What Percentage of Poker Players Are Profitable?

Nobody measures this well, and anyone quoting one clean number is selling certainty they don't have. The defensible ranges: among tracked online regulars, roughly 30 percent show a profit after rake in database samples; across everyone who plays, the honest estimate is a small minority — perhaps 10 to 20 percent ahead at any snapshot, fewer over a lifetime. The figures floating around — 5 percent, 10 percent, 30 percent — are all real in the sense that someone derived each of them from something, and all misleading in the sense that they measure different populations with different biases. This piece lays out every serious source I can find, what each one actually measured, and where the honest answer is that no one knows. That admission, as it turns out, is the most useful data point on the page.

Every Number in Circulation, With Its Bias Attached

Start with the claims you've seen, traced back to what produced them.

"About 30 percent of players are winners." This comes from tracking-database analyses. The most commonly cited is an examination of a Hold'em Manager database — 609 players with at least 10,000 hands each — which found roughly 30 percent profitable after rake, with about 18 percent winning solidly and around half losing at moderate-to-significant rates. Independent filtering of PokerTracker databases by the strategy writer BlackRain79 lands in the same neighborhood: on the order of 30 percent showing any profit, with maybe 10 percent winning enough to matter. The biases here run in both directions and they are not small. Ten thousand hands is far too few to separate skill from luck — a chunk of that 30 percent are losers on a heater and some losers are winners in a downswing. Tracked pools oversample serious, studying players, which inflates the winner share relative to the whole player population; but the samples also exclude rakeback and promotions, which quietly flips some small losers into net winners. Call it a genuine estimate of one thing only: the share of regulars in tracked online pools showing profit over short samples.

"Only 5 to 10 percent win." This is the number most often said at tables and least often sourced. I could not trace it to any dataset — it appears to be folklore, plausibly descended from someone's intuition about meaningful winners rather than any measurement. It may even be roughly right for lifetime profitability across all players. But it is a vibe with a percent sign, and it should be quoted as such.

"Most tournament players lose." This one has real academic footing. Steven Levitt and Thomas Miles of the University of Chicago analyzed the entire 2010 World Series of Poker: the 720 players identifiable in advance as high-skilled earned an average return on investment of about +30.5 percent, while the other 31,000-plus entrants averaged −15.6 percent, losing over $400 per event. Two things follow. Skill is real and persistent — the gap is enormous. And the average tournament entrant loses, reliably, before travel and expenses are counted. The study's limits: one series, one year, gross of costs, and "identified as skilled beforehand" builds in its own selection.

What the Academic Literature Actually Measures

Here is the finding underneath the findings, and it is the honest core of this piece: the academic literature on poker mostly does not measure profitability at all. It measures behavior — and what the behavior data shows quietly reframes the whole question.

The largest dataset ever examined belongs to Ingo Fiedler and Ann-Christin Wilcke, who tracked over two million player identities on the biggest online site across six months around 2009-2010. Their headline finding was not about winners. It was that the median player barely plays: seven sessions and under five hours total across half a year, at tiny stakes. The player population is a vast, thin crowd of dabblers wrapped around a small core of regulars — Fiedler's related work found the most active tenth of customers generating around 90 percent of operator revenue. This is why every "percentage of players" claim is unstable before it starts: the denominator is mostly people whose entire poker career is an evening or two, whose results are a coin flip minus rake, and who no tracking database will ever see.

Sociologist Kyle Siler's study — conducted at Cornell — of 27 million online hands, published in the Journal of Gambling Studies in 2010, adds the cruelest detail: win frequency and win rate diverge. Players who won the most pots tended to lose the most money — many small wins, punctuated by occasional large losses their memory underweights. The implication for this essay's question is sharp: most players cannot even self-report accurately whether they're profitable, because the texture of losing poker feels like winning most nights.

And the contemporary literature has, if anything, retreated from the question. The most recent large behavioral study — "Second Session at the Virtual Poker Table," published in the Journal of Gambling Studies in 2022, tracking 2,489 online players from 2015 to 2017 — is careful, modern, and about activity patterns, not profit distributions. Operators hold the profitability data, and operators do not publish it. There is no peer-reviewed answer to "what percentage of poker players are profitable." That gap is not an accident; a business built on the losses of the many has no incentive to count them in public. So every number in this piece is an inference from partial data, including mine.

The Rake Is the Hidden Variable

One structural fact does more to set the percentage than the skill distribution does, and most discussions of this question skip it entirely: poker among the players is zero-sum, but poker in a cardroom is not. The rake removes money from the pool on every pot, which makes the game negative-sum for the player population as a whole — and that arithmetic, not talent, is what drags the winner share below half.

Run the logic from the ground up. In a home game with no rake, roughly half the money in play sits ahead at any given time by construction — someone holds every chip. Introduce rake and the break-even line rises above zero skill: you now must be better than the table by more than the toll to profit. And the toll is heaviest exactly where most people play. Measured the way win rates are measured, effective rake at online micro-stakes commonly runs somewhere around eight to eleven big blinds per hundred hands, easing to five or six at mid-stakes and a couple at high stakes — ranges, not constants, varying by site and format. Sit with the micro-stakes figure: a genuinely strong small-stakes player generating eight big blinds per hundred of real edge — a player who would be a clear long-term winner in a rake-free world — can be taxed to exactly zero and never see it happen, because the money comes out of pots before he touches them. Live poker's time-and-drop structures do the same at low stakes, where a few dollars a hand against small pots is a brutal effective rate.

So the question "what percentage of players are profitable" is partly a question about pricing, and its answer moves when the price does. The share of winners is not a constant of nature. It is a policy outcome — which is worth remembering whenever the number gets used to flatter or shame anyone's talent.

A Defensible Synthesis, Segment by Segment

With every bias on the table, here is my honest synthesis. These are estimates and I will label them as estimates; where a segment is unmeasurable I will say so instead of inventing precision.

Everyone who ever plays poker — home games, casino tourists, app dabblers: unknowable, and mostly unmeaningful, since the median "player" logs a handful of hours. Over a lifetime of raked play, a small minority finish ahead; something in the 10-to-20-percent range at any snapshot is consistent with the data above, and lifetime the number is lower. Nobody has measured it and nobody can.

Online cash-game regulars (tracked, 10,000+ hands): roughly 30 percent showing profit after rake in the samples we have, before rakeback. Stretch the sample to serious volume — hundreds of thousands of hands — and the share almost certainly shrinks as variance washes out, while rakeback pulls a band of high-volume break-even players back above zero. Solid winners — the ones a training site would put in a testimonial: plausibly 10 to 15 percent of tracked regulars, which is a far smaller slice of all accounts.

Live regulars at low and mid stakes: probably the highest winner share of any raked segment — the pools are softer, the floor of the average opponent lower, and the solver-era flattening that homogenized online play arrived late and unevenly to live rooms. But this is exactly the segment where no dataset exists at all: no trackers, no databases, self-reported results only, from the demographic most motivated to misreport. I believe a meaningfully higher fraction of serious live regulars beat their games than online regulars beat theirs. I cannot prove it, and neither can anyone selling you the live dream.

Tournament players: the average entrant loses — Levitt and Miles put the non-elite field at roughly −15 percent ROI before expenses — and the winner distribution is savagely top-heavy, with variance horizons measured in years even for genuinely skilled players.

And the losing scenarios deserve their own plain paragraph, because this is people's money. "Profitable" is not the same as paid. A winning rate at small stakes can amount to a few dollars an hour — real skill, taxed by rake to the order of pocket money — and it comes bundled with costs the graphs never show: downswings that outlast bankrolls, expenses and taxes, games that dry up, and the documented tendency (Siler again) to misread your own results in your favor. Most people who set out to make money at poker do not, and a fraction lose amounts that matter to their lives. Any answer to this page's question that doubles as an income promise is a lie; if the question you're actually holding is whether to keep pushing at the game at all, that question gets its own honest treatment.

Your Own Number Needs a Sample Before It Means Anything

Which brings the question home, because nobody searches this phrase out of census curiosity. The question under the question is am I one of them? — and here the measurement problem turns personal. The same variance that ruins the population statistics ruins your self-assessment. A true win rate of a few big blinds per hundred hides under a standard deviation of ninety or a hundred per hundred hands; at that ratio, tens of thousands of hands are needed before a profit graph is distinguishable from a lucky coin, and a live player at thirty hands an hour can grind a year without reaching statistical daylight. Your last three winning months are, mathematically, almost no evidence. So are your last three losing ones. Nearly everyone at your table is running on a sample that proves nothing, in whichever direction their memory prefers.

That's not a counsel of despair; it's a measurement problem, and measurement problems have tools. The Verdict exists for exactly this — it takes your actual results and sample size and tells you what they do and do not yet prove, which is a stranger kind of honesty than a flattering percentage. The population number, whatever it is, was never going to answer the question you came here carrying. Your own number can — once it has an N behind it.

The wider arithmetic — the rake, the flattened field, and what still pays in a game where the answers got cheap — is a book called The Church of GTO. The whole thing is on the shelf.