Example
A strategy that truly wins 50% has about a 13% chance of winning 7 or fewer of its next 20 trades.
In practice
Sample size tells you how far to trust your numbers. Ten trades can show a great win rate by luck alone; a hundred start to show what the strategy really does.
Traders hold back from changing a strategy until enough trades have accumulated, and judge it over blocks of trades instead of day by day.
What to watch for
The trades must come from the same rules in similar conditions. Changing the method partway through, or leaving out trades you would rather forget, makes the sample worthless.