \\ 0 \quad &\frac{8}{5} \quad \frac{64}{5} \\ 0 \quad 0 &\quad 4 \quad 28 \end{align*}

American Call-Put Parity

The relation between call and put options in general can be shown using Jensen's inequality. We have

E(ϕ(X))≥ϕ(E(X))E(ϕ(X∣G))≥ϕ(E(X∣G))\begin{align*} \mathbb E(\phi(X)) &\geq \phi(\mathbb E(X)) \\ \mathbb E(\phi(X \vert \mathcal G)) &\geq \phi(\mathbb E(X | \mathcal G)) \end{align*}

The payout functions for both call and put options are convex. We aim to show that Ut=VtU_t = V_t for t=T,T−1,…,1,0t = T, T - 1, \dots, 1, 0, using induction. The statement is true for t=Tt = T.

ASsume Us=VsU_s = V_s for s=t,t+1,…,Ts = t, t + 1, \dots, T, then

Ut−1=max⁡(Yt−1,11+rE∗(Ut∣FT−1))=max⁡(Yt−1,11+rE∗(Vt∣FT−1))\begin{align*} U_{t - 1} &= \max \left(Y_{t - 1} , \frac{1}{1 + r} \mathbb E^*(U_t \vert \mathcal F_{T - 1})\right)\\ &= \max \left(Y_{t - 1} , \frac{1}{1 + r} \mathbb E^*(V_t \vert \mathcal F_{T - 1})\right)\\ \end{align*}

Substituting Vt=1(1+r)T−τE∗(Vt∣Ft)V_t = \frac{1}{(1 + r)^{T - \tau}} \mathbb E^*(V_t \vert \mathcal F_t),

E∗(Vt∣Ft−1)=1(1+r)T−tE∗(VT∣Ft−1)=1(1+r)T−tE∗((ST−K)+∣FT−1)≥1(1+r)T−t(E∗(ST∣Ft−1)−K)+∵Jensen’s inequality=1(1+r)T−t((1+r)T−t+1St−1−K)+=(1+r)(St−1−K(1+r)T−t+1)+≥(1+r)(St−1−K)+=(1+r)Yt−1\begin{align*} \mathbb E^*(V_t \vert \mathcal F_{t - 1}) &= \frac{1}{(1 + r)^{T - t}} \mathbb E^*(V_T \vert \mathcal F_{t - 1}) \\ &= \frac{1}{(1 + r)^{T - t}} \mathbb E^*((S_T - K)^+ \vert \mathcal F_{T - 1}) \\ &\geq \frac{1}{(1 + r)^{T - t}} \left(\mathbb E^*(S_T \vert \mathcal F_{t - 1}) - K\right)^+ & \because \text{Jensen's inequality}\\ &= \frac{1}{(1 + r)^{T - t}} ((1 + r)^{T - t + 1}S_{t - 1} - K)^+ \\ &= (1 + r)\left( S_{t - 1} - \frac{K}{(1 + r)^{T - t + 1}}\right)^+ \\ &\geq (1 + r) (S_{t - 1} - K)^+ \\ &= (1 + r) Y_{t - 1} \end{align*}

Using this result above, we have

Ut−1=max⁡(Yt−1,11+rE∗(Vt∣FT−1))=11+rE∗(Vt∣FT−1)=Vt−1\begin{align*} U_{t - 1} &= \max \left(Y_{t - 1} , \frac{1}{1 + r} \mathbb E^*(V_t \vert \mathcal F_{T - 1})\right)\\ &= \frac{1}{1 + r} \mathbb E^*(V_t \vert \mathcal F_{T - 1}) = V_{t - 1} \end{align*}

Finite Market Models

The binomial model we have discussed previously is an example of a finite market model. These structures can be generalised to markets with multiple stocks, bonds and contingent claims. The price evolution need not be binary We shall derive results for pricing similarly like before for this general market. To do the analysis, the notation for probability space, portfolio, value, trading strategies and arbitrage has to be defined -

Probability Space

Again, we consider a finite probability space (Ω,F,P)(\Omega, \mathcal F, P), that is #(Ω)<∞)\#(\Omega)< \infty) with F=P(Ω)\mathcal F = \mathcal P(\Omega) (the power set) and P(A)=∑w∈AP(w),A∈FP(A) = \sum_{w \in A} P(w), A \in \mathcal F.

We shall assume that every event has a non-zero probability. If there are events with zero-probability, they are not relevant to our discussion and we can simply prune these points.

A filtration F0⊂F1⊂⋯⊂FT\mathcal F_0 \subset \mathcal F_1 \subset \cdots \subset \mathcal F_T represents the history of the market or the evolution of prices of assets in the market. Typically, FT=F\mathcal F_T = \mathcal F.

Assets in portfolio

Now in this setup, we consider the assets - asset 0 bond and assets 1,2,…,d1, 2, \dots, d stocks. At any time tt, the prices are given by

St=(St0,St1,…,STd),t=0,1,…,TS_t = (S^0_t, S^1_t, \dots, S^d_T), t = 0, 1, \dots, T

Where St0S^0_t is the bond-price and not random and StUS_t^U for i>0i > 0 represent the stock prices.

Each sequence {Sti}t=0T\{S_t^i\}_{t = 0}^T is Ftmeasurable\mathcal F_t measurable, that is (St)(S_t) is adapted to the filtration.

Trading strategy

After defining the prices and the structure of the portfolio, we define a trading strategy for this market -

ϕ={ϕt:t=1,…,T}\phi = \{\phi_t: t = 1, \dots, T\}

where ϕt=(ϕt0,ϕt1,…ϕtd)\phi_t = (\phi_t^0, \phi_t^1, \dots \phi_t^d) is a vector of holdings representing the assets held at time tt. That is, ϕtI\phi_t^I is the number of shares of asset ii to be held on day tt.

The basic assumptions with our trading strategies are

  • The trading strategy is predictable - ϕti∈Ft−1\phi_t^i \in \mathcal F_{t - 1} for all ii and t=1,2,…,Tt = 1, 2, \dots, T.
  • Self-financing - ϕt⋅St=ϕt+1⋅St\phi_t \cdot S_t = \phi_{t + 1} \cdot S_t (note that these are vector dot products ϕt⋅St=∑i=1dϕtiSti\phi_t \cdot S_t = \sum_{i = 1}^d \phi_t^i S_t^i).

Value of the portfolio

The value of the portfolio at time t=0t = 0 is given by V0(ϕ)=ϕ1⋅S0V_0(\phi) = \phi_1 \cdot S_0 and for time t>0t > 0, we have Vt(ϕ)=ϕt⋅StV_t(\phi) = \phi_t \cdot S_t. For this market, the change in the value of portfolio across a day is given by ΔVt(ϕ)\Delta V_t(\phi)

ΔVt(ϕ):=Vt(ϕ)−Vt−1(ϕ)=ϕt⋅St−ϕt−1⋅St−1=ϕt⋅(St−St−1)=ϕt⋅ΔSt\begin{align*} \Delta V_t(\phi) &:= V_t(\phi) - V_{t - 1}(\phi) \\ &= \phi_t \cdot S_t - \phi_{t - 1}\cdot S_{t - 1} \\ &= \phi_t \cdot (S_t - S_{t - 1}) \\ &= \phi_t \cdot \Delta S_t \end{align*}

So the value at any time tt can now be defined as

Vt(ϕ)=V0(ϕ)+∑u=1t∑i=0dϕuiΔSui∵Telescoping sum=V0(ϕ)+∑u=1tϕu⋅ΔSu\begin{align*} V_t(\phi) &= V_0(\phi) + \sum_{u = 1}^t \sum_{i = 0}^d \phi_u^i \Delta S_u^i \quad \quad\because \quad \text{Telescoping sum} \\ &=V_0(\phi) + \sum_{u = 1}^t \phi_u \cdot \Delta S_u \end{align*}

for t=0,1,…,Tt = 0, 1, \dots, T

To help us derive results in the later sections, we define the following notation for discounted stock prices -

St∗=St/St0=(1,St1∗,…,Std∗)\begin{align*} S_t^* &= S_t /S^0_t \\ &= (1, S_t^{1*}, \dots, S_t^{d*}) \end{align*}

It essentially divides the stock prices with the bond price at time tt. Similarly we can define discounted value of a portfolio.

We can derive that the telescopic sum property still holds with the discounted values for the portfolio. That is,

VT∗(ϕ)=V0∗(ϕ)+∑u=1tϕu⋅ΔSu∗V_T^*(\phi) = V_0^*(\phi) + \sum_{u = 1}^t \phi_u \cdot \Delta S_u^*

where ΔSu∗=Su∗−Su−1∗\Delta S_u^* = S_u^* - S_{u - 1}^*. This property will be useful for extracting Martingales from the price evolution tree. Specifically, we have the following lemma

Lemma 1 If (Mt:t=0,1,…,T)(M_t: t = 0, 1, \dots, T) is a martingale and (H1,H2,…,HT)(H_1,H_2, \dots, H_T) is a predictable sequence of random variables then

Xt:=∑u=1THuΔMuX_t := \sum_{u = 1}^T H_u \Delta M_u

is a martingale!

Arbitrage

The expressions are very similar to before wherein an arbitrage opportunity (risk-free profit) is a trading strategy ϕ\phi such that

V0(ϕ)=0VT(ϕ)≥0P(VT(ϕ)>0)>0\begin{align*} V_0(\phi) &= 0 \\ V_T(\phi) &\geq 0 \\ P(V_T(\phi) > 0) &> 0 \end{align*}

The last condition essentially ensures that there is atleast one case where the profit is non-zero. Previously, we had used expectation to ensure this using the p∗p^* model derived using the binomial model. However, we want a general scenario here and an expectation requires an associated probability distribution.

The market is set to be viable if there are no arbitrage opportunities. We can show that the binomial model considered previously is viable if d<1+r<ud < 1 + r < u. So, when is a market non-viable? Consider Ω={w1,w2},T=1,d=2\Omega = \{w_1, w_2\}, T = 1, d = 2 with the stock price evolution S00=S10=1S_0^0 = S_1^0 = 1, S01=S02=1S_0^1 = S_0^2 = 1, and

( w_1 ) ( w_2 )
( S_1 ) 1 2
( S_2 ) 1 3

The trading strategy ϕ={(0,−1,1)}\phi = \{(0, -1, 1)\} has V1(ϕ)≥0V_1(\phi) \geq 0 in all entries and >0>0 in at least one entry.

Notice the subtleties in this example. Previously, in the CRR model, we focused on finding an arbitrage-free pricing. We ensured there is no arbitrage in the market with this design. However, considering a trading strategy with only stocks in a bullish market, there is always an opportunity for arbitrage.

For trading strategies with only stocks, the problem of finding a trading strategy boils down to solving some linear inequalities.

Martingale Measure

As noted earlier, we will assume that P(w)>0P(w) > 0 or each w∈Ωw \in \Omega. For a new measure (Ω,F,Q)(\Omega, \mathcal F, Q), we say that QQ is equivalent to PP (Q∼PQ \sim P) if for all A∈FA \in \mathcal F

Q(A)>0  ⟺  P(A)>0Q(A) > 0 \iff P(A) > 0

With this infrastructure, we are aiming to find a new probability QQ (like p∗p^* in CRR model) to prove properties in a market.

Definition. A probability QQ on (Ω,F)(\Omega, \mathcal F) is a martingale measure provided {STi∗:t=0,1,…,T}\{S_T^{i*}: t = 0, 1, \dots, T\} is a Q−martingaleQ-martingale for each stock I=1,…,dI = 1, \dots, d.

An equivalent martingale measure is a probability QQ such that Q∼PQ \sim P and QQ is a martingale measure. With these tools, we now define the first fundamental theorem of asset pricing.

First Fundamental Theorem of Asset Pricing

A market is viable if and only if there exists at least one Equivalent Martingale Measure (EMM).

To understand this better, consider the previous example of a market that is not viable. Suppose we have a martingale measure QQ, then

EQ(Si1)=S0i\mathbb E^Q(S_i^1) = S_0^i

for I=1,2I = 1, 2 using the martingale property. Letting a=Q(w1),b=Q(w2)a = Q(w_1), b = Q(w_2), we get the linear equations

a+2b=1a+3b=1}b=0\left. \begin{align*} a + 2b &= 1\\ a + 3b &= 1 \end{align*} \right\} b = 0

Since P(w2)>0⇏Q(w2)>0P(w_2) > 0 \not\Rightarrow Q(w_2) > 0, there cannot exist an Equivalent Martingale Measure in this market.

Proof

(⟸\Longleftarrow) Let QQ be an EMM and ϕ\phi is a strategy making an arbitrage. That is,

V0(ϕ)=0,VT(ϕ)≥0 V_0(\phi) = 0, V_T(\phi) \geq 0

Since {St}t=1T\{S_t\}_{t = 1}^T is a martingale under QQ, the sequence ∑i=1tϕuI⋅ΔSui∗\sum_{i = 1}^t \phi_u^I \cdot \Delta S_u^{i*} for t=0,1,…,Tt = 0, 1, \dots, T is a QQ-martingale.

VT∗(ϕ):=VT(ϕ)/ST0=V0∗+∑u=1Tϕu⋅ΔSu∗=0+∑u=1T∑i=1dϕuI⋅ΔSui∗\begin{align*} V^*_T(\phi) &:= V_T(\phi)/S_T^0 \\ &= V_0^* + \sum_{u = 1}^T \phi_u \cdot \Delta S_u^* \\ &= 0 + \sum_{u = 1}^T \sum_{i = 1}^d \phi_u^I \cdot \Delta S_u^{i*} \end{align*}

Vt∗V_t^* is also a martingale under QQ, implying EQ(VT∗(ϕ))=EQ(V0∗(ϕ))=0\mathbb E^Q(V_T^*(\phi)) = \mathbb E^Q(V_0^*(\phi)) = 0. Since P∼QP \sim Q, P(VT(ϕ)>0)=0P(V_T(\phi) > 0) = 0.

(  ⟹  )(\implies) Now, we show the opposite direction, that is for a viable market, there exists an EMM.

We have #(Ω)<∞\#(\Omega) < \infty, where Ω={w1,w2,…,wn}\Omega = \{w_1, w_2, \dots, w_n\}. Consider a random variable X:Ω→RX: \Omega \to \mathbb R, then we have (X(w1),X(w2),…,X(wn)∈Rn)(X(w_1), X(w_2), \dots, X(w_n) \in \mathbb R^n) and a distribution QQ with probabilities (Q(w1),…,Q(wn))(Q(w_1), \dots, Q(w_n)). Then, we have

EQ(VT∗(ϕ))=∑k=1nQ(wk)VT∗(ϕ)(wk)=Q⋅VT∗(ϕ)\mathbb E^Q(V^*_T(\phi)) = \sum_{k = 1}^n Q(w_k) V^*_T(\phi) (w_k) = Q \cdot V^*_T(\phi)

It suffices to show that there exists a QQ such that Q⊥{VT∗(ϕ):ϕ is a trading strategy with V0(ϕ)=0}⊂Rn}Q \perp \{V_T^*(\phi): \phi \text{ is a trading strategy with } V_0(\phi) = 0\} \subset \mathbb R^n\}.

We have seen before that as a result of a telscopric sum formulation, VT∗(ϕ)=V0∗(ϕ)+∑u=1Tϕu⋅ΔSu∗V^*_T(\phi) = V_0^*(\phi) + \sum_{u = 1}^T \phi_u \cdot \Delta S_u^*.

Lemma 2

Suppose a M=(M1,…,Mn)M = (M_1, \dots, M_n) is an adapted sequence such that

E(∑u=1Tqu⋅ΔMu)=0;ΔMk=Mk−Mk−1\mathbb E\left(\sum_{u = 1}^T q_u \cdot \Delta M_u\right) = 0; \quad \Delta M_k = M_k - M_{k - 1}

for all predictable (q1,…,qT)(q_1, \dots, q_T), then MM is a martingale.

Proof. To show, E(Mt∣Ft−1)=Mt−1\mathbb E(M_t \vert \mathcal F_{t - 1}) = M_{t - 1} for t=1,2,…,Tt = 1, 2, \dots, T. Equivalently, E(Mt⋅1A)=E(Mt−1⋅1A)\mathbb E(M_t \cdot \mathbb 1_A) = \mathbb E(M_{t - 1} \cdot \mathbb 1_A) for all A∈Ft−1A \in \mathcal F_{t - 1}.

If X,YX, Y are both G\mathcal G-measurable and E(X⋅1A)=E(Y⋅1A)\mathbb E(X \cdot \mathbb 1_A) = \mathbb E(Y \cdot \mathbb 1_A) for all A∈GA \in \mathcal G, then P(X=Y)=1P(X = Y) = 1 (Consider {X<Y}=A\{X < Y\} = A).

In our context X=E(Mt∣Ft−1)X = \mathbb E(M_t \vert \mathcal F_{t - 1}) and Y=Mt−1Y = M_{t - 1}, leading to (ΔMt⋅1A)=0\mathbb (\Delta M_t \cdot \mathbb 1_A) = 0. Fixing tt and A∈Ft−1A \in \mathcal F_{t - 1}, let

qu={1Au=t0u≠tq_u = \begin{cases} \mathbb 1_A & u = t \\ 0 & u \neq t \end{cases}

Consequently,

∑u=1Tqu⋅ΔMu=1A⋅(Mt−Mt−1)\begin{align*} \sum_{u = 1}^T q_u \cdot \Delta M_u = \mathbb 1_A \cdot (M_t - M_{t - 1}) \end{align*}

Lemma 3 If we have, qti∈Ft−1q_t^i \in \mathcal F_{t - 1} for i=1,…,di = 1, \dots, d amd t=1,…,Tt = 1, \dots, T, then for any c∈Rc \in \mathbb R, there exists ϕt0\phi^0_t for t=1,…,Tt = 1, \dots, T such that

ϕ={(ϕt0,ϕt1,…,ϕtd):t=1,…,T}\phi = \{(\phi_t^0, \phi_t^1, \dots, \phi_t^d): t = 1, \dots, T\}

is a self-financing trading strategy and V0(ϕ)=cV_0(\phi) = c.

Proposition. Suppose Q∼PQ \sim P, such that EQ(VT∗(ϕ))=0\mathbb E^Q(V_T^*(\phi)) = 0 for all trading strategies ϕ\phi with V0(ϕ)=0V_0(\phi) = 0, then QQ is an EMM. That is, we need to show St1∗S^{1*}_t is a Q-martingale. From lemma 2, it is equivalent to showing EQ(∑u=1Tqu⋅ΔSu1∗)=0\mathbb E^Q(\sum_{u = 1}^T q_u \cdot \Delta S_u^{1*}) = 0 for all (q1,…,qT)(q_1, \dots, q_T). Consider the following strategy

ϕti={qti=10i=2,3,…,dϕt0 from Lemma 3\phi^i_t = \begin{cases} q_t & i = 1 \\ 0 & i = 2, 3, \dots, d \\ \phi_t^0 \text{ from Lemma 3} \end{cases}

Then,

∑u=1TquΔMu=1A(Mt−Mt−1)VT∗(ϕ)=∑i=0d∑u=1T(ϕui⋅Sui∗)=∑u=1Tqu⋅ΔSu1∗\begin{align*} \sum_{u = 1}^T q_u \Delta M_u &= \mathbb 1_A (M_t - M_{t - 1}) \\ V^*_T(\phi) &= \sum_{i = 0}^d \sum_{u = 1}^T (\phi_u^i \cdot S_u^{i*}) \\ &= \sum_{u = 1}^T q_u \cdot \Delta S_u^{1*} \end{align*}

Now, for all trading strategies with the self-financing property, the set L:=VT∗(ϕ):ϕ with V0(ϕ)=0⊂RnL:= {V_T^*(\phi): \phi \text{ with } V_0(\phi) = 0} \subset \mathbb R^n is a linear subspace of Rn\mathbb R^n. Similarly, consider the set D:={Y∈Rn:Y≥0 with Y≠0}D := \{Y \in \mathbb R^n : Y \geq 0 \text{ with } Y \neq 0\}.

Our goal boils down to show that a viable market exists if and only if L∩D=ϕL \cap D = \phi. It is equivalent to show L∩F=ϕL \cap F = \phi for F:={Y∈D:∑i=1nYk=1}F:= \{Y \in D: \sum_{i = 1}^n Y_k = 1\}.

Assume viability -

We have noted that LL is a subspace of Rn\mathbb R^n. Also, FF is compact and convex.

Lemma. Separating Hyperplane Theorem For a vector subspace L⊆RnL \subseteq \mathbb R^n and a compact convex set FF, if L∩F=ϕL \cap F = \phi, then there exists a ζ∈Rn\zeta \in \mathbb R^n, ζ≠0\zeta \neq 0 such that

L⊂{t:t⋅ζ=0}F⊂{y:y⋅z>0}\begin{align*} L &\subset \{t: t \cdot \zeta = 0\} \\ F &\subset \{y: y \cdot z > 0\} \end{align*}

Proof. Let G:=F−L={f−l:f∈F,l∈L}G:= F - L = \{f - l: f \in F, l \in L\}, then GG is convex and closed and 0∉G0 \not \in G. We choose the vector ζ∈G\zeta \in G that minimizes ∥x∥\|x\|, x∈Gx \in G. Now,

β:=inf⁡{∥x∥:x∈G}\beta := \inf \{\|x \|: x \in G \}

then for any sequence xn∈Gx_n \in G such that ∥xn∥\|x_n\| converges to β\beta (∥xn∥↘β\|x_n\| \searrow \beta) , using the parallelogram law

∥xn−xm∥2=2(∥xn∥2+∥xm∥2)−4∥xn+xm2∥2\begin{align*} \|x_n - x_m\|^2 &= 2(\|x_n\|^2 + \|x_m\|^2) - 4\left\|\frac{x_n + x_m}{2}\right\|^2 \\ \end{align*}

The terms ∥xn∥2,∥xm∥2\|x_n\|^2, \|x_m\|^2 converge to β2\beta^2 and the average ∣xn+xm2∣2\left|\frac{x_n + x_m}{2}\right|^2 goes to β2\beta^2 leading to ∥xn−xm∥2→0\|x_n - x_m\|^2 \to 0 for m,n→∞m, n \to \infty and xn→ζx_n \to \zeta. Therefore, {xn}\{x_n\} form a Cauchy sequence convergent to some ζ\zeta. Since GG is closed, this shows the existence of such ζ∈G\zeta \in G.

Now, for any x∈Gx \in G , x⋅ζ≥∥ζ∥2x \cdot \zeta \geq \|\zeta\|^2. How do we show this? Consider any arbitrary vector formed by x,ζx, \zeta, ∥αx+(1−α)ζ∥2−∥ζ∥2≥0\|\alpha x + (1 - \alpha)\zeta\|^2 - \|\zeta\|^2 \geq 0 (a quadratic in α\alpha, and 0≤α≤10 \leq \alpha \leq 1). For α→0\alpha \to 0, we get

−2∥ζ2∥+2(x⋅ζ)≥0-2 \|\zeta^2\| + 2(x\cdot \zeta) \geq 0

Now, in particular, since 0∈L0 \in L, if f∈Gf \in G, then f∈Ff \in F. Consequently, if f∈Ff \in F, then for any λ∈R,l∈L\lambda \in \mathbb R, l \in L, f−λl∈Gf - \lambda l \in G. We get

f⋅ζ≥λ(l⋅ζ)+∥ζ∥2f\cdot \zeta \geq \lambda (l \cdot \zeta) + \|\zeta\|^2

With the limits of λ\lambda, we can conclude that l⋅ζ=0l \cdot \zeta = 0. □\square

How does this theorem help us show L∩F=ϕL \cap F = \phi for a viable market? Consider the EMM Q(wk)=ζkc>0Q(w_k) = \frac{\zeta_k}{c} > 0. c=∑i=1nζic = \sum_{i = 1}^n \zeta_i. For a fixed kk, consider

Yi={1i=k0i≠kY_i = \begin{cases} 1 & i = k \\ 0 & i \neq k \end{cases}

Then for Y∈FY \in F, Y⋅ζ>0Y \cdot \zeta > 0. Let ϕ\phi be a trading strategy with V0(ϕ)=0V_0(\phi) = 0, then

EQ(Vt∗(ϕ))=Q⋅VT∗(ϕ)=1c[ζ⋅Vt∗(ϕ)⏟∈L]=0\begin{align*} \mathbb E^Q(V_t^*(\phi)) &= Q\cdot V^*_T(\phi) \\ &= \frac{1}{c} \left[\zeta \cdot \underbrace{V_t^*(\phi)}_{\in L}\right] = 0 \end{align*}

Completeness of a Market

A merket is complete if for every contingent claim X∈FTX \in \mathcal F_T, there exists a trading strategy ϕ\phi that can be replicate the claim with VT(ϕ)≡XV_T(\phi) \equiv X.

Second Fundamental Theorem of Asset Pricing

Assuming the market is viable, then the market is complete if and only if there is a unique EMM for the market.

Proof.

  1. (  ⟹  )(\implies) Assume that the market is complete. Let P1∗,P2∗P_1^*, P_2^* be the distinct EMMs. Then for a A∈FTA \in \mathcal F_T, define X:=1AX:= \mathbb 1_A. Due to completeness, ∃ϕ\exists \phi with VT(ϕ)=1AV_T(\phi) = \mathbb 1_A.

    Pi∗(A)=Ei∗(1A)=E1∗(VT(ϕ))=ST0S00⋅V0(ϕ)\begin{align*} P_i^*(A) &= \mathbb E_i^*(\mathbb 1_A) \\ &= \mathbb E_1^*(V_T(\phi)) = \frac{S^0_T}{S^0_0} \cdot V_0(\phi) \end{align*}

    Therefore, P1∗(A)=P2∗(A)P_1^*(A) = P_2^*(A) for all A∈FTA \in \mathcal F_T implying that P1∗=P2∗P_1^* = P_2^*.

  2. (⟸\Longleftarrow ) Assuming that there is a unique EMM, we need to show that the market is complete. We do this by proving the contrapositive - a market is incomplete implies that there is no unique EMM for the market. We define

    L:={VT∗(ϕ):ϕ self-financing trading strategies }L:= \{V_T^*(\phi): \phi \text{ self-financing trading strategies }\}

    Now, we show that the market is incomplete if and only if L⊊RnL \subsetneq \mathbb R^n and ∃z≠0\exists z \neq 0 such that z⊥Lz \perp L. Consider the EMM QQ for a viable market. Then, define Q~(wk)=Q(wk)+czk\tilde Q(w_k) = Q(w_k) + c z_k for some c>0c > 0. Then,

    ∑k=1nQ~(wk)=∑k=1nQ(wk)+c(z⋅1)=1+0\sum_{k = 1}^n \tilde Q(w_k) = \sum_{k = 1}^n Q(w_k) + c (z \cdot \bf{1}) = 1 + 0

    Is 1∈L\bf{1} \in L? Yes, the strategy of investing in one bond yields a discounted value of 11 at TT.

    Now, it is left to show that there exists a cc such that Q~(wk)>0\tilde Q(w_k) > 0. Consider the following formulation

c:=12max⁡1≤k≤n(∥zk∥Q(wk))c:= \frac{1}{2 \max_{1 \leq k \leq n} \left(\frac{\|z_k\|}{Q(w_k)}\right)}

Then,

12c=max⁡k(∥zk∥Q(wk))Q(wk)≥2c∥zk∥Q~(wk)=Q(wk)+czk≥2c∥zk∥+czk>0\begin{align*} \frac{1}{2c} &= \max_k \left(\frac{\|z_k\|}{Q(w_k)}\right) \\ Q(w_k) &\geq 2c \|z_k\| \\ \tilde Q(w_k) &= Q(w_k) + cz_k \\ &\geq 2c \|z_k\| + cz_k > 0 \end{align*}

Finally, to show that Q~\tilde Q is an EMM,

EQ~(VT∗(ϕ))=?0 ifV0(ϕ)=0=EQ(VT∗(ϕ))+cz⋅VT∗(ϕ)∵z⊥L=0\begin{align*} \mathbb E^{\tilde Q}(V_T^*(\phi)) &\stackrel{?}{=} 0 \quad \text{ if} V_0(\phi) = 0\\ &= \mathbb E^Q (V_T^*(\phi)) + c z \cdot V_T^*(\phi) \quad \because z \perp L \\ &= 0 \end{align*}

This proves the second fundamental theorem of asset pricing.

Note that if Q1,Q2Q_1, Q_2 are two EMMs, then a linear combination of these αQ1(A)+(1−α)Q2(A)\alpha Q_1(A) + (1 - \alpha) Q_2(A) for A∈FA \in \mathcal F and 0<α<10<\alpha<1 is also an EMM.

Example

For T=1,Ω={w1,w2,w3}T = 1, \Omega = \{w_1, w_2,w_3\}, let S10=S00=1S_1^0 = S_0^0 = 1, S01=2S_0^1 = 2 and S11=(135)TS_1^1 = \begin{pmatrix}1 & 3 & 5\end{pmatrix}^T. To check if the market is viable, we try and solve for an EMM on this market - Let Q=(a,b,c)Q = (a, b, c), then

a+b+c=1,a>0,b>0,c>0a+3b+5c=2\begin{align*} a + b + c = 1, a > 0,b>0,c>0\\ a + 3b + 5c = 2 \end{align*}

Solving, we get

Q(c)=(12+c,12−2c,c)Q^{(c)} = \left(\frac{1}{2} + c, \frac{1}{2} - 2c, c\right)

for 0<c<140 < c < \frac{1}{4}. This shows that the market is viable. Also, the market is not viable since there are multiple EMMs possible.

For a contingent claim XX, we have

E(c)(X∗)=(12+c)x1+(12−2c)x2+cx3=x1+x22+c(x1+x3−2x2)\begin{align*} \mathbb E^{(c)}(X^*) &= \left(\frac{1}{2} + c\right)x_1 + \left(\frac{1}{2} - 2c\right)x_2 + cx_3 \\ &= \frac{x_1 + x_2}{2} + c (x_1 + x_3 - 2x_2) \end{align*}

That means, when x1+x3−2x2x_1 + x_3 - 2x_2 is 00, then the expectation is independent of cc - XX can be replicated in some cases even if the market is incomplete!

Let us try an replicate such an XX. Consider the bond holdings α\alpha and stock holdings β\beta.

αS11+βS10=Xα(135)+β(111)=(x1x2x3)\begin{align*} \alpha S_1^1 + \beta S_1^0 &= X \\ \alpha \begin{pmatrix}1 \\ 3 \\ 5\end{pmatrix} + \beta \begin{pmatrix}1 \\ 1 \\ 1\end{pmatrix} &= \begin{pmatrix}x_1 \\ x_2 \\ x_3\end{pmatrix} \end{align*}

XX is replicable iff X∈span((135),(111))≡X⊥(1−21)X \in \text{span}(\begin{pmatrix}1 & 3 & 5\end{pmatrix}, \begin{pmatrix}1 & 1 & 1\end{pmatrix}) \equiv X \perp \begin{pmatrix}1 & -2 & 1\end{pmatrix}. This again yields x1+x3−2x2x_1 + x_3 - 2x_2.

Theorem. In a viable and complete market, the number V_0(\phi) is the no-arbitrage price of a European Call Clain X = V_T(\phi).

Note. We have seen this before in the context of a binomial model, but this theorem is for a general market with discrete possibilities. In our previous proof, we did not use any binomial properties but rather showed the result using martingale properties. The same proof follows for this as well.

How do we conclude such results in a general case? Even when the market is incomplete, how do we find contingent claims which can be replicated?

Consider the set M={Q:Q is an EMM}≠ϕM = \{Q: Q \text{ is an EMM}\} \neq \phi. Then for any measure in this set,

EQ(VT∗(ϕ))=V0∗(ϕ)=V0(ϕ)S00\begin{align*} \mathbb E^Q(V_T^*(\phi)) &= V_0^*(\phi) \\ & = \frac{V_0(\phi)}{S_0^0} \end{align*}

Now, for a contingent claim X, we have

V+(X):=inf⁡{V0(ϕ):VT(ϕ)≥X}=min⁡{V0(ϕ):VT(ϕ)≥X}EQ(X∗)≤EQ(VT∗(ϕ))ST0=V0(ϕ)ST0S00≤1S00V+(X)\begin{align*} V_+(X)&:= \inf \{V_0(\phi):V_T(\phi)\geq X\} \\ &= \min \{V_0(\phi):V_T(\phi)\geq X\} \\ \hline \mathbb E^Q(X^*) &\leq \mathbb E^Q(V_T^*(\phi)) \sout{S_T^0} = V_0(\phi) \frac{\sout{S_T^0}}{S_0^0} \\ &\leq \frac{1}{S_0^0} V_+(X) \end{align*}

Similarly,

V−(X):=sup⁡{V0(ϕ):VT(ϕ)≤X}=max⁡{V0(ϕ):VT(ϕ)≤X}V−(X)≤S00EQ(X∗)≤V+(X)\begin{align*} V_-(X)&:= \sup \{V_0(\phi):V_T(\phi)\leq X\} \\ &= \max \{V_0(\phi):V_T(\phi)\leq X\} \\ \hline \\ V_-(X) &\leq S_0^0 \mathbb E^Q(X^*) \leq V_+ (X) \end{align*}

Suppose XX is replicable, then X=VT(ϕ~)X = V_T(\tilde \phi) for some ϕ~\tilde \phi. Then,

V0(ϕ~)≥V+(X),V0(ϕ~)≤V−(X)V−(X)≤V+(X)\begin{align*} V_0(\tilde\phi) \geq V_+(X), V_0(\tilde\phi) \leq V_-(X) \\ V_-(X) \leq V_+(X) \end{align*}

leading to V+(X)=V−(X)V_+(X) = V_-(X)!

Does the converse hold? In the previous example, say we have a contingent claim X=(1925)=(S11)2X = \begin{pmatrix}1 & 9 & 25\end{pmatrix} = (S_1^1)^2. Now consider a strategy ϕ=(α,β)\phi = (\alpha, \beta), we have

V0(ϕ)=2α+βV11(ϕ)≤X for V−(X)α(135)+β(111)≤(1925)\begin{align*} V_0(\phi) &= 2\alpha + \beta \\ V_1^1(\phi) &\leq X \quad \text{ for } V_-(X) \\ \alpha \begin{pmatrix} 1 \\ 3 \\5 \end{pmatrix} + \beta \begin{pmatrix} 1 \\ 1 \\ 1\end{pmatrix} &\leq \begin{pmatrix}1 \\9 \\ 25\end{pmatrix} \end{align*}

Solving these inequalities, we get V−(X)=5V_-(X) = 5, and similarly, V+(X)=7V_+(X) = 7. Let us check if XX can be replicated

E(c)(X)=5+8c,0<c<14\mathbb E^{(c)}(X) = 5 + 8c, 0 < c< \frac{1}{4}

It is not possible for the expectation to be independent of cc. Also note that the range of the above expression is exactly (5,7)(5, 7).

We have seen that V+(X)≥V−(X)V_+(X) \geq V_-(X) for a contingent claim XX.

  • If C0>V+(X)C_0 > V_+(X) then there is a seller's arbitrage. How? There exists a strategy ϕ\phi such that VT(ϕ)≥XV_T(\phi) \geq X and V0(ϕ)=V+(X)<C0V_0(\phi) = V_+(X) < C_0.

  • Similarly. when C0<V−(X)C_0 < V_-(X) there is a buyer's arbitrage.

Theorem. A contingent claim XX is replicable if and only if V+(X)=V−(X)V_+(X) = V_-(X)

Proof.

  • (  ⟹  \implies) We saw that previously

  • (⟸\Longleftarrow) Assume V+(X)=V−(X)V_+(X) = V_-(X). Let ϕ+\phi^+ be a trading strategy such that VT(ϕ+)≥XV_T(\phi^+) \geq X and ϕ−\phi^- is such that VT(ϕ−)≥XV_T(\phi^-) \geq X. However, V0(ϕ+)=V+(X)=V−(X)=V0(ϕ−)V_0(\phi^+) = V_+(X) = V_-(X) = V_0(\phi^-) and consequently VT(ϕ+)−VT(ϕ−)≥0V_T(\phi^+) - V_T(\phi^-) \geq 0. Let Q∈MQ \in M be any martingale measure -

    EQ(VT∗(ϕ+)−VT∗(ϕ−))=V0∗(ϕ+)−V0∗(ϕ−)=(V0(ϕ+)−V0(ϕ−))/S00=0\begin{align*} \mathbb E^Q(V_T^*(\phi^+) - V_T^*(\phi^-)) &= V_0^*(\phi^+) - V_0^*(\phi^-) \\ &= (V_0(\phi^+) - V_0(\phi^-))/S_0^0 = 0 \end{align*}

    Therefore, VT(ϕ+)≡VT(ϕ−)≡XV_T(\phi^+) \equiv V_T^(\phi^-) \equiv X implying that ϕ+\phi^+ can be used to replicate XX.

    Suppose XX is not replicable, then there is ϕ\phi such that VT(ϕ)≤XV_T(\phi) \leq X and V0(ϕ)=V+(X)V_0(\phi) = V_+(X).

    Claim. P(VT(ϕ)<X)>0P(V_T(\phi) < X) > 0. If it is equal to 00, then VT(ϕ)≡XV_T(\phi) \equiv X.

    So, any market price for XX with C0∉(V−(X),V+(X))C_0 \not \in (V_-(X), V_+(X)) yields an arbitrage.

Theorem. XX is replicable if and only if EQ(X)\mathbb E^Q(X) is the same for all Q∈MQ \in M.

Proof.

  • (  ⟹  )(\implies) X=VT(ϕ)X = V_T(\phi) then

    EQ(X)=EQ(VT(ϕ))=V0(ϕ)ST0S00\mathbb E^Q(X) = \mathbb E^Q(V_T(\phi)) = V_0(\phi) \frac{S_T^0}{S_0^0}
  • (⟸)(\Longleftarrow) Assume for contradition XX is not replicable. As we've constructed before, let L={VT(ϕ):ϕ is a trading strategy}L = \{V_T(\phi): \phi \text{ is a trading strategy}\}. Since XX is not replicable, X∉LX \not \in L. Let X=X0+ZX = X_0 + Z such that X0∈LX_0 \in L and Z≠0,⊥LZ \neq 0, \perp L. For any Q∈MQ \in M, we define Q~(wk)=Q(wk)+cZk\tilde Q(w_k) = Q(w_k) + c Z_k. Again, we choose a cc such that Q~>0\tilde Q > 0 and Q~∈M\tilde Q \in M. Then, since EQ(X)\mathbb E^Q(X) is the same for all Q∈MQ \in M, we have

    EQ(X)=EQ~(X)Q⋅X=Q~⋅X=Q⋅X+cZ⋅X∴Z⋅X=0\begin{align*} \mathbb E^Q(X) &= \mathbb E^{\tilde Q}(X) \\ Q \cdot X &= \tilde Q \cdot X = Q \cdot X + c Z \cdot X \\ &\therefore Z \cdot X = 0 \end{align*}

    implying that X∈LX \in L giving a contradiction.

Dual Formulae

As a result of the above theorem, we can give alternate formulae to V+(X)V_+(X) and V−(X)V_-(X) (dual comes from linear programming)

V+(X)=sup⁡{EQ(X):Q∈M}V−(X)=inf⁡{EQ(X):Q∈M}\begin{align*} V_+(X) &= \sup \{\mathbb E^Q(X): Q \in M\} \\ V_-(X) &= \inf \{\mathbb E^Q(X) : Q \in M\} \end{align*}

Random Walk

In this section, we shall extend our theory beyond finite market models, wherein the price evolution is described using a random walk.

Let Xn=ξ1+⋯+ξnX_n = \xi_1 + \dots + \xi_n where E(ξk)=μ\mathbb E(\xi_k) = \mu and ξ\xi are iid for n=0,1,2,…n = 0, 1, 2, \dots. Let Fn=σ(ξ1,…,ξn)\mathcal F_n = \sigma(\xi_1, \dots, \xi_n) and Fn⊥ ⁣ ⁣ ⁣⊥ξ+n+k\mathcal F_n \perp \!\!\! \perp \xi+{n + k} for k≥1k \geq 1. Then, consider set of random variables Mn=Xn−nμM_n = X_n - n \mu.

Claim. MnM_n is a martingale with respect to Fn\mathcal F_n.

Proof.

E[Mn+1∣Fn]=E[Xn+1∣Fn]−(n+1)μ=E[Xn+ξn+1∣Fn]−(n+1)μ=Xn+μ−(n+1)μ=Mn\begin{align*} \mathbb E[M_{n + 1} \vert \mathcal F_n] &= \mathbb E[X_{n + 1} \vert \mathcal F_n] - (n + 1)\mu\\ &= \mathbb E[X_n + \xi_{n + 1} \vert \mathcal F_n] - (n + 1)\mu\\ &= X_n + \mu - (n + 1)\mu = M_n \end{align*}

Assuming that var(ξk)=σ2var(\xi_k) = \sigma^2 is finite, consider Yn=Mn2−nσ2Y_n = M_n^2 - n\sigma^2.

Claim. YnY_n is a martingale with respect to Fn\mathcal F_n when μ=0\mu = 0

Proof.

E[Yn+1∣Fn]=E[Mn+12∣Fn]−(n+1)σ2=E[(Mn+ξn+1)2∣Fn]−(n+1)σ2=Mn2+Mnμ+σ2−(n+1)σ2=Yn\begin{align*} \mathbb E[Y_{n + 1} \vert \mathcal F_n]&= \mathbb E[M^2_{n + 1} \vert \mathcal F_n] - (n + 1)\sigma^2\\ &= \mathbb E[(M_n + \xi_{n + 1})^2 \vert \mathcal F_n] - (n + 1)\sigma^2\\ &= M_n^2 + M_n \mu + \sigma^2 - (n + 1)\sigma^2 = Y_n \end{align*}

Optional Stopping

Let τ\tau be a random variable representing stopping time. Then, we have

E[Mτ]=E(M0)\mathbb E[M_\tau] = \mathbb E(M_0)

provided τ(w)≤N\tau(w) \leq N. That is, Mn∧τM_{n \wedge \tau} is a martingale..

Single Random Walk

We have the following setup

Xn=x+ξ1+⋯+ξnX_n = x + \xi_1 + \dots + \xi_n

where ξk\xi_k are i.i.d. Let P(ξ=1)=pP(\xi= 1) = p and P(ξk=−1)=q=1−pP(\xi_k = -1) = q = 1 - p where 0<p<10 < p < 1. We then have

μ=2p−1,σ2=1−(2p−1)2\mu = 2p - 1, \sigma^2 = 1 - (2p - 1)^2

Let τb:=min⁡(n≥0:Xn=b)\tau_b := \min(n \geq 0: X_n = b) and τ=min⁡(τa,τb)\tau = \min(\tau_a, \tau_b).

Claim. Suppose p>12p > \frac{1}{2} and μ>0\mu > 0 then

Px(τb<∞)=1b≥xPx(τa<∞)<1a<x\begin{align*} P_x(\tau_b < \infty) = 1 \quad b \geq x \\ P_x(\tau_a < \infty) < 1\quad a < x \end{align*}

Proof. We construct a martingale of the form Mn=αXnβnM_n = \alpha^{X_n} \beta^n for some α>0\alpha > 0.

For α>0\alpha > 0, we let 1β=E(αξk)=αp+1α(1−p)\frac{1}{\beta} = \mathbb E(\alpha^{\xi_k}) = \alpha p + \frac{1}{\alpha}(1 - p). Then,

E(Mn+1∣Fn)=αXnβn+1E(αξn+1∣Fn)=αXnβn+11β=αXnβn=Mn\begin{align*} \mathbb E(M_{n + 1} \vert \mathcal F_n) &= \alpha^{X_n} \beta^{n + 1} \mathbb E(\alpha^{\xi_{n + 1}} \vert \mathcal F_n) \\ &= \alpha^{X_n} \beta^{n + 1} \frac{1}{\beta} = \alpha^{X_n} \beta^n = M_n \end{align*}

That concludes the proof.

Claim. When p≥12p \geq \frac{1}{2}, then Px(τb<∞)=1P_x(\tau_b < \infty) = 1 where τa=min⁡(n≥0:Xn=a)\tau_a = \min(n \geq 0: X_n = a) (=∞=\infty if no such nn). Note that we are not dealing with finite timeframe anymore.

The graph of β\beta looks like

Proof. So for b≥Xb \geq X, we choose 0<α<10 < \alpha < 1 such that β>1\beta > 1, then

αX=EX(Mn∧τb)=EX(αXn∧τbβn∧τb)≥EX(αbβn∧τb)≥αbEx(βn∧τb;τb=∞)=αbβnPX(τb=∞)\begin{align*} \alpha^X &= \mathbb E_X(M_{n \wedge \tau_b}) \\ &= \mathbb E_X(\alpha^{X_{n \wedge \tau_b}} \beta^{n \wedge \tau_b}) \\ &\geq \mathbb E_X (\alpha^b \beta^{n \wedge \tau_b}) \geq \alpha^b \mathbb E_x(\beta^{n \wedge \tau_b}; \tau_b = \infty) \\ &= \alpha^b \beta^n P_X(\tau_b = \infty) \end{align*}

Letting n→∞n \to \infty, we infer PX(τb=∞)=0P_X(\tau_b = \infty) = 0.

Claim. When p>12p > \frac{1}{2}, then Px(τa<∞)=(1p)x−aP_x(\tau_a < \infty) = \left(\frac{1}{p}\right)^{x - a}.

Proof. For qp<α<1\frac{q}{p} < \alpha < 1, 0<β<10 < \beta< 1

αX+EX(Mn∧τa)=EX(αXnβn;n<τa)+EX(αXτaβτa;n≥τa)≤αaβnPX(n<τa)+αaEX(βτa;τa≤n)\begin{align*} \alpha^X &+ \mathbb E_X(M_{n \wedge \tau_a} ) \\ &= \mathbb E_X(\alpha^{X_n}\beta^n ; n < \tau_a) + \mathbb E_X(\alpha^{X_{\tau_a} \beta^{\tau_a}}; n \geq \tau_a) \\ &\leq \alpha^a \beta^n P_X(n < \tau_a) + \alpha^a\mathbb E_X (\beta^{\tau_a}; \tau_a \leq n ) \\ \end{align*}

The second term can be simplified using the Monotone Convergence Theorem -

EX(βτa;τa<n)=EX(βτa;τa<∞)=αx−a\begin{align*} \mathbb E_X (\beta^{\tau_a}; \tau_a < n) &= \mathbb E_X(\beta^{\tau_a}; \tau_a < \infty) \\ &=\alpha^{x - a} \end{align*}

Letting α↘q/p\alpha \searrow q/p, β↗1\beta \nearrow 1 and

(qp)x−a=Ex(1;τa<∞)=PX(τa<∞)\left(\frac{q}{p}\right)^{x - a} = \mathbb E_x( 1; \tau_a < \infty) = \mathbb P_X(\tau_a < \infty)

In summary, we have for p>12p > \frac{1}{2}

PX(τa<∞)=(qp)x−a<1PX(τb<∞)=1\begin{align*} P_X(\tau_a < \infty) &= \left(\frac{q}{p}\right)^{x - a} < 1\\ P_X(\tau_b < \infty) &= 1 \end{align*}

For p=12p = \frac{1}{2},

PX(τb<∞)=PX(τa<∞)=1P_X(\tau_b < \infty) = P_X(\tau_a < \infty) = 1

So irrespective of pp, PX(τa∧τb<∞)=1P_X(\tau_a \wedge \tau_b < \infty) = 1.

Example

When PX(τa<τb)=ρb−ρxρb−ρaP_X(\tau_a < \tau_b) = \frac{\rho^b - \rho^x}{\rho^b - \rho^a} where Mn=(qp)Xn=ρM_n = \left(\frac{q}{p}\right)^{X_n} = \rho (p≠qp \neq q)

Note that Mn=XnM_n = X_n when p=qp = q.

Let τ=τa∧τb\tau = \tau_a \wedge \tau_b when p≠qp \neq q,

ρX=EX(Mn∧τ)\rho^X = \mathbb E_X(M_{n \wedge \tau})

Using the Dominated Convergence Theorem,

ρX=EX(Mτ)\rho^X = \mathbb E_X(M_\tau)

and we get

ρX=ρAPX(τa<τb)+ρbPx(τb<τa)1=PX(τa<τb)+PX(τb<τa)\begin{align*} \rho^X &= \rho^A P_X(\tau_a < \tau_b) + \rho^b P_x(\tau_b < \tau_a) \\ 1 &= P_X(\tau_a < \tau_b) + P_X(\tau_b < \tau_a) \end{align*}

Example

Let Mn=X2−nM_n = X^2 - n, with M0=X2M_0 = X^2. Then, for p=12,σ2=1,μ=1p = \frac{1}{2}, \sigma^2 = 1, \mu = 1, τ=τa∧τb\tau = \tau_a \wedge \tau_b, we have

X2=EX(Xn∧τ2)−E(n∧τ)X2+E(n∧τ)=EX(Xn∧τ2)\begin{align*} X^2 &= \mathbb E_X (X^2_{n \wedge \tau}) - \mathbb E(n \wedge \tau) \\ X^2 + \mathbb E(n \wedge \tau) &= \mathbb E_X (X^2_{n \wedge \tau}) \\ \end{align*}

Letting n→∞n \to \infty and using Monotone convergence theorem on the left and Dominated convergence theorem on the right, we get

X2+EX(τ)=EX(Xτ2)X2+EX(τ)=a2b−xb−a+b2x−ab−a\begin{align*} X^2 + \mathbb E_X(\tau) &= \mathbb E_X(X_\tau^2) \\ X^2 + \mathbb E_X(\tau) &= a^2\frac{b - x}{b - a} + b^2 \frac{x - a}{b - a}\\ \end{align*}

resulting in EX(τa∧τb)=(X−a)(b−X)\mathbb E_X(\tau_a \wedge \tau_b) = (X - a)(b - X).

Brownian Motion (Wiener Process)

Brownian motion is defined as a stochastic process defined on continuous time - W=(Wt,t≥0)W = (W_t, t \geq 0). It is defined on a probability space (Ω,F,P)(\Omega, \mathcal F, P) and has the following properties -

  • The initial value is assumed to be W0=0W_0 = 0

  • Independent increments - Given a set of times t0,…,tnt_0, \dots, t_n such that 0=t0<t1<⋯<tn0 = t_0 <t_1 < \cdots < t_n then Wt1−Wt0,…,Wtn−Wtn−1W_{t_1} - W_{t_0}, \dots, W_{t_{n}} - W_{t_{n - 1}} are independent.

  • Gaussian increments - For 0≤s≤t0 \leq s \leq t, Wt−Ws∼N(0,t−s)W_t - W_s \sim \mathcal N(0, t - s).

  • For a fixed ω∈Ω\omega \in \Omega, the function t→Wt(ω)t \to W_t(\omega) is continuous.

Equivalently, a Brownian motion is a stochastic process (wt:t≥0)(w_t: t \geq 0) is a Gaussian process with E(Wt)=0,cov(Ws,Wt)=s∧t\mathbb E(W_t) = 0, cov(W_s, W_t) = s \wedge t for all s,ts, t and t→Wtt \to W_t is continuous.

This is equivalent to the third property, for s<ts < t,

E[WsWt]=E[Ws(Ws+(Wt−Ws))]=E[Ws2]+E[Ws(Wt−Ws)]\begin{align*} \mathbb E[W_sW_t] &= \mathbb E[W_s(W_s + (W_t - W_s))] \\ &= \mathbb E[W_s^2] + \mathbb E[W_s(W_t - W_s)] \\ \end{align*}

since Ws−W0W_s - W_0 is a Gaussian distribution and independent increments property,

E[WsWt]=E[Ws2]+E[Ws(Wt−Ws)]=s2+E[Ws]E[Wt−Ws]=s2\begin{align*} \mathbb E[W_sW_t] &= \mathbb E[W_s^2] + \mathbb E[W_s(W_t - W_s)] \\ &= s^2 + \mathbb E[W_s]\mathbb E[W_t - W_s] = s^2 \end{align*}

Symmetries

Brownian motion is a unique stochastic process that has some symmetry properties. That is, transformations that yield a Brownian motion when applied on a Brownian motion. Let (Wt)(W_t) be a Brownian motion, then the so are the following

  • Scaling - 1cWct\frac{1}{\sqrt{c}} W_{ct} for c>0c> 0

  • Reflection - −Wt-W_t

  • Time Inversion - tW1/ttW_{1/t} with lim⁡t→0tW1/t=0\lim_{t \to 0} tW_{1/t} = 0

  • Time Shift - W~t=Wt+s−Ws\tilde W_t = W_{t + s} - W_s, t≥0t\geq 0 for a fixed ss. This comes in handy to define Markov processes.

  • Time Reversal - For a T>0T > 0, W~t=W~T−t−WT\tilde W_t = \tilde W_{T - t} - W_T for 0≤t≤T0 \leq t \leq T.

More Properties

  • Define the sigma field Fs:=σ(Wu:0≤u≤s)\mathcal F_s:=\sigma(W_u : 0 \leq u \leq s), then (Wt+s−Ws)t≥0⊥ ⁣ ⁣ ⁣⊥Fs(W_{t + s} - W_s)_{t \geq 0} \perp\!\!\!\perp F_s - This is the Markov Property for Brownian Motion.

    Brownian paths are not smooth

    Wt+h−Wt=O(h)W_{t + h - W_t} = \mathcal O(\sqrt{h})

    roughly whereas if ff is a C′C' function,

    then f(t+h)−f(t)=O(h)f(t + h) - f(t) = \mathcal O(h). h≫h\sqrt{h} \gg h as h→0h \to 0. A Brownian motion path has a property called fractal and it is nowhere differentiable.

    Also,

    ∑k=12n[W(k2−nt)−W((k−1)2−nt]2→t\sum_{k = 1}^{2^n} [W(k2^{-n} t) - W((k - 1)2^{-n} t]^2 \to t

    as n→∞n \to \infty. For a smooth function ff, doing the same thing yields 2−nt→02^{-n}t \to 0. This observation forms one of the fundamental building blocks for stochastic calculus.

  • Another peculiar property of Brownian motion has is -

    Wtt→0lim⁡t→∞sup⁡Wt=∞lim⁡t→∞inf⁡Wt=−∞\begin{align*} \frac{W_t}{t} &\to 0 \\ \lim_{t \to \infty} \sup W_t &= \infty \\ \lim_{t \to \infty} \inf W_t &= -\infty \\ \end{align*}
  • Since Brownian motion is basically the continuous version of random walks, it has some Martingale properties. Specifically, the following processes are martingales

    WtWt2−texp⁡(zWt−z2t/2) for z∈C\begin{align*} W_t \\ W_t^2 - t \\ \exp\left(zW_t - z^2t /2\right) \text{ for } z \in \mathbb C \end{align*}

    The last result comes from the properties of moment generating functions - for a random variable X∼N(0,1)X \sim \mathcal N(0, 1), we get E(ezX)=ez2/2\mathbb E(e^{zX}) = e^{z^2/2}. Then,

    E(ezWt∣Fs)=E(ezWsez(Wt−Ws)∣Fs)=ezWsE(ez(Wt−Ws))=ezWsez2(t−s)/2\begin{align*} \mathbb E(e^{zW_t} \vert \mathcal F_s) &= \mathbb E(e^{zWs} e^{z(W_t - W_s)} \vert \mathcal F_s) \\ &= e^{zW_s} \mathbb E(e^{z(W_t - W_s)}) = e^{zW_s} e^{z^2 (t - s)/2} \end{align*}
  • Wiener also studied the integration properties for Brownian motion

    Mtf=∫0tf(u)dWu=Wtf(t)−∫0tWuf′(u)du\begin{align*} M_t^f = \int_0^t f(u) dW_u = W_t f(t) - \int_0^t W_u f'(u) du \end{align*}

    for a f∈C′f \in C' and non-random. The above expression comes from integration by parts and is called as the Wiener's integral. Also, MtfM_t^f is a martingale (This is easy to show).

  • Doob's inequality - Let (Mt)(M_t) be a martingale with right-continuous paths. Then, for each t>0t > 0,

    E[sup⁡0≤s≤tMs2]≤4E[Mt2]\mathbb E[\sup_{0 \leq s \leq t} M_s^2] \leq 4\mathbb E[M_t^2]
  • Passage times - Let τb=min⁡(t:Wt=b)\tau_b = \min(t: W_t = b), then

    Ex(e−ατb)=exp⁡(−2α∣b−x∣)\mathbb E_x(e^{-\alpha \tau_b}) = \exp(-\sqrt{2\alpha} \vert b - x\vert)

    where Ex\mathbb E_x means we use x+Wtx + W_t (we are changing the initial mean value of WtW_t).

    Proof. Fix λ>0\lambda > 0 and x<bx < b. Since exp⁡(λWt∧τb−λ2(t∧τb)/2)\exp(\lambda W_{t \wedge \tau_b} - \lambda^2(t \wedge \tau_b)/2) is a martingale, and is ≤eλb\leq e^{\lambda b}, we get

    eλx=Ex(eλWt∧τb−λ2/2(t∧τb))as t→∞  ⟹  Ex(eλWτb−λ2/2τb)∵Dominated Convergence Theorem=eλbEx(e−λ2/2τb)\begin{align*} e^{\lambda x} &= \mathbb E_x(e^{\lambda W_{t \wedge \tau_b} - \lambda^2/2 (t \wedge \tau_b)}) \\ &\text{as } t\to \infty \implies \mathbb E_x(e^{\lambda W_{\tau_b} - \lambda^2/2\tau_b}) \quad \because \text{Dominated Convergence Theorem} \\ &= e^{\lambda b} \mathbb E_x(e^{-\lambda^2/2 \tau_b}) \end{align*}

    Taking λ=2α\lambda = \sqrt{2\alpha}, we get the statement of the theorem. Observe that when α→0\alpha \to 0, the RHS becomes 11. Therefore, Px(τb<∞)=1P_x(\tau_b < \infty) = 1 as α→0\alpha \to 0. That is, Brownian motion will eventually get to any level!

  • Wiener's integral - Formally,

    Mtf:=f(t)Wt−∫0twsf′(s)dsM_t^f := f(t) W_t - \int_0^t w_s f'(s)ds

    It is useful to know the distribution of this process. Thinking of the above integral as Riemann's sum, we can see that MtfM_t^f has the distribution N(0,σ2)\mathcal N(0, \sigma^2). The variance is calculated as

    E[(Mtf)2]=f(t)2t+∫0t∫0t(u∧v)f′(u)f′(v)dudv⏟(2)−2f(t)∫0tuf′(u)du⏟(3)(2)  ⟹  2∫0t(∫0uf′(v)dv)uf′(u)du∵ by symmetry=2∫0t[f(t)−f(u)]uf′(u)du(2)+(3)  ⟹  −2∫0tuf(u)f′(u)du=−tf(t)2+∫0tf(u)2duE[(Mtf)2]=∫0t[f(u)]2du\begin{align*} \mathbb E[(M_t^f)^2 ] &= f(t)^2 t + \underbrace{\int_0^t \int_0^t (u \wedge v) f'(u) f'(v) du dv}_{(2)} - \underbrace{2f(t) \int_0^t uf'(u) du}_{(3)} \\ (2) &\implies 2\int_0^t \left(\int_0^u f'(v) dv \right) u f'(u) du \quad\because \text{ by symmetry}\\ &= 2\int_0^t [f(t) - f(u) ]u f'(u) du \\ (2) + (3) &\implies -2\int_0^t uf(u)f'(u) du \\ &= -t f(t)^2 + \int_0^t f(u)^2 du \\ \mathbb E[(M_t^f)^2 ] &= \int_0^t [f(u)]^2 du \end{align*}

    This is a very useful result. In fact, using two different functions -

    MtfMtg−∫0tf(u)g(u)duM_t^f M_t^g - \int_0^t f(u)g(u) du

    is also a martingale! An easier way to show this would be using polarization - manipulate (Mtf+Mtg)2−(Mtf−Mtg)2(M_t^f + M_t^g )^2 - (M_t^f - M_t^g)^2. Also, Mtf±Mtg=Mtf±gM_t^f \pm M_t^g = M_t^{f \pm g}.

Donsker's Invariance Principle

A discrete random walk can be represented as Sn=ξ1+⋯+ξnS_n = \xi_1 + \cdots + \xi_n. When nn becomes very large, the magnitude of the random walk scales by 1n\frac{1}{\sqrt{n}}. This is due to the Central Limit Theorem - lim⁡n→∞Sn/n∼N(0,1)\lim_{n \to \infty} S_n/\sqrt{n} \sim \mathcal N(0, 1). It is a useful fact to note while visualizing the random walk to fit the evolution appropriately in space. Also,

Wn(t)={Sknt=kn,k=0,1,…,nlinearly interpolate otherwiseW_n(t) = \begin{cases} \frac{S_k}{\sqrt{n}} & t = \frac{k}{n}, k = 0, 1, \dots, n \\ \text{linearly interpolate otherwise}\end{cases}

Again, from the CLT, lim⁡n→∞Wn(t)→N(0,t)\lim_{n \to \infty} W_n(t) \to \mathcal N(0, t). The above formulation takes care of appropriately representing large random walks in a bounded space (probabilistically). Alternately, WnW_n can be understood as a function from the space C([0,1]→R)C([0, 1] \to \mathbb R) - continuous functions from [0,1][0, 1] to R\mathbb R.

Now, we are interested in performing operations FF on these random walks -

lim⁡n→∞E[F(Wn)]→E[F(W)]\lim_{n \to \infty} \mathbb E[F(W_n)] \to \mathbb E[F(W)]

for all bounded continious functions F:C→RF: C \to \mathbb R (maps each walk WnW_n to a real number). Here, WW is a standard Brownian motion. All the above equation is saying is that the discretized formulation WnW_n is equivalent to a standard Brownian motion for practical purposes when nn is large.

For example, let F(x)=∫01x(t)dtF(x) = \int_{0}^1 x(t) dt. Then,

∣F(x)−F(y)∣≤∫01∣x(t)−y(t)∣dt≤max⁡t∣x(t)−y(t)∣=∥x−y∥∞\begin{align*} \vert F(x) - F(y) \vert &\leq \int_0^1 \vert x(t) - y(t) \vert dt \\ &\leq \max_t \vert x(t) - y(t) \vert = \|x - y\|_\infty \end{align*}

Binomial Model as a Random Walk

We have St=St−1⋅ξtS_t = S_{t - 1} \cdot \xi_t where P(ξt=1)=p,P(ξt=−1)=1−pP(\xi_t = 1) = p, P(\xi_t = -1) = 1 - p. How do we model this as a difference equation?

St−St−1=St−1(ξt−1)≡dStSt=ξt−1\begin{align*} S_t - S_{t - 1} &= S_{t - 1} (\xi_t - 1) \\ \equiv \frac{dS_t}{S_t} &= \xi_t - 1 \end{align*}

The RHS should be equated to a⋅dt+b⋅dWta \cdot dt + b \cdot dW_t for extending it to random walks. The change in the stock price is represented as a composition of a constant value and a random variable. Interpreting this as an integral, we get the expression

St−S0=∫0tSu(a)du+∫0tSubdWu\begin{align*} S_t - S_0 = \int_0^t S_u (a) du + \int_0^t S_u b dW_u \end{align*}

In this expression, we know how to evaluate Su(a)duS_u(a)du (integrate with respect to time, sum of values as time evolves), but how do we make sense of SubdWuS_u b dW_u? We evaluate the above expression using Ito's Stochastic Integral.

Intuition - If you think about stock price, it has an inherent drift and randomness. Using this intuition, we get

>dSt=μStdt+σStdWt>> d S_t = \mu S_t dt + \sigma S_t dW_t >

The first integral with respect to dudu captures the drift whereas the second term captures the randomness.

Ito's Stochastic Integral

How do we calculate the integral ∫0tYsdWs\int_0^t Y_s dW_s? From Wiener's analysis, we have seen that this expression is a martingale. One approach to evaluate the expression is to it as a combination of Riemman's sums of step functions -

  1. Let Yt=1(u,v](s)⋅HY_t = 1_{(u, v]}(s) \cdot H where H∈L2(Fu)H \in L^2(\mathcal F_u)

    Mt=∫0tYsdWs={0,t≤uH(Wt−Wu),u≤t≤vH(Wv−Wu),t≥v=H(Wt∧v−Wt∧u)M_t = \int_0^t Y_s dW_s = \begin{cases} 0, &t \leq u \\ H(W_t - W_u), &u \leq t \leq v \\ H(W_v - W_u), & t \geq v \end{cases} = H(W_{t \wedge v} - W_{t \wedge u})

    Claim. MM is a continuous martingale and E[Mt2]=E[∫0tYs2ds]\mathbb E[M_t^2] = \mathbb E[\int_0^t Y_s^2 ds]

    Proof. The continuous part is trivial since WtW_{t} is continuous. We need to show that E[Mt∣Fs]?=Ms\mathbb E[M_t \vert \mathcal F_s] \substack{?}{=} M_s for s<ts< t

    E[Mt∣Fs]=E[H(Wt−Wu)∣Fs]=E[HWt∣Fs]−E[HWu∣Fs]=HWs−HWu=Ms\begin{align*} \mathbb E[M_t \vert F_s] &= \mathbb E[H(W_t - W_u) \vert F_s] \\ &= \mathbb E[HW_t \vert F_s] - E[HW_u \vert F_s] \\ &= HW_s - HW_u = M_s \end{align*}

    due to the martingale property of WtW_t. Finally, for the second moment, we have

    E(Mt2)=E[H2(Wt∧v−Wt∧u)]={0t≤uE[H2(Wt−Wu)2],u≤t≤vE[H2(Wv−Wu)2],t≥u={0t≤uE[H2](t−u),u≤t≤vE[H2](v−u),t≥u=E[H2(t∧u−t∧u)]=E[∫0tH21(u,v](s)ds]\begin{align*} \mathbb E(M_t^2) &= \mathbb E[H^2(W_{t \wedge v} - W_{t \wedge u})] \\ &= \begin{cases} 0 & t \leq u \\ \mathbb E[H^2(W_t - W_u)^2], & u \leq t \leq v \\ \mathbb E[H^2(W_v - W_u)^2], & t\geq u \end{cases} \\ &= \begin{cases} 0 & t \leq u \\ \mathbb E[H^2](t - u), & u \leq t \leq v \\ \mathbb E[H^2](v - u), & t\geq u \end{cases} \\ &= \mathbb E[H^2(t \wedge u - t \wedge u)] = \mathbb E\left[\int_0^t H^2 1_{(u, v]} (s)ds\right] \end{align*}
  2. Now, we need to add such YkY_k's for the required interval (Riemman sum). Ys=∑k=1n1(uk,vk](s)HkY_s = \sum_{k = 1}^n 1_{(u_k, v_k]} (s) H_k and Hk∈L2(Fuk)H_k \in L^2 (\mathcal F_{u_k}) -

    ∫0tYsdWs:=∑k=1n∫0tYs(k)dWs\int_0^t Y_s dW_s := \sum_{k = 1}^n \int_0^t Y_s^{(k)} dW_s

    Since YsY_s is a continuous martingale in tt. Also, the Ito's Isometry property states that

    E((∫0tYsdWs)2)=E(∫0tYs2ds)\mathbb E((\int_0^t Y_s dW_s)^2) = \mathbb E(\int_0^t Y_s^2 ds)

    This is easy to show considering non-overlapping intervals. Hint. Consider conditional expectation using Tower law.

    We have seen this in context of Wiener integrals, but the integrand was considered to be deterministic in that case. Here, we generalize the notion by considering integrands that can be stochastic processes as well.

  1. Progressively Measurable - Consider a finite segment of time 0≤t≤T0 \leq t \leq T. Then, a stochastic process Xt(ω)X_t(\omega) can be viewed as a function Ω×[0,T]→R\Omega \times [0, T] \to \mathbb R. Then, the integral ∫0tXs(ω)ds\int_{0}^t X_s(\omega)ds can be viewed as a marginal integral of a joint function. The fact that ∫0tXs(ω)ds∈Fs\int_0^t X_s(\omega) ds \in \mathcal F_s (the integral is measurable) comes from Fubini's theorem in Measure theory. Specifically, the process Xt(ω)X_t(\omega) is on the domain Ω×[0,t]\Omega \times [0, t] and is consequently measurable on Ft⊗B[0,t]\mathcal F_t \otimes \mathcal B[0, t]. Here B\mathcal B refers to Borel Measurability. (Xt)(X_t) is said to be adapted to (Ft)(\mathcal F_t)

    Claim. If XtX_t is adapted and t→Xt(ω)t \to X_t(\omega) is left-continuous for all ω∈Ω\omega \in \Omega, then XX is progressively measurable.

    In the previous approach to integrate stochastic processes, we considered Ls\mathcal L_s class of integrands. That is, we assumed that the integrands can be written as a sequence of step functions. Now, we generalise this notion for any integrands that are progressive measurable

    L={Y:Y is progressively measurable and E(∫0TYS2ds)<∞}\mathcal L = \{Y: Y \text{ is progressively measurable and } \mathbb E\left(\int_0^T Y_S^2ds\right) < \infty\}

    A key point to note is that Ls\mathcal L_s is dense in L\mathcal L (we will not be proving this). That means, given Y∈LY \in \mathcal L, there exists a sequence of integrands (Y(n))n≥1(Y^{(n)})_{n \geq 1} belonging to Ls\mathcal L_s such that

    lim⁡n→∞E[∫0T(Ys−Ys(n))2ds]→0\lim_{n \to \infty} \mathbb E\left[\int_0^T (Y_s - Y_s^{(n)})^2 ds\right] \to 0

    Since each (Y(n))(Y^{(n)}) is continuous and a martingale, we will try to show that Mt(n)=∫0tYs(n)dWsM_t^{(n)} = \int_0^t Y_s^{(n)} dW_s is a Cauchy sequence converging to the required integral of YsY_s (note that (Y(n))(Y^{(n)}) is Cauchy) -

    E[Mt(n)−Mt(m)]=E[∫0t[Ys(n)−Ys(m)]]≤E[∫0T[Ys(n)−Ys(m)]]→0\begin{align*} \mathbb E\left[ M_t^{(n)} - M_t^{(m)}\right] &= \mathbb E\left[\int_0^t [Y_s^{(n)} - Y_s^{(m)}]\right] \\ &\leq \mathbb E\left[\int_0^T [Y_s^{(n)} - Y_s^{(m)}]\right] \to 0 \end{align*}

    Therefore, (Mt(n))n≥1(M_t^{(n)})_{n \geq 1} is Cauchy in L2(Ft)\mathcal L^2(\mathcal F_t). Consequently, there exists Mt∈Lw(Ft)M_t \in \mathcal L^w(\mathcal F_t) such that Mt(n)→MtM_t^{(n)} \to M_t. Then, using Doob's inequality

    E(sup⁡0≤t≤T∣Mt(n)−Mt(m)∣2)≤4E[(MT(n)−MT(m))2]\begin{align*} \mathbb E\left(\sup_{0 \leq t \leq T} \vert M_t^{(n)} - M_t^{(m)}\vert^2 \right) \leq 4 \mathbb E[(M_T^{(n)} - M_T^{(m)})^2] \end{align*}

    we can say that (Mt)0≤t≤T(M_t)_{0 \leq t \leq T} is a path-continuous martingale and E[Mt2]=E[∫0tYs2ds]\mathbb E[M_t^2] = \mathbb E[\int_0^t Y_s^2 ds] for all t∈[0,T]t \in [0, T] from Ito's isometry.

  2. An alternate way to view the integral is to formulate it as a Riemann integral. If Y∈LY \in \mathcal L is such that s→Ys(w)s \to Y_s(w) is left-continuous for all ω∈Ω\omega \in \Omega then

    ∫0tYsdWs=lim⁡n∑k=0n−1Yknt(Wk+1nt−Wknt)\int_0^t Y_s dW_s = \lim_n \sum_{k = 0}^{n - 1} Y_{\frac{k}{n}t}\left(W_{\frac{k + 1}{n} t} - W_{\frac{k}{n} t}\right)

    is also a limiting sum of Riemann sums (converges in L2L^2). If the above limit converges, then Ys∈L2Y_s \in \mathcal L^2.

    Example - Consider ∫0tWsdWs\int_0^t W_s dW_s . W∈LW \in \mathcal L. How?

    E[∫0TWs2ds]=∫0TE(Ws2)ds=∫0Tsds<∞\mathbb E\left[\int_0^T W_s^2 ds \right] = \int_0^T \mathbb E(W_s^2) ds = \int_0^T sds < \infty

        from Fubini's theorem. Also, since WW is         progressive, it is integrable.

Quadratic Variation

Quadratic variation of a Brownian motion is an important tool in stochastic calculus. Simply put, it is defined as

lim⁡n→∞Qt(n)=lim⁡n→∞∑k=0n−1(Wk+1nt−Wknt)2=t\lim_{n \to \infty} Q_t^{(n)} = \lim_{n \to \infty} \sum_{k = 0}^{n - 1} (W_{\frac{k + 1}{n} t} - W_{\frac{k}{n}t})^2 = t

Since the increments are independent are from the distribution N(0,tn)\mathcal N(0, \frac{t}{n}), the variance of Qt(n)Q_t^{(n)} is

E((Qt(n))2)=3n(tn)2+n(n−1)(tn)2\mathbb E((Q_t^{(n)})^2 ) = 3n \left(\frac{t}{n}\right)^2 + n(n - 1)\left(\frac{t}{n}\right)^2

implying Var(Qt(n))=2nt2→0Var(Q_t^{(n)}) = \frac{2}{n} t^2 \to 0 as n→∞n \to \infty.

Going back to our example, writing the integral as Riemann sums -

Wt2=∑k=0n−1(Wk+1nt2−Wknt2)∵W02=0=2∑k=0n−1Wknt(Wk+1nt−Wknt)+Qt(n)∵telescopic sum=2∫0tWSdWs+t\begin{align*} W_t^2 &= \sum_{k = 0}^{n - 1} (W^2_{\frac{k + 1}{n} t} - W^2_{\frac{k }{n} t}) \quad \because W_0^2 = 0\\ &= 2 \sum_{k =0}^{n - 1} W_{\frac{k}{n}t} (W_{\frac{k + 1}{n} t} - W_{\frac{k }{n} t}) + Q_t^{(n)}\quad \because \text{telescopic sum} \\ &= 2\int_0^t W_S dW_s + t \end{align*}

since the above is left-continuous, letting n→∞n \to \infty, we get the above expression.

We have derived

Wt2−t=∫0t2WsdWsW_t^2 - t = \int_0^t 2W_s dW_s

We can check Ito's isometry using this.

Also, the above expression in reminiscent of the fundamental theorem of calculus. Using this, we now state the following theorem

Theorem (Ito's Formula/Lemma) For a function f∈C2f \in C^2 and ∣f′(x)∣≤c0ec1∣x∣\vert f'(x) \vert \leq c_0 e^{c_1\vert x \vert} for constants c0,c1c_0, c_1,

f(Wt)=f(W0)+∫0tf′(Ws)dWs+12∫0tf′′(WS)dsf(W_t) = f(W_0) + \int_0^t f'(W_s)dW_s + \frac{1}{2} \int_0^t f''(W_S) ds

This forms the basis for the fundamental theorem for stochastic calculus.

Localization

We define the set

Lloc:={Y:Y is progressive and P(∫0tYs2ds<∞)=1}\mathcal L_{loc} := \{Y: Y \text{ is progressive and } P\left(\int_0^t Y_s^2ds < \infty\right) = 1\}

This is a weaker condition than E[∫0tYS2ds]<∞\mathbb E[\int_0^t Y_S^2 ds] < \infty, and it allows us to generalize the notion of ∫0tYsdWs\int_0^t Y_s dWs to more integrands.

Given a Y∈LlocY \in \mathcal L_{loc}, we define the stop times τ(n)=inf⁡{t:∫0tYs2ds>n}∧T\tau(n) = \inf \{t: \int_0^t Y_s^2 ds > n\} \wedge T. Then,

∩n{τ(n)<T}⊂{∫0TYs2ds=∞}\cap_n \{\tau(n) < T\} \subset \left\{\int_0^T Y_s^2 ds = \infty\right\}

so P(τ(n)<T)→0P(\tau(n) < T) \to 0 as n→∞n \to \infty. Furthermore, defining Ys(n)=1(0,τ(n)](s)YsY_s^{(n)} = \mathbb 1_{(0, \tau(n)]} (s) Y_s, we get

∫0T(Ys(n))2ds≤n2\int_0^T (Y_s^{(n)})^2 ds \leq n^2 implying that Y(n)∈LY^{(n)} \in \mathcal L. Also, the martingale ∫0tYs(n)dWs\int_0^t Y_s^{(n)} dW_s is well-defined for each n=1,2,…n = 1, 2, \dots.

They are also consistent - For m<nm < n, τ(m)<τ(n)\tau(m) < \tau(n) and

∫0tYs(m)dWs=∫0tYs(n)dWs\int_0^t Y_s^{(m)} dW_s = \int_0^t Y_s^{(n)} dW_s

if t∈[0,τ(m)]t \in [0, \tau(m)].

Since P(τ(n)<T)→0P(\tau(n) < T) \to 0 as n→∞n \to \infty, and on ∪n{τ(n)=T}\cup_n \{\tau(n) = T\}, we have

It=lim⁡n∫0tYs(n)dWs,0≤t≤TI_t = \lim_n \int_0^t Y_s^{(n)}dW_s, 0 \leq t \leq T

and It∧τ(n)=∫0tYs(n)dWsI_{t \wedge \tau(n)} = \int_0^t Y_s^{(n)} dW_s is a martingale.

This process II, which we use as our definition of ∫0tYsdWs\int_0^t Y_s dW_s is an instance of a local martingale.

A local martingale is an adapted process M=(Mt)M = (M_t) for which there is an increasing sequence (σ(n))(\sigma(n)) of stopping times such that P(σ(n)<T)→0P(\sigma(n) < T) \to 0 as n→∞n \to \infty and Mt∧σ(n)M_{t \wedge \sigma(n)} is a martingale.

When the martingales and local martingales have continuous paths, (σ(n))(\sigma(n)) can be taken to be σ(n)=inf⁡(t:∣Mt∣n)∧T\sigma(n) = \inf(t: \vert M_t \vert n) \wedge T. In particular it can always be arranged that ∣Mt∧σ(n)∣≤Kn\vert M_{t \wedge \sigma(n)} \vert \leq K_n for all tt.

Ito's Process

An Ito's process is a continuous adapted process X=(Xt)X = (X_t) of the form

Xt=X0+∫0tYsdWs+∫0tbsds,0≤t≤TX_t = X_0 + \int_0^t Y_s dW_s + \int_0^t b_s ds, 0 \leq t \leq T

where Y∈LlocY \in \mathcal L_{loc} and b=(bs)b = (b_s) is progressive with ∫0T∣bs∣ds<∞\int_0^T \vert b_s\vert ds < \infty.

Revisiting the notion of quadratic variation, the quadratic variation of XX is formally defined as ⟨X⟩\langle X\rangle -

⟨X⟩t=∫0tYs2ds,0≤t≤T\langle X\rangle_t = \int_0^t Y_s^2 ds, 0 \leq t \leq T

For a non-random continuous function gg, the process Gt=∫0tg(s)dsG_t = \int_0^t g(s)ds has 00 quadratic variation.

Theorem (Ito Formula II) Let f:R×[0,T]f: \mathbb R \times [0, T] be of class C2,1C^{2, 1}. If XX is an Ito's process as defined above, then so is f(Xt,t)f(X_t, t) with

f(Xt,t)=f(X0,0)+∫0tfx(Xs,s)dXs+∫0tft(Xs,s)ds+12∫0tfxx(Xs,s)d⟨X⟩sf(X_t,t) = f(X_0, 0) + \int_0^t f_x(X_s, s)dX_s + \int_0^t f_t(X_s, s)ds + \frac{1}{2} \int_0^t f_{xx} (X_s, s) d\langle X\rangle_s

where dXs=YsdWs+BsdsdX_s = Y_s dW_s + B_s ds and d⟨X⟩s=Ys2dsd\langle X\rangle_s = Y^2_s ds.

Example

Consider the case of Stochastic exponential,

For a Y∈LlocY \in \mathcal L_{loc} we define Mt=∫0tYsdWsM_t = \int_0^t Y_s dW_s (local martingale). Define

Zt=exp⁡(Mt−12⟨M⟩t),0≤t≤TZ_t = \exp(M_t - \frac{1}{2} \langle M\rangle_t), 0\leq t\leq T

By Ito's formula for f(x)=exf(x) = e^x and Ito's process Xt=Mt−12⟨M⟩tX_t = M_t - \frac{1}{2} \langle M\rangle_t, we have

Zt=1+∫0tzsdMs≡dZt=ZtdMt  &  Z0=1Z_t = 1 + \int_0^t z_s dM_s \equiv dZ_t = Z_t dM_t \;\&\; Z_0 = 1

Again, ZZ is a local martingale.

  • Since (Zt)(Z_t) is a strictly positive local martingale, it's a supermartingale (Fatou's Lemma)

  • If ∣Ys(w)∣≤K\vert Y_s(w)\vert \leq K for all s,ws, w, then Z⋅Y∈LZ\cdot Y \in \mathcal L and ZZ is a true martingale.

Black-Scholes Model

As usual, consider two assets to be traded given by

  • Bond
Bt=ert,r>0dBt=rBtdt,B0=1\begin{align*} B_t &= e^{rt}, r > 0 \\ dB_t &= rB_t dt, B_0 = 1 \end{align*}
  • Stock
St=S0exp⁡(σWt+(μ−σ2/2)t)dSt=St(σdWt+μdt)From Ito’s formula\begin{align*} S_t &= S_0 \exp(\sigma W_t + (\mu - \sigma^2/2)t)\\ dS_t &= S_t (\sigma dW_t + \mu dt) \quad \text{From Ito's formula} \end{align*}

Note that σWt+μt\sigma W_t + \mu t is a Brownian motion with volatility σ>0\sigma > 0 and drift μ∈R\mu \in \mathbb R

A trading strategy is a pair ϕ=(α,β)\phi = (\alpha, \beta) of progressive processes (αt)(\alpha_t), (βt)(\beta_t) such that ∫0Tαs2ds<∞\int_0^T \alpha_s^2 ds < \infty and ∫0T∣βs∣ds<∞\int_0^T \vert \beta_s\vert ds < \infty (they are progressively measurable). α\alpha represents the holdings in stock and β\beta represents the holdings in bond. At any time tt,

Vt(ϕ)=αtSt+βtBtV_t(\phi) = \alpha_t S_t + \beta_t B_t

For a trading strategy to be self-financing,

Vt(ϕ)−V0(ϕ)=∫0tαudSu+∫0tβudBu,0≤t≤TV_t(\phi) - V_0(\phi) = \int_0^t \alpha_u dS_u + \int_0^t \beta_u dB_u, 0 \leq t\leq T

where dSu=Su(σdWu+μdu)dS_u = S_u(\sigma dW_u + \mu du) and dBu=rerududB_u = re^{ru}du. In particular, Vt(ϕ)V_t(\phi) is an Ito process.

The importance of α\alpha being progressively measurable comes in handy here, since we want σαSu∈Lloc\sigma \alpha S_u \in \mathcal L_{loc} . That is because SuS_u is positive, ∫0Tαu2Su2du<∞  ⟺  ∫0Tαu2du<∞\int_0^T \alpha_u^2 S_u^2 du < \infty \iff \int_0^T \alpha_u^2 du < \infty

Once again, we define the discounted processes

Vt∗(ϕ)=Vt(ϕ)/Bt=Vte−rtSt∗(ϕ)=St(ϕ)/Bt=Ste−rt\begin{align*} V_t^*(\phi) &= V_t(\phi) /B_t = V_t e^{-rt} \\ S_t^*(\phi) &= S_t(\phi) /B_t = S_t e^{-rt} \\ \end{align*}

and consequently,

Vt∗(ϕ)−V0(ϕ)=∫0tαudSu∗V_t^*(\phi) - V_0(\phi) = \int_0^t \alpha_u dS^*_u

Product Rule for Ito's Calculus

Suppose we have two Ito's processes Xt=X0+∫0tHsdWs+∫0TbsdsX_t = X_0 + \int_0^t H_s dWs + \int_0^T bs_ds + Yt=Y0+∫0tKsdWs+∫0TcsdsY_t = Y_0 + \int_0^t K_s dWs + \int_0^T cs_ds, then their product is also an Ito's process

d(XtYt)=XtdYt+YtdXt+d⟨X,Y⟩td(X_tY_t) = X_t dY_t + Y_t dX_t + d\langle X,Y\rangle_t

where ⟨X,Y⟩t=∫0tHsKsds\langle X, Y \rangle_t = \int_0^t H_sK_s ds is the quadratic covariation.

Arbitrage

An arbitrage is a self-financing trading strategy ϕ\phi is such that V0(ϕ)=0V_0(\phi) = 0 and VT(ϕ)≥0V_T(\phi) \geq 0 with P(VT(ϕ)>0)>0P(V_T(\phi) > 0)> 0.

The binomial model is viable and complete. However, it is easy to create arbitrage opportunities in Black-Scholes model -

It=∫0t1T−sdWs,0≤t<T⟨I⟩t=log⁡TT−t↗∞ as t↗T\begin{align*} I_t &= \int_0^t \frac{1}{\sqrt{T - s}} dW_s, 0\leq t < T \\ \langle I \rangle_t &= \log\frac{T}{T - t} \nearrow \infty \text{ as } t \nearrow T \end{align*}

The quadratic variation represents the energy of the martingale in a way. We further define τq=inf⁡(t:It=a)∧T\tau_q = \inf (t: I_t = a) \wedge T for some a>0a > 0 (P(τa<T)=1P(\tau_a < T) = 1). Now, the trading strategy is given by

αt=1t≤T1T−t1Stβt=It∧τa−αtSt\begin{align*} \alpha_t &= \mathbb 1_{t \leq T} \frac{1}{\sqrt{T - t}} \frac{1}{S_t} \\ \beta_t &= I_{t \wedge \tau_a} - \alpha_t S_t \end{align*}

Let r=0,Bt=1r = 0, B_t = 1 for simplicity. Then,

Vt(ϕ)=It∧τa\begin{align*} V_t(\phi) = I_{t \wedge \tau_a} \end{align*}

implying V0(ϕ)=0,VT(ϕ)=Iτa=a>0V_0(\phi) = 0, V_T(\phi) = I_{\tau_a} = a > 0.

Equivalent Martingale Measure

When μ=0\mu = 0, we have St=S0eσWt−σ2t/2S_t = S_0 e^{\sigma W_t - \sigma^2t/2} is a martingale. Then, PP itself is an EMM. When μ≠0\mu \neq 0, we need to find a QQ on (Ω,FT)(\Omega, \mathcal F_T) that is equivalent to PP. How do we check equivalence here? Q(A)>0  ⟺  P(A)>0Q(A) > 0 \iff P(A)> 0 for all A∈FtA \in \mathcal F_t.

Girsanov's Theorem

Let H∈LlocH \in \mathcal L_{loc} and Mt=∫0tHsdWsM_t = \int_0^t H_s dW_s and ⟨M⟩t=∫0tHs2ds\langle M \rangle_t = \int_0^t H_s^2 ds, then

Λt:=exp⁡(Mt−12⟨M⟩t)=1+∫0tΛsdMs\Lambda_t := \exp(M_t - \frac{1}{2} \langle M\rangle_t) = 1 + \int_0^t \Lambda_s dM_s

is continuous in tt , positive, local martingale and a supermartingale. We have E(Λt)≤1\mathbb E(\Lambda_t) \leq 1, t∈[0,T]t \in [0, T]. The process is a martingale when E(Λt)=1\mathbb E(\Lambda_t) = 1.

Now, the theorem states that suppose E(Λt)=1\mathbb E(\Lambda_t) = 1 for all t∈[0,T]t \in [0, T] (  ⟺  E(ΛT)=1  ⟺  Λt\iff \mathbb E(\Lambda_T) = 1 \iff \Lambda_t is a martingale ). We define

Q(A):=EP(1AΛT),A∈FTQ(A) := \mathbb E^P (\mathbb 1_A \Lambda_T), A \in \mathcal F_T

then Q∼PQ \sim P and

Nt:=∫0tYsdWsN_t := \int_0^t Y_s dW_s

for Y∈LlocY \in \mathcal L_{loc} is a local martingale with respect to PP and

N~t=Nt−∫0tYsHsds\tilde N_t = N_t - \int_0^t Y_s H_s ds

is a local martingale with respect to QQ.

Finally, ⟨N⟩t=⟨N~⟩t\langle N \rangle_t = \langle \tilde N\rangle_t. In particular, consider

W~t=Wt−∫0tHsds\tilde W_t = W_t - \int_0^t H_s ds

then W~s\tilde W_s is a local martingale with respect to QQ and ⟨W~⟩=t\langle \tilde W \rangle = t. It can be shown that any process with quadratic variation equating to tt is a Brownian motion!

Using this theorem, we try and define a martingale for St∗S_t^*

St∗=S0exp⁡(σWt+(μ−r−σ2/2)t)S_t^* = S_0 \exp (\sigma W_t + (\mu - r - \sigma^2/2)t)

Let Ht=θ=μ−rσH_t = \theta = \frac{\mu - r}{\sigma}, then S∗S^* is a martingale with respect to QQ.

By Ito's formula,

dSt∗St∗=σ{dWt+μ−rσ⏟θdt}\frac{dS^*_t}{S_t^*} = \sigma\left\{dW_t + \underbrace{\frac{\mu - r}{\sigma}}_{\theta} dt \right\}

Using Girsanov's theorem with Hs=−θH_s = - \theta, P∗=ΛtPP^* = \Lambda_t P is an EMM.

St∗=S0+∫0tSu∗dW~uS_t^* = S_0 + \int_0^t S_u^* d \tilde W_u

so St∗S_t^* is a local martingale. Furthermore,

E∗((St∗)2)=E∗(exp⁡(2σW~t−σ2t))S02.=e4σ2t/2−σ2tS02=S02eσ2t\begin{align*} \mathbb E^*((S_t^*)^2) &= \mathbb E^*(\exp(2\sigma \tilde W_t - \sigma^2 t))S_0^2. \\ &= e^{4\sigma^2 t/2 - \sigma^2 t}S_0^2 = S_0^2 e^{\sigma^2t} \end{align*}

Consequently,

E∗[∫0T(St∗)2dt]=S02∫0Teσ2tdt=S02eσ2T−1σ2<∞\mathbb E^*[\int_0^T (S_t^*)^2 dt] = S_0^2 \int_0^T e^{\sigma^2 t}dt = S_0^2 \frac{e^{\sigma^2T} - 1} {\sigma^2} < \infty

implying that St∗∈Lloc(P∗)S_t^* \in \mathcal L_{loc}(P^*) and (St∗)(S_t^*) is a P∗P^*-martingale.

Corollary If ϕ\phi is a self-financing trading strategy, then Vt∗(ϕ)V_t^*(\phi) is a p∗p^*-local martingale.

Proof.

Vt∗(ϕ)=V0(ϕ)+∫0tαudSu∗=V0(ϕ)+∫0t(αuSu∗)dW~u\begin{align*} V_t^*(\phi) &= V_0(\phi) + \int_0^t \alpha_u dS_u^* \\ &= V_0(\phi) + \int_0^t (\alpha_u S_u^*) d\tilde W_u \end{align*}

and αuSu∗∈Lloc(P∗)\alpha_uS_u^* \in \mathcal L_{loc}(P^*) by path continuity.

Definition. A self-financing trading strategy is admissible provided Vt∗(ϕ)V_t^*(\phi) is a P∗P^*-martingale (not just a local martingale).

Also, if E∗[∫0Tαs2(Ss∗)2ds]<∞\mathbb E^*[\int_0^T \alpha_s^2 (S_s^*)^2ds] < \infty, then ϕ\phi is admissible.

Martingale Representation Theorem

If (Mt)(M_t) is a P∗P^*-martingale (such that t→Mtt \to M_t is right continuous) then there exists a unique η=(ηt)0≤t≤T∈Lloc\eta = (\eta_t)_{0 \leq t \leq T} \in \mathcal L_{loc} such that

Mt=M0+∫0tηsdW~s,0≤t≤TM_t = M_0 + \int_0^t \eta_s d\tilde W_s, 0 \leq t \leq T

Corollary. Given X∈L1(FT,P∗)X \in L^1(\mathcal F_T, P^*), there is a unique admissible ϕ\phi such that E∗(X∗∣Ft)=Vt∗(ϕ)\mathbb E^*(X^* \vert \mathcal F_t) = V_t^*(\phi) for 0≤t≤T0 \leq t\leq T (X∗=e−rTXX^* = e^{-rT} X).

Proof. Define a right continuous P∗P^*-martingale MM by

Mt:=E∗(X∗∣Ft)M_t := \mathbb E^*(X^* \vert \mathcal F_t)

By the martingale representation theorem, there exists a unique η\eta such that

Mt=M0+∫0tηudW~u=M0+∫0tαudSu∗M_t = M_0 + \int_0^t \eta_u d \tilde W_u = M_0 + \int_0^t \alpha_u dS_u^*

where αu:=ηu/Su∗∈Lloc\alpha_u := \eta_u/S_u^* \in \mathcal L_{loc}. Define βt:=Mt−αtSt∗\beta_t := M_t - \alpha_t S_t^* which is continuous. Then, ϕ:=(αt,βt)\phi := (\alpha_t, \beta_t) is a self-financing trading strategy and

Vt∗(ϕ)=αtSt∗+βt=MtV_t^*(\phi) = \alpha_t S_t^* + \beta_t = M_t

is a martingale and hence, admissible. Also,

V0(ϕ)=M0=E∗[X∗∣F0]=E∗[X∗]V_0(\phi) = M_0 = \mathbb E^*[X^* \vert \mathcal F_0] = \mathbb E^*[X^*]

Therefore, each X∈L1(P∗)X \in L^1(P^*) is replicable!

Pricing a European Contingent Claim

Theorem. The unique no arbitrage price of the European claim X∈L1(P∗)X \in L^1 (P^*) is Vt(ϕ)=e−r(T−t)E∗(X∣Ft)V_t(\phi) = e^{-r(T - t)} \mathbb E^*(X \vert \mathcal F_t).

The European call option is given by X=(ST−K)∗X = (S_T - K)^*

Let us compute E∗(X∗∣Ft)\mathbb E^*(X^* \vert \mathcal F_t) - Firstly, we have

ST∗=St∗exp⁡(σ(W~T−W~t)−σ22(T−t))S_T^* = S_t^* \exp(\sigma(\tilde W_T - \tilde W_t) - \frac{\sigma^2}{2} (T - t))

and W~T−W~t∼N(0,T−t)⊥ ⁣ ⁣ ⁣⊥p∗Ft\tilde W_T - \tilde W_t \sim \mathcal N(0, T - t) \perp\!\!\!\perp_{p^*} \mathcal F_t . Renaming W~T−W~t\tilde W_T - \tilde W_t as T−tZ\sqrt{T - t} Z where Z∼N(0,1)Z \sim \mathcal N(0, 1) -

E∗((ST∗−K∗)+∣Ft)=E[(St∗exp⁡(σZT−t−σ22(T−t))−K∗)+]E∗[g(ST∗)∣Ft]=E[g(xeαz−α2/2)∣Ft]\begin{align*} \mathbb E^*((S^*_T - K^*)^+ \vert \mathcal F_t) &= E[(S_t^*\exp(\sigma Z\sqrt{T - t} - \frac{\sigma^2}{2}(T - t)) - K^*)^+] \\ \mathbb E^*[g(S_T^*) \vert \mathcal F_t]&= \mathbb E[g(xe^{\alpha z - \alpha^2/2}) \vert \mathcal F_t] \end{align*}

where x=St∗x = S_t^* and α=σT−t\alpha = \sigma \sqrt{T - t}. Let us work with this expression. Letting ϕ(z)=12πe−z2/2\phi(z) = \frac{1}{\sqrt{2\pi}}e^{-z^2/2} we get E∗[g(ST∗)∣Ft]=f(St∗,t)\mathbb E^*[g(S^*_T) \vert \mathcal F_t] = f(S_t^*, t) where

f(x,t)=∫−∞∞g(xeαz−α2/2)ϕ(z)dzf(x, t) = \int_{-\infty}^{\infty} g(xe^{\alpha z - \alpha^2/2})\phi(z) dz

Now,

g(xeαz−α2/2)={xeαz−α2/2−K∗z>l0z≤lg(xe^{\alpha z - \alpha^2/2}) = \begin{cases} xe^{\alpha z - \alpha^2/2} - K^* & z > l \\ 0 & z \leq l \end{cases}

where l=Lα+α2l = \frac{L}{\alpha} + \frac{\alpha}{2}. and L=log⁡(K∗/x)L = \log(K^*/x). Then, we get f(x,t)=(1)+(2)f(x, t) = (1) + (2) where

(1)=x∫l∞eαz−α2/2ϕ(z)dz=x∫l~∞12πe−y2/2dy=xΦ(α2−Lα)(2)=−K∗∫l∞ϕ(z)dz=−K∗Φ(−l)\begin{align*} (1) &= x\int_l^\infty e^{\alpha z - \alpha^2 /2} \phi(z) dz \\ &= x \int_{\tilde l} ^\infty \frac{1}{\sqrt{2\pi}} e^{-y^2/2}dy = x \Phi(\frac{\alpha}{2} - \frac{L}{\alpha}) \\ (2) &= -K^*\int_l^\infty \phi(z) dz \\ &= -K^* \Phi(-l) \end{align*}

Finally, we get

C∗(t)=f(x,t)=xΦ(log⁡(x/K∗)σT−t+σT−t2)−K∗Φ(log⁡(x/K∗)σT−t−σT−t2)C^*(t) = f(x, t) = x \Phi\left(\frac{\log{(x/K^*)}}{\sigma \sqrt{T - t}} + \frac{\sigma \sqrt{T - t}}{2}\right) - K^* \Phi\left(\frac{\log(x/K^*)}{\sigma \sqrt{T - t}} - \frac{\sigma \sqrt{T - t}}{2}\right)

Let us now try to replicate the option using a trading strategy -

We have seen that

Vt(ϕ)=ertf(St∗,t)V_t(\phi) = e^{rt} f(S_t^*, t)

We have

f(x,t)=xΦ(log⁡(x/K∗)σT−t+σT−t2)−K∗Φ(log⁡(x/K∗)σT−t−σT−t2)f(x, t) = x \Phi\left(\frac{\log{(x/K^*)}}{\sigma \sqrt{T - t}} + \frac{\sigma \sqrt{T - t}}{2}\right) - K^* \Phi\left(\frac{\log(x/K^*)}{\sigma \sqrt{T - t}} - \frac{\sigma \sqrt{T - t}}{2}\right)

where X=(St−K)+,K∗=Ke−rT,Φ=X = (S_t - K)^+, K^* = Ke^{-rT}, \Phi = standard normal CDF. Here, ff is smooth in the sense that it is twice differentiable in St∗S_t^* and differentiable in tt. Using Ito's formula, since Vt∗(ϕ)V_t^*(\phi) is a martingale,

Vt∗(ϕ)=V0∗(ϕ)+∫0t∂f∂x(Su∗,u)⏟αudSu∗+⋯V_t^*(\phi) = V_0^*(\phi) + \int_0^t \underbrace{\frac{\partial f}{\partial x} (S_u^*, u)}_{\alpha_u} dS_u^* + \cdots

The remaining integral terms cancel out to 00

Why?

Consequently,

αt=∂f∂x(St∗,t)=Φ(log⁡(St∗/K∗)σT−t+σT−t2)βt=Vt∗(ϕ)−αtSt∗\begin{align*} \alpha_t &= \frac{\partial f}{\partial x} (S_t^*, t) = \Phi\left(\frac{\log(S_t^*/K^*)}{\sigma \sqrt{T - t}} + \frac{\sigma \sqrt{T - t}}{2}\right) \\ \beta_t &= V_t^*(\phi) - \alpha_t S_t^* \end{align*}

We can check that when ST>K,αt→1S_T > K, \alpha_t \to 1, and ST<K,αt→0S_T < K, \alpha_t \to 0.

Let X∗=g(ST∗)X^* = g(S_T^*), then, VT∗(ϕ)=f(St∗,t)=E∗(g(St∗)∣Ft)V_T^*(\phi) = f(S_t^*, t) = \mathbb E^*(g(S_t^*) \vert \mathcal F_t) and lim⁡t↑Tf(x,t)=g(x)\lim_{t \uparrow T} f(x, t) = g(x)

Also,

∂f∂t+12σ2x2∂2f∂x2≡0\frac{\partial f}{\partial t} + \frac{1}{2} \sigma^2 x^2 \frac{\partial^2 f}{\partial x^2} \equiv 0

for (x,t)∈(0,∞)×[0,T)(x, t) \in (0, \infty) \times [0, T). This PDE has a unique solutions when gg has certain properties.

Black-Scholes Formula

Consider the market with the assets St=(St0,St1,…,Std)S_t = (S_t^0, S_t^1, \dots, S_t^d) where St0=Bt=e∫0trsdsS_t^0 = B_t = e^{\int_0^t r_s ds} and (rs)0≤s≤T(r_s)_{0 \leq s \leq T} is a progressive process (rs≥0,∫0trsds<∞r_s \geq 0, \int_0^t r_s ds < \infty). This formulation maintains dSt0St0=rtdt\frac{d S_t^0}{S_t^0} = r_t dt.

  • The stocks have drifts μt1,…,μtd\mu_t^1 , \dots, \mu_t^d which are progressive processes and ∫0T∣μti∣dt<∞\int_0^T \vert \mu_t^i \vert dt < \infty for all ii.

  • The volatilities are represented as a matrix σT=[σ1σ2…σd]\sigma^T = \begin{bmatrix} \sigma^1 &\sigma^2 & \dots & \sigma^d \end{bmatrix} (d×nd \times n) where σtij\sigma_t^{ij} is progressive and ∫0T∥σti∥2dt<∞\int_0^T \|\sigma_t^i\|^2 dt < \infty.

  • W=(W1,…,Wn)W = (W^1, \dots, W^n) are independent 1-dimensional standard Brownian Motions.

Using these elements, the price process is defined as

Sti=S0iexp⁡(∫0tσui⋅dWu−12∫0t∣σui∣2du+∫0tμuidu)S_t^i = S_0^i \exp\left(\int_0^t \sigma_u^i \cdot dW_u - \frac{1}{2} \int_0^t \vert \sigma_u^i \vert ^2 du + \int_0^t \mu_u^i du\right)

Then, the using the basic definitions, a trading strategy is ϕt=(ϕt0,…,ϕtd)\phi_t = (\phi_t^0, \dots, \phi_t^d) and

Vt(ϕ)=ϕt⋅St=V0(ϕ)+∫0tϕu⋅dSu∵self financing≡Vt∗(ϕ)−V0(ϕ)=∑i=1d∫0tϕuidSui,∗\begin{align*} V_t(\phi) &= \phi_t \cdot S_t\\ &= V_0(\phi) + \int_0^t \phi_u \cdot dS_u \quad \because \text{self financing} \\ \equiv V_t^*(\phi) - V_0(\phi) &= \sum_{i = 1}^d \int_0^t \phi_u^i dS_u^{i, *} \end{align*}

Also, given any strategy ϕ~=(ϕ~t1,…,ϕ~td)\tilde \phi = (\tilde \phi_t^1, \dots, \tilde \phi_t^d) which is progressive and integrable and c∈Rc \in \mathbb R, representing the initial holdings, there exists ϕt0\phi_t^0 such that

ϕt=(ϕt0,ϕ~t1,…,ϕ~td)\phi_t = (\phi_t^0, \tilde \phi_t^1, \dots, \tilde \phi_t^d) is a self-financing trading strategy V0(ϕ)=cV_0(\phi) = c.


A tame trading strategy ϕ\phi there exists c>0c > 0 , Vt∗(ϕ)≥−cV_t^*(\phi) \geq -c, ∀t∈[0,T]\forall t \in [0, T].

Consequently, Vt∗(ϕ)+c  ⟹  Vt∗(ϕ)V_t^*(\phi) + c \implies V_t^*(\phi) is a QQ-supermartingale.

Proposition - If there is an ELMM QQ then no tame trading strategy can be an arbitrage. Proof. Suppose you have a tame trading strategy with VT∗(ϕ)≥0V_T^*(\phi) \geq 0 and V0∗(ϕ)=0V_0^*(\phi) = 0. Due to supermartingale property,

EQ(VT∗(ϕ))≤EQ(V0∗(ϕ))=V0(ϕ)=0\begin{align*} E^Q(V_T^*(\phi)) &\leq \mathbb E^Q(V_0^*(\phi)) \\ &= V_0(\phi) = 0 \end{align*}

Theorem. (FFTAP)

  • No free lunch with vanishing risk  ⟺  \iff there exists an ELMM

ELMM via Girsanov’s theorem - For any Q∼PQ \sim P there exists a Λt\Lambda_t, such that dQdP∣Ft=Λt\left. \frac{dQ}{dP} \right\vert_{\mathcal F_t} = \Lambda_t. This is equivalent to saying

Q(A)=EP(1AΛt)Q(A) = \mathbb E^P(\mathbb 1_A \Lambda_t)

For all A∈FtA \in \mathcal F_t. Here, Λt\Lambda_t is a PP-martingale. We want the form Λt=exp⁡(Mt−12⟨M⟩t))\Lambda_t = \exp(M_t - \frac{1}{2}\langle M\rangle_t)) for the Girsanov’s theorem. Consider the following - Using Ito’s theorem

d(log⁡(Λt))=1ΛtdΛt−121Λt2d⟨Λ⟩td(\log(\Lambda_t)) = \frac{1}{\Lambda_t}{d \Lambda_t} - \frac{1}{2} \frac{1}{\Lambda_t^2} d\langle\Lambda\rangle_t

Resembling that dMt=1ΛtdΛtdM_t = \frac{1}{\Lambda_t} d\Lambda_t.

Using Martingale representation theorem, Mt=−∫0tθs⋅dWsM_t = - \int_0^t \theta_s \cdot dW_s. With Q=exp⁡(MT−12⟨M⟩T)PQ = \exp(M_T - \frac{1}{2} \langle M\rangle_T)P, Girsanov’s theorem gives us

W~t=Wt+∫0tθsds\tilde W_t = W_t + \int_0^t \theta_s ds

is a QQ-Brownian motion. Going back to our Black-Scholes model,

Sti,∗=S0iexp⁡(∫0tσsi⋅dWs−12∫0t∣σsi∣2ds+∫0t(μsi−rs)ds)S_t^{i,*} = S_0^i \exp(\int_0^t \sigma_s^i\cdot dW_s - \frac{1}{2} \int_0^t \vert \sigma_s^i\vert^2 ds + \int_0^t (\mu_s^i - r_s) ds)

We want to choose θ\theta such that W~s\tilde W_s captures the drift term. That is, let −σsi⋅θs+μsi−rs=0-\sigma_s^i \cdot \theta_s + \mu_s^i - r_s = 0, then

Sti,∗=S0iexp⁡(∫0tσsi⋅dW~s−12∫0t∣σsi∣2ds)S^{i,*}_t = S_0^i \exp(\int_0^t \sigma_s^i \cdot d \tilde W_s - \frac{1}{2} \int_0^t \vert \sigma_s^i \vert^2 ds)

Simply put, we are trying to find a QQ-martingale given σsi,μsi,rs\sigma_s^i, \mu_s^i, r_s for all ω,s\omega, s in terms of θs\theta_s. In other words, we are checking if μs−rs1\mu_s - r_s {\bf 1} is in the column span of σs\sigma_s. We are working our way backwards to find an ELMM QQ using the Black-Scholes model. Using θs\theta_s, we get (Λ)t(\Lambda)_t. But when is this a PP-martingale? A super-martingale with same initial and final expectations is a martingale.

Novikov’s condition - If Ep[e12⟨M⟩T]<∞\mathbb E^p[e^{\frac{1}{2} \langle M\rangle_T}] < \infty, then Λt=exp⁡(Mt−12⟨M⟩t)\Lambda_t = \exp(M_t - \frac{1}{2} \langle M\rangle_t) for all 0≤t≤T0 \leq t \leq T is a PP-martingale.

Theorem. If n=dn = d, and σs\sigma_s is invertible for all ω,s\omega, s and Ep[e12∫0T∣θs∣2ds]<∞\mathbb E^p [e^{\frac{1}{2} \int_0^T \vert \theta_s\vert^2 ds}] < \infty, then QQ is an ELMM.

With that, we now state the Second Fundamental Theorem of Asset Pricing for continuous models -

Theorem (SFTAP). Suppose there exists an ELMM P∗P^*. Then, the market is complete if the ELMM is unique. By complete, we mean that for all contingent claims X∗∈L1(P∗)X^* \in L^1 (P^*), there is a self-financing trading strategy such that X=VT(ϕ)X = V_T(\phi).

Proof.

(  ⟹  )(\implies) Let QQ be another ELMM, then for A∈FTA \in \mathcal F_T, we define 1A=X∗∈L1(P∗)\mathbb 1_A = X^* \in L^1(P^*). That is, X=ST01AX = S_T^0 \mathbb 1_A (let S00=1S_0^0 = 1).

X∗=VT∗(ϕ)E∗(VT∗(ϕ))=E∗(V0∗(ϕ))=V0(ϕ)=P∗(A)\begin{align*} X^* &= V_T^*(\phi) \\ \mathbb E^*(V_T^*(\phi)) &= \mathbb E^*(V_0^*(\phi)) = V_0(\phi) = P^*(A) \end{align*}

Also, Vt∗(ϕ),0≤t≤TV_t^*(\phi), 0\leq t\leq T is a QQ-local martingale

Vt∗=V0(ϕ)+∫0tϕsdSs∗V_t^* = V_0(\phi) + \int_0^t \phi_s dS_s^*

Since the claim is replicable and X∗≥0X^* \geq 0, Vt∗(ϕ)=E∗(X∗∣Ft)≥0V_t^*(\phi) = \mathbb E^*(X^* \vert \mathcal F_t) \geq 0. Since Vt∗(ϕ)V_t^*(\phi) is a QQ-local martingale, it is also a QQ-supermartingale -

P∗(A)=V0(ϕ)≥EQ(VT∗(ϕ))=EQ(X∗)=Q(A)P^*(A) =V_0(\phi) \geq \mathbb E^Q(V_T^*(\phi)) = \mathbb E^Q(X^*) = Q(A)

We have P∗(A)≥Q(A)P^*(A) \geq Q(A) for all A∈FTA \in \mathcal F_T. Also, 1−P∗(A)≥1−Q(A)1 - P^*(A) \geq 1 - Q(A) as well, implying P∗=QP^* = Q.

(⟸)(\Longleftarrow) Assume P∗P^* is the only ELMM. Let X∗∈L1(P∗)X^* \in L^1(P^*). The martingale

Mt=E∗(X∗∣FT),0≤t≤TM_t = \mathbb E^*(X^* \vert \mathcal F_T), 0 \leq t \leq T

is right-continuous (continuous). From the martingale representation theorem, there is a unique progressive process ηu=(ηu1,…,ηud)\eta_u = (\eta_u^1, \dots, \eta_u^d) such that

Mt=M0+∫0tηu⋅dSu∗M_t = M_0 + \int_0^t \eta_u \cdot dS_u^*

Now, correspondingly, there exists a unique self-financing trading strategy ϕ=(ϕt0,ηt1,…,ηtd)\phi = (\phi_t^0, \eta_t^1, \dots, \eta_t^d) such that V0(ϕ)=M0V_0(\phi) = M_0 and

X∗=V0(ϕ)+∫0Tϕu⋅dSu∗X^* = V_0(\phi) + \int_0^T \phi_u \cdot dS_u^*

How does the proof conclude here?


From Girsanov's theorem, we have seen that

Sti,∗=S0iexp⁡(∫0tσsi⋅dW~+∫0t∣σsi∣2ds+∫0tμsidst)S_t^{i,*} = S_0^i \exp\left(\int_0^t \sigma_s^i \cdot d\tilde W_ + \int_0^t \vert \sigma_s^i \vert^2 ds + \int_0^t \mu_s^i dst\right)

where σsi\sigma_s^i is the iith row of σs\sigma_s which itself is a (d×n)(d \times n) matrix. We typically assume d≤nd \leq n meaning that we want enough randomness to represent the stocks.

Then, we have the following result -

Lemma. The market is complete iff d=nd = n and σ(w)\sigma(w) is invertible

Proof.

(⟸)(\Longleftarrow) This is trivial to show

(  ⟹  )(\implies) Let kt(w)∈n(σt(w))k_t(w) \in \mathcal n(\sigma_t(w)) where n\mathcal n represents the nullspace operation. Also, let ∥kt(w)∥≤1\|k_t(w) \|\leq 1 and ∥kt(w)∥>0\|k_t(w) \| > 0 if n(σt(w))≠{0}\mathcal n(\sigma_t(w)) \neq \{0\}. Then, let

X=ST0(1+∫0Tku⋅dW~u)X = S_T^0 (1 + \int_0^T k_u \cdot d \tilde W_u)

Since the second term is always a local martingale, and ∥k∥≤1\|k\|\leq 1, it's norm is finite and therefore is a martingale. Also, E∗(X∗)=1\mathbb E^*(X^*) = 1.

The market being complete means, X∗=1+∫0Tϕu⋅dSu∗X^* = 1 + \int_0^T \phi_u \cdot dS_u^* for some trading strategy ϕ\phi. Equating both the equations, we get

∫0T(kt−∑i=1dϕtiSti,∗σti)⋅dW~t=0∫0T∣kt−∑i=1dϕtiSti,∗σti∣2=0kt=∑i=1dϕtiSti,∗σti\begin{align*} \int_0^T \left(k_t - \sum_{i = 1}^d \phi_t^i S_t^{i, *} \sigma_t^i\right) \cdot d\tilde W_t &= 0 \\ \int_0^T \left\vert k_t - \sum_{i = 1}^d \phi_t^i S_t^{i, *} \sigma_t^i\right\vert^2 &= 0 \\ k_t = \sum_{i = 1}^d \phi_t^i S_t^{i, ^*} \sigma_t^i \end{align*}

implying that the vector kt(w)k_t(w) is in the column space of σt(w)T\sigma_t(w) ^T. For a matrix BB, the orthogonal subspace of R(BT)\mathcal R(B^{T}) denoted by [R(BT)]⊥=n(B)[\mathcal R(B^T)]^{\perp} = \mathcal n(B). Therefore, kt(w)⊥n(σt(w))k_t(w) \perp \mathcal n(\sigma_t(w)) implying that kt(w)=0k_t(w) = 0!

How do we conclude n=dn = d? If d<nd < n, then the matrix would not be full rank, implying d=nd = n.


Theorem If a market is viable and complete, then the no-arbitrage price of XX (assuming XX is integrable ∈L1(P∗)\in L^1(P^*)) at time tt is

Vt(ϕ)=St0E∗(X∗∣Ft)V_t(\phi) = S_t^0 \mathbb E^*(X^* \vert \mathcal F_t)

The proof is similar to the ones we have shown before, and this theorem completes the whole picture of a continuous market.


Consider the Black-Scholes model where we had

X∗=g(ST∗)X^* = g(S_T^*)

We had calculated E∗(g(ST∗)∣Ft)=f(St∗,t)\mathbb E^*(g(S_T^*) \vert \mathcal F_t) = f(S_t^*, t) where

f(x,t)=∫−∞∞g(xexp⁡(αz−α2/2)),x>0=∫0∞g(y)ϕ(log⁡(y/s)α+α2)dyαy\begin{align*} f(x, t) &= \int_{-\infty}^\infty g(x\exp(\alpha z - \alpha^2/2)), x> 0\\ &= \int_0^\infty g(y) \phi\left(\frac{\log(y/s)}{\alpha} + \frac{\alpha}{2}\right) \frac{dy}{\alpha y} \end{align*}

where α=σT−t\alpha = \sigma \sqrt{T - t} and ϕ(z)=12πe−z2/2\phi(z) = \frac{1}{\sqrt{2\pi}} e^{-z^2/2}. Assume ∣g(y)∣≤C∣y∣p\vert g(y) \vert \leq C \vert y\vert^p for some p,C>0p, C > 0. That is, we want the contingent claim to have a polynomial growth.

We have (x,t)∈(0,∞)×[0,T)(x, t) \in (0, \infty) \times [0, T), and letting y=xexp⁡(αz−α2/2)y = x\exp(\alpha z - \alpha^2/2) we get

∂f∂t+12σ2x2∂2f∂x2=0\frac{\partial f}{\partial t} + \frac{1}{2} \sigma^2 x^2 \frac{\partial^2 f}{\partial x^2} = 0

Let us delve into an example to understand this better - consider the calculation for E[WT4∣Ft]\mathbb E[W_T^4 \vert \mathcal F_t] Note that this process is a martingale. Now, we present two methods to perform this calculation -

Method 1

We say that this expression is equivalent to evaluating a martingale function - =f(Wt,t)= f(W_t, t). We need the property =f(Wt,t)=E[WT4∣Ft]= f(W_t, t) = \mathbb E[W_T^4 \vert \mathcal F_t], and we choose f(x,t)=x4+6x2(T−t)+3(T−t)2f(x, t) = x^4 + 6x^2(T - t) + 3(T - t)^2 so that it has the following property -

fx=4x2+12x(T−t)fxx=12x2+12(T−t)ft=−6x2−6(T−t)\begin{align*} f_x &= 4x^2 + 12x(T - t) \\ f_{xx} &= 12x^2 + 12(T - t) \\ f_t &= -6x^2 - 6(T - t) \end{align*}

We observe that ft+12fxx=0f_t + \frac{1}{2} f_{xx} = 0. From Ito's formula

f(Wt,t)=f(0,0)+∫0tfxdWs+∫0tftds+∫0r12fxxds⏟=0f(W_t,t) = f(0, 0) + \int_0^t f_x dW_s + \underbrace{\int_0^t f_t ds + \int_0^r \frac{1}{2} f_{xx} ds}_{= 0}

Therefore, f(Wt,t)f(W_t, t) is a local martingale. We can check that the integrand fxx=4Ws3+2Ws(T−s)f_{xx} = 4W_s^3 + 2W_s(T - s) does not integrate to infinite (is a finite value), and consequently, f(Wt,t)f(W_t, t) is a martingale.

We see that MT=WT4M_T = W_T^4. Let the process Nt=E[WT4∣Ft] N_t =\mathbb E[W_T^4 \vert \mathcal F_t] - this by itself is also a martingale. Consider the following lemma -

Lemma. Suppose MtM_t and NtN_t are martingales with right continuous paths and Mt=NTM_t = N_T a.s., then P(Mt=Nt,∀t∈[0,T])=1P(M_t = N_t, \forall t \in [0, T]) = 1.

Proof. Mt=E[MT∣Ft]=E[Nt∣Ft]=NtM_t = \mathbb E[M_T \vert \mathcal F_t] = \mathbb E[N_t \vert \mathcal F_t] = N_t

This implies that MtM_t and NtN_t agree on all rational times (the way we define martingales for Brownian motions). However, since the processes are right continuous , we get

Mt(w)=lim⁡s↓t,s∈QMs(w)=lim⁡s↓t,s∈QNs(w)=Nt(w)M_t(w) = \lim_{s \downarrow t, s \in Q} M_s(w) = \lim_{s \downarrow t, s \in Q}N_s(w) = N_t(w)

for an arbitrary t∈[0,T]t \in [0, T].

Method 2

Consider the Ito's formula for Wt4W_t^4

Wt4=∫0t4Ws2dWs+∫0t6Ws2dsE[Wu2∣Ft]=Wt2+u−t,t<uE[WT2∣Ft]=∫0t4Ws2dWs⏟Martingale+∫0t6Ws2ds+E[∫0T6Ws2ds∣Ft]=Wt4+∫tT6E[Ws2∣Ft]∵Fubini’s theorem=Wt4+6∫tT[Wt2+s−t]ds=Wt4+6Wt2(T−t)+3(T−t)2\begin{align*} W_t^4 &= \int_0^t 4W_s^2 dWs + \int_0^t 6W_s^2 ds \\ \mathbb E[W_u^2 \vert \mathcal F_t] &= W_t^2 + u - t, t < u \\ \mathbb E[W_T^2 \vert \mathcal F_t] &= \underbrace{\int_0^t 4W_s^2 dWs}_{\text{Martingale}} + \int_0^t 6W_s^2 ds + \mathbb E\left[\int_0^T 6W_s^2 ds \vert \mathcal F_t\right] \\ &= W_t^4 + \int_t^T 6\mathbb E[W_s^2 \vert \mathcal F_t] \quad \because \text{Fubini's theorem} \\ &= W_t^4 + 6\int_t^T [W_t^2 + s - t] ds \\ &= W_t^4 + 6W_t^2(T - t) + 3(T - t)^2 \end{align*}

Both boil down to the same sort of an expression that satisfy the heat equation! But how do we guess the f(.,.)f(., .) in the first method?


Okay, now suppose we have an arbitrary function gg and we have to calculate E[g(WT)∣Ft]\mathbb E[g(W_T) \vert \mathcal F_t]. We can do the following -

E[g(Wt+(WT−Wt))∣Ft]=E[g(x+zT−t)]∣x=Wt\begin{align*} \mathbb E[g(W_t + (W_T - W_t)) \vert \mathcal F_t ] = E[g(x + z\sqrt{T - t})]\vert_{x = W_t} \end{align*}

Since WT−WtW_T - W_t is independent of Ft\mathcal F_t and WtW_t is known at Ft\mathcal F_t, we can remove the conditionality in the expectation!

Then, we say [g(x+T−tx]∣x=Wt=f(Wt,t)\mathbb [g(x + \sqrt{T - t} x]\vert_{x = W_t} = f(W_t, t). For g(x)=x4g(x) = x^4, we can expand the binomial expression to get ff. Furthermore,

f(x,t)=∫−∞∞g(y)pT−t(x,y)dy\begin{align*} f(x, t) = \int_{-\infty}^{\infty} g(y) p_{T - t} (x, y) dy \end{align*}

where pu(x,y)=12πue(y−x)2/2up_u(x, y) = \frac{1}{\sqrt{2\pi u}} e^{(y - x)^2/2u} . This satisfies the heat equation, namely

∂pu∂u=12∂2pu∂x2,∀y\frac{\partial p_u}{\partial u} = \frac{1}{2} \frac{\partial^2 p_u}{\partial x^2}, \quad \forall y

and consequently we can check that the dual is satisfied for ff as well!

∂f∂u+12∂2f∂x2=0\frac{\partial f}{\partial u} + \frac{1}{2} \frac{\partial^2 f}{\partial x^2}= 0

resulting in f(Wt,t)f(W_t, t) always being a martingale!

This works for any gg that does not grow too fast (polynomial growth - sufficient not necessary), and it does not have anything to do with smoothness of gg either!

Also, lim⁡t↑Tf(x,t)=g(x)\lim_{t \uparrow T} f(x, t) = g(x) as long as gg is continuous at xx.


This indeed is a very powerful result. Let us see how it can be used. Consider another example - Suppose we have μ=r=0\mu = r = 0, and a σ\sigma for a price process StS_t of the form St=S0exp⁡(σWt−σ2t/2)S_t = S_0 \exp\left(\sigma W_t - \sigma^2 t/2\right). Let the contingent claim be

X=g(∫0tStdt⏟IT)X = g\left(\underbrace{\int_0^t S_t dt}_{I_T}\right)

That is, the contingent claim is on the total value is a function of the total value accumulated till time TT.

Now, E[g(IT)∣Ft]\mathbb E[g(I_T) \vert \mathcal F_t] is calculated as -

IT=It+∫tTS0exp⁡(σWu−σ2u/2)du=It+S0exp⁡(σ(Wt+(Wu−Wt))−σ2u/2)du∵similar trickas before=It+S0eσWt∫0Teσ(Wu−Wt)e−σ2udu=It+S0eσWt∫0T−teσW^se−σ2s/2ds∵W^s=Wt+s−Wt,s=u−t=It+St∫0T−teσW^se−σ2s/2⏟⊥ ⁣ ⁣ ⁣⊥Ft\begin{align*} I_T &= I_t + \int_t^T S_0 \exp(\sigma W_u - \sigma^2 u /2)du \\ &= I_t + S_0\exp(\sigma (W_t + (W_u - W_t)) - \sigma^2 u /2)du \quad \because \text{similar trickas before} \\ &= I_t + S_0 e^{\sigma W_t} \int_0^T e^{\sigma (W_u - W_t)} e ^{-\sigma^2 u} du \\ &= I_t + S_0e^{\sigma W_t} \int_0^{T - t} e^{\sigma \hat W_s} e^{-\sigma^2 s/2} ds \quad \because \hat W_s = W_{t + s} - W_t, s = u - t \\ &= I_t + S_t \underbrace{\int_0^{T - t} e^{\sigma \hat W_s } e^{-\sigma^2 s/2}}_{\perp\!\!\! \perp\mathcal F_t} \end{align*}

So, E[g(It)∣Ft]=ψ(St,It,t)\mathbb E[g(I_t) \vert \mathcal F_t] = \psi(S_t, I_t, t) where ψ(x,y,t)=E[g(y+x∫0T−tSs)]\psi(x, y, t) = \mathbb E[g(y + x \int_0^{T - t} S_s)] if S0=1S_0 = 1. The closed form expression depends on gg in this case. Again, using Ito's formula on ψ(St,It,t)\psi(S_t, I_t,t) we get

ψ(St,It,t)=ψ(S0,0,0)+∫0tψi′(Su,Iu,u)dSu+∫0tψ2′Sudu+∫0tψ3′di+12∫0tψ1′′σ2Su2du\psi(S_t, I_t, t) = \psi(S_0, 0, 0) + \int_0^t \psi_i'(S_u, I_u, u)dS_u + \int_0^t \psi_2' S_u du + \int_0^t \psi_3'di + \frac{1}{2} \int_0^t \psi_1'' \sigma^2 S_u^2 du

Using this formulation, we will now look at Asian Call Options

Asian Call Option

The contingent claim is given by

X=(1T∫0TSudu−K)+X = \left(\frac{1}{T} \int_0^T S_u du - K\right)^+

and obtain g(y)=(YT−K)+g(y) = (\frac{Y}{T} - K)^+. The no-arbitrage price is given by the martingale

VT=E[(1TIT−K)+∣Ft]=E[(1TIt+1T∫tTSudu−K)+∣Ft]\begin{align*} V_T &= \mathbb E\left[\left(\frac{1}{T} I_T - K\right)^+ {\Huge \vert} \mathcal F_t\right] \\ &= \mathbb E\left[\left(\frac{1}{T} I_t + \frac{1}{T} \int_t^T S_u du- K\right)^+ {\Huge \vert} \mathcal F_t\right] \end{align*}

We'll assume μ=r=0\mu = r = 0 for simplicity. For stock price processes we have seen that

ST=S0exp⁡(σWt−σ2T2)=Stexp⁡(σ(WT−Wt)−σ22(T−t))S_T = S_0 \exp(\sigma W_t - \frac{\sigma^2 T}{2}) = S_t \exp(\sigma(W_T - W_t) - \frac{\sigma^2}{2}(T - t))

So the idea is to factor out StS_t this way in the expectation as well

VT=StE[(1T∫tTSuStdu−(K−1TItSt))+∣Ft]=Stϕ(Xt,t)\begin{align*} V_T &= S_t \mathbb E\left[\left(\frac{1}{T} \int_t^T \frac{S_u}{S_t}du - \left(\frac{K - \frac{1}{T}I_t}{S_t}\right)\right)^+ {\Huge \vert} \mathcal F_t\right] \\ &= S_t \phi(X_t, t) \end{align*}

where

Xt=K−1TItStϕ(x,t)=E[(1T∫0T−tS^udu−x)+]S^u=exp⁡(σ(Wu+t−Wt)−uσ2/2)\begin{align*} X_t &= \frac{K - \frac{1}{T}I_t}{S_t} \\ \phi(x, t) &= \mathbb E\left[\left(\frac{1}{T} \int_0^{T - t} \hat S_u du - x\right)^+\right] \\ \hat S_u &= \exp(\sigma(W_{u + t} - W_t) - u\sigma^2/2) \end{align*}

The expectation of this integral is not trivial. Therefore, we resort to representing the equations as a differential equation. We know Vt=Stϕ(Xt,t)V_t = S_t\phi(X_t, t) is a martingale from the way it has been constructed. Let us use Ito's formula to represent it. Furthermore, it can be shown that Xt=St−1(K−1TIt)X_t = S_t^{-1} (K - \frac{1}{T} I_t) is an Ito's process. Why? ItI_t is an Ito's process and St−1=S0−1exp⁡(−σWt+σ2t/2)S_t^{-1} = S_0^{-1} \exp(-\sigma W_t + \sigma^2t/2) is also an Ito's process - multiplying Ito's processes gives an Ito process. Using Ito's formula -

d(St−1)=1St(σdWt+σ2t)dXt=−St−11TStdt+(k−1TIt)d(St−1)\begin{align*} d(S_t^{-1}) &= \frac{1}{S_t} \left(\sigma dW_t + \sigma^2 t\right)\\ dX_t&= -S_t^{-1} \frac{1}{T} S_t dt + (k - \frac{1}{T} I_t ) d(S_t^{-1}) \end{align*}

The quadratic variation of XtX_t is 00, and therefore the product rule does not have any covariation terms.

Finally, we get

dVt=St[ϕt−1Tϕx+σ22Xt2ϕxx]dt+martingale terms\begin{align*} dV_t = S_t \left[\phi_t - \frac{1}{T} \phi_x + \frac{\sigma^2}{2} X_t^2 \phi_{xx}\right]dt + \text{martingale terms} \end{align*}

We have

ϕ(x,T)=(−x)+=x−ϕt−1Tϕx+σ2x22ϕxx=0\begin{align*} \phi(x, T) = (-x)^+ = x^{-} \\ \phi_t - \frac{1}{T} \phi_x + \frac{\sigma^2 x^2}{2}\phi_{xx} = 0 \end{align*}

The stochastic differential equation comes from the fact that VtV_t is a martingale, and with this we have essentially captured the process in terms of a single spatial random variable. Previously, we had

E(g(IT)∣Ft)=ψ(St,It,t)\mathbb E(g(I_T) \vert \mathcal F_t) = \psi(S_t, I_t, t)

where

ψt+xψy+σ2x22ψxx=0ψ(x,y,T)=(yT−K)+\psi_t + x \psi_y + \frac{\sigma^2 x^2}{2} \psi_{xx} = 0 \\ \psi(x, y, T) = \left(\frac{y}{T} - K\right)^+

In fact, with brutal exercise, we can show that

ψ(x,y,y)=xϕ(TK−yxT,t)\psi(x, y, y) = x \phi\left(\frac{TK - y}{xT}, t\right)

Note that we were able to do such a simplification for a specific nature of gg - here gg is a straight-line function (gk(cy)=cgk/c(y)g_k(cy) = cg_{k/c}(y)). Something similar can be done when g(y)=ypg(y) = y^p as well.

Geometric Asian Call

A continual version of geometric mean is given by

GT=exp⁡(1T∫0Tlog⁡Stdt)G_T = \exp\left(\frac{1}{T}\int_0^T \log S_t dt \right)

So, the contingent claim is given by

X=(exp⁡(1T∫0Tlog⁡Sudu)−K)+Alsolog⁡Su=log⁡S0+σWu−σ2uu\begin{align*} X &= \left(\exp\left(\frac{1}{T} \int_0^T \log S_u du\right) -K \right)^+ \\ \text{Also}& \log S_u = \log S_0 + \sigma W_u - \frac{\sigma^2u}{u} \end{align*}

Once again, assume μ=r=0\mu = r = 0. The no-arbitrage price is given by

E[X∣Ft]=E[[S0exp⁡(σTJt+σT∫tTWudu−σ2T4)−K]+∣Ft]\mathbb E[X \vert \mathcal F_t] = \mathbb E\left[\left[S_0\exp\left(\frac{\sigma}{T} J_t + \frac{\sigma}{T}\int_t^TW_u du - \frac{\sigma^2 T}{4}\right) - K\right]^+ {\Huge \vert} \mathcal F_t\right]

where Jt=∫0tWuduJ_t = \int_0^t W_u du. Then,

E[X∣Ft]=E[[S0exp⁡(σTJt+σT(T−t)Wt+σT∫0T−tW^sds−σ2T4)−K]+∣Ft]\begin{align*} \mathbb E[X \vert \mathcal F_t] = \mathbb E\left[\left[S_0\exp\left(\frac{\sigma}{T} J_t + \frac{\sigma}{T}(T-t) W_t + \frac{\sigma}{T}\int_0^{T - t}\hat W_s ds - \frac{\sigma^2 T}{4}\right) - K\right]^+ {\Huge \vert} \mathcal F_t\right] \end{align*}

where W^s=Wt+s−Wt\hat W_s = W_{t + s} - W_t and J^T−t=∫0T−tW^udu∼N(0,(T−t)3/3)\hat J_{T - t} = \int_0^{T - t} \hat W_u du \sim \mathcal N(0, (T - t)^3/3)

The above expression is then equal to ψ(Wt,Jt,t)\psi(W_t, J_t, t)

where

ψ(x,y,y)=E[(S0exp⁡(σTy+σT(T−t)x−σ2T4+σTJ^T−t)−K)+]\psi(x, y, y) = \mathbb E\left[\left(S_0 \exp\left(\frac{\sigma}{T} y + \frac{\sigma}{T}(T - t) x - \frac{\sigma^2T}{4} + \frac{\sigma}{T} \hat J_{T - t}\right) - K \right)^+\right]

this is very similar to the expectation we have in Black-Scholes - E[(xexp⁡(αx−α22)−K)+]\mathbb E[(x \exp(\alpha x - \frac{\alpha^2}{2}) - K)^+] and the final expression can be derived from there.


Consider another example where X=g(max⁡0≤t≤TSt)X = g(\max_{0 \leq t \leq T} S_t) , then the no-arbitrage price is

E[X∣Ft]=Yt=f(Zt⏟Ito,At⏟adapted and r.c.,t)  ⟹  dYT=f1′dZt+f2′dAt+f3′dt+f′′d⟨Z⟩−−(1)\begin{align*} \mathbb E[X \vert \mathcal F_t] &= Y_t = f(\underbrace{Z_t}_{\text{Ito}}, \underbrace{A_t}_{\text{adapted and r.c.}}, t)\\ &\implies dY_T = f_1' dZ_t + f_2' dA_t + f_3' dt + f''d \langle Z\rangle -- (1) \end{align*}

where At=max⁡0≤u≤tSuA_t = \max_{0 \leq u \leq t} S_u (the Lebesgue measure of this set is 0).

AT=At∨St(max⁡u≤u≤TSuSt)=At∨Stmax⁡0≤v≤T−texp⁡(σWt+v−σWt⏟σW^v−σ22(v))=At∨StA^T−t⏟⊥ ⁣ ⁣ ⁣⊥Ft\begin{align*} A_T &= A_t \vee S_t \left(\max_{u \leq u \leq T} \frac{S_u}{S_t}\right) \\ &= A_t \vee S_t \max_{0 \leq v \leq T- t}\exp(\underbrace{\sigma W_{t + v} - \sigma W_t}_{\sigma \hat W_v} - \frac{\sigma^2}{2}(v))\\ &= A_t \vee S_t \underbrace{\hat A_{T - t}}_{\perp \!\!\! \perp \mathcal F_t} \end{align*}

So, we get Vt=ψ(St,At,t)=E[At∨(StA^T−t)]V_t = \psi(S_t, A_t, t) = \mathbb E[A_t \vee (S_t \hat A_{T - t})] (note that Ft\mathcal F_t goes away).

Using Ito's formula (equation (1) above) ,

dVt=ψ1′dSt⏟martingale+ψ2′dAt+ψ3′dt+ψ′′σ2S2dt\begin{align*} dV_t = \underbrace{\psi_1' dS_t}_{\text{martingale}} + \psi_2' dA_t + \psi_3' dt + \psi'' \sigma^2 S^2 dt \end{align*}

dVtdV_t should be a martingale by definition, therefore, ψ2′dAt+ψ3′dt+ψ′′σ2S2dt=0\psi_2' dA_t + \psi_3' dt + \psi'' \sigma^2 S^2 dt = 0. Since dAtdA_t and dtdt are independent

∂ψ∂At=0 when dAt=0  ⟹  St=At  ⟹  {(St,At,t):St≤At,0≤t≤T}∂ψ∂t+12σ2x2∂2ψ∂x2=0,x<y\begin{align*} \frac{\partial \psi }{\partial A_t} = 0 \text{ when } dA_t = 0 &\implies S_t = A_t \implies \{(S_t, A_t, t): S_t \leq A_t ,0\leq t\leq T\} \\ \frac{\partial \psi}{\partial t} + \frac{1}{2} \sigma^2 x^2 \frac{\partial^2 \psi}{\partial x^2} = 0&, x<y \end{align*}

Generally, such PDEs are solved by curated guesses, and there is no algorithmic procedure to determine these.


Lemma: X,YX, Y are independent martingales, then XtYtX_tY_t is a martingale. Applies to local martingale as well

Proof. Let W=(W1,…,Wn)W = (W^1, \dots, W^n) be an nn-dimensional Brownian motion.

Mt=∫0tHs⋅dWNt=∫0tKs⋅dW⟨Mt,NT⟩:=∫0t∑i(HsiKsi)ds=∫0tHs⋅Ksds\begin{align*} M_t &= \int_0^t H_s \cdot dW \\ N_t &= \int_0^t K_s \cdot dW \\ \langle M_t, N_T \rangle &:= \int_0^t \sum_i (H_s^i K_s^i) ds \\ &= \int_0^t H_s \cdot K_s ds \end{align*}

To show E[XtYt⋅1C]=E[XsYs⋅1C]\mathbb E[X_t Y_t \cdot 1_C] = \mathbb E[X_s Y_s \cdot 1_C]. Let FtX=σ(XS:s≤t),FtY=σ(Ys:s≤t)\mathcal F_t^X = \sigma(X_S: s \leq t), \mathcal F_t^Y = \sigma(Y_s: s \leq t). Then FtX,Y=σ(FtX,FtY)\mathcal F_t^{X, Y} = \sigma(\mathcal F_t^X, \mathcal F_t^Y)

This is true for C=A∩BC = A \cap B for A∈FsX,B∈FSYA \in \mathcal F_s^X, B \in F_S^Y. From Monotone Class Theorem, if such a relation is true on the Pi class (closed under intersection), then it is true for all CC.

Corollary. ⟨X,Y⟩≡0\langle X, Y \rangle \equiv 0


With these results, we can now work with multiple stocks (dd) using the Black-Scholes equation with underlying nn-Brownian motion.

Consider this example - -d=1,n=2,r=0d = 1, n = 2, r = 0. We have dSt=St(βdt+g(Zt)dWt1),S0>0dS_t = S_t(\beta dt + g(Z_t)dW_t^1), S_0 > 0 where dZt=Zt(γdt+δdWt2),Z0=1dZ_t = Z_t (\gamma dt + \delta dW_t^2), Z_0 = 1, β,γ∈R,δ>0,g(x)≥ϵ>0\beta, \gamma \in \mathbb R, \delta > 0, g(x) \geq \epsilon > 0.

Claim. There exists at least one ELMM but this market is not complete.

Proof. Choose a martingale

MTθ:=−∫0tβg(Zs)dWs′+θWt2M_T^\theta := -\int_0^t \frac{\beta}{g(Z_s)} dW_s' + \theta W_t^2

for θ∈R\theta \in \mathbb R. The density process is chosen as Λt=exp⁡(Mtθ−12⟨Mθ⟩t)\Lambda_t = \exp(M_t^\theta - \frac{1}{2} \langle M^\theta \rangle_t) as suggested by the Girsanov's theorem. Now, we need to show two things - MtθM_t^\theta is a martingale and Λt\Lambda_t satisfied Novikov's condition.

MtθM_t^\theta is a local martingale because the individual terms in the expression are martingales. The quadratic variation of MtθM_t^\theta is given by

⟨Mθ⟩t=∫0tβ2g(Zs)2ds+θ2t≤β2ϵ2t+θ2t\begin{align*} \langle M^\theta \rangle_t &= \int_0^t \frac{\beta^2}{g(Z_s)^2}ds + \theta^2 t \\ &\leq \frac{\beta^2}{\epsilon^2}t + \theta ^2 t \end{align*}

Since the quadratic variation of MtθM_t^\theta is bounded by a constant, MtθM_t^\theta is a martingale. Also, from Novikov's condition, Λt\Lambda_t is a martingale as well.

Now, according to Girsanov's theorem,

W~t1=Wt1−⟨W1,Mθ⟩t=Wt1+∫0tβg(Zs)ds\begin{align*} \tilde W_t^1 &= W_t^1 - \langle W^1, M^\theta\rangle_t \\ &= W_t^1 + \int_0^t \frac{\beta}{g(Z_s)}ds \end{align*}

is a QθQ^\theta-Brownian motion.

From the previous result, W~t2=Wt2−θt\tilde W^2_t = W_t^2 - \theta t is a QQ-Brownian motion.

dSt=St(βdt−g(Zt)dW′)=Stg(Zt)dW~t′\begin{align*} dS_t &= S_t (\beta dt- g(Z_t)dW') \\ &= S_t g(Z_t) d \tilde W'_t \end{align*}

So, StS_t is a QθQ^\theta-local martingale, implying that QθQ^\theta is a ELMM.

Now, to show that the market is not complete, we'll explicitly construct a martingale that cannot be replicated. Letting θ=0\theta = 0, we get W~T2=WT2\tilde W_T^2 = W_T^2. To replicate it, we have

W~t2=∫0tαudSu\begin{align*} \tilde W_t^2 = \int_0^t \alpha_u dS_u \\ \end{align*}

Let us now calculate the quadratic variation -

⟨W~2⟩t=t∵Q-Brownian motion=⟨W~2,W~2⟩t=⟨W~2,∫0tαuSug(Zs)dW~u′⟩t=∫0tαuSug(Zu)d⟨W~2,W~1⟩u=0\begin{align*} \langle \tilde W^2 \rangle_t &= t \quad \because Q\text{-Brownian motion} \\ &= \langle \tilde W^2, \tilde W^2 \rangle_t = \langle \tilde W^2, \int_0^t \alpha_u S_u g(Z_s) d \tilde W_u' \rangle _t \\ &= \int_0^t \alpha_u S_u g(Z_u) d \langle \tilde W^2, \tilde W^1 \rangle_u = 0 \end{align*}

because ⟨W~2,W~′⟩t=⟨W2,W1⟩t=0\langle \tilde W^2, \tilde W' \rangle_t = \langle W^2, W^1 \rangle_t = 0. This is a contradiction, and therefore, the market cannot be replicated.

∼fin∼\sim fin \sim
Type to search the whole site.