Drawdown and risk of ruin, explained with 2,000 simulations
A strategy can be profitable on average and still take you through a drawdown you cannot sit through. The expectancy tells you where you end up on average. The drawdown tells you what the road there looks like.
Two numbers to know
- Maximum drawdown: the largest fall of your balance from a previous peak, in percent.
- Risk of ruin: here, the chance of hitting a drawdown you have decided you will not tolerate, for example 20%.
What the risk per trade does
The results below come from the drawdown simulator on this site: 2,000 runs of 200 trades each, a 45% win rate, winners of 2R, a fixed percentage risk on the current balance, and a 20% drawdown limit.
| Risk per trade | Typical max drawdown | Worst 5% of runs | Chance to hit 20% |
|---|---|---|---|
| 1% | 8.8% | 15.4% | 0.5% |
| 2% | 17.1% | 29.2% | 33.8% |
| 3% | 25.0% | 40.9% | 80.8% |
The strategy and its edge (+0.35R per trade) are the same in every row. Only the risk per trade changed, and a 20% drawdown went from rare to likely.
A small edge is fragile
Keep the risk at 1%, but lower the win rate to 35% with the same 2R winners. The expectancy falls to +0.05R, the typical drawdown roughly doubles to 16.7%, and the chance of hitting 20% rises to 31.9%. A strategy that is barely positive on paper feels very different in practice.
Recovering takes more than you lost
| Drawdown | Gain needed to recover |
|---|---|
| 10% | 11.1% |
| 20% | 25% |
| 30% | 42.9% |
| 50% | 100% |
That is why keeping drawdowns small matters more than squeezing out a slightly higher average return.
What the simulation leaves out
It assumes independent trades, a fixed reward to risk and no costs. Real trading is usually worse: losses cluster when market conditions change, and spreads, commission and slippage add up. If you run an automated system, the gap between a simulation and live results has its own causes. I wrote about those on my developer blog: Why my trading bot loses money in live trading.