Free, no login
Bets to prove an edge
Type a win rate, an American price, and a confidence level, and the sample size calculator gives you a bet count. At -110 and a 54.476% win rate, the edge is 4%. At 95% confidence that sample is about 2,260 bets.
Assumptions
Win rate is the chance a bet wins. Odds are American. Edge is the average profit on a 1 unit stake. A win pays the price. A loss costs 1 unit. It assumes every bet is independent, with a flat stake and a steady edge. At 95% confidence, the sample is the point where your edge reaches two standard errors (z = 2). 90% uses 1.64. 99% uses 2.58. If your bets are correlated or your edge changes, you need a bigger sample than this.
Example
Win rate 54.476%, odds -110, 95% confidence. Edge is 4%. Variance per 1 unit bet is about 0.904. Bets needed is about 2,260.
Runs in your browser. Nothing is sent to EdgeTilt. For your own math, not betting advice.
FAQ
What do I type?
Win rate in percent, American odds, and 90%, 95%, or 99% confidence.
What is the edge?
Average profit per 1 unit stake at that win rate and price.
Why does 95% use 2, not 1.96?
This page uses two standard errors, the same rule as n = 4 x variance / edge squared.
What if the edge is zero or negative?
Then there's no sample size. Your win rate doesn't beat the price.