Why sample size matters before you trust a statistic

Trading statistics are measured on the trades you took, and those trades are a sample. The fewer there are, the less the numbers can tell you — and the easier it is to read a pattern into noise.

Small samples swing

Seven wins from ten trades is a 70% win rate. Two more losses make it 58%. Nothing about the setup changed; the estimate simply had very little to stand on. The same is true of expectancy, profit factor and every breakdown by setup, weekday or hour — splitting your trades into groups makes each group smaller.

Reading an interval

A confidence interval gives a range of values consistent with the trades you have. A win rate of 52% with an interval of 37–67% says: the true rate could plausibly be anywhere in that range. As you add trades, the interval narrows. When two setups’ intervals overlap heavily, the data has not yet shown that one is better than the other.

Practical rules

• Be sceptical of any conclusion drawn from a handful of trades.
• Prefer longer windows when comparing setups or times of day.
• Look for the interval or grade, not just the headline number.
• Treat a surprising result as a question to keep watching, not an answer.

This is also why TradeDrift’s AI coach receives every rate together with its sample size and grade — so a claim built on three trades is labelled as such before anyone reads it.