Every poker player has a story about a brutal downswing. But how big do downswings actually get for a winning player, and how long can you run bad before it means something is wrong? To answer that, we simulated 30,000 online cash game players, each playing 500,000 hands, and measured every swing.
Key findings
- A solid winner at 5 bb/100 had a median worst downswing of 31 buy-ins over 500,000 hands. 96% of them suffered a 20 buy-in downswing at some point.
- The same player's longest stretch without a new high was typically about 69,000 hands, and one in ten went 128,000 hands or more.
- After 10,000 hands, 29% of these genuinely winning players were still showing a loss.
- Starting with 20 buy-ins, 8% of 5 bb/100 winners went broke. With 30 buy-ins, 2%.
- A small winner at 2 bb/100 had a median worst downswing of 46 buy-ins, and 40% had a downswing of 50 buy-ins or more.
How the simulation works
We modeled three groups of 10,000 players with true win rates of 2, 5 and 10 big blinds per 100 hands. Every player had a standard deviation of 90 bb/100, a typical figure for 6-max no-limit hold'em. Each player played 500,000 hands, simulated in blocks of 100 hands using a normal distribution. A "buy-in" is 100 big blinds.
We measured each player's biggest peak-to-trough downswing, their longest stretch between new all-time highs, whether they were losing at different points, and whether they ever lost a given number of buy-ins from the start. Simulating in 100-hand blocks slightly understates the very largest swings, so treat these figures as a floor, not a ceiling.
How big is the worst downswing?
| Win rate | Median worst downswing | Worst 10% of players | Chance of a 30+ buy-in downswing | Chance of a 50+ buy-in downswing |
|---|---|---|---|---|
| 2 bb/100 | 46 buy-ins | 74+ buy-ins | 92% | 40% |
| 5 bb/100 | 31 buy-ins | 44+ buy-ins | 53% | 4.8% |
| 10 bb/100 | 20 buy-ins | 28+ buy-ins | 5.6% | 0.04% |
Every single simulated player, even the 10 bb/100 crushers, had at least one 10 buy-in downswing. If you play a lot of hands, a 10 buy-in losing stretch says almost nothing about your skill.
How long can you go without a new high?
| Win rate | Median longest breakeven stretch | Worst 10% of players |
|---|---|---|
| 2 bb/100 | 153,650 hands | 322,200+ hands |
| 5 bb/100 | 68,800 hands | 128,100+ hands |
| 10 bb/100 | 29,700 hands | 49,400+ hands |
For a recreational player who plays 10,000 hands a month, a typical 5 bb/100 winner should expect a stretch of around half a year at some point where they never beat their previous best.
How often are winners showing a loss?
| Hands played | 2 bb/100 | 5 bb/100 | 10 bb/100 |
|---|---|---|---|
| 10,000 | 41% | 29% | 14% |
| 50,000 | 31% | 11% | 0.7% |
| 100,000 | 25% | 4.1% | 0.02% |
| 250,000 | 13% | 0.3% | 0% |
| 500,000 | 5.6% | 0% | 0% |
A quarter of genuine 2 bb/100 winners were still losing after 100,000 hands. Win rates are noisy: with a standard deviation of 90 bb/100, your measured win rate after 100,000 hands is still only accurate to about ±5.6 bb/100 (95% confidence). This is why a few months of results tell you very little about your true win rate.
How many buy-ins do you need?
We also checked how often each player lost a given number of buy-ins from their starting point during the 500,000 hands (their risk of ruin), and compared it with the standard formula used in our risk of ruin calculator.
| Starting bankroll | 2 bb/100 | 5 bb/100 | 10 bb/100 |
|---|---|---|---|
| 10 buy-ins | 59% (formula 61%) | 28% (29%) | 7.8% (8.5%) |
| 20 buy-ins | 36% (37%) | 8.2% (8.5%) | 0.7% (0.7%) |
| 30 buy-ins | 21% (23%) | 2.2% (2.5%) | 0.07% (0.06%) |
| 50 buy-ins | 7.4% (8.5%) | 0.2% (0.2%) | 0% |
The simulation closely matches the formula. The formula assumes you play forever, so it comes out slightly higher than a 500,000-hand simulation. The lesson: the bankroll you need depends heavily on your win rate. A 2 bb/100 player with 30 buy-ins still has about a one-in-five chance of going broke if they never move down.
What this means for you
- Have a move-down rule. These risk-of-ruin figures assume you never move down. Moving down after a large downswing cuts your real risk dramatically. Our bankroll management guide explains how, and our stake guides such as bankroll for NL50 give move-up and move-down points.
- Expect 20+ buy-in downswings if you play online volume, even as a good winner. Decide your stop-loss and move-down rules before they happen, and review your big hands with the odds calculator instead of judging by results.
- Don't judge your game by a month of results. Judge it by your decisions and by a large sample. Keep tracking every session, especially the losing ones.
- Know your own numbers. Your real win rate and standard deviation decide everything above. Plug them into the variance calculator and risk of ruin calculator.
Limitations
Real players aren't simulations. Win rates change as you improve, move stakes or tilt; standard deviation varies with game type and playing style; and live games play far fewer hands per hour. The simulation also ignores rakeback and bonuses. Use the figures as a guide to what normal variance looks like, not as a prediction for any individual.
Method: 3 × 10,000 simulated players, 500,000 hands each, SD 90 bb/100, normally distributed 100-hand blocks, fixed random seed. You're welcome to cite these figures with a link to this page.