Blog · Stats

How many hands before you know you are a winner? The truth about sample size

You've played 3,000 hands, your tracker shows +12 BB/100, and you're wondering whether it's time to move up. Or the opposite: you're at -8 and wondering whether you're simply bad. In both cases, the honest answer is the same — that number tells you almost nothing. Not because your tracker is wrong, but because poker produces results so scattered that it takes a huge volume before a winrate becomes information. Here's how much, why, and what to judge yourself on in the meantime.

Why a winrate takes so long to stabilize

Your BB/100is an average. And an average is only reliable if the values it's made of don't stray too far from one another. In poker, it's the exact opposite: most hands cost or win you a fraction of a big blind, and every now and then a 100 or 200 big blind pot comes along and crushes everything. A single set over set, a single flush draw that gets there on the river, and your average over a few thousand hands flips.

That spread is measured by the standard deviation, also expressed in BB/100. It depends on your style: an aggressive player who plays a lot of big pots has a higher standard deviation than a tight one. Above all, remember the mathematical rule that follows: the precision of your winrate only improves with the square rootof the number of hands. To cut your margin of error in half, doubling your volume isn't enough — you have to multiply it by four.

What your BB/100 really proves, step by step

To make this concrete, let's assume a standard deviation of 90 BB/100 — a plausible order of magnitude in 6-max cash game; yours may be higher or lower. We then compute the range in which your true winrate sits with 95% confidence. These are calculation illustrations, not statistics from real players:

The lesson isn't that you have to wait for 100,000 hands before making any decision. It's that your results can't be your steering wheel in the short and medium term. That's also why we recommend a real validated volume before changing stakes, as explained in our guide to bankroll from NL2 to NL10.

The stats that do stabilize fast

Good news: your frequency stats don't follow the same slow pace. VPIP and PFR measure a binary choice (you enter the pot or you don't) that comes up every hand, without pot size playing any part. With a true VPIP of 22% and 1,000 hands played, the margin of error is already down to about ± 2.6 points. After a few thousand hands, you know quite precisely whether you play too many hands or too few — long before you know whether you're winning.

Be careful, though, with stats for rare situations. A squeeze or check-raise rate is only computed over the times the spot actually came up. If you've only had 200 squeeze opportunities, a displayed 8% rests on 16 squeezes, and your true rate could sit anywhere between 4 and 12%. The practical rule: always look at the number of opportunities behind a percentage, never the percentage alone.

What to judge yourself on while you build volume

Since results are too noisy, shift your attention to what isn't. Three indicators are much faster and more honest than raw BB/100:

Your frequencies compared to reference ranges. A 35% VPIP in 6-max over 5,000 hands is a real problem, whatever your results were over the same period. Conversely, frequencies consistent with a solid strategy remain a good sign even during a downswing.

All-in EV adjusted results.Most trackers can recompute your winnings by replacing the actual outcome of all-ins with their expected value. It doesn't remove all variance — draws that get there before the all-in still count — but it strips out the most violent part.

The quality of your decisions in big pots. Reviewing your biggest losses by asking whether the line was correct, not whether you got lucky, tells you more about your level than a month of results. If you're going through a rough patch, variance may explain everything — but only that review lets you know for sure.

What about tournaments?

In MTTs, the problem is even sharper. A large share of a profitable player's winnings comes from a few rare deep runs, so it takes several hundred tournaments before an ROI becomes reliable. Over 50 tournaments, a single final table can turn a mediocre player into an apparent crusher, or its absence can make a very good player doubt themselves. We break this mechanism down in our article on ROI and variance in tournaments.

Frequently asked questions