Return to player, house edge and provably fair, compared

When a studio builds a slot, it defines the mathematical model — the symbol distribution, the paytable, the feature frequencies — and from that model the long-run return to player follows arithmetically. An independent testing laboratory checks that the delivered game matches the declared model and that its random number generator behaves as claimed. The certificate belongs to the game, and it travels with the game.

A casino licenses that game and hosts it. In the ordinary case it cannot alter the mathematics. What some studios do offer is a game supplied in more than one configuration, with the operator choosing which to run — and where that happens, the figure shown in the game's own information panel still reflects the configuration you are actually playing. That panel, not a comparison table, is the authoritative source for any specific title.

This is why we publish no percentage for any operator on this site. A published RTP belongs to a title and to the studio that built it; attributing it to a casino misrepresents where it came from, and the numbers are not hard to find at their real source. A brand that carries Play'n GO offers Play'n GO's published figures on Play'n GO titles, identical to every other venue carrying the same titles.

House edge is what remains after return to player, and the two always sum to one hundred per cent. A game returning 96 per cent has a 4 per cent house edge. Neither figure tells you anything about a session; both are long-run averages over a very large number of rounds, and the variance around them at any realistic number of spins is far larger than the difference between two games a percentage point apart.

That has a practical consequence for how much weight to place on the figure at all. Choosing a game with 97 per cent over one with 96 per cent shifts the expected cost of a hundred-unit turnover from four units to three. Volatility — how the return is distributed across sessions — will swing your actual outcome by orders of magnitude more than that, and volatility is disclosed far less consistently than return to player is.

For crash-style and dice originals the arithmetic is usually stated as a house edge directly, often one per cent, and the game's help page normally says so. It is the same statement in the other direction, and it is the operator's own number in that case, because the operator built the game.

Counting how many of the 24 published lists name each studio gives a workable map of where the certified content sits. If you care about a particular studio's published figures, this tells you how widely available its games are across BC.Game alternatives.

The final column is the one to act on. In every case the answer is the same and it is not a comparison site: the return figure for a given title is published by the studio and is shown inside the game, normally in the information or paytable panel reached from the game's own menu.

BC.Game describes its provably fair implementation as a pre-generated hash chain: ten million hashes generated in advance and released in reverse order, one per round, so that a round's result exists before the round begins. The verification code is published on GitHub. That is a materially different design from the seed-pair scheme most alternatives run, in which a server seed and a client seed are combined per round and the server seed is revealed afterwards.

The practical difference is what you are checking and when. Under a seed-pair scheme you verify one round at a time, after the fact, by confirming that the revealed server seed hashes to the commitment you were shown beforehand. Under a chain, each round's hash must produce the previous round's hash, so the entire history is linked: a single altered result breaks the chain for every round after it, and anyone can walk the chain backwards to the published head.

Neither scheme replaces an RNG certificate, and neither is replaced by one. A laboratory certificate is a statement by a third party that a game's declared model is the model it delivers. A provably fair scheme is a cryptographic commitment that lets you check individual outcomes yourself, without trusting anyone, but it says nothing about the house edge built into the game. Eighteen of the 40 alternatives in our library carry a provably fair tag; that tag records that a scheme exists, not that we have audited it.

We will not publish an average return figure for a casino. Such a number would require the return of every title, weighted by how much is actually staked on each, and that staking data is not published by any operator here. What circulates instead is either a single game's figure presented as the venue's, or a number with no derivation at all.

We will not rank brands by return to player. Two brands carrying the same studios offer the same games with the same published figures; a ranking between them would be measuring their marketing, not their mathematics. Where brands genuinely differ is in catalogue composition, and that is set out on the providers comparison rather than dressed up as a percentage here.

What we will say is which studios each brand carries, which we have read from operators' own sites, and where the authoritative figure for any title lives. If a page offers you more precision than that without showing you a source, the extra precision was added rather than found.