Why is this stopping time the result of intersections instead of unions?

66 Views Asked by At

On page 54 of the book "Basic Stochastic Processes" of Brzezniak and Zastawniak, author proposes this example:

A coin is tossed repeatdely and you win or lose 1 pound depending on which way it lands. You start with 5 pounds and you will play until you have no money or if you have 10 pounds. If $X_{n}$ is the amount of money at $n$ toss, then the time when the game stops is: $$\tau = min\{n: X_{n} = 0, X_{n} = 10 \}$$

Prove $\tau$ is a stopping time.

Author gives the following solution: $X_{n}$ is adapted to the filtration $F_{n} = \sigma(X_{1}, X_{2}, ... X_{n})$ and $\tau$ is n stopping time because: $$\{\tau = n\} = \{0<X_{1}<10\} \cap \{0<X_{2}<10\} ... \cap \{0<X_{n-1}<10\} \cap \{X_{n} = 0\} \cap \{X_{n} = 10 \} \in F_n$$

What I don't understand is why the proof isn't like this: $$\{\tau = n\} = \{0<X_{1}<10\} \cup \{0<X_{2}<10\} ... \cup \{0<X_{n-1}<10\} \cap \{X_{n} = 0\} \cap \{X_{n} = 10 \} \in F_n$$

Because it will include all the times when I didn't have 10 or 0 pounds. $F_{n}$ is closed under intersections and unions, so my proposed solution will prove $\tau$ is a stopping time too.

1

There are 1 best solutions below

3
On BEST ANSWER

With probability $1$ the process reaches one of the stop states. That is a fact about random walks that has nothing to do with filtrations.

Intersections are used in the logical formula because to stop at time $= n$ the process should not stop at time $1$ AND not stop at time $2$ AND not stop at .... time $n-1$. The purpose of the formula is to show that stopping at time $\leq t$ depends only on information in the filtration at indices $\leq t$.