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.
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.
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.