Simple buy-in count
A 5,000 bankroll with a 100 buy-in contains 50 buy-ins, without implying that 50 is safe.
Poker Bankroll Calculator
Convert a bankroll into buy-ins or solve a positive-edge risk model for a target failure probability. The poker bankroll calculator reports mathematical scenarios, not a universal stake rule or financial guarantee.
Convert bankroll or solve a risk model
Basic mode is a direct conversion. The risk model is an approximation, not a financial or profit guarantee.
How the result is built
More Bankroll Reduces Risk but Does Not Remove Variance
Under a fixed positive-edge Brownian approximation, modeled risk declines with bankroll but remains conditional on win-rate and variance assumptions.
Current bankroll3,000 bb
Estimated current risk2.35%
Target risk5%
Modeled target bankroll2,397 bb
| Bankroll (bb) | Estimated risk |
|---|---|
| 0 | 100% |
| 750 | 39.16% |
| 1,500 | 15.34% |
| 2,250 | 6.01% |
| 3,000 | 2.35% |
| 3,750 | 0.92% |
| 4,500 | 0.36% |
| 5,250 | 0.14% |
| 6,000 | 0.06% |
Basic conversion has no risk claim
Bankroll divided by buy-in answers only how many entries or stacks the amount contains. It says nothing about whether that count is sufficient.
Model mode needs a positive edge
The exponential approximation uses win rate, standard deviation, and target risk in consistent big-blind units. Small changes in uncertain win rate can move the result sharply.
Real bankroll paths are not fixed
Game selection, changing stakes, withdrawals, correlated sessions, and estimation error violate the stationary model. Treat outputs as stress-test scenarios.
Worked examples
Lower risk target
Holding edge and variance fixed, demanding a lower ruin probability increases the modeled bankroll.
Frequently asked questions
Does the calculator recommend a stake?
No. It converts or estimates from your assumptions.
Why is win rate required in model mode?
The risk formula depends on a positive expected increment.
Are tournament buy-ins modeled by the cash formula?
No. Tournament return distributions need a discrete model.
Can variance estimates be wrong?
Yes. Historical samples may be noisy and future games may differ.