Downswing or Leak? How to Tell From Your Own Poker Numbers
How big downswings get for winning players, how often a winner is still behind after 50,000 hands, and the signs in your own results that point to a leak instead.
By PokerRoll7 min read
Every player in a long losing stretch asks the same question: is this variance, or am I doing something wrong? The question matters, because the answers point in opposite directions. A downswing calls for patience and an intact bankroll plan; a leak calls for changing something.
Your own results can answer more of it than gut feeling can, and less than most players hope. This article shows how big normal downswings are, how long a winner can stay behind, and which patterns in your numbers point to a leak.
Key takeaways
- Large downswings are a normal part of winning poker. A 3 bb/100 cash winner has roughly a one-in-two chance of ever dropping 10 buy-ins below any given point.
- A genuine winner can be behind after 50,000 hands more than one time in five.
- Your overall win rate cannot separate a downswing from a leak until the sample is huge. Look at where and when the losses happen instead.
- Leaks usually start with a change and concentrate in one part of your results. Downswings are spread across everything.
How big downswings get for winning players
The same approximation that gives risk of ruin answers a more useful question during a downswing: from where I am now, how likely is it that I will at some point be this many buy-ins lower? For cash players with a standard deviation of 90 bb/100:
| Win rate | 10 buy-ins down | 20 buy-ins down | 30 buy-ins down | 50 buy-ins down |
|---|---|---|---|---|
| 1.5 bb/100 | 69.0% | 47.7% | 32.9% | 15.7% |
| 3 bb/100 | 47.7% | 22.7% | 10.8% | 2.5% |
| 5 bb/100 | 29.1% | 8.5% | 2.5% | 0.21% |
Read the middle row. A solid 3 bb/100 winner has about a 48% chance of a 10 buy-in downswing starting from any given moment, about 23% for 20 buy-ins and about 11% for 30. And a career has many moments to start from: every new high is a fresh starting point. Over a few hundred thousand hands, a 20 or 30 buy-in downswing is close to certain even for a strong player.
That is the first thing to check. If your downswing is 15 buy-ins and you win around 3 bb/100, nothing unusual is happening.
A winner can be behind for a long time
The second check is the length. Here is where a 3 bb/100 winner's total lands after each number of hands:
| Hands | Expected result | Middle 95% of results | Chance of being behind |
|---|---|---|---|
| 10,000 | +300 bb | -1,500 bb to +2,100 bb | 37% |
| 25,000 | +750 bb | -2,000 bb to +3,500 bb | 30% |
| 50,000 | +1,500 bb | -2,400 bb to +5,400 bb | 23% |
| 100,000 | +3,000 bb | -2,600 bb to +8,600 bb | 15% |
| 250,000 | +7,500 bb | -1,300 bb to +16,300 bb | 5% |
After 50,000 hands, the same winning player is behind about 23% of the time. After 100,000, about 15%. For an online player at four tables that is months of grinding; for a live player at 30 hands an hour, 100,000 hands is more than 3,000 hours, which is several years. A losing year is entirely compatible with being a winning player.
Why your overall win rate cannot settle it
The natural move is to look at your win rate and see whether it is negative. The trouble is the error bars. The 95% range around an observed win rate is ±1.96 × standard deviation / √(hands / 100):
- a player at −2 bb/100 over 20,000 hands could truly be anywhere from about −14.5 to +10.5 bb/100;
- after 100,000 hands at −2 bb/100, the range is still about −7.6 to +3.6.
In both cases the data cannot tell a losing player from a winning one. That is not a flaw in your tracking; it is how much noise poker has. So instead of asking the total, ask the details. The hourly rate calculator shows the range for your own sample.
A quick test: is this downswing unusual for you?
If you trust your long-term win rate, you can ask how unusual the recent stretch is. Take a player who has won 3 bb/100 over a large sample and has just lost 15 buy-ins over the last 20,000 hands. A result that bad or worse happens about one time in twenty for a 3 bb/100 winner: unpleasant, not alarming. Losing 30 buy-ins over the same 20,000 hands happens about one time in 400. That is unusual enough to look for a cause.
What a leak looks like in your numbers
Leaks leave patterns that pure variance does not. These are the ones worth checking, roughly in order of how often they turn out to be the answer:
- It started with a change. A new stake, a new game or format, a new venue or site, more tables, a new schedule, a stressful month away from the table. Compare your results before and after. Variance has no reason to start on the day you moved up.
- It is concentrated. Your results by stake, format, venue, day of the week, time of day or session length are flat everywhere except one place. A downswing tends to be spread across all of them. Only trust segments with a reasonable sample, because small ones will always contain a terrible one by chance.
- It shows up in behaviour, not just results. Sessions get longer after losses, rebuys per session go up, you play later into the night, or you sit at a higher stake right after a losing session. These are signs of tilt, and they show up in a session log before they show up in your win rate.
- It persists in the details. The same kind of hand keeps costing you, in your notes or hand reviews. A downswing is many different bad outcomes; a leak is the same mistake repeating.
If none of these turns up anything and the downswing fits within the ranges above, the most likely explanation is variance, and the right response is to keep playing your normal game within your bankroll plan.
Tournament downswings are bigger and longer
Tournament results swing far more than cash, because most entries lose a buy-in and a few pay many times over. For a 20% ROI player in fields of 100 to 200, with a standard deviation of about 4 buy-ins per tournament:
| Tournaments | Expected result | Middle 95% of results | Chance of being behind |
|---|---|---|---|
| 100 | +20 buy-ins | -58 buy-ins to +98 buy-ins | 31% |
| 250 | +50 buy-ins | -74 buy-ins to +174 buy-ins | 21% |
| 500 | +100 buy-ins | -75 buy-ins to +275 buy-ins | 13% |
| 1,000 | +200 buy-ins | -48 buy-ins to +448 buy-ins | 6% |
After 250 tournaments the same winning player is behind about 22% of the time, and in fields of a thousand or more the figures are much worse. A tournament player who has not cashed for weeks has learned very little about their game, and the right check is the same as for cash: where the losses come from, and whether anything changed.
What to do during a downswing
- Check the bankroll plan first. If the bankroll has crossed your move-down line, move down now. If it has not, the downswing is not a bankroll problem yet.
- Look for the patterns above, using your session log rather than memory.
- Review hands, not results. Pick the spots that cost the most and ask whether the decision was right given what you knew.
- Protect your decisions. A stop-loss or a fixed session length removes the hands you play worst. A short break is cheaper than a tilted session.
- Change one thing at a time, so that the next stretch of results can tell you whether it worked.
Track so you can tell the difference
Every check in this article needs data most players do not keep: results by stake, format, venue and time of day, session lengths, rebuys and notes on how you played. A poker session log covers what to record. PokerRoll keeps that log for live and online play, with rebuys, notes and session lengths, and breaks your results down by format, day of the week and time of day, so the next time a downswing comes, the question "variance or leak" starts from numbers instead of from the last three sessions.
Frequently asked questions
- How do I know if I am in a downswing or just losing?
- Look at where the losses come from and when they started. A downswing is spread across your games and sits within the range your long-term win rate allows. A leak tends to start with a change, such as a new stake, game or schedule, and to concentrate in one part of your results. Your overall win rate alone cannot tell the two apart for a long time.
- How many buy-ins is a normal downswing?
- For a cash player winning 3 bb/100 with a standard deviation of 90 bb/100, the chance of ever falling 10 buy-ins below any given point is about 48%, 20 buy-ins about 23% and 30 buy-ins about 11%. Over a career with many such starting points, downswings of 20 to 30 buy-ins are expected, not exceptional.
- Can a winning poker player be behind after 100,000 hands?
- Yes. A player whose true win rate is 3 bb/100 at a standard deviation of 90 bb/100 is still behind after 100,000 hands about 15% of the time, and after 50,000 hands about 23% of the time.
- Should I move down during a downswing?
- Move down when your bankroll crosses the line your plan set for it, not because of how the downswing feels. If the plan says you still have enough buy-ins, keep playing your normal game; if it says move down, do it straight away rather than waiting for one more session.
- How long do poker downswings last?
- Long enough to feel endless. Break-even stretches of tens of thousands of hands are normal for winning cash players, which can mean months of live play. Measure a downswing in hands or hours played, not in days, and compare it with the ranges your win rate allows.
Calculators used in this article
- Poker Hourly Rate Calculator
Turn bb/100 into an hourly rate across tables and stakes, add rakeback, and see the range your true win rate could sit in after the hands you have played.
- Risk of Ruin Calculator
How likely is your bankroll to bust at your ROI and variance, and how many buy-ins you need to bring that risk under 5% or 1%.
- Poker Bankroll Calculator
Work out how many buy-ins you need for cash games, sit & gos and tournaments from your win rate or ROI, how much your results swing and the risk of going broke you accept.
Get these numbers from your own sessions
PokerRoll is a free bankroll tracker and session log for live and online poker. It turns the sessions you log into hourly rate, ROI and swings by stake and format.