- 📈 Results are noise at any sample you play — a 5bb/100 winner finishes down over 10,000 hands about a third of the time.
- ⚠️ The graph cannot separate a downswing from a leak — both look identical for months.
- 💡 Decisions are checkable on one hand — that is the only feedback that arrives fast enough to act on.
You lost four buy-ins on Tuesday and you want to know what it means. The honest answer is that it means almost nothing, and the reason it means nothing is arithmetic rather than consolation.
The gap between a winning player and a losing one is small compared to how much the cards move around in any evening. That single fact is behind most of the bad decisions players make about their own game.
Why the graph lies to everyone
No-limit hold'em has a standard deviation somewhere around 90 big blinds per 100 hands. A good win rate is 5. The noise is roughly eighteen times the signal, per block of a hundred hands, and it only shrinks with the square root of how many blocks you play.
What that does to a month
Over 10,000 hands, a genuine winner is down about one time in three. Over 100,000, about one time in twenty. Almost nobody plays enough hands for their results to be evidence of anything on the timescale they actually make decisions.
| Hands played | Expected | Chance of finishing down | Signal |
|---|---|---|---|
| 10,000 | +500bb | ~29% | ↓ meaningless |
| 100,000 | +5,000bb | ~4% | → suggestive |
| 1,000,000 | +50,000bb | <1% | ↑ real |
SOURCE : 5bb/100 win rate, 90bb/100 standard deviation · normal approximation
What to read instead
Decisions. Whether the call had the equity the price demanded. Whether the bet size made sense against the range you put them on. Whether the fold gave up more than it saved. Each of those is answerable on a single hand, tonight, without waiting for a hundred thousand more.
- Export the session while it is fresh, before your room's retention setting deletes it.
- Find the three biggest pots you lost and the three you won. That is where the money moved.
- Check the price on every call against the equity you actually had, not the equity you hoped for.
- Write down one thing to do differently next session. One, not five.
« The hand you misplayed feels exactly like the hand you played well, once you already know how it ended. »
— the reason self-review is hard
FAQ
How many hands before my win rate means something?
More than most people play before drawing conclusions. At a 90bb/100 standard deviation you need six figures of hands before the confidence interval around your win rate is narrow enough to act on.
Does a downswing mean I am playing badly?
It is not evidence either way, which is precisely the problem. The only reliable way to tell is to look at the decisions rather than the outcomes.
Should I move down in stakes after a bad month?
Move down for bankroll reasons, which are arithmetic, not for confidence reasons, which are noise. If the buy-ins left in your roll have dropped below your own threshold, move. If they have not, the month is not information.
Sources
- Standard deviation figures: typical reported ranges for 6-max no-limit hold'em, 80 to 100bb/100.
- Probabilities computed with the variance calculator on this site.
