How to use this tool

Poker is a game of skill, but in the short run it's a game of variance. PLO amplifies that variance dramatically: a typical winning NLHE player runs a standard deviation of 60–80 bb/100, while a typical winning PLO player runs 100–130 bb/100. Same win rate, much wider range of outcomes. The simulator above helps you see exactly how wide.

1. Enter your inputs

Win rate (bb/100).This is how many big blinds you expect to win per 100 hands, on average. A breakeven player is 0. A solid winning PLO player is typically 3–7 bb/100. A crusher is 10+. If you don't know your true win rate, start at 5 and see how the distribution changes when you drop to 2 or 0 — that's what uncertainty actually costs you.

Standard deviation (bb/100). This is how much your results bounce around per 100 hands. For online 6-max PLO, 110–120 is typical. Live PLO and heavily multiway games can push 130+. If you play mostly heads-up or a tighter style, you might be closer to 100. Compare to NLHE (60–80) to see the structural difference between the games.

Sample size. The total hands in your run. 10k is a single session. 100k is roughly three months of serious volume. 500k+ is the sample at which a true win rate starts to emerge with reasonable confidence in PLO.

2. Read the chart

The solid dark green line is the median outcome: half the simulated trials land above, half below. The inner band (25th to 75th percentile) contains the middle half of all outcomes. The outer band (5th to 95th) contains 90% of them — the edges of this band are your realistic best-case and worst-case scenarios at this volume.

The faint dashed line is the expected value— what your results would be if variance didn't exist. Notice that even at a healthy 5 bb/100 win rate, the 5th percentile line at 100k hands is usually still deeply negative. That's not a bug. That's PLO.

3. Interpret the summary

Probability of profit: what fraction of trials end above zero. If this is less than 80% at your sample size, you should expect to have losing stretches even though your play is winning.
10+ buy-in downswing odds:the chance that, during the run, your peak-to-trough drawdown exceeds 1,000 bb. For most winning PLO players at 100k+ hands, this is north of 50%. A 10 buy-in downswing is not the exception — it's the expectation.
Median max drawdown: the typical worst dip you experience during the run. Your bankroll needs to comfortably survive at least this.
95th percentile drawdown:the worst 1-in-20 scenario. This is what "running really bad" looks like for a winning player — and why 30–40 buy-in bankrolls aren't padding, they're insurance.

What the simulator actually does

Under the hood, we run 1,000 independent Monte Carlo trials. Each trial is split into 100 checkpoints. At each checkpoint, we sample a chunk of hands from a normal distribution with mean winRate * chunkHands / 100 and standard deviation stdDev * sqrt(chunkHands / 100), then add that to the running total. We track each trial's peak and the deepest drawdown from that peak. After all trials finish, we sort the cumulative values at each checkpoint to extract percentiles, and compute summary stats across the full set.

The Central Limit Theorem makes this a good approximation: the sum of many independent poker hands is (to a very close approximation) normally distributed, even though individual hand outcomes are wildly non-normal. For sample sizes above ~2,000 hands the approximation is essentially perfect; at very small samples the real distribution has fatter tails than the simulation shows.

Choosing a bankroll from this tool

A reasonable rule of thumb: your bankroll should be roughly 1.5–2× the 95th percentile drawdown you see in this simulator at a sample size matching how long you plan to play at this stake. That gives you a margin of safety for both the simulated variance and the things the simulator can't capture (tilt, leaks that emerge during downswings, unexpected life expenses, etc.).

For more detail on building a bankroll plan, see Bankroll Management for PLO. For the theory of why PLO variance is so much nastier than NLHE, read Variance in PLO: Why It Feels So Brutal.

Things the simulator does not model

Changing win rates. Real players improve (and regress) over time. The simulator assumes a constant true win rate for the entire run.
Tilt and leaks. The simulator assumes perfect execution. Real downswings often compound with tilt-induced mistakes.
Game selection changes. If you move up or down stakes mid-sample, the effective win rate and std dev change. Run separate simulations for each stake.
Fat-tail events. The normal approximation is excellent for typical outcomes but slightly understates the probability of extreme runs. Real PLO has slightly fatter tails than the simulator shows — another reason to pad your bankroll.

PLO Variance Simulator