← All tools

Win-Rate Confidence Interval Calculator

Is your win rate real, or just small-sample noise?

Must be at least 1.
Observed win rate
95% confidence interval (true win rate)
Significantly above 50%?
Trades needed for ±3% margin

What this tells you

A trader with a 70% win rate across 10 trades will get cocky. The Wilson confidence interval says their true win rate could be anywhere from roughly 40% to 89%. With that sample, you can't even rule out that they're worse than coin-flip.

The same trader after 200 trades with a 56% win rate has a confidence interval of roughly 49%-63% — much tighter, and probably real edge.

This is the math that separates "small-sample lucky" from "actual skill." Most retail traders fail to distinguish them, which is why most retail traders blow up after their first lucky run.

What "significantly above 50%" means

If the entire 95% confidence interval is above 50%, the win rate is statistically distinguishable from coin-flip. You can be reasonably confident the system has real positive edge.

If the interval straddles 50%, you don't have enough data to claim edge. The system might be good. It might be bad. You can't tell from this sample.

Methodology

This calculator uses the Wilson score interval, which is more accurate than the basic normal approximation, especially for small samples and extreme win rates. The interval gives a range where the true win rate sits with 95% confidence.

The "trades needed for ±3% margin" estimate uses the standard sample-size formula for a binomial proportion at 95% confidence: roughly n ≈ 1067 × p(1-p) for a 3% margin.

Sanity check on published win rates. If a signal service claims an 85% win rate over 20 trades, plug it in. The 95% CI is wide enough that the true win rate could plausibly be 60%. If a service claims 55% over 1,000 trades, the CI is tight — that's real. The math doesn't lie. Tools like SultraxAI publish full sample sizes alongside win rates so you can run this calculation yourself.