Why we rank drawdown above win rate
A high win rate is easy to show and easy to misread. We judge a strategy first by how deep its losses go, and by how the engine stops them going deeper.
QuantFX2 min read
The win rate trap
A 70% win rate sounds good. If the average loss is three times the average win, the strategy still loses money: seven wins of one unit and three losses of three units leave it two units down for every ten trades.
Why drawdown matters more
Systematic strategies fail in their tails. One run of correlated losses can undo months of gains, and the deeper the drawdown, the harder the climb back. A 50% loss needs a 100% gain to recover.
That is why, in client deployments, drawdown limits live in the engine. They are set per client before go-live, checked on every trade and every equity update, and they trigger automatically. Nobody gets to second-guess them in the middle of a losing streak.
Our own master strategy account is a research environment and runs with wider limits. Its month-by-month history, including the losing month, is on the track record page.
The limits we set
For a typical deployment:
- A daily loss limit, usually 3% to 5% of equity.
- Position size that scales down as equity falls.
- A hard stop at a set total drawdown, usually 8% to 10%.
- No trading during high-impact news events.
An example
| Strategy | Win rate | Drawdown | Result |
|---|---|---|---|
| A | 80% | -15% | Rejected |
| B | 55% | -6% | Accepted |
Both strategies measured against an 8% total drawdown limit.
Strategy A wins more often and still fails, because a 15% drawdown breaches the limit. Strategy B wins just over half its trades and stays inside it.
In practice
Position size is capped by recent volatility, so exposure drops when ATR spikes. When equity falls 5%, position sizes are cut by half.
We only start tuning a strategy for returns once it stays inside its limits.