Sample Size in Poker: Why Your Results Mean Less Than You Think

The Uncomfortable Truth About Short Samples

Most poker players evaluate their game far too early. After 50 tournaments with a positive ROI, they feel confident they are winning players. After 100 tournaments running bad, they question their fundamentals and change their strategy. In both cases, the sample is too small to support either conclusion.

MTT poker has extreme variance compared to other poker formats. The top-heavy nature of prize pool distributions means that a large portion of your expected profit comes from a small number of deep runs and final table finishes. These events are rare by definition – and whether you run deep in any given 100-tournament stretch is almost entirely determined by luck, not skill.

A player with a genuine long-run ROI of +10% can produce results anywhere from -30% to +60% over 100 tournaments without any of those outcomes being statistically unusual. The distribution of possible short-run results is enormous. Only as sample size grows does the actual ROI range narrow toward its true value.

How Many Tournaments Do You Actually Need?

• Under 200 tournaments: Almost entirely noise. Any ROI figure in this range tells you almost nothing about your true skill level. Do not adjust stakes, formats, or strategy based on results this small.
• 200-500 tournaments: Directionally informative. A strong positive trend is weakly encouraging. A strong negative trend warrants review – but not panic. Could still be variance.
• 500-1,000 tournaments: Becoming meaningful. At this range, persistent positive ROI is more likely to reflect genuine edge than lucky variance.
• 1,000+ tournaments: Statistically significant. At this sample size at a consistent buy-in level, your ROI is a reliable indicator of your true win rate. Conclusions drawn here are actionable.

These thresholds assume you are playing consistent buy-in levels and formats. Mixing $11 turbos with $215 deep stacks in the same sample produces a number that is difficult to interpret, because the variance profiles and field compositions are so different.

What Invalidates a Sample

Sample size is not just about quantity. A large sample can still be misleading if the conditions within it are inconsistent. Three factors most commonly invalidate a poker sample:

Changing buy-in levels mid-sample. Moving from $22 to $55 events halfway through a 500-tournament stretch creates a mixed dataset. The ROI figure reflects two different competitive environments blended together.

Changing networks or formats significantly. Each network has a different player pool and each format has a different variance profile. A sample that spans major changes in where or how you play is not a clean measurement of any single thing.

Re-entry and multi-entry tournaments. In re-entry events, a single session may involve 2-3 buy-ins. Counting re-entries as separate entries inflates your sample size relative to unique tournament sessions – which overstates statistical significance.

Using Sample Size Correctly

The practical implication is straightforward: hold your conclusions loosely until the numbers are large enough to support them. Do not move up stakes after a 100-tournament heater. Do not abandon a strategy after a 150-tournament cold stretch. Do not conclude you have found your best format after 80 events in it.

This does not mean ignoring short-run results. A -30% ROI over 300 tournaments is worth investigating even if not statistically conclusive. The key is to investigate rather than react. Review hand histories, analyse decisions, look at spot-specific data. Do not restructure your entire approach based on a sample that variance alone could plausibly explain.