Poker Topic for 2/28/2024: Poker Sample Sizes

Last month I compared live and online poker by looking at possible winrates. To recap, live poker is far more profitable than online when looking at the metric of big blinds won per 100 hands played. The competition in online poker is far tougher than in live poker, and it’s not even close.

Where online poker has the edge over live poker is volume. It’s possible to see far more hands when playing online: 12 times as many for a player who’s 4-tabling a 6max game. Online poker also revealed the importance of taking into account sample size when analyzing results, to determine that a current observed winrate isn’t just the result of positive variance (running good).

Given the relatively smaller winrates in online poker and the fluctuations inherent in the game, a large sample size of hands is needed: many online players recommend at least 100,000 hands for a reasonable sample size, if not more! (By the way, it’s worth playing around with a poker variance calculator with different sample sizes of hands to look at possible outcomes given a particular winrate.)

Now, let’s take an online poker pro who plays 360 hands an hour, for 40 hours a week. It would take around 7 weeks to reach 100,000 hands. A live player, playing 30 hands per hour, would need over 3,300 hours to reach 100,000 hands! Let’s say that this live player plays 40 hours a week; she would need 83 weeks, over a year and a half, to reach that many hands! And that’s assuming a nonstop grind, with no breaks or time off. It would take years for a recreational player who only plays a few hours per month to amass a reasonable sample size of hands!

Now, I did point out last week that live poker can be more than 3 times as profitable as online poker, as far as winrates go, so that would reduce the sample size required. So what kind of sample size is reasonable for live poker?