Free tool
Backtest sample size calculator
A win rate on its own tells you almost nothing. This puts error bars around it, so you can see whether your result is evidence of an edge or just a small sample behaving like small samples do.
What does my result prove?
How many trades do I need?
Trades needed to rule out a coin flip
At 95% confidence. The weaker the edge, the more evidence it takes to establish.
| If your true win rate is | You need about |
|---|---|
| 52% | 2,402 trades |
| 55% | 385 trades |
| 58% | 151 trades |
| 60% | 97 trades |
| 65% | 43 trades |
| 70% | 25 trades |
The method
This uses the Wilson score interval, introduced by Edwin B. Wilson in 1927. It is preferred over the more familiar Wald interval because it stays accurate at small sample sizes and near the extremes — precisely the case when you are judging forty trades rather than four thousand.
"Proof" here means the interval no longer contains 50%. That is a deliberately low bar: it establishes that you are probably not flipping a coin, not that the strategy is profitable after costs. Readthe full explanation for why expectancy matters more than win rate, and how splitting results by session or pair quietly destroys your sample.
Wilson, E. B. (1927).Probable Inference, the Law of Succession, and Statistical Inference.Journal of the American Statistical Association 22(158), 209–212.
Frequently asked questions
How many trades do I need to prove my strategy works?
It depends entirely on the size of your edge. To show a 55% win rate is not a coin flip at 95% confidence takes about 385 trades. A 60% win rate needs about 97. A 52% win rate needs roughly 2,400. The weaker the edge, the more evidence is required to establish it — which is the opposite of how most traders allocate their testing time.
What is a Wilson confidence interval?
It is a method for putting error bars around a proportion, introduced by Edwin B. Wilson in 1927. Unlike the more familiar Wald interval, it stays accurate at small sample sizes and near 0% or 100%, which is exactly the situation traders are in when judging a few dozen trades.
Why does my 70% win rate not count as proof?
Because a high win rate over few trades is easy to produce by chance. Seven wins from ten trades gives a 95% interval of roughly 35% to 91% — consistent with a losing strategy. The interval narrows as the sample grows; the observed percentage alone tells you almost nothing without it.
Should I be measuring win rate at all?
For most strategies, no — expectancy is the better metric, because win rate ignores the size of wins and losses. A 40% win rate at 3R beats a 70% win rate at 0.5R. Win rate is used here because proportions have clean, well-understood confidence intervals; treat the result as a lower bound on how much data you need.