optional stopping: Nonlinear Function
Created: August 29, 2022
Modified: August 29, 2022

optional stopping

This page is from my personal notes, and has not been specifically reviewed for public consumption. It might be incomplete, wrong, outdated, or stupid. Caveat lector.

If xtx_t is a martingale and τ\tau is a stopping time, then any of the following conditions implies that E[xτ]=E[x0]\mathbb{E}[x_{\tau}] = \mathbb{E}[x_0]:

  1. The stopping time is bounded (almost surely) by some constant τ\tau.
  2. The stopping time has finite expectation, and the expected martingale increments are bounded (there is some cc such that E[xt1xtx0,,xt]c\mathbb{E}\left[\left|x_{t-1} - x_t\right| |x_0, \ldots, x_t \right] \le c almost surely for all tt).
  3. The values up to the stopping point are bounded (there is some cc such that xmin(t,τ)c|x_{\min(t, \tau)}| \le c almost surely for all tt).

This establishes that there is no successful betting strategy (one that makes money in expectation) for a gambler with a finite lifetime (the first condition) or a house limit on bets (the second condition).