Why a Small House Edge Costs More Than a Player Expects

What a House Edge Measures

The Gambling Commission's consumer guidance on gaming machine payouts, a page last updated on 16 June 2021, defines the term in its own words: “This is called the house edge, and it gives a measure of the percentage that a casino would expect to keep, on average, from each hand or spin, given normal patterns of play.” That page addresses gaming machines rather than online slots, and it is precise about the companion figure: “Percentage Return To Player (%RTP) can be thought of as part of the money paid to play the machine that is then given back to the player as prizes.” It adds that “The %RTP is an average achieved over a significant number of game plays and not each time the gaming machine is played.”, and it states the measurement window: “Average %RTP is generally measured over 10,000 or 100,000 games for compensated machines and more games than this for random machines, depending on their category.”

Every Bet Starts From the Same Position

The same page states that “The odds of achieving a win in the current game remain constant and are not affected by wins or losses in previous games.” For remote games the Commission's technical standards turn that into a licence requirement. RTS 7 requires that “Random number generation and game results must be ‘acceptably random’.” and that “Adaptive behaviour (that is, a compensated game) is not permitted.” Its guidance spells out the scope: “Restricting adaptive behaviour prohibits automatic or manual interventions that change the probabilities of game outcomes occurring during play.” Where a game simulates several devices, “Where multiple physical devices are simulated the probabilities of each outcome should be independent of the other simulated devices.” And each outcome is drawn against fixed odds: “The mapping of the random inputs to game outcomes should be in accordance with prevailing probabilities, pay tables, etc.” A staking pattern therefore cannot change the probabilities it is betting into.

The Quantity That Grows With the Number of Bets Is Turnover

What accumulates across a long run of bets is the total amount staked. The Commission's live return to player monitoring guidance, last updated on 29 January 2025, defines that total: “The total of all stakes made on the game, this will include reinvested winnings awarded during play.” Measured against it, “The GGY of a game will be the turnover minus win.”, and “By dividing the win and turnover figures generated from a game you can determine the actual RTP.” The expected amount a game retains is the total staked multiplied by the share it is designed to keep. Ten bets of one unit stake ten units in total, so the expected loss across them is ten units multiplied by the house edge. It is stake multiplied by number of bets multiplied by house edge — not one stake shrinking by a percentage each round.

The Commission's own worked example shows the scale on a real measurement: “For example, if after one month of play a game, designed with a 91.68% RTP, has accrued £1,200,000 of turnover and £1,085,000 in wins the RTP can be calculated as follows:” The division it performs is “1,085,000 / 1,200,000 = .9042”, and it concludes: “Therefore this game has achieved an actual RTP of 90.42%, which is below the designed RTP.”

Why a Short Run Shows Nothing

The same guidance explains why a handful of spins carries no information about a return figure: “Measuring a game with only a small amount of play will be pointless as the tolerance will be too large for meaningful results.” Volume is what narrows the band — “The tolerance will be wider when only a limited amount of play has been measured, but as the volume of play increases the tolerance will decrease.” — and only at scale do the two figures meet: “After a significant amount of plays the actual RTP should be very close or equal to the theoretical RTP.” The Commission publishes the tolerances. For the game in its example, designed at 91.68% with a volatility of 5.6: “So if 400,000 games had been played to accrue turnover and win figures of the example then the allowable tolerance will be 1.75 above or below 91.68.” At that volume, “The game could return between 89.93% and 93.43% and still be considered to be performing as expected.” Four hundred thousand plays is a volume for a game across all of its players, not for one person's session.

Decisions Reduce a House Edge and Do Not Remove It

The Commission's definition of the house edge is framed on “normal patterns of play”, and where skill is involved its RTS 3 guidance ties the published return figure to the strategy assumed: “return to player percentage should be calculated using either the auto-play strategy or a standard/published strategy”. Skill does not make an outcome certain in either direction: “Where games involve an element of skill, every outcome described in the virtual event rules or artwork should be possible, that is, the customer should have some chance of achieving an advertised outcome regardless of skill.” Choosing one game over another, or one bet inside a game over another, changes which house edge applies to the money staked. It does not change the fact that each result is drawn against fixed probabilities. A decision can reduce a house edge; it never eliminates it.

Why Staking Systems Cannot Change the Odds

Staking strategies such as the Martingale or the D’Alembert do not work. The theory behind them is that a player should increase the stake after a loss, on the basis that a win is more likely next time. What they ignore is the Commission's own sentence quoted above — “The odds of achieving a win in the current game remain constant and are not affected by wins or losses in previous games.” — and the licence requirement behind it, that “Adaptive behaviour (that is, a compensated game) is not permitted.” The outcome of a casino game bet is independent of the outcomes before it, and a licensed game may not be built any other way.

A bet on black is no more likely to win because the ball fell on red on the last spin than it would be if the ball had landed on black. These systems all fall prey to the gambler’s fallacy: it does not matter if the ball lands on red eight times in a row, because an unusual run does not make red less likely on the next spin. The odds stay the same.

When playing casino games, it is worth remembering that the first goal is entertainment. Over the long term, the chances of beating the house are remote, and knowing when to stop — whether a session is up or down — is worth more than any staking rule.