What 3% Rake Does to Win Rates at 50,000 Hands
A 3% rake in online poker is not a footnote to the game; it is a co-author of every result you post. At 50,000 hands of $1/$2 no-limit hold'em, a player who wins 4 bb/100 before rake sees that edge compressed to roughly 1.6 bb/100 after the house takes its cut, a reduction of about 60%. The difference between a winning and a break-even player at that volume is often not skill but the size of the rake, how it is capped, and whether the site charges by pot or by hand.
The Arithmetic of 3% Over 50,000 Hands
Start with the unit that matters: bb/100. At $1/$2, one big blind is $2. A win rate of 4 bb/100 means $8 per 100 hands, or $4,000 over 50,000 hands before rake. Now apply a 3% rake. The critical question is 3% of what. Most cash games rake a percentage of the pot, capped, and only when the pot reaches a minimum size. A 3% uncapped rake on every pot is not the same as 3% capped at $3.
Assume a typical structure: 3% of the pot, capped at $3, no flop no drop. In a $1/$2 game, the average pot that sees a flop might be $18 to $24. At $20, 3% is $0.60. If you play 50,000 hands and see a flop in 22% of them, that is 11,000 raked pots. At an average $0.60 per pot, you pay $6,600 in rake. Your pre-rake win of $4,000 becomes a loss of $2,600. That is not a rounding error; it is the difference between a bankroll and a bust.
But not every hand is raked at the same rate. Pots under $5 often go unraked or are raked at a lower tier. Pots above $100 hit the $3 cap, which is only 3% at exactly $100; at $200, the effective rake is 1.5%. This regressive structure means high-volume, small-pot players feel the rake more than those who play fewer, larger pots. A 3% headline rate is therefore a weighted average that depends entirely on your pot-size distribution.
The Cap as a Hidden Variable
The cap is where sites compete, and where win rates diverge. A $3 cap at $1/$2 is standard, but some sites cap at $2.50 or $2, and a few legacy rooms still use a $1 cap for certain stakes. Over 11,000 raked pots, the difference between a $3 cap and a $2 cap is $11,000 in saved rake — more than double the pre-rake win in the example above. This is why serious players treat the cap, not the percentage, as the primary number. A 5% rake capped at $2 is often cheaper than a 3% rake capped at $3 for the same player pool.
There is also the question of when the rake is taken. Some sites rake preflop, some only after the flop, and a few take a percentage of the blinds in tournaments. In cash games, the no-flop-no-drop rule is a significant subsidy to aggressive players who win pots before the flop. If you win 1,500 pots preflop at 50,000 hands, and those pots average $6, you pay zero rake on them under a no-flop-no-drop regime. Under a rake-always regime, you pay $270 on those same pots. That $270 is 6.75% of your pre-rake win.
Win Rate Decay by Player Type
The effect of 3% rake is not uniform. It punishes marginal winners and high-frequency players disproportionately. Consider three archetypes over 50,000 hands at $1/$2:
- The tight-aggressive winner: 5 bb/100 pre-rake, sees 20% of flops, average raked pot $22. Pre-rake win: $5,000. Raked pots: 10,000. Rake at $0.66 per pot: $6,600. Net: -$1,600.
- The loose-aggressive winner: 6 bb/100 pre-rake, sees 30% of flops, average raked pot $18. Pre-rake win: $6,000. Raked pots: 15,000. Rake at $0.54 per pot: $8,100. Net: -$2,100.
- The break-even player: 1 bb/100 pre-rake, sees 24% of flops, average raked pot $16. Pre-rake win: $1,000. Raked pots: 12,000. Rake at $0.48 per pot: $5,760. Net: -$4,760.
The break-even player loses nearly five times their pre-rake win to the rake. The loose-aggressive player, despite a higher pre-rake win rate, loses more total dollars than the tight-aggressive player because they pay rake more often. This is the opposite of the intuition that aggression is always rewarded. In a raked game, frequency is a tax multiplier.
The Rakeback Offset
Most sites return a portion of rake as rakeback, typically 25% to 40% for high-volume players. At 35% rakeback, the tight-aggressive player in the example above gets $2,310 back, turning a -$1,600 net into +$710. The loose-aggressive player gets $2,835 back, turning -$2,100 into +$735. The break-even player gets $2,016 back, still leaving -$2,744. Rakeback is not a bonus; it is a partial rebate that changes the effective rake from 3% to roughly 1.95%. Even at that reduced rate, the break-even player cannot survive.
This is why the effective rake, not the advertised rake, is the number that matters. A site with 3% rake and 40% rakeback has an effective rake of 1.8%. A site with 2.5% rake and 20% rakeback has an effective rake of 2%. The first site is cheaper despite a higher headline rate. Players who compare only the percentage are comparing the wrong number.
What 50,000 Hands Actually Represents
Fifty thousand hands is not a large sample in poker. It is roughly 500 to 800 hours of live play, or 250 to 400 hours of online play at four tables. For a professional playing 30 hours a week, it is three to four months of work. The rake you pay over that period is not an abstraction; it is rent. At $1/$2, a full-time player paying $6,000 to $8,000 in rake over 50,000 hands is paying more in rake than many players earn in profit.
The variance is also brutal. A 4 bb/100 winner pre-rake has a standard deviation of roughly 80 bb/100 per 100 hands. Over 50,000 hands, the standard error of the win rate is about 3.6 bb/100. That means a player with a true 4 bb/100 pre-rake win rate could easily post a negative result over 50,000 hands even before rake. Add a 3% rake that costs 2.4 bb/100, and the margin for error disappears. The rake does not just reduce the mean; it makes the confidence interval cross zero more often.
This is the central problem with rake discussions that focus on the percentage. The 3% is a nominal rate applied to a distribution of pots that varies by player, stake, and site. The real question is not "what is the rake?" but "what is the rake on the pots I actually win?" A player who wins 60% of their pots without showdown pays rake on those pots. A player who wins 40% at showdown pays rake on larger pots. The same 3% produces different effective rates.
The Implication for Game Selection
If 3% rake turns a 4 bb/100 winner into a 1.6 bb/100 winner, then game selection is not about finding the softest table. It is about finding the table where the rake structure and the player pool interact to produce the highest effective win rate. A table with a $2 cap and no-flop-no-drop is worth more than a table with a $3 cap and a 40% rakeback deal, because the cap applies to every pot while rakeback applies only to the rake you pay. The cap is a ceiling on your tax; rakeback is a rebate on the tax you already paid.
The open question is whether the 3% rake is sustainable for the player pool. If the average winning player at $1/$2 is actually a loser after rake, the pool shrinks. The recreational players who fund the ecosystem leave when they stop winning. The professionals leave when the rakeback no longer covers the rake. What remains is a game where the only consistent winner is the house, and the 3% is not a fee on the game but the game itself. The next time you see a 3% rake, ask not what it costs per pot, but what it costs per 50,000 hands. The answer is usually your entire edge.