How many hands is enough?
Poker statistics settle far more slowly than people assume, and they do not settle at the same speed as each other.
The filter changes how often each spot comes up and how often players take it, so chances, margin and hands to quote move with it. The band each statistic is judged against, the same one behind the confidence bar on every profile, does not.
What your sample is worth
Drag to your own hand count.Chances: how often this pool offers the spot (VPIP 98 per 100 hands here, Fold vs 3Bet about 2.1).Margin: how far the figure can sit from the player’s true rate, 19 times in 20.Confidence: the chance of landing inside the statistic’s band. The profile bar fills to it.Hands to quote it: hands until three in four.
| Stat | Chancesin 10,000 hands | Margin95%, in points | Confidenceof landing inside its band | At this sample | Hands to quote ituntil three in four |
|---|---|---|---|---|---|
| VPIP | 9,769 | ±0.9 | 100% | quotable | 1,290 |
| PFR | 9,769 | ±0.7 | 96% | quotable | 3,158 |
| Limp | 9,769 | ±0.5 | 100% | quotable | 649 |
| 3Bet | 3,560 | ±0.8 | 79% | solid | 8,610 |
| Fold vs 3Bet | 210 | ±6.6 | 59% | a direction | 19,942 |
| 4Bet | 210 | ±4.2 | 36% | not evidence | 59,068 |
| Steal | 2,212 | ±1.9 | 89% | solid | 5,311 |
| Fold vs Steal | 860 | ±3.2 | 72% | a direction | 11,356 |
| Cold Call | 1,289 | ±2.0 | 91% | quotable | 4,677 |
| CBet flop | 663 | ±3.7 | 64% | a direction | 15,685 |
| Fold vs CBet | 481 | ±4.4 | 56% | a direction | 22,367 |
| WTSD | 2,736 | ±1.8 | 63% | a direction | 16,733 |
| W$SD | 907 | ±3.2 | 63% | a direction | 16,167 |
| WWSF | 2,736 | ±1.8 | 60% | a direction | 18,381 |
| AFq | 5,522 | ±1.2 | 89% | solid | 5,145 |
Is this gap real?
The other way round: not how long a statistic takes, but whether the difference in front of you survives the sample it was measured on. This is the test the radar’s bands and the verdict panel already apply — here you can move it.
Where these come from
Margins are the standard formula for a percentage over n chances. It was checked on real players by splitting their hands into two interleaved halves, and the halves agree as closely as the formula predicts.
Methodology has the full argument. What a value is worth covers the same fifteen statistics.