The loss that ends an account is almost never the one in the plan. Premium-selling backtests are built from calm months — the visible hump of small wins — while the rare, oversized loss barely appears in the sample. The tail doesn't shrink because you ignored it; you just can't see it.
Short-premium tails have three engines that fire together: gap moves that jump straight through strikes (no stop-loss can exit at a price that never printed), correlations rising toward one so ten 'diversified' short puts become one market bet, and implied-volatility spikes that reprice every short option against you at once.
The interactive figure hides the far-left tail behind a 'what you modeled / what exists' toggle, then lets a position-size slider scale the damage. The same five simulated years of markets are a bruise at 20–30% of the account and an ending at 100% — sizing is the only dial that changed.
Margin is a hidden multiplier: buying-power relief lets you hold the same tail with a fifth of the cash, so traders who believe they're 30% allocated are often past 100% in crash terms. Measure exposure by crash-day loss, never by the margin required on a quiet day.
Educational, not investment advice.