Poker variance simulator
pokalab.comHow sure are you of your winrate?
Your true winrate, from your sample
pokalab.comWhat standard deviation should you use?
Regulars' standard deviation by stake and room (bb/100)
| Stake | GGPoker | iPoker | WPN | 888poker |
|---|---|---|---|---|
| NL25 | 76 69–84 | — | — | — |
| NL50 | 78 71–85 | 85 79–91 | 75 69–80 | 86 80–92 |
| NL100 | 81 74–87 | 87 80–94 | 75 69–81 | 89 85–94 |
| NL200 | 82 76–88 | 90 83–96 | 75 70–82 | 91 85–98 |
| NL500 | 83 78–89 | 91 87–96 | — | 94 86–97 |
| NL1000 | 90 84–97 | — | — | — |
| Stake | Typical | Middle half |
|---|---|---|
| NL25 | 76 | 69–84 |
| NL50 | 78 | 71–85 |
| NL100 | 81 | 74–87 |
| NL200 | 82 | 76–88 |
| NL500 | 83 | 78–89 |
| NL1000 | 90 | 84–97 |
Standard deviation in bb/100 of regulars’ results after rake: the typical regular in bold, then the range of the middle half of regulars. A dash: too few regulars with that many hands. PokerStars is not included yet: a defect in how its all-in cash-outs are recorded slightly understates its swings, and it is fixed in our next data refresh.
pokalab.comHow it works
- The model. Results over 100 hands follow a normal distribution with a mean equal to your winrate and a spread equal to your standard deviation, one block of hands independent of the next. It is the model every variance calculator uses, and it holds well for a regular’s results over thousands of hands.
- The simulation. 2,000 runs of the hands you chose, drawn 100 hands at a time (coarser past 200,000 hands). The bands show where 70% and 95% of the runs are at each point; “New runs” draws another set. Downswings and stretches without a new high are measured on those steps, so they are, if anything, slightly understated.
- The exact parts. The expected result, the ranges, the chance to be down and the observed winrate come from the formula, not the simulation. Risk of ruin is exp(−2 × winrate × bankroll ÷ SD²), with the winrate and SD in bb/100 and the bankroll in bb: the chance of ever losing the whole bankroll, playing forever at that winrate.
- The link. Your inputs are in the page’s address: copy it to share a result.
Method note. The standard deviation presets come from real-money NLHE cash hands in Pokalab’s database of 379M real hands: GGPoker, iPoker, WPN and 888poker, 6-max with four to six players dealt, no ante; no simulated hands and no user uploads. A regular is a player of the REG family in our segmentation, judged on his no-ante history on that room; each one with 20,000 hands or more at a stake gives one value: ten times the standard deviation of his result per hand, after rake. The table shows the middle value and the middle half of those players, room by room; the rooms’ hands come from different years. Measured in October 2026 and re-measured independently from the hand records.
Questions
What is a normal standard deviation in poker?
For 6-max no-limit hold’em cash, the middle half of regulars measures between 69 and 98 bb/100, depending on the stake and the room; the typical NL50 regular on GGPoker sits at about 78. The room matters more than the stake: see the table above.
How many hands until I know my winrate?
More than most players think. At 3 bb/100 with a standard deviation of 78, it takes about 183,000 hands to be 95% sure you are a winner at all. The calculator gives the number for your own inputs under “Hands to prove it”.
What bankroll do I need?
It depends on your winrate as much as on the stake. At 3 bb/100 and an SD of 78, a bankroll of about 31 buy-ins keeps the risk of ruin under 5%; halve the winrate and you need twice as much.
Why do my results look worse than my winrate?
Because the median run of a winning player spends long stretches without a new high, and deep downswings are normal. Look at the downswing table: what feels like running bad is usually inside the 70% band.