The Martingale Transform and Optional Stopping Formula in MathFinance 345/Stat390

As a seasoned financial writer, I'm excited to dive into the world of MathFinance 345/Stat390 and explore the concepts of martingale transforms and optional stopping formulas.

Martingale Transforms

Let's start with the martingale transform. The martingale transform is defined as follows: (Y · Z)n = Z0 + n X j=1 Yj(Zj −Zj−1). This definition suggests that the martingale transform depends on a sequence of random variables Yn, and it takes these Yn values to determine the martingale Z.

For instance, consider a martingale relative to a filtration (Fn)n≥0, where Fn = {1} for all n ≥ 0. In this case, we have (Y · Z)0 = 0 + 0 X j=1 Yj(0 − 0−1), which equals 0.

As the martingale transform depends on individual random variables Yn, the martingale property is not preserved under different filtrations. However, if we apply a stopping time τ to the martingale sequence (Zn)n≥0, we can examine its behavior.

Optional Stopping

When we introduce a stopping time τ, we are essentially asking when certain events occur within a given timeframe. In this context, the martingale transform becomes a martingale relative to the filtration (Fn)∧τ, where Fn∧τ = {1} for all n ≥ 0.

Using the result from part (a), we can show that the sequence Yn is predictable under the stopping time τ. This means that we can predict future values of Yn based on past information.

Bonds and Optional Stopping

Now let's apply this to a bond market scenario. Suppose we have a zero-coupon bond with maturity M, denoted as Bt,M = Bt,M(ω) for t ≤ M. We want to find the riskless asset in this market.

Using Doob's Optional Stopping Formula, which states that EZτ∧n = 1 for every n = 1, 2, ..., , T, we can deduce that (Zt)0≤t≤T is a martingale relative to the usual filtration. Specifically, we have Zt = Bt,T Qt−1 j=0 Bj,j+1.

As this process continues, we can conclude that Sn −n(p − q) is a martingale. Using the Optional Stopping Formula, with the stopping time τ being the minimal integer n > 0 such that Sn = +B or Sn = −A, we can show that (4) ESτ = (p −q)Eτ.

Gambler's Ruin

Finally, let's revisit the classic gambler's ruin problem. Suppose we have a game where two players take turns betting, and the probability of winning is p > 1/2, while the probability of losing is q = 1 − p < 1/2. We want to determine the probability that Slim exhausts his initial fortune B before Fats exhausts his initial fortune A.

Using the Optional Stopping Formula with τ being the minimal integer n > 0 such that Sn = +B or Sn = −A, we can show that (3) E(Zτ) = 1. This implies that the probability of Slim winning is p, while the probability of Fats winning is q.

To solve for the probability that Fats wins the game, we can use the equality in equation (3). We have (p −q)Eτ = (p −q)(1), which simplifies to Eτ = 1. This means that the expected duration of the game is equal to 1.

Conclusion

In conclusion, MathFinance 345/Stat390 has provided us with a wealth of insights into martingale transforms and optional stopping formulas. By understanding these concepts, we can better navigate complex financial markets and make more informed investment decisions.