How many trades are enough before a result means anything
Posted: Sat Oct 31, 2026 6:47 am
Fifty trades with a high win rate feels convincing, but the sample is far too small. Randomness can produce long winning streaks, and the confidence interval around a 60 percent win rate on 50 trades is wide enough to include a break-even or losing strategy.
For scalping, where each trade has a small expected value, the needed sample is larger. A few hundred trades is the bare minimum, and a thousand is better, especially when the edge is a fraction of a pip per trade.
Costs add noise too. Variation in spread and slippage affects results, so a small sample might look good simply because it landed in a quiet period.
Splitting the sample into chunks and checking consistency helps. If performance is strong in some months and terrible in others, the average hides instability.
Patience is hard when each trade takes minutes and a thousand is a few weeks of work, but conclusions drawn too early waste money.
What trade count was needed before confidence in a strategy became reasonable?
For scalping, where each trade has a small expected value, the needed sample is larger. A few hundred trades is the bare minimum, and a thousand is better, especially when the edge is a fraction of a pip per trade.
Costs add noise too. Variation in spread and slippage affects results, so a small sample might look good simply because it landed in a quiet period.
Splitting the sample into chunks and checking consistency helps. If performance is strong in some months and terrible in others, the average hides instability.
Patience is hard when each trade takes minutes and a thousand is a few weeks of work, but conclusions drawn too early waste money.
What trade count was needed before confidence in a strategy became reasonable?