Reference · 18+

Casino games, and how the mathematics behind them works

Every casino game is a piece of applied probability with a margin built into the rules, and the margin is a property of the rules rather than of luck, the venue or the time of day.

The margin is in the rules

A casino game is a piece of applied probability with a margin engineered into the payouts. That sentence contains everything else on this page.

Take a bet that wins one time in thirty-seven. A fair payout would be thirty-six to one plus the stake returned. Pay thirty-five to one instead and the game has a defined margin, which will assert itself with increasing reliability the more the bet is repeated. Nothing about that requires anybody to cheat, the outcomes to be non-random, or the player to be unlucky. The margin is arithmetic and it is published.

Everything else, the theme, the interface, the sound, the studio that made it, is presentation. The rules are the game.

Return and edge are one number

Two vocabularies describe the same thing and the difference is only convention.

Slots are usually described by return to player: the proportion of everything staked that the game is designed to pay back over a very large number of plays. Table games are usually described by house edge: the proportion the game is designed to keep. A return of ninety-six per cent and an edge of four per cent are identical statements.

Both are long-run design properties, established across a volume of play far larger than anybody personally performs. This is the point at which the number is most often misused. It is a valid basis for comparing two games and an invalid basis for predicting an evening, because the variance around it over any realistic number of plays is very large. A game designed to return ninety-six per cent will take everything staked in a session often enough that the design figure is no comfort at all.

Volatility, which matters more day to day

Return tells you the long-run margin. Volatility tells you how bumpy the road is, and for a player it is the more consequential figure.

A low-volatility game returns small amounts frequently. A high-volatility game returns rarely and in larger amounts. Two games with the same design return can produce completely different experiences and completely different rates of balance depletion, because what empties an account is not the margin but the size of the swings relative to the money available.

The practical consequence is that a modest balance on a high-volatility game is likely to be gone before the long-run figure has any chance to be relevant. Anybody choosing a game on the return figure alone, without looking at volatility, is optimising the wrong variable.

Randomness, testing, and what it does not mean

Software games use a random number generator, and operators licensed in Great Britain must have it and the stated return figures independently tested. Details of what licensing obliges are under UK casino sites.

Testing establishes that the game behaves as described: the outcomes are independent, the distribution matches the specification, the return figure is real. That is worth having and it is frequently misread as a form of reassurance about the odds. It is not. A tested game with a four per cent edge has a four per cent edge, verified.

The corollary addresses the commonest belief among players, which is that a game tightens after a win. It cannot. Independent outcomes have no memory: the generator does not know the previous result, the account balance, or how long somebody has been playing. Runs of wins and runs of losses are exactly what independence produces, and the human tendency to read a pattern into them is a well-documented feature of how people process randomness rather than evidence about the machine.

Where the differences between games actually are

If the object is to understand what makes one game a worse proposition than another, three things matter and the theme is not among them.

The first is the game type. Well-played table games generally carry a much lower edge than slots. That is not a marketing claim but a consequence of their structure, and it is why bonus terms weight table games down or exclude them, as described under no deposit bonuses.

The second is the rule variant. Within a single game, a change to one rule can move the edge by more than the entire difference between two different games, and tables sitting side by side under the same name frequently differ. The rules are in the game information panel. Reading them takes less time than a hand takes to play.

The third, in games involving decisions, is correct play. Strategy in such a game does not create an advantage; it recovers the quoted edge, which assumes correct play. Playing badly does not leave the margin where it was, it increases it, and in some games substantially.

Jackpots, and where the money for them comes from

A progressive jackpot is worth understanding as a mechanism, because it changes the arithmetic of the game carrying it in a way that is not visible in the headline return figure.

The jackpot is funded by a levy on every stake placed on the game across all the operators offering it. That contribution is part of the game’s overall return, so a game with a large progressive prize typically has a lower base return than an equivalent game without one: the money in the prize pool has been taken out of the ordinary paying cycle. A player who never wins the jackpot, which is almost every player, is therefore playing a tighter game than the advertised total return implies.

Whether that is a bad deal depends on what the player wants. The total return figure may be perfectly respectable, and the chance of a very large outcome is real rather than notional. But it is concentrated in an event of very low probability, so the practical experience of the game is worse than an equivalent non-jackpot title while the theoretical return is the same or better.

The general principle applies beyond jackpots. Any feature that concentrates return into rare events makes the everyday experience of a game worse without changing its design percentage, and the design percentage is the number that gets advertised.

Why the interface is designed the way it is

Two features of modern casino software are worth naming, because they are engineering decisions with measurable effects on how much gets staked and they are not obvious while playing.

The first is speed. The number of rounds per hour is the multiplier on the house edge: a four per cent margin costs four per cent of turnover, and turnover is stakes multiplied by rounds. A game that resolves in two seconds rather than six produces three times the turnover in the same hour, and therefore three times the expected cost, without the stake changing at all. Auto-play, quick-spin settings and short animations all move in the same direction.

The second is the treatment of partial returns. On slot software generally, a round that returns less than the stake is presented with the same celebratory sound and animation as a round that returns more. A fifty pence return on a one pound stake is a loss of fifty pence, and it is frequently displayed as a win with the amount counting up. This is documented in the literature on gambling design as a losses-disguised-as-wins effect, and its consequence is that a session can feel roughly break-even while the balance falls steadily.

Neither of these is concealed and neither is illegal. Both are reasons that the subjective sense of how a session went is a poor guide to what happened, and both are arguments for looking at the account history rather than trusting the recollection. Licensed operators must make that history available.

There is a related point about the figures themselves. The return percentage is a property of the game, not of the session, but the number of rounds is a property of the session and is under the player’s control. Of the two variables that determine the expected cost of an evening, the one that can actually be changed is the one nobody talks about.

What a player actually controls

The honest summary is short.

The margin is fixed by the rules and is not affected by anything a player does, except in games with decisions, where poor play makes it worse. The results over any realistic number of plays are dominated by variance and are not predictable. The only variables under a player’s control are which game, which rule variant, how much is staked and for how long.

Of those, the amount is the one that determines outcomes, and it is the one most easily decided in advance and least reliably decided during play. Licensed operators must provide deposit limits for that reason.

The duration is the second, and it is the one this page has argued is systematically underestimated. Expected cost is the margin multiplied by turnover, and turnover is stakes multiplied by rounds, so an hour of fast play at a small stake can cost more than a short session at a large one. Anybody budgeting for an evening by thinking about stake size alone is working with half the equation.

Gambling involves risk of loss and nothing here is advice to gamble. Anybody under eighteen must not. Anybody who wants to stop should use the national self-exclusion scheme rather than a page like this one, and BeGambleAware provides advice independently of the industry.

Sources

• Online casino — Wikipedia
• Return to player — Wikipedia
• Gambling Commission
• BeGambleAware

Checked

2026-08-25

REF/CG-001 · 4 sources