Poker Downswing Odds: We Simulated 70 Billion Hands
We simulated 700,000 poker careers of 100,000 hands each — 70 billion hands, dealt hand by hand. A solid 3bb/100 winner (σ 100bb/100, 6-max cash) has an 81.1% chance of a 20+ buy-in downswing, a 39.2% chance of 30+ buy-ins, goes a median 40,173 hands without a new equity high at some point, and is still losing money after 100,000 hands in 17.1% of careers. Downswings are not a sign you play badly — they are the mathematical price of volume.
Every bankroll article repeats the same folklore numbers. We wanted real ones, so we ran the per-hand model behind our variance calculator at scale: 100,000 independent 100,000-hand careers for each win rate from 1 to 10bb/100 — 70 billion hands in total — and recorded every peak-to-trough downswing, every breakeven stretch, and how often genuine winners were still in the red. The full table is a free CSV download.
How likely is a downswing at your win rate?
Each cell is the percentage of 100,000-hand careers that hit at least one peak-to-trough downswing of that depth (a buy-in is 100bb). Read your win rate's row and accept the left half of it as a near-certainty:
| Win rate | ≥ 20 buy-ins | ≥ 30 buy-ins | ≥ 50 buy-ins | ≥ 70 buy-ins | ≥ 100 buy-ins |
|---|---|---|---|---|---|
| 1 bb/100 | 90.8% | 58.2% | 15.3% | 2.8% | 0.13% |
| 2 bb/100 | 86.5% | 48.6% | 9.9% | 1.4% | <0.1% |
| 3 bb/100 | 81.1% | 39.2% | 5.8% | 0.66% | <0.1% |
| 4 bb/100 | 74.4% | 30.7% | 3.5% | 0.29% | <0.1% |
| 5 bb/100 | 67.7% | 23.6% | 1.9% | 0.13% | <0.01% |
| 7 bb/100 | 52.6% | 12.4% | 0.53% | <0.1% | <0.01% |
| 10 bb/100 | 31.1% | 4.1% | <0.1% | <0.01% | <0.01% |
The two headlines: first, 20 buy-in downswings happen to almost everyone — 86.5% of 2bb/100 winners and even 31.1% of elite 10bb/100 winners hit one inside 100k hands. Second, win rate is the best downswing insurance that exists: a 50+ buy-in disaster hits 15.3% of 1bb/100 grinders but only 1.9% of 5bb/100 winners. And remember these are 100,000-hand windows — keep playing for a million hands and the deep-downswing columns keep climbing.
How deep is a "normal" worst downswing?
Every career has some worst downswing. Here is how deep that single worst stretch ran — the median career, the average, and the unlucky 5% tail:
| Win rate | Median worst downswing | Average | Worst 5% of careers |
|---|---|---|---|
| 1 bb/100 | 32.6 BI | 35.7 BI | 63.8+ BI |
| 2 bb/100 | 29.6 BI | 32.4 BI | 57.6+ BI |
| 3 bb/100 | 27.1 BI | 29.5 BI | 51.6+ BI |
| 4 bb/100 | 24.9 BI | 27.1 BI | 46.7+ BI |
| 5 bb/100 | 23.2 BI | 25.2 BI | 42.7+ BI |
| 7 bb/100 | 20.4 BI | 21.9 BI | 35.8+ BI |
| 10 bb/100 | 17.3 BI | 18.4 BI | 29.0+ BI |
This is the table that should set your bankroll expectations. The median3bb/100 career's worst downswing is 27.1 buy-ins — but medians are not what bankrolls are for. One career in twenty runs 52+ buy-ins deep, and you do not get to choose which career you are living in.
How long will you go without making money?
A downswing is measured in buy-ins; the experienceof one is measured in hands. We tracked each career's longest breakeven stretch — the most hands between two consecutive new equity highs:
| Win rate | Median longest stretch | Average | Worst 5% of careers |
|---|---|---|---|
| 1 bb/100 | 53,655 hands | 56,898 hands | 98,714+ hands |
| 2 bb/100 | 46,523 hands | 51,126 hands | 96,900+ hands |
| 3 bb/100 | 40,173 hands | 45,436 hands | 92,867+ hands |
| 4 bb/100 | 34,830 hands | 40,254 hands | 86,271+ hands |
| 5 bb/100 | 30,519 hands | 35,680 hands | 77,701+ hands |
| 7 bb/100 | 23,819 hands | 27,930 hands | 59,705+ hands |
| 10 bb/100 | 17,255 hands | 19,983 hands | 41,053+ hands |
Sit with the 2bb/100 row: the median career contains a 46,523-hand stretch with zero net profit — months of regular play — and the unlucky tail goes 96,900+ hands. If anything these numbers are slightly kind, because stretches still running when a career hit 100,000 hands were counted at their current length. This is the statistic that ends poker careers: not the money, the time.
How often is a true winner still losing?
| True win rate | Losing after 10k hands | Losing after 50k hands | Losing after 100k hands |
|---|---|---|---|
| 1 bb/100 | 45.9% | 41.3% | 37.6% |
| 2 bb/100 | 42.1% | 32.9% | 26.4% |
| 3 bb/100 | 38.5% | 25.1% | 17.1% |
| 4 bb/100 | 34.5% | 18.6% | 10.3% |
| 5 bb/100 | 30.7% | 13.2% | 5.8% |
| 7 bb/100 | 24.1% | 5.9% | 1.3% |
| 10 bb/100 | 15.9% | 1.2% | <0.1% |
A genuine 5bb/100 crusher — a win rate most regs never reach — still shows a losing graph after 10,000 hands 30.7% of the time, and after 50,000 hands 13.2% of the time. Flip it around and the same table is a warning about overconfidence: if a quarter of true 2bb/100 winners are losing after 100k hands, plenty of losing players are winning over samples that size too. Your database cannot tell you which one you are; our win rate calculator at least tells you how wide the honest error bars are.
What this means for bankroll size
The classic advice — 30 buy-ins minimum, 50 for comfort — falls straight out of the first two tables. A 3bb/100 winner's median worst downswing is 27 buy-ins, and roughly one career in twenty runs 52+ deep: a 30 buy-in roll survives the median career with room to move down; a 50 buy-in roll survives the ugly tail. Concretely:
- ✓ Confident 5bb/100+ at the micros: 30 buy-ins works, but only with move-down discipline — 23.6% of 100k-hand careers at this win rate still hit a 30+ buy-in slide.
- ✓ Typical 2-3bb/100 reg: hold 40-50 buy-ins. The 30+ BI risk is 48.6%-39.2% per 100k hands, and you will live through a 40,173-hand breakeven stretch on the way.
- ✓ Unproven win rate: assume the 1bb/100 row — a 58.2% chance of a 30+ buy-in downswing — and keep 50+ buy-ins or pick stakes where redepositing is trivial.
Set your own rules in the bankroll calculator, stress-test your exact win rate and standard deviation in the variance simulator, and check what your sample actually proves with the win rate calculator.
Every number assumes your win rate stays fixed while you swing. Real players tilt, table-select worse, and jump stakes mid-downswing — which turns a survivable 27 buy-in downswing into a busted roll. The simulation is the best case. Bankroll rules exist to keep you close to it.
Methodology
- • 100,000 careers × 100,000 hands for each win rate (1, 2, 3, 4, 5, 7, 10 bb/100) — 70,000,000,000 hands, simulated hand by hand. Each hand is drawn from a normal distribution with mean winRate/100 and standard deviation 100/√100 bb — the same per-hand model as our variance simulator, with σ = 100bb/100 (typical 6-max NLHE cash).
- • Downswing = peak-to-trough equity drop, tracked at single-hand resolution; one buy-in = 100bb. Breakeven stretch = most hands between consecutive new equity highs (stretches unfinished at hand 100,000 count at their current length, so true stretches are slightly longer).
- • Reproducible: deterministic sfc32 PRNG, seed "pokerpro-downswing-v1", generated 2026-07-21. Simulated losing-after-N-hands probabilities were verified against the closed form Φ(−WR·√(N/100)/σ) to within 0.6 percentage points before publication.
- • Limits: the normal model has no fat tails, no tilt, no rake changes, and a fixed win rate — real downswings run deeper when play degrades. Results are per 100,000-hand window; longer careers see deeper extremes. Higher-σ formats (HU, PLO, MTTs) are substantially swingier than every table here.
Common questions
How likely is a 20 buy-in downswing in poker?
Very likely — for almost everyone. In our simulation of 100,000 careers per win rate (100,000 hands each, stddev 100bb/100), 81.1% of true 3bb/100 winners hit at least one 20+ buy-in downswing, 67.7% of 5bb/100 winners did, and even 10bb/100 crushers hit one 31.1% of the time. A 20 buy-in downswing is not evidence you are a losing player — it is the expected cost of playing enough hands.
How long do poker downswings last?
Longer than most players can emotionally tolerate. Measured as the longest stretch without setting a new equity high, the median for a 3bb/100 winner over 100,000 hands was 40,173 hands, and the unlucky 5% of careers stayed below a previous peak for 92,867 hands or more. Even at 10bb/100 the median longest breakeven stretch was 17,255 hands.
Can a winning player lose over 50,000 hands?
Yes, routinely. In the simulation a true 2bb/100 winner was still losing money after 50,000 hands in 32.9% of careers and after 100,000 hands in 26.4% of careers. At 5bb/100 those numbers fall to 13.2% and 5.8%. This is why a graph over 20k-50k hands proves very little about your true win rate — our win rate calculator puts confidence intervals on your actual sample.
Does a higher win rate protect you from downswings?
Partially. Going from 1bb/100 to 10bb/100 cuts the chance of a 30+ buy-in downswing from 58.2% to 4.1% in a 100,000-hand career — a huge drop. But no realistic win rate makes downswings rare: the median worst downswing for a 10bb/100 player was still 17 buy-ins. Skill shrinks downswings; it never removes them.
What game do these downswing numbers assume?
6-max no-limit hold'em cash with a standard deviation of 100bb/100, the typical tracker value. Full-ring runs closer to 80bb/100 (slightly smaller swings), fast-fold around 105, heads-up 140+, and PLO 150+. Downswing risk scales up fast with standard deviation, which is why tournament and PLO players need far deeper bankrolls than these tables suggest.
Where can I download the raw downswing data?
The full table (all win rates, downswing probabilities, breakeven stretches, and losing-after-N-hands probabilities) is a free CSV at pokerpro.tools/downloads/poker-downswing-odds.csv, licensed CC BY 4.0 — cite this article if you republish. The generator script is deterministic (seed "pokerpro-downswing-v1"), so every number is reproducible.
Tools for this job
Keep reading
New to poker? Take the free 15-minute crash course →