Drawdown is not symmetric and every decision made during one is made badly. This works out what the hole actually costs in gain, and how long the current return takes to fill it.
| Loss from peak | Gain to recover |
|---|---|
| 5% | +5.3% |
| 10% | +11.1% |
| 15% | +17.6% |
| 20% | +25.0% |
| 30% | +42.9% |
| 40% | +66.7% |
| 50% | +100.0% |
| 60% | +150.0% |
| 70% | +233.3% |
| 80% | +400.0% |
| 90% | +900.0% |
| 95% | +1,900.0% |
Five percent down needs five point three up — close enough to feel fair, which is why the first one is easy. It stops feeling fair somewhere around 30%: you are down 30 and you need 42.9. By 50 it is double. By 80 the account needs to multiply by five, and nobody who has lost 80% of an account trades the same way afterwards.
This is the arithmetic behind a rule that sounds like a platitude: cut it early. Not because losing is bad, but because the gain required is convex in the loss.
From the data. The best result in the 372 batch was +1,041% with a 96% drawdown. To get there it had to be down 96% first, which meant surviving a moment where it needed a 25-fold gain to recover. That is not a strategy, that is a sequence of coin flips where the account had to win all of them. The five best performers in the batch all sat between 76% and 99% drawdown — the ranking by return and the ranking by risk are close to the same list, which is on the results page.
Peak-to-current equity only measures what has been closed. Of the 22 configurations that survived an accounting screen across three timeframes, realised profit summed to +$63,537 and the positions they were still holding would have marked at −$28,131. One of them, 0444, showed a profit while carrying open positions that were 74.1% underwater on lot count.
So: if your platform shows a drawdown of 10%, check it against the floating loss on open trades before you believe it. An account that has not closed its losers has not measured its drawdown. That audit is here.
Related: losing streaks · per-trade expectancy · where the money ended up