High win rate, tiny payoff: the shape of a hidden blow-up

[Amanhasnoname]

Scalping strategies that win 90 percent of the time feel great. Each trade makes a pip or two. Then one loss takes back fifty trades' worth of profit.

That payoff structure is common in grid, martingale and tight-target systems. The equity curve climbs smoothly for weeks, which builds confidence, and then drops sharply.

The risk is that the loss distribution has a fat tail. News, a gap, or a slow trend can turn a position that normally recovers into a runaway. When stops are wide or non-existent, one event can wipe out the account.

Judging such systems by win rate is misleading. Better measures are the worst drawdown, the largest single loss relative to average win, and how results behave in volatile periods.

A way to test is to run the strategy through historical shocks, such as the 2015 franc event, Brexit night and the 2020 March volatility, and see what happens.

If the result is ruin, the smooth curve was borrowed from the future.

A telling diagnostic is the ratio of the largest loss to the average win. When a single loss equals more than thirty average wins, the strategy is selling insurance, and the question is whether the premium collected is worth the day the policy pays out.

Have you tested your system against extreme events?