Return to Player: What RTP Really Tells You

Return to player, or RTP, is the percentage of all money staked on a game that the game pays back — measured over millions of rounds, not over your session. A slot with an RTP of 96% keeps, on average and across every player who ever touches it, four cents of every euro wagered.

That is the definition. The rest of this page is about what the number does not tell you, which is where almost all of the confusion lives.

RTP and house edge are the same fact, written twice

The house edge is simply what is left over: house edge = 100% − RTP. A 96% slot has a 4% edge. A blackjack table at 99.5% has an edge of 0.5%. Casinos and studios tend to publish RTP because a big number reads better than a small one, but the two are interchangeable.

Where the wording matters is scope. RTP describes a game. House edge is usually quoted for a bet. In games with one wager the two are the same thing said differently; in games with many wagers they come apart completely, which is covered further down.

How the figure is actually produced

RTP is not measured by watching real players. It is calculated from the game’s own mathematics — the symbol weightings on each reel, the paytable, the trigger frequency of every feature — and then confirmed by simulation, typically over billions of rounds, before an independent test laboratory certifies that the built game matches the maths its designer submitted.

Two consequences follow, and both are routinely misunderstood.

It is an average, not a schedule. A 96% slot does not owe you 96 cents back per euro this evening, this week, or this year. It has no memory of what it has paid, no concept of being “due”, and no mechanism to correct a run in either direction. Machines are not loose or tight by mood — we wrote about that separately.

Your session is statistically invisible. A few hundred spins against a simulation of several billion is not a small sample; it is no sample at all. Everything you will personally experience happens inside the noise around the average, which is why the next number matters at least as much.

Volatility: the number that describes your evening

Two slots can both return 96% and feel nothing alike. Volatility — sometimes called variance — describes how that return is distributed.

• Low volatility. Frequent small wins. The balance drains slowly and predictably. Longer play for the same money.
• High volatility. Long dry stretches punctuated by rare large wins. Same long-run return, radically wider swings, and a much larger bankroll needed to survive to the point where the average means anything.

If return to player tells you what the game costs, volatility tells you how it will feel while it costs it. A player who reads only the RTP and picks a high-volatility game with a modest bankroll has chosen a game that will almost certainly end early.

The part nobody advertises: one game, several RTPs

Here is the fact that changes how you should read every return to player figure you see online.

Studios routinely ship the same slot in multiple RTP configurations — a headline version at, say, 96.5%, and lower-paying builds of the identical game at 94.5% or lower — and the operator chooses which one to run. Same title, same artwork, same reels, same name in the lobby. Different return.

This is why an RTP quoted on a review site, or even on the studio’s own promotional page, is not automatically the RTP of the game you are about to open. The authoritative figure is the one inside that operator’s copy of the game, in its info or paytable screen.

It is also exactly why we do not print a return figure unless we can verify it. Across our catalogue of 161 games, 34 carry a published RTP that we could check. The other 127 have the field empty. That is not an oversight and we would rather it looked incomplete than looked confident and wrong.

What our own numbers look like

Of those 34 verified figures:

The gap between the top and bottom of that list is 5.5 percentage points — on a €4,000 turnover, the difference between losing about €31 and losing about €250, on average. It costs nothing to check which one you are playing.

Note also that the table games sit at the top and the slots below them. That ordering is not a coincidence: games where the player makes decisions, or where the maths is simple enough to be checked by hand, are priced tighter than games where it is not.

Typical return to player by game type

Return to player varies far more between game types than between titles inside a type. These are the ranges you should expect before you open anything:

Our own catalogue confirms the roulette line exactly: we hold European Roulette at 97.30% and American Roulette at 94.74%, from the same studio. Identical game, one extra pocket, and 2.56 percentage points of your money. Choosing the European wheel is the single easiest return-to-player decision available in any casino.

Note too that within baccarat the spread between the best and worst bet on the same table is over thirteen points, which brings us to the games where one figure will not do.

Where a single RTP figure stops meaning anything

Two of the most popular casino games cannot be summarised by one number at all, and any site that gives you one is guessing.

Craps. The return depends entirely on which bet you place. The free odds bet behind the line returns 100%; the “any seven” box returns 83.3%. An average across them would describe a player who does not exist. We break down every craps bet by house edge separately.

Video poker. The return lives in the paytable, which the operator configures. The same Jacks or Better game pays 99.54% or 95.00% depending on two rows you can read before your first hand — see video poker strategy.

Both are the same lesson as the multiple-RTP problem above: the number belongs to the configuration in front of you, not to the name on the lobby tile.

Using return to player without fooling yourself

Four things it is genuinely good for:

1. Comparing two games before you play. Higher is better, all else equal, and the difference is real money over time.
2. Costing a bonus. Wagering requirements are paid in house edge: turnover × (100% − RTP) is what clearing an offer costs on average. That single line is the whole of wagering requirements.
3. Spotting a game that is quietly expensive. Anything below 94% is being priced for a player who is not checking.
4. Sanity-checking a casino’s marketing. An operator advertising “up to 98%” is quoting its best game, not its lobby.

And two things it is not: a prediction of your session, and a reason to keep playing. No RTP is above 100% in practice, so every game on every site is, in the long run, a cost. The number tells you how much that cost is — nothing more optimistic than that.

Four things people get wrong about RTP

“The casino can turn the RTP down when I start winning.” No. Changing the configuration of a certified game is not a switch behind the counter; it means loading a different certified build, and in regulated markets it is logged. What is true is the point made above — the operator chose which build to install before you arrived, and may never have told you which.

“A machine that has not paid out is due.” Each round is independent. The game has no record of what it has paid and no mechanism to catch up. This is the gambler’s fallacy, and it is the single most expensive misunderstanding in the building.

“Higher return to player always means a better game for me.” Better priced, yes. Better suited, not necessarily. A 97% high-volatility slot will end a small bankroll faster than a 95.5% low-volatility one, because you will not survive long enough for the average to apply.

“RTP includes the bonus round.” It does — and that is the catch. On many modern slots a large share of the advertised return sits inside features you may never trigger. The headline figure is the whole game, base and bonus together, which is why base-game play can feel so much worse than the number suggests.

The return the RTP figure never covers

Everything above describes a property of the game. Whether you actually receive the money is a property of the operator, and no RTP figure measures it. A 97% slot returns nothing at all if the balance never leaves the casino.

That second number has no label and never appears in a paytable, but it is the one that decides what you end up with: how fast withdrawals clear, whether an account gets frozen at payout, whether a bonus clause is used to void a win.

Self-custody removes the question rather than answering it. On Keyhouse your balance sits in your own wallet, so there is no operator step between a winning round and the money being yours — settlement happens as the round ends, and you wait on network confirmation, not on a cashier. Its catalogue runs a typical 95% to 98% RTP, from studios like BGaming, Yggdrasil, Thunderkick and Playson. Judge the game by its RTP; judge the platform by whether it can hold your balance at all.

How to check it yourself, in thirty seconds

Open the game, find the info, help or “i” button, and look for the paytable or game rules screen. The RTP is stated there, in the operator’s own build, and that figure is the one that applies to you. If it is absent, or if the operator does not let you see it before depositing, treat the absence as information.

The same principle runs through everything on this site. Our crypto casino comparison checks each operator’s licence in the regulator’s own register rather than repeating what the operator claims, and publishes its unresolved-complaint count. And if you want the version of this argument where the player can verify each individual round rather than the long-run average, that is provably fair gambling.

Set your limits before you deposit: see responsible gambling.

Catalogue figures — 34 of 161 games with a verifiable published RTP, median 96.95%, range 93.75% to 99.23%, and 15 of 105 slots with a median of 96.29% — measured on this site on 15 August 2026. Where a studio or operator does not publish a return we can check, we leave the field empty rather than estimate.