How many trades are enough to trust a scalping result?

[Amanhasnoname]

A hundred trades with a 60 percent win rate feels convincing. Statistically, it's weak evidence.

With a fair coin, getting 60 heads in 100 flips happens a few percent of the time. A strategy with no edge could produce that result. Add the fact that many strategies were tested before this one was chosen, and the odds of a lucky streak go up.

For scalping, the useful numbers are bigger. Several hundred trades give a rough idea. A thousand or more is better, spread across different market conditions.

Average profit per trade matters more than win rate. The confidence interval around it can be estimated from the standard deviation of trade results. If the interval includes zero, the edge isn't proven.

Splitting the trades into halves and checking that both are profitable is a quick sanity test. So is looking at results by month.

Patience with sample size is unglamorous but essential.

How many trades do you require before believing a result?