Poker variance calculator

Updated 12 September 2026

Cash games, online or live. Put in your win rate and your volume and see the whole spread of results it can produce — including the downswing you are probably in the middle of right now.

The short answer. A winning poker player loses for far longer than feels reasonable. At 5 bb/100 with a standard deviation of 100 bb/100, you finish 50,000 hands down about 13% of the time, and a 30 buy-in downswing is a normal career event rather than a disaster. Use the tool below to put your own numbers in. If your result lands inside the middle 70% of outcomes, it is variance and the answer is to keep studying. If it lands outside the 95% band, it is probably a leak.

Your numbers

Standard deviation is how much your results bounce around, in the same unit as your win rate. Typical: Online 6-max 75–120 · Online full ring 60–80 · Live 70–100 · PLO 6-max 120–160 bb/100. Your tracker shows yours; if you do not have one, leave it at 100.

50,000 hands ≈ 333 hours at 150 hands an hour (online, 2 tables).

$1,500 is 30.0 buy-ins at NL50 (one buy-in = 100 bb = $50.00).

Compare your actual result (optional)

What that means

Over 50k hands at 5.0 bb/100 you expect +$1,250 (25.0 buy-ins) — but you still finish in the red 13% of the time, and a $1,500 roll (30.0 buy-ins at NL50) carries a 5% risk of ruin.

+$1,250Expected result · 2,500 bb · 25.0 buy-ins
87%Chance you finish ahead over 50k hands
5%Risk of ruin on 30.0 buy-ins (simulated 2.8%)
$1,498Bankroll for a 5% risk of ruin · 30.0 buy-ins
+$3.75Per hour at NL50, 150 hands/hour
17.1 BITypical worst downswing inside this sample
$0$2.5k020k40k
  • Expected value
  • Middle 70%
  • Middle 95%
  • 20 sample runs
Dollars up the side, hands along the bottom. Every grey line is one possible version of the same 50,000 hands, played by the same winning player.

Where you could finish

LuckResult in bbIn dollarsBuy-ins
Best 1%7,702+$3,85177.0
Top 5%6,178+$3,08961.8
Top 25%4,008+$2,00440.1
Median2,500+$1,25025.0
Bottom 25%992+$4969.9
Bottom 10%-366−$183-3.7
Worst 1%-2,702−$1,351-27.0

Read it as: if a thousand players with your exact win rate each played these 50,000 hands, the luckiest ten would finish near the top row and the unluckiest ten near the bottom one.

How likely is a downswing?

DownswingIn dollarsPeak to trough in 50k handsThat far below where you started
10 buy-ins$50095%33%
20 buy-ins$1,00036%10%
30 buy-ins$1,5008.4%2.7%
50 buy-ins$2,5000.6%0.1%

Two different questions. Peak to trough is the drop from your best point to your worst — the one that feels like a downswing. Below where you started is how often you are that far behind your starting bankroll, which is the one that busts you.

How long until you know your win rate?

To pin your win rate down to ±1 bb/100 with 95% confidence takes 3.8M hands at a standard deviation of 100. Within ±2 bb/100 takes 960.4k; within ±3 bb/100, 426.9k. That is why a 20,000-hand sample tells you almost nothing on its own, and why studying beats staring at your graph.

If your true win rate is lower than you think

True win rateRisk of ruin on 30.0 buy-insBankroll for 5% risk of ruin
1.5 bb/10041%$4,993 (100 buy-ins)
2.5 bb/10022%$2,996 (60 buy-ins)
5.0 bb/1005%$1,498 (30 buy-ins)
7.5 bb/1001.1%$999 (20 buy-ins)

Risk of ruin assumes you keep playing this stake all the way to zero. In real life you move down, which is why a disciplined player with a thin roll rarely busts — and why the number is a warning light, not a prophecy. See the bankroll calculator for the stake that fits what you actually have.

What this calculator is doing

Poker results are a random walk with a slope. The slope is your win rate, measured in big blinds per 100 hands (bb/100) — the amount you expect to win, on average, every hundred hands. The randomness around the slope is your standard deviation, measured in the same unit. The calculator takes those two numbers and plays your sample out thousands of times, in blocks of 100 hands, recording where each run finished and how far it fell along the way.

Two facts drive everything else on this page. First, the expected profit grows in a straight line with the hands you play, but the swing around it only grows with the square root of them. Play four times as many hands and your expected win is four times bigger while the uncertainty only doubles. That is why volume is the only thing that converts an edge into money, and why nobody can judge their own play from a weekend.

Second, the swing is enormous compared with the edge. A strong online winner makes 5 bb/100 and carries a standard deviation of around 100 bb/100 — twenty times larger. Over 10,000 hands the edge is worth 500 bb and the noise is worth roughly 1,000 bb, so the noise simply drowns the signal. The edge only wins the argument at hundreds of thousands of hands.

Downswings: what is actually normal

The downswing table above answers two different questions, and mixing them up is the most common mistake players make. The peak-to-trough figure is the drop from your best ever point to your worst point afterwards. That is the number that feels like a downswing, and it is always bigger than people guess: a 20 buy-in peak-to-trough drop is close to routine for a winning 6-max player inside 100,000 hands.

The second figure, how often you are that far below where you started, is the one that busts bankrolls, because it is measured from the money you actually had. A 30 buy-in peak-to-trough dip after you have already banked 40 buy-ins of profit is uncomfortable. The same 30 buy-in dip starting from day one, on a 30 buy-in roll, ends your career at that stake. Same variance, completely different consequence — which is the entire argument for bankroll management.

Risk of ruin, and why it reads worse than it is

Risk of ruin is the chance of losing your whole bankroll before your edge pulls you clear. The classic Malmuth–Grossman formula behind the tile above assumes something untrue about you: that you keep firing at the same stake all the way to zero. Real players move down a level when the roll thins out, which both shrinks each bet and buys time for the edge to show up. Treat risk of ruin as a measure of how much stress your roll is under, not as a prediction.

What is not negotiable is the sensitivity. The formula divides by your win rate and squares your standard deviation, so halving your assumed win rate doubles the bankroll you need, and a looser style that lifts your standard deviation by 40% roughly doubles it again. Most bankroll disagreements are really disagreements about win rate, and almost nobody knows their own to better than a couple of big blinds.

Live poker: fewer hands, same mountain

A live table deals about 27 hands an hour, against roughly 75 an hour per online table. So a live player grinding 30 hours a month sees about 9,700 hands a year — less than a two-table online player gets through in a fortnight. Live win rates in big blinds are much higher (good $1/$2 regulars talk about 8–15 bb/100), which is why live players can run on fewer buy-ins. But the number of hands needed to prove a win rate does not care how long those hands took. A live professional can be two years into a career and still hold a sample that proves nothing.

That is the real cost of live variance: not the size of the swings but how slowly the truth arrives. Switch the toggle to Live above and the tool prices everything per hour and per month, which is how live players actually think about money.

Variance or a leak? The test that matters

Enter your real result and the hands it took. The tool compares it with the spread of outcomes for the win rate you believe you have, and puts you in one of three places.

  • Inside the middle 70%. There is nothing to explain. This is what your win rate looks like from the inside. Keep your volume up and keep studying.
  • Outside 70%, inside 95%. A genuinely cold run. Variance still covers it, but this is the sensible moment to review hands rather than wait it out.
  • Outside 95%. One in twenty results land here by chance, so it is not proof — but the odds now favour the simpler explanation. Assume a leak until you have looked properly.

The reason that distinction is worth having is that the two situations need opposite responses. Variance punishes people who change their game to chase it. A leak punishes people who wait for it to pass. The calculator cannot tell you which spots you are misplaying — but it can tell you whether to go looking.

Where to go next

Frequently asked questions

What is a poker variance calculator?

It is a simulator. You give it your win rate in big blinds per 100 hands, your standard deviation (how much your results bounce around) and how many hands you want to play. It then plays that stretch of poker thousands of times over and shows you the spread of results: the best and worst runs, the chance you finish in profit, the size of downswing you should expect, and the chance a bankroll of a given size does not survive.

How long can a winning poker player lose for?

Longer than almost anyone expects. At 5 bb/100 with a standard deviation of 100 bb/100 — a solid online 6-max winner — you finish 50,000 hands in the red about 13% of the time, and a 30 buy-in downswing happens to that player often enough to be a normal career event. Losing stretches of 20,000 to 40,000 hands are routine. Set the calculator to your own numbers and read the downswing table; it is the single most calming thing in poker.

What is a normal standard deviation in poker?

For no-limit Hold’em cash games: roughly 75–120 bb/100 online 6-max, 60–80 bb/100 online full ring, and 70–100 bb/100 live. Pot-limit Omaha runs far higher, around 120–160 bb/100. A looser, more aggressive style pushes your number up; a tight, straightforward style pulls it down. Any tracker (PokerTracker, Hold’em Manager) reports yours directly, and it matters more than people think: doubling your standard deviation quadruples the bankroll you need for the same risk of ruin.

What is risk of ruin in poker?

The chance that you lose your entire bankroll before your edge rescues you, assuming you keep playing the same stake all the way down. The calculator uses the standard Malmuth–Grossman formula and cross-checks it against the simulation. It is a useful warning light, not a prophecy: in real life you move down a level when your roll shrinks, and a player who moves down almost never actually busts.

How many hands do I need to know my win rate?

To know your true win rate within ±1 bb/100 at 95% confidence takes roughly 3.8 million hands at a standard deviation of 100. Within ±2 bb/100 it is around 960,000, and within ±3 bb/100 about 430,000. That is the uncomfortable truth of a 20,000-hand sample: it is compatible with being a small winner and with being a small loser at the same time.

Is my downswing variance or a leak?

Enter your actual result and the hands it took in the tool above. If it lands inside the middle 70% of outcomes for your assumed win rate, the maths has nothing to explain — that is ordinary variance, and the answer is to keep studying. If it lands outside the 95% band, variance alone very rarely stretches that far, and you should treat it as a leak until you have looked for one.

Does this work for live poker?

Yes. Switch the toggle to Live and the tool converts hours into hands at about 27 hands an hour, prices everything at live stakes ($1/$2 through $5/$10) and shows your expectation per hour and per month. Live win rates in big blinds are much higher than online ones, but you play a fraction of the hands, so in dollars per hour the swings feel slower and in hands they are just as long.

Does it handle tournaments?

Not yet. This version covers cash games only, because tournament variance is a different shape: most of your profit arrives in a handful of deep runs, so the distribution is skewed rather than bell-shaped and the bankroll numbers are far larger. Using a cash model on tournaments would flatter you badly, so we would rather leave it out than get it wrong.

Public maths, no secret sauce. Everything here is the standard statistics of a random walk — the normal approximation for the bands and percentiles, the Malmuth–Grossman formula for risk of ruin, and a seeded Monte Carlo for the sample paths and drawdowns. Identical inputs give identical output every time, so a shared link always shows the same picture.

Variance is the price of a real edge. Make sure the edge is actually there — drill the spots you get wrong.

Ready to train?

Play real spots, see the solver’s answer and what your choice cost. Free to start, no real-money play, 18+.

iPhone: coming soon