Why sample size is the one number nobody checks

Every statistic on a signal page — Sharpe ratio, win rate, profit factor — is an estimate, and estimates get noisier the fewer data points they're built from. A coin that's genuinely fair (50/50) can land 8 heads in 10 flips without being a trick coin; it's just a small sample doing what small samples do. Trade records work the same way. A strategy with no real edge at all can string together 15-20 winning trades through ordinary variance, and every stat calculated from that run — Sharpe, drawdown, profit factor — will look genuinely excellent, because the math has no way to distinguish "real edge" from "got lucky so far."

The more trades in the sample, the more those numbers converge toward what the strategy's actual, long-run behavior looks like. Twenty trades barely constrains the estimate at all. A hundred starts to. A thousand is a different kind of evidence entirely.

A strategy live for three weeks with 15 closed trades and a flawless equity curve isn't lying to you — it just hasn't been tested by enough market conditions yet to say anything reliable about what happens next. The track record and the strategy's actual edge are two different things, and small samples are exactly where they're most likely to diverge. A platform's own "newly opened account, results may be of a random nature" warning, when it shows up, is saying precisely this.

This cuts both ways: a strategy with a rough patch early on and only a handful of trades hasn't necessarily proven anything is wrong either. The honest position with a small sample is "not enough evidence yet" in either direction, not "great" or "bad."

There's no single number where a track record suddenly becomes trustworthy, but the pattern with n=15, n=50, n=100+ trades is consistent enough to build a rule around: below ~30 trades, treat every other stat on the page as provisional. Between 30 and 100, the numbers are informative but still have real error bars around them. Above 100, the stats start to mean roughly what they claim to mean — still not a guarantee, but no longer mostly noise.