Glossary

The share of everything ever staked on a game that the game returns, averaged over a number of plays far larger than anyone will ever make.

RTP is a property of a game’s design, calculated from its paytable and the probability of each outcome. It is not measured from what players experienced, and it is not a promise about any session.

The figure is a limit. As the number of plays grows without bound, the share returned converges on it. It says nothing about where a given session sits on the way there, and no realistic number of sessions is close to the limit.

A 96% RTP does not mean 96 cents comes back per dollar staked. It means that across all money ever staked on that game, by everyone, the long-run share returned approaches 96%.

Where this is used: RTP, Expected balance calculator, Volatility

The share of each amount staked that the game is expected to keep, defined as 1 − RTP.

House edge and RTP are the same fact stated from opposite sides. A 96% RTP is a 4% house edge; a 4% house edge is a 96% RTP. Neither number is more fundamental than the other.

The edge applies to every amount staked, not to a starting balance. This is why a wagering requirement is expensive: the edge is charged again on each pass, and a playthrough requirement forces many passes.

Because the edge is a per-stake rate, changing bet size does not change it. A larger stake multiplies both the expected loss and the spread of results; it does not shift the rate.

Where this is used: RTP, Expected balance calculator, Responsible play

The total credited wagering that must be completed before a balance becomes eligible for redemption, usually stated as a multiple.

A requirement is written as a multiple of some base — a purchase amount, a bonus amount, or the sum of both. Which base applies changes the answer substantially, and the headline multiple rarely says which one it is.

The multiple alone means nothing. A 10x requirement at 96% RTP with full contribution is expected to cost about 40% of the balance: 10 × the balance is staked, and 4% of each amount staked is the house edge, so 10 × 4% = 40%.

Contribution changes that figure sharply. At 20% contribution the same 10x requirement needs five times the amount actually staked to earn the same credited wagering — 50 × the balance — and 50 × 4% is 200%, which is to say the balance is expected to be gone before the requirement is met.

Where this is used: Playthrough rules, Playthrough calculator, Game contribution, Worked examples

A common informal name for a playthrough requirement; this site uses “playthrough requirement” or “wagering requirement” instead.

The three terms describe the same mechanic. “Rollover” is avoided here because it suggests a balance being turned over once, which understates what the requirement asks for.

A 10x requirement does not roll a balance over ten times in any recoverable sense. It requires ten times the balance to be staked, and the house edge is charged on each of those stakes.

When an operator’s own terms use the word, this site quotes the operator’s wording and states the mechanic in its own terms alongside.

Where this is used: Playthrough rules, Playthrough calculator

The percentage of an amount staked on a given game that counts toward a playthrough requirement.

Contribution is the multiplier between what is staked and what is credited. At 100% the two are equal. At 20%, five units must be staked for every one unit credited.

Because the house edge applies to the amount actually staked rather than the credited amount, low contribution multiplies the cost of a requirement by the reciprocal of the contribution rate. Twenty percent contribution makes a requirement five times more expensive in expectation.

Contribution rates are operator rules, not mathematics. This site publishes them only where they were read from an operator’s own terms, and leaves them empty otherwise.

Where this is used: Game contribution, Playthrough calculator, Playthrough rules

The amount that counts toward a playthrough requirement, which equals the amount actually staked only when contribution is 100%.

Separating these two numbers is the reason this site exists. A requirement is expressed in credited wagering; the cost of meeting it is driven by the amount actually staked.

Credited wagering = amount actually staked × contribution rate. Rearranged: the amount that must be staked = requirement ÷ contribution rate.

Promotional copy usually states the requirement and the contribution rate in different places, or states the requirement and omits contribution entirely. The interaction is where the cost hides.

Where this is used: Playthrough calculator, Game contribution, Worked examples

A cap on the size of a single stake that will count toward a playthrough requirement.

Where a cap applies, stakes above it may count partially, may not count at all, or may void the promotion — which of those applies is an operator rule and differs between operators.

The cap sets a floor on how long meeting a requirement takes, because it bounds the credited wagering earned per play. It does not change the expected cost, which depends on the total amount staked rather than how that total is divided.

This site publishes a cap only where it was read from an operator’s own terms.

Where this is used: Playthrough rules, Playthrough calculator

The average result of a play, weighted by the probability of each outcome, over very many repetitions.

Expected value is an average over a distribution, not a forecast of any single result. For a game with a house edge, the expected value of each amount staked is negative and stays negative regardless of stake size, session length, or game choice.

It is additive across plays: staking an amount twice has twice the expected loss of staking it once. That additivity is what makes playthrough requirements expensive in a way the headline multiple does not show.

An expected value can be a number no individual play can produce. That is not a defect; it describes the centre of a distribution, not a possible outcome.

Where this is used: Expected balance calculator, Volatility, Responsible play

The average balance remaining after a stated amount of wagering, over very many repetitions of the same wagering.

Expected balance = starting balance − (amount actually staked × house edge). It follows directly from the house edge being a per-stake rate.

The figure describes the centre of a distribution of outcomes. Real sessions scatter around it, and the width of that scatter is set by volatility rather than by RTP.

This site always names it the expected balance. Possessive phrasing — describing it as the balance a reader will hold — states a certainty the mathematics does not support.

Where this is used: Expected balance calculator, Variance explorer, Volatility

A measure of how widely individual results scatter around their average, independent of what that average is.

Variance is the expected squared deviation from the mean. Its square root, the standard deviation, is in the same units as the result and is usually the more readable figure.

Two games with identical RTP can have completely different variance. Holding the long-run return fixed while making wins rarer and correspondingly larger raises variance without touching RTP.

Over a session, variance dominates the experience. The average is a statement about the limit; the scatter is what actually happens on the way there.

Where this is used: Volatility, Variance explorer

The informal name for how widely a game’s results scatter around its average return, usually published as a label rather than a number.

Volatility and variance describe the same property. “Volatility” is the label a game maker publishes; variance is the quantity a model computes.

Labels such as low, medium and high are not standardised between game makers, so the same word can describe materially different distributions. This site publishes a volatility label only where a source states it, and never infers one.

A high-volatility game with a good RTP can still produce long stretches with no meaningful return. The published average and the typical session are different things.

Where this is used: Volatility, Variance explorer, RTP

The square root of variance, expressed in the same units as the result it describes.

Because it shares units with the underlying quantity, standard deviation is the readable form of variance: a spread of ±2 SC means something directly, where a variance of 4 SC² does not.

For a sum of independent plays, standard deviation grows with the square root of the number of plays while the expected total loss grows in proportion to it. Over enough plays the expected loss outgrows the spread, which is why long play makes the average outcome more reliable and the total loss larger.

A single standard deviation is not a bound. Results outside one standard deviation are common, and the shape of the distribution matters as much as its width.

Where this is used: Variance explorer, Volatility, Methodology

The share of plays that return anything at all, regardless of how much.

Hit frequency counts returns; it does not weigh them. A game returning a fraction of the stake on a third of plays has a high hit frequency and may still have a poor RTP.

It is closely related to how a session feels and only loosely related to what a session costs. A high hit frequency with small returns produces a slow, steady decline rather than a flat line.

Hit frequency and RTP together constrain the size of returns: for a fixed RTP, more frequent returns must on average be smaller.

Where this is used: Volatility, RTP

What a purchase is worth once the wagering attached to it has been priced in, rather than what its headline states.

A package’s headline compares an amount received against a price. That comparison ignores the requirement attached to the amount, which is where most of the cost sits.

Priced properly, the value of a package is the redeemable amount expected to survive the required wagering, compared against the price paid. The requirement, the contribution rate and the house edge all enter that calculation.

Two packages with identical headline ratios can differ substantially in value once their requirements differ. Comparing headlines compares the wrong quantity.

Where this is used: Package value calculator, Compare package scenarios, Worked examples

The class of promotional entry used in sweepstakes casino models that may become eligible for redemption after conditions are met.

Sweeps Coins are the side of a dual-currency model that carries redemption eligibility. The other side is Gold Coins, which does not.

Whether a given balance is eligible, what conditions apply, and what a redemption yields are operator rules that differ between operators and change over time. This site publishes such rules only where they were read from an operator’s own terms.

The mathematics on this site is independent of the currency name. A playthrough requirement costs what it costs whether the unit is called a coin, a credit or a currency.

Where this is used: Playthrough rules, About

The play-only side of a dual-currency sweepstakes model, carrying no redemption eligibility.

Gold Coins exist to allow play without the entry mechanics that attach to Sweeps Coins. Amounts staked in Gold Coins do not produce redeemable balances.

Because they carry no redemption eligibility, Gold Coin balances are outside the scope of the calculations on this site: there is no redeemable amount for a requirement to gate.

Packages are frequently priced as a Gold Coin purchase with Sweeps Coins included. In that structure the price attaches to the Gold Coins and the redeemable side arrives alongside, which is what makes package value awkward to read from a headline.

Where this is used: Package value calculator, About

The published version of the calculation a figure came from, shown next to the figure so a number can be traced to the model that produced it.

Every calculator on this site states its method and version. Worked examples are pinned to a version, and each one carries exactly the figures the calculator gives at that version.

If a model changes in a way that moves a published number, the example is re-derived at the new version. Nothing here is left standing out of step with the tool that produced it.

Versions change when the model changes, not when copy changes. A version bump is a statement that a figure computed before may differ from one computed after.

Where this is used: Methodology, Worked examples

The result that the average outcome per play converges on its expected value as the number of plays grows without bound.

This is what makes RTP meaningful: it is the limit the average return per amount staked approaches. It is also routinely misread.

The law describes the average per play, not the running total. As play continues, the average return per stake settles toward the expected value while the total amount won or lost spreads further apart in absolute terms.

Convergence is slow and has no schedule. There is no number of plays after which results are owed a correction.

Where this is used: RTP, Responsible play, Variance explorer

Plays whose outcomes carry no information about each other, which is the standard assumption for the games these calculations describe.

Independence means the probability of each outcome is the same on every play regardless of what came before. A run of losing plays does not raise the chance of the next one returning anything.

The belief that it does — that a game is “due” — is the gambler’s fallacy. It is a claim about dependence between plays that the design of these games does not support.

Independence is also what allows expected values to be added across plays and variances to be added across plays, which is what makes the arithmetic on this site tractable.

Where this is used: Responsible play, Variance explorer, Methodology