The sample problem: rare payoffs, thin evidence

the options sample problem

Options evaluation has a distinctive sample problem: the payoffs that matter most - the large gains a long option occasionally makes, the tail losses a seller occasionally takes - are rare, so a backtest of any normal length contains few of them, and the probability-weighted result rests heavily on outcomes the sample barely sampled. The distribution's tails, which dominate an option's value, are exactly where the evidence is thinnest.