[adenadel]

For martingales (stochastic processes whose expected value given the current value is the current value) there is no strategy that does better (on average) than the (expected) initial value (Doob's optional stopping theorem).

https://en.wikipedia.org/wiki/Optional_stopping_theorem