Ten trades tell you almost nothing — pure luck routinely produces seven wins out of ten from a coin flip. More trades shrink the share of luck in your results. There's no magic number, but the direction is clear: dozens of trades are weak evidence, hundreds are meaningfully better — and the quality of those trades matters as much as the count.
Why do ten trades tell you almost nothing?
Because luck impersonates skill effortlessly at small sizes. Flip a fair coin ten times: getting seven or more heads happens about one run in six — that's plain arithmetic, no market needed. Now translate: a strategy with zero real edge, winning purely by chance, will show a "70% win rate" over ten trades about one attempt in six. If you test a handful of ideas, one of them will look brilliant by accident almost every time. That's the trap of the short backtest: it doesn't measure your strategy, it measures which of your coin flips happened to land well. Ten trades can't separate a good strategy from a lucky one, and it can't separate a bad strategy from an unlucky one either.
How does adding trades shrink the luck share?
Luck doesn't scale; a real edge does. Flip that same fair coin 100 times and getting 70 or more heads is vanishingly rare — far beyond one in a thousand. So a 70% win rate over 100 trades is very hard for pure chance to fake, while the same rate over 10 trades is a shrug. The mechanism in plain words: lucky streaks and unlucky streaks partly cancel each other as trades pile up, so the average drifts toward what the strategy actually is. Every added trade dilutes the noise and concentrates the signal. That's the entire reason sample size matters — not statistics for their own sake, but making luck's disguise progressively harder to wear.
So is there a magic number of trades?
No — and anyone quoting one precisely is inventing it. What can honestly be offered is a direction:
| Sample size | What it can tell you |
|---|---|
| ~10 trades | Almost nothing — luck can fake any result |
| A few dozen | A rough sketch; big streaks still distort everything |
| Low hundreds | The shape becomes worth taking seriously |
| Several hundred+ | Luck's share is small — if the data is honest |
Treat this as a rough guide, not a law. The practical move: prefer strategies and timeframes that generate enough trades to judge. A rule that fires twice a year may be wonderful, but you won't have evidence either way for a very long time.
