RTP is set by the casino, not the game — and almost nobody publishes theirs

What the number actually means

RTP is the proportion of total stakes a game returns to players across an enormous number of spins. At 96%, four pence of every pound staked is the house edge, and the remaining ninety-six pence is redistributed — mostly to a handful of accounts that hit something large.

The word "staked" is the part people miss. It is not four pence of every pound deposited: it is four pence of every pound wagered, and the same pound gets wagered repeatedly. Deposit £100, play it through five times, and you have staked £500 — the expected cost is £20, not £4. This is why session length matters more than deposit size, and why a low-volatility game can drain an account as effectively as a high-volatility one.

The number also says nothing about how the money comes back. Two games at identical RTP can feel completely different: one pays small amounts constantly, the other pays nothing for four hundred spins and then pays a great deal. That is volatility, it is a separate property, and it decides what a given evening feels like far more than RTP does.

The setting is the casino's choice

Studios ship many slots with more than one configurable return — a game might be available at 96.5%, 95%, and 94%, with identical artwork, identical maths in every other respect and no visible difference in the lobby. The operator picks which build to run.

This is the single most valuable piece of consumer information in the whole market, and it is almost never published. A player comparing "the same game" at two casinos can be playing two different games, and the difference of two percentage points doubles the house edge from 4% to 6% — meaning the same session costs half as much again.

Where an operator does publish it, the figure normally appears inside the game information panel rather than in any comparison anyone can see from outside. That is why our rating method separates two things that are usually conflated: the studio default, which is a property of the game, and the setting at this operator, which is a property of the casino. We record them in different fields, and we never substitute the first for the second.

Across the operators in our catalogue, the honest summary is bleak: not one publishes per-game RTP figures on pages a visitor can read without an account, and most publish no figure of any kind. That is scored as zero on our transparency criterion, and it is zero for almost the whole market rather than for one unlucky brand.

How to check the RTP of a game you are about to play

• Open the game's information panel — the "i" or paytable icon — inside the lobby. Where an operator publishes anything, it is here.
• Compare it against the studio's own published range for that title. A figure below the top of the range means a lower build is running.
• Check it again after a lobby update. Builds get swapped, and nothing announces it.
• Treat "up to 96%" in marketing copy as meaning nothing at all. The words "up to" are doing the work.
• If no figure is available anywhere, that is information too: it means the cost of playing is not disclosed.

What RTP does not protect you from

It does not make a session predictable. Over a few hundred spins, the spread of outcomes is enormous and the theoretical return is barely visible — which is precisely why gambling remains entertaining and why the maths remains undefeated.

It does not apply to bonus play in the way most players assume. Wagering a bonus through means turning the balance over many times, and each turn costs the house edge. A 35× requirement on a £100 bonus means £3,500 of turnover, and at 96% that costs £140 in expectation — more than the bonus. Our <a href="/bonus-wagering-calculator">wagering calculator</a> puts a pound figure on any offer.

And it does not describe live dealer games, where the equivalent number is the house edge of the specific bet you are making. On a roulette table, a single-zero wheel takes 2.7% of every stake and a double-zero wheel takes 5.26% — the same game, the same table layout, twice the cost. In blackjack the figure depends on the rules of the particular table and on whether you play the correct strategy, which most players do not.

Volatility: the half of the picture RTP leaves out

Two slots can both return 96% and feel like completely different games. RTP tells you the long-run average; volatility tells you how the money arrives, and for anyone playing for an evening rather than a lifetime, volatility is the number that describes the experience.

A low-volatility slot pays small amounts often. Your balance drifts down slowly, sessions last a long time on a given deposit, and a big win is essentially off the table. A high-volatility slot pays rarely and heavily: long stretches of nothing punctuated by the occasional large hit, and a much wider range of possible outcomes over any short session.

The consequence people miss is that high volatility makes the advertised RTP almost irrelevant to a single session. A 96% game whose returns are concentrated in a one-in-twenty-thousand event returns 96% across millions of spins and something close to zero across your Saturday evening. The average is real; it is just not a description of what happens to you.

This is also why "I picked the highest RTP game and still lost" is not a contradiction. RTP is a property of the game over an enormous sample. Your sample is not enormous.

If you want sessions that last, choose lower volatility and a stake that gives you several hundred spins. If you are buying a small chance at a large outcome, choose high volatility and accept that the likeliest result is losing the stake — which is a legitimate choice, provided it is the one you meant to make.

Where the number comes from, and who verifies it

A slot's RTP is not an observation, it is a specification. The studio designs the maths, calculates the theoretical return over the full cycle of outcomes, and then has it verified by an independent testing house before the game can be supplied to a licensed market.

In Britain, testing is a licensing requirement rather than a marketing badge: game software supplied to British licensees has to be tested against the Commission's technical standards by an approved test house. Names you will see include eCOGRA, GLI, iTech Labs and BMM. A studio can tell you which one certified a title, and a casino can tell you which one audits its operation, but neither is obliged to publish it prominently — and most do not.

This is a criterion in our own method for that reason. We look for a named testing house on the operator's pages, and we score zero when we cannot find one. That is a finding about disclosure, not an accusation that the games are untested; the testing is a condition of the licence either way.

The distinction worth holding onto: certification confirms the game behaves as specified. It does not tell you which specification the casino chose to run — which is the next section, and the part that costs players money.

The same game, two casinos, two returns

Many studios ship a title in several RTP configurations — commonly something like 96.5%, 94.5% and 92% — and the operator picks which one to run. The game looks identical: same name, same artwork, same features, same paytable presentation. The return is not identical.

The gap is not trivial. Over £1,000 of stakes, a 96.5% setting costs an expected £35 and a 92% setting costs £80. Over a year of regular play, that difference is larger than most players' entire bonus intake.

A well-run operator publishes the RTP of each title in the game information panel, which every British-licensed casino must make available. The panel is usually reachable from an "i" icon inside the game, and it states the actual configured return for that game at that casino — not the studio's headline.

That is the check worth doing, and it takes ten seconds: open the game, open the information panel, read the RTP line. If a casino makes that panel hard to reach, or the figure is absent, treat it as a data point about the operator.

It is also why our method records return-to-player at two levels — the studio default on the game, and the figure actually configured at each operator — rather than pretending one number describes both. Where we have not read the operator's figure, the review says so instead of substituting the studio's.

Table games, live dealer, and the numbers that beat slots

Slots occupy a narrow band: roughly 92% to 97% at British operators, with 96% the common centre. Table games sit well above it. Blackjack played to basic strategy runs at a house edge below 1%, which is a return above 99%. European roulette is 97.3%. Baccarat on the banker bet is about 98.9%.

Two things stop that from being free money. The first is that basic strategy is a discipline, not an instinct: the edge figures assume every decision is correct, and casual play gives most of it back. The second is bonus weighting — the games with the lowest edge contribute least toward wagering requirements, or nothing at all, precisely because the maths would otherwise favour the player.

Live dealer versions carry the same underlying odds as their digital counterparts, with two practical differences: the pace is slower, which reduces total turnover per hour and therefore total expected loss per hour, and side bets are far more prominent. Side bets are where live tables make their margin, and their house edge is frequently in double digits.

The honest summary is that game choice moves the expected cost of an hour by a factor of ten or more, and almost nobody chooses on that basis. If you want to reduce the cost of playing without reducing the time you spend, that is the lever.

Bonus buys, and the RTP that changes when you pay for it

A bonus-buy feature lets you pay a multiple of your stake — typically 60 to 100 times — to enter the free-spin round immediately. It is popular, it is fast, and it usually runs at a different RTP from the base game.

Sometimes that difference is upward: several studios configure the bought round slightly above the base RTP, because the player is committing a large stake in a single transaction. Sometimes it is downward. Either way the figure in the information panel normally lists both, and they are not the same number.

What does not change is the volatility, and buying the feature concentrates it violently. A hundred-times stake spent in a single round means an evening's budget can resolve in twenty seconds. If a bought feature costs more than you would stake in an hour of ordinary play, it is not a shortcut to the same experience — it is a much larger bet wearing the same interface.

Bonus buys are also frequently excluded from wagering contribution, or banned outright while a bonus is active. Using one with an active bonus is one of the routine ways winnings get voided.

Reading an RTP claim on a casino page

• A stated average RTP across the lobby is close to meaningless unless the operator says how it was calculated and over what period. Averaged across which games, weighted by what?
• A published per-game figure in the in-game information panel is meaningful, and it is the one that binds. If it disagrees with the studio's published default, the panel is the truth for that casino.
• A monthly payout percentage — "our players won 96.4% last month" — is an observation about actual play, not the configured RTP. It moves with what people happened to play and with a small number of large wins, and it can be above 100% in a month when a jackpot lands.
• A testing-house logo without a linked report is decoration. A linked, dated report is evidence. The difference is one click and almost nobody provides it.
• No published figure at all is common and is a reason to check the in-game panel yourself before staking, rather than a reason to assume the worst.

What this means for choosing where to play

RTP transparency is worth six points in our rating method, which places it above the licence check and level with the largest single bonus criterion. That weighting is deliberate: over any sustained period, the configured return costs or saves a player far more than a welcome offer is worth.

Across this catalogue, published RTP policy is the exception. Most operators publish nothing at the site level, which scores zero on that criterion — and, as everywhere on this site, that is a finding about what is published rather than a claim about what is configured.

If you take one habit from this guide, make it the information panel. It is available in every game at every British-licensed casino, it states the number that actually applies to you, and it is the only figure in this entire subject that is neither an average nor a promise.

Questions

Read next