Winrate confidence

The margin of error around a measured winrate, which shrinks only with the square root of the hands played.

Last updated October 16, 2026

What it means: A winrate from a sample is an estimate of the true one. The confidence interval is the range the true winrate probably lies in, usually at 95%.
How it is calculated: Winrate ± 1.96 × standard deviation ÷ √(hands ÷ 100), with both the winrate and the standard deviation in big blinds per 100 hands.
Worked example: At 5bb per 100 over 20,000 hands with a standard deviation of 80, the interval is -6.1 to 16.1. Proving a winrate above zero at 95% takes about 98,000 hands.
Why it matters: It shows why results over a few thousand hands say almost nothing about skill, and why graded decisions are a faster measure of how well you play.

Formula and worked example: How sure is a winrate? in poker math

Lessons that teach it: How sure is a winrate? · lessons other than the free one need an invite