Poker Variance Explained: Sample Size, Win Rate and Swings

Poker variance is the reason a winning player can lose for three months and a losing player can book the best year of their life. Understanding it is not an academic exercise — it is the difference between correctly diagnosing a leak and quitting a profitable game because of noise.

Quick answer: Poker variance is the statistical spread of results around your true win rate. It is measured by standard deviation, typically 80–100 big blinds per 100 hands in online cash games. Because variance is large relative to typical win rates, it takes hundreds of thousands of hands before observed results reliably reflect actual skill.

What Variance Actually Measures

Your true win rate is the long-run average you would earn per 100 hands if you played the same games forever with the same skill edge. Your observed win rate is what your database shows. Variance is the gap between the two, and it is enormous.

In a typical online six-max no-limit game, standard deviation runs around 90 big blinds per 100 hands. A strong regular might have a true win rate of 5 bb/100. The noise term is eighteen times the size of the signal. That ratio is the entire reason poker feels unfair in the short run.

How Many Hands Do You Actually Need?

The standard error of your observed win rate shrinks with the square root of your sample. Concretely, with a standard deviation of 90 bb/100:

• 10,000 hands: 95% confidence interval of roughly ±18 bb/100. Essentially meaningless.
• 100,000 hands: roughly ±5.6 bb/100. A 5 bb/100 winner still cannot rule out being a breakeven player.
• 500,000 hands: roughly ±2.5 bb/100. Now the sample says something.
• 1,000,000 hands: roughly ±1.8 bb/100. Reasonably conclusive.

The uncomfortable implication: most players never reach a sample large enough to confirm their win rate at their current stake before they move up, change format, or the game pool shifts underneath them. Your graph is a rough estimate, not a verdict.

Tournament Variance Is Far Worse

Multi-table tournaments compress most of their prize pool into the top 1% of finishes. That produces a distribution with a long right tail and an enormous number of zeros. Standard deviation in MTTs is commonly expressed in buy-ins rather than big blinds, and figures of 150–250% ROI standard deviation per tournament are typical for large fields.

The practical consequence is that a genuinely profitable tournament player with a 20% ROI can and routinely does run 20,000 tournaments without a major score, or endure downswings of 200 buy-ins or more. This is not a hypothetical — it is the base case. Understanding ICM strategy improves the quality of your late-stage decisions, but it cannot reduce the structural variance of the format.

What Variance Means for Bankroll

Bankroll requirements are a direct function of variance, not of ego. Higher variance formats demand deeper rolls for the same risk of ruin:

• Full-ring cash: 20–30 buy-ins is often adequate for a modest winner.
• Six-max cash: 30–50 buy-ins.
• Heads-up or high-variance short-stack formats: 50–100 buy-ins.
• Large-field MTTs: 200–300 buy-ins at minimum, and more if tournaments are your primary income.

These numbers are not arbitrary. They fall out of risk-of-ruin calculations that combine win rate, standard deviation and roll size. Our bankroll management guide walks through the arithmetic so you can size your own roll rather than copying someone else's.

Separating Variance From Real Leaks

The hard question during a downswing is whether you are running badly or playing badly. Results cannot answer it. Process metrics can:

All-in adjusted winnings

Comparing actual winnings to all-in expected value isolates the luck component of your big pots. A large negative gap points toward variance; a small gap during a losing stretch points toward strategy.

Non-showdown winnings

Non-showdown winnings are largely a function of your decisions, not card distribution. A deteriorating non-showdown line during a downswing is a strong signal that tilt or overfolding has crept in.

Hand review volume

If you have not reviewed hands in six weeks, assume your play has drifted. Regular structured review is the only reliable defence, and our poker training videos cover the review frameworks that professionals actually use.

Managing the Psychology of Variance

Knowing that a 30-buy-in downswing is statistically unremarkable does not make it feel unremarkable. The players who survive long careers build systems rather than relying on willpower: fixed stop-loss rules, session length caps, scheduled breaks after losing stretches, and a clear separation between bankroll and living expenses.

Tilt is not a character flaw — it is a predictable response to a high-noise environment. Treat it as an engineering problem. Our poker mental game resources cover the concrete routines that keep decision quality flat when results are not.

Frequently Asked Questions

How many hands until I know my win rate?

Roughly 500,000 hands for a usable estimate in cash games, and well over a million for high confidence. Anything under 100,000 hands carries a confidence interval wider than most realistic win rates.

What is a normal downswing?

For cash games, 15–25 buy-ins is entirely ordinary for a winning player. For large-field tournaments, 100–200 buy-ins without a significant score is within normal expectation.

Does playing more tables increase variance?

It increases variance per unit of time but not per hand. More tables means more hands per hour, so swings arrive faster in calendar terms even though the per-hand distribution is unchanged — and win rate per hand often falls as table count rises.

Can I reduce variance without reducing win rate?

Partially. Game selection, avoiding marginal high-variance spots and choosing lower-variance formats all reduce swings. Deliberately playing more passively to "reduce variance" usually costs more in win rate than it saves in comfort.

Is variance different online versus live?

The per-hand distribution is similar, but live players see far fewer hands per hour, so downswings measured in hands take much longer in calendar time. Live samples are correspondingly less conclusive.

Final Thoughts

Poker variance is not an obstacle to playing well — it is the condition under which all poker is played. The players who last are the ones who size their bankroll to survive it, measure their game with process metrics rather than results, and keep studying through the stretches where nothing works.

Start by getting your fundamentals audited. Our beginner poker guide covers the base, and the best online poker sites comparison will help you find the volume you need to build a meaningful sample.