Poker Downswing Calculator: Is This Normal?
Is this downswing normal? Enter how far down you are and over how many hands. This works out how often a downswing that deep happens at the win rate you think you have, how likely the stretch is if you are break-even instead, and whether the answer is keep playing, move down or check your hands. Cash games, in big blinds per 100.
Example filled in: 15 buy-ins down over 30,000 hands, with 30 buy-ins left and no firm win rate. Change anything.
Step 1: How many hands since the peak of your graph?
Since the peak of your graph.
Step 2: How far down are you?
1 buy-in = 100 big blinds.
Step 3: How big is your bankroll now?
At the stake you are playing.
Step 4: What is your win rate?
Your honest estimate, not your best month. In big blinds won per 100 hands ().
How much does your game swing? (, in bb/100)
Leave it at typical unless you track it.
Out of 100 runs of 30,000 hands at 2 bb/100, 62 go at least 15 buy-ins down (±2.1 points). An ordinary downswing: keep playing your game.
- 62 go at least 15 buy-ins down
- 38 never go that deep
A normal downswing
A drawdown at least this deep happens in about 62% of runs at the win rate you entered, and the median worst drawdown over 30,000 hands is 17.3 buy-ins. Nothing here says you have stopped being a winner.
- Keep playing, keep the same stake
- Review hands as usual, not because of the graph
- If the swing is affecting how you play, take a break for that reason, not this one
Sampled · 2,000 runs · ±2.1 points · seeded, so the same inputs always give the same answer
Exact for this model · a normal tail, not sampled: the two ending chances.
Runs 2,000 simulated runs of 30,000 hands in your browser. Nothing is uploaded.
Why
Over 30,000 hands a 2 bb/100 winner expects to be up 6 buy-ins, but luck moves the total by about 17.3 buy-ins either way. So ending 15 buy-ins down or worse happens 11.3% of the time to that winner, and 19.3% to a break-even player. Both are ordinary enough that this stretch cannot tell you which one you are.
That is variance: the swing of luck around your true win rate.
How deep, how often
| Inside 30,000 hands | Runs | Margin |
|---|---|---|
| A 10+ buy-in downswing | 92.2% | ±1.2 points |
| A 20+ buy-in downswing | 36.3% | ±2.1 points |
| A 30+ buy-in downswing | 9.4% | ±1.3 points |
| A 40+ buy-in downswing | 2.0% | ±0.6 points |
| A 50+ buy-in downswing | 0.1% | ±0.1 points |
| A 70+ buy-in downswing | 0.0% | none of 2,000 runs |
| A 100+ buy-in downswing | 0.0% | none of 2,000 runs |
Sampled from the same 2,000 seeded runs as the answer.
What the 70 billion hand study says
Downswings feel personal and are mostly arithmetic. Across 70 billion simulated hands, 81.12% of genuine 3bb/100 winners hit a 20+ buy-in downswing inside 100,000 hands, and 38.47% of their 10,000-hand stretches end below break-even.
Variance, or a leak?
Finishing 15 buy-ins down over 30,000 hands happens 11.3% of the time to a 2 bb/100 winner, and 19.3% of the time to a break-even player. Both are ordinary enough that this stretch does not tell you which one you are.
These two are exact for this model (a normal tail probability), not simulated. The depths above come from 2,000 simulated runs, seeded so the same inputs always give the same answer.
How big is a normal poker downswing?
Bigger than almost everyone expects. In a study of 70 billion simulated hands, 81.12% of players with a genuine 3bb/100 win rate hit a downswing of 20 buy-ins or more within 100,000 hands, and 5.85% hit one of 50 buy-ins or more. The median worst stretch for that player is 27.1 buy-ins, and their longest run without making a new high is a median of 40,173 hands.
| True win rate | 20+ buy-in downswing | 50+ buy-in | Median worst | Still down after 100k |
|---|---|---|---|---|
| 1bb/100 | 90.79% | 15.35% | 32.6 BI | 37.56% |
| 2bb/100 | 86.46% | 9.88% | 29.6 BI | 26.4% |
| 3bb/100 | 81.12% | 5.85% | 27.1 BI | 17.07% |
| 4bb/100 | 74.44% | 3.45% | 24.9 BI | 10.29% |
| 5bb/100 | 67.67% | 1.91% | 23.2 BI | 5.75% |
| 7bb/100 | 52.59% | 0.53% | 20.4 BI | 1.29% |
| 10bb/100 | 31.14% | 0.07% | 17.3 BI | 0.07% |
Per 100,000 hands, at a standard deviation of 100bb/100. Full table and the free CSV: poker downswing odds.
Why you cannot tell variance from a leak by looking
A break-even player and a 3bb/100 winner produce overlapping graphs for tens of thousands of hands: 38.47% of a real winner's 10,000-hand stretches finish below break-even, and 25.07% of their 50,000-hand stretches do. That is why this calculator shows how likely your stretch is under both assumptions instead of announcing which one you are. The honest answer from results alone is usually “not enough hands to say”, and the way past it is to look at decisions rather than at the graph.
When to move down
Moving down is a bankroll decision, not a punishment. The common rule is to keep 30 or more buy-ins for the stake you play and to drop a level when you fall under about 20, because the risk of losing the roll grows faster than the extra money per hand at the higher stake. Nothing about that rule depends on whether the downswing was bad luck or bad play, which is what makes it useful while you are still deciding. The bankroll calculator tracks the buy-in count, and the variance simulator draws the sample paths this tool summarises.
When to take a break
Take a break if the downswing has changed how you play: chasing losses, playing longer sessions than planned, moving up to win it back. That is a reason found in your own behaviour, not in the graph. If your sessions are logged, the poker journal will show whether your worst results cluster with late hours or bad moods. If the swing has not changed your game, there is nothing here that says stopping helps.
Cash games only
These figures are for cash games, where results per hand are close to normally distributed. Tournament swings are far wilder and need a different model: most of the return sits in rare big finishes, so downswings of 100 buy-ins or more are ordinary for a winning tournament player. Do not read these numbers as tournament numbers.
Wondering whether the site is dealing fairly instead? The arithmetic is in is online poker rigged? (the short version is that online play deals about eleven times more hands per hour, so it delivers eleven times the coolers in the same evening).
Frequently asked questions
How big is a normal poker downswing?›
Much bigger than most players expect. Over 70 billion simulated hands, 81.12% of genuine 3bb/100 winners hit a downswing of 20 buy-ins or more within 100,000 hands, with a median worst stretch of 27.1 buy-ins. A 1bb/100 winner's median worst downswing is 32.6 buy-ins.
Am I running bad or am I a losing player?›
From results alone, usually you cannot tell. A real 3bb/100 winner finishes below break-even in about 38% of 10,000-hand stretches and about 17% of 100,000-hand stretches. This calculator shows how likely your stretch is if your win rate is what you think it is, and how likely it is if you are break-even. When the two are close, the honest answer is that the sample is too small. Check your decisions instead.
When should I move down in stakes?›
When your bankroll falls below about 20 buy-ins for the stake. It is a bankroll rule, independent of whether the downswing was luck or leaks: below that point the chance of losing the whole roll rises faster than the extra money per hand at the higher stake.
Should I take a break during a downswing?›
Take a break if the downswing has changed how you play: longer sessions than planned, chasing losses, moving up to get even. That is a reason in your behaviour, not in the graph. If your game has not changed, stopping does not improve your results.
Does this work for tournaments?›
No. It models cash-game results, which are close to normally distributed per hand. Tournament returns come mostly from rare large finishes, so downswings of 100 buy-ins or more are normal for winning tournament players and these figures would understate them badly.
How is the calculation done?›
Each hand is drawn from a normal distribution with your win rate as the mean and your standard deviation as the spread, the same model as the 70-billion-hand study. The chance of a result at least this bad is a closed-form normal probability. The depth of the worst drawdown inside a stretch has no closed form, so it is simulated over 2,000 runs with a fixed seed, and the simulation is tested against the published study's figures.
Related tools
Simulate sample paths from your win rate and stddev. See real variance.
Tournament downswings from a real payout structure: bands, drawdown, dry runs.
Upload 50+ hands. AI returns your top 3 leaks with money lost.
One hand, worked through.
Occasional strategy emails: one interesting hand, broken down with equity and range against range. Free, unsubscribe any time.
Unlimited AI hand reviews and every cheatsheet: PRO from $9/mo