The most common picture in a micro-stakes regular's database: plus four bb/100 before rake, minus one after. The player looks for a mistake in their play, while the main expense sits outside the game. Rake cannot be fixed with technique, but it can be counted and adjusted for. The only opponent at the table who wins every hand is the room.
How much rake takes
The exact number depends on the room, the stake and the table structure, but the order of magnitude is the same everywhere.
| Stake | Rake in a typical database, bb/100 |
|---|---|
| NL10 | 8–11 |
| NL25 | 6–8 |
| NL50 | 5–7 |
| NL100 | 4–5 |
| NL200 | 3–4 |
For comparison: the most expensive mistake in a database review usually costs between half and one bb/100. At NL25 the rake equals six to eight such mistakes at once. So the first question for any database below NL100 is: how much of my result does the room take? Usually the answer is: more than your most expensive leak, your second leak and your third put together.
Why rake hits your style differently
Rake is taken from the pot, which means it is paid by whoever plays big pots with many participants more often. Three consequences follow.
First, limped pots and multiway pots at micro stakes are more expensive than they look: the pot grows, the rake grows, and your share of it is split three ways. Second, 3bet pots against one opponent are raked to the cap almost always, and the cap in bb at low stakes is higher than at high ones. Third, small pots that end on the flop give the room the least. A player who takes many small pots without showdown pays less rake per hundred hands than a player with the same winrate earned in big pots. The room loves those who get to the river. It feeds them, and it shears them.
This does not mean big pots should be avoided. It means that when comparing two lines at micro stakes, the difference in rake can outweigh the difference in EV before rake.
What changes the number and what does not
What does not work: fitting stats to a "norm". If you pull your cbet or 3bet toward the averages, the rake will not change, and the EV may fall.
What works:
- Rakeback and loyalty programs. At NL10–NL50 a return of 20–30% of rake equals two to three bb/100. That is more than fixing any single leak will give. If you play without rakeback where it exists, that is the first finding in your database, and the only one you can fix in a single evening.
- Choice of room and format. Two rooms at the same stake can differ by a bb/100 or more purely through the rake cap and its size. Fast-fold and regular tables are raked differently too.
- Discipline in multiway pots. Calling third into a pot with hands that realize poorly does worse than a fold at micro stakes than at NL200, precisely because of rake.
- Preflop without extra calls from the SB. A call from the SB creates a pot where you are out of position, with a third player behind you and rake on top. At micro stakes some of these calls become unprofitable purely because of rake.
How to look at your result
Split the winrate into three numbers: before rake, rake, after rake. If before rake you are in the plus by more than the rake, the game pays, and the review looks for where to add. If the plus before rake is less than the rake, the question is not only about play: first rakeback and table selection, then spots.
And check the sample. On 100,000 hands the winrate interval is about six bb/100 either way, so the difference between "plus one" and "minus one" after rake is often inside the noise. The rake in the same database, however, is counted exactly, to the cent. Of all the numbers in a database, it is the most reliable. The one thing the room never gets wrong.