The standard advice to a brand-new player has barely changed in twenty years. Play the free tournaments first. Then drop a few dollars at the smallest cash game you can find. Learn there, where mistakes are cheap, and move up when you’ve proved you can beat it.
It’s sensible-sounding advice, and most of it survives contact with arithmetic. But not all of it, and the part that fails is the part people lean on hardest: the idea that you’re building something. You aren’t building a bankroll at these levels. The money is a rounding error in both directions. What you’re actually buying with your time is reps, feedback, and a particular kind of opponent, and those three things are worth very different amounts depending on which free or near-free game you pick.
So the right question isn’t “can I win at a freeroll.” It’s narrower. What does an hour there teach you, compared to an hour somewhere else, and what does that hour cost?
What an hour in a freeroll actually pays
Run the numbers on a typical one. Say 3,000 entrants and a $100 prize pool, which is roughly the scale of the daily free tournaments the big sites still run. Average equity per player is three and a third cents. Suppose you are genuinely three times better than that field, which is a heroic assumption for someone learning. Your expected return is about ten cents.
Now count the hours. Late registration, a slow early level, a couple of hours to reach the vicinity of the money. Call it two hours of sitting. You’ve earned a nickel an hour, before you factor in that most of those two hours end in a bustout and nothing at all.
The micro-stakes comparison isn’t flattering either. A solid winner at NL2, where the blinds are a penny and two cents, might make five big blinds per hundred hands. That’s ten cents per hundred hands. Four tables at once gets you maybe 300 hands an hour, so thirty cents. Nobody is paying rent. Both options are, financially, free.
That’s liberating once you accept it. If the money is noise, you should choose between these games purely on what they teach, and you can stop agonizing over whether to fire a bluff for eleven cents.
Count the decisions, not the dollars
Here is where freerolls and micro cash games stop looking similar.
In that two-hour freeroll, you’ll be dealt somewhere around 140 hands at a single nine-handed table. You’ll fold three-quarters of them before the flop. You’ll see maybe twenty flops. Of those, perhaps eight will present a decision where thinking harder would have changed your answer. Call it four meaningful spots per hour.
Four tables of NL2 over the same two hours deals you roughly 600 hands. You’ll see something like 140 flops. Thirty or forty of those will be genuinely non-obvious. That’s a learning rate four to five times higher, on identical money.
There’s a second difference that matters more. Freeroll opponents paid nothing, so early on they behave like it. Stacks go in with ace-rag on the first level because busting costs a player nothing but the walk back to the lobby. You can’t practice reading a bet when the bet carries no information. In a cash game, even a two-dollar buy-in is money the player chose to put at risk, and a surprising amount of real poker behavior shows up at that price.
What freerolls do teach well is short-stack tournament play. Huge fields and fast structures mean you spend most of your tournament life between eight and eighteen big blinds, deciding whether to shove or fold. That’s a narrow skill, but it’s a real one, and it’s the skill that pays in low buy-in tournaments later. If tournaments are your goal, freerolls are legitimate practice for the exact situation you’ll face most.
The rake at NL2 takes a bigger bite than the players do
New players worry about the regulars at their table. They should worry about the house.
Micro-stakes rake is typically five percent of the pot with a cap, and the cap is what decides whether that’s tolerable. One 2026 comparison of the major rooms put the caps at around half a big blind at the smallest limits, rising to eight big blinds at $0.25/$0.50, with effective rake at NL50 landing near nine big blinds per hundred hands on one major site. Read that number against a strong winner’s edge of eight to ten big blinds per hundred. Roughly half the money a good player extracts goes straight out the door.
Work one pot. At NL2, a standard single-raised pot that reaches the flop might be forty cents. Five percent is two cents, a full big blind, though the cap will usually knock it to a penny. You pay that on every pot that sees a flop, win or lose. Over a session you’re paying it dozens of times.
This has a practical consequence. Marginal pots are worse than they look, because the rake is deducted before you see any of it. Thin value bets and speculative calls that break even in a theoretical, rake-free world are small losers in practice. Against the opponents you’ll meet at NL2, who call far too much, the correction is mostly to bet your good hands bigger and bluff less, which is also roughly the right adjustment against calling stations at any level.
You cannot climb out of NL2 on winnings alone
This is the number that ends the bankroll-building fantasy, and it’s worth doing slowly.
A conservative bankroll is thirty buy-ins. At NL2, that’s $60. At NL10, where a buy-in is $10, it’s $300. So the gap you need to cross is $240.
Suppose you’re a strong NL2 winner at eight big blinds per hundred hands. The big blind is two cents, so you’re making sixteen cents per hundred hands. Divide $240 by $0.16. You need 150,000 hands. Four-tabling at 300 hands an hour, that’s 500 hours.
Five hundred hours is three months of full-time work to earn the right to play a game where a buy-in costs ten dollars. Nobody has that patience, and nobody should. The honest conclusion is that you fund your move up from outside the game, or you accept that NL2 is a classroom and not a job. Rakeback changes the arithmetic a little, since some rooms return a flat share to everyone rather than reserving it for high-volume players, but it moves 500 hours to maybe 400. It doesn’t move it to 50.
Where freerolls still earn their place
Three situations, and they’re real.
The first is having nothing to deposit. If your choice is a freeroll or play money, take the freeroll every time. Play-money opponents are indistinguishable from a random number generator. Freeroll opponents are at least playing for a prize, which produces recognizable behavior around the bubble and at the final table.
The second is ticket ladders. Many freerolls pay out in tournament entries rather than cash, and those tickets feed into bigger events. The field you beat for a $5 ticket is weaker than the field you’d face if you bought in directly. That’s a genuine edge, and it compounds in a way the cash prize never does.
The third is bubble practice. Freerolls have enormous fields and tiny min-cashes, which creates intense pressure at the money bubble for almost no actual money. You get to feel the pull of folding your way into a payout, and to practice ignoring it, at zero cost. That instinct is expensive to learn in a $50 tournament.
The case for starting one level higher than feels safe
The usual advice says start as low as possible. If you can comfortably afford $300, there’s a decent argument for starting at NL10 instead.
Rake hurts proportionally less as you move up. The player pool at NL10 is bad but recognizably playing poker, so the reads you develop will transfer. And the stakes are high enough that losing a buy-in registers, which is what makes you actually review the hand afterward.
The counterargument holds if money is tight, and it holds completely. Playing scared is more expensive than playing one level too low. A player who folds a good hand because the pot has become frightening is burning more than the rake ever took. Pick the level where a lost buy-in annoys you and nothing more.