Payout Rate: How the Weighted Average of a Game Mix Is Built, and Which Testing House Signs It
Two casinos can be equally generous and publish payout rates a full point apart.
A game’s return to player is designed. A casino’s payout rate is measured. Almost every comparison of online casino payout rates is built on that difference without noticing it. That one difference is the whole subject, and nearly every comparison of online casino payout rates gets built on top of it without noticing.
An individual slot is configured to return 96% over its modelled lifetime, and an independent testing house certifies that before the game is deployed. A casino cannot be configured. What it publishes at the end of a month is what actually happened: everything staked, everything returned, divided. The result is a weighted average of whatever its players chose to play, and it moves with their choices rather than with the operator’s intentions.
The weighted average, worked through
Take two operators with identical games, identically configured. The only difference is what their customers played.
Two operators, the same games at the same RTP, different game mix
| Game category | Typical RTP | Casino A share of stake | Casino B share of stake |
|---|---|---|---|
| Slots | 96.0% | 70% | 40% |
| Table games | 99.5% | 20% | 45% |
| Live dealer | 98.5% | 10% | 15% |
| Published payout rate | 96.95% | 97.95% |
Casino A returns 96.95%. Casino B returns 97.95%. A full percentage point separates them, and nothing about either operator’s generosity differs by a hundredth of a point. The gap is entirely the game mix: table games return two to three points more than slots, so the operator whose players prefer them publishes the better headline.
That is the trap in every ranking built on the figure. The metric rewards having customers who play blackjack. An operator could raise its published figure without touching a single game, simply by attracting a different audience, and an operator that markets slots aggressively will always look worse than one that does not.
| Category | Value |
|---|---|
| Blackjack, basic strategy | 99.5% |
| Baccarat, banker bet | 98.94% |
| Live dealer, typical | 98.5% |
| Roulette, European | 97.3% |
| Slots, typical online | 96% |
| Roulette, American | 94.74% |
Typical modelled return by game category, the inputs to the weighted average above. The spread between slots and table games is what makes a house-wide payout rate a statement about game mix rather than about the operator. Source: Category ranges compiled from the studios’ published per-game RTP; individual titles vary, and each game’s own information panel is authoritative over any figure printed elsewhere.
Read that chart against the table above and the mechanism is obvious. The weighted average can only land between the highest and lowest bars, and where it lands is decided by which bars carry the volume.
Why the game mix makes the figure jump around
Those monthly figures are noisier than their two decimal places suggest, and the noise has two separate sources.
The first is the mix moving. A month with a heavily promoted slot release shifts stake towards the lowest bar on the chart and pulls the average down. A month with a live dealer promotion does the reverse. Neither tells you anything about how the games pay.
The second is that a single outcome can dominate a small denominator. A progressive jackpot paying seven figures into one month’s slot stake volume can lift that category above 100% on its own, and a category above 100% is not evidence of generosity. It is evidence that one person got lucky inside the reporting window, and the following month will fall back below the modelled figure to compensate.
This is why a figure quoted for a single month, without the period stated and without a series to compare it against, is close to meaningless. A twelve-month series from the same operator is a real piece of information. One month is a coin landing.
What a payout rate has to carry, and which testing house signed it
A payout figure is only as good as its provenance, and the useful ones state all four of these:
- The period. A rate with no stated month or quarter cannot be checked and cannot be compared with anything.
- The scope. House-wide, or per category, or per game. A house-wide figure and a slots-only figure differ by a point or more for reasons that have nothing to do with the operator.
- The auditor. In licensed markets an independent testing house verifies the underlying data and is named on the report. An unnamed auditor is not an auditor. What the licence behind that chain actually obliges an operator to do is narrower than the badge implies — the Malta licence is worked through separately, including the payout and player-funds rules that come with it.
- The denominator. Total stake, not deposits. Money won and restaked counts again each time it goes through, which is why payout rates look high next to what a player recognises from their own account.
A figure missing the period and the auditor is a marketing claim wearing the costume of a measurement. That is worth saying plainly, because payout rate is one of the few gambling numbers with a genuine verification chain behind it, and the existence of that chain is what unlicensed operators borrow when they print a percentage on a homepage.
The number that actually applies to you
Here is the uncomfortable part. Even a properly audited, properly scoped, twelve-month house-wide payout rate describes the aggregate behaviour of a population of players, and you are not that population.
If you only play blackjack, the house-wide rate is irrelevant to you: it has been dragged down by other people’s slot spins. If you only play one slot, the figure that governs your play is that game’s certified RTP, stated in its own information panel, and the house average has no bearing on it at all.
So the comparison worth making between operators is not the headline. It is whether the specific games you intend to play are deployed at their highest certified configuration, because several studios ship the same title in more than one version and an operator chooses which to run. That check happens on the game screen, one game at a time, and it is covered in more detail under what RTP means and how it is verified. The titles where the ceiling is highest, and the conditions attached to each, are listed under the highest RTP slots.
Where the return is not audited at all but derived from the geometry of the game itself, the arithmetic is public and needs nobody’s certificate. That is the odds of the common games, where a European roulette wheel’s 97.3% follows from thirty-seven pockets and nothing else.