Tail-concentrated (jackpot)

That's the mechanic behind tail-concentrated profit, also called jackpot risk. A strategy can look profitable for years while doing nothing useful on all but a few sessions, and the backtest has no idea it's a fluke, because a total return doesn't show you where the return came from.

This is the most seductive failure mode we see, because the headline return looks spectacular no matter how it was built. A curve built on four wild days and a curve built on four hundred quiet ones can print the same total on a marketing slide. Nothing in the summary statistics warns you which one you're looking at. You have to open the curve up and look at the distribution of daily returns, not just the number at the end.

We test for it directly: strip out the top 10 percent of winning days and re-run the strategy on what's left. A strategy with real edge keeps working, the return drops, the curve gets less exciting, but it stays positive and the shape holds. A jackpot strategy collapses the moment you remove its best days, because those days were never a sample of ongoing performance. They were the whole performance.

Tail-concentrated profit is our third most common reject reason across the audit, behind having no gross edge at all and riding market beta in disguise. It's also the one that fools people longest. A fat Sharpe ratio and a smooth-looking curve can both survive with a handful of outlier days sitting in the sample, and the statistics don't automatically flag concentration. You have to go looking for it.

The practical problem is timing, not just statistics. You cannot schedule a jackpot day in advance. Live, a strategy that depends on catching a handful of specific moves a year is exactly the strategy most likely to miss one of them to a weekend gap, a broker outage, or a fill that lands worse than the backtest assumed. Miss the best day and the strategy that looked untouchable on paper spends the rest of the year clawing back toward flat, often deep in drawdown the whole way.

A curve resting on five green candles is not a system. It's a photograph of five lucky days wearing a system's clothes.