RTP: the first number worth reading

Every slot publishes a return to player figure, shortened to RTP, and it is the most useful single number on the game’s information page. Written as a percentage, it models the share of everything staked that the game hands back to players across a very long run. A game at 96 percent returns, on the model, £96 of every £100 pushed through it over its lifetime. The figure describes the game’s whole life, not your Tuesday evening, and holding that distinction in mind is most of the work of reading slots sensibly.

Across the market, published returns cluster into three working bands. The low band runs from 94 to 95.9 percent, the middle band from 96 to 96.9 percent, and the top band covers 97 percent and above. On the model this page uses, drawn as a distribution across the current catalogue, the middle band holds the largest single share at 44 percent, with the low band at 31 percent and the top band at 25. The chart below shows that spread, and the table underneath adds what each band tends to travel with.

What the figure does promise is a price. House edge is simply 100 minus RTP, and edge turns into expected cost at a rate you can budget: at 96 percent, every £10 staked costs 40p in expectation, at 95 percent it costs 50p, and at 97 percent it costs 30p. Turnover is the multiplier that matters. A hundred spins at 20p is £20 of turnover and 80p of expected cost at the middle band’s better end. Bonus wagering multiplies the same arithmetic: £336 of wagering at a 4 percent edge carries £13.44 of expected loss, whatever the balance started with.

What the figure does not promise is anything about a single session. RTP is a long-run average across every stake and every player, and short-run variance dwarfs it: a 96 percent game can swallow a £50 bankroll in an hour, and it can hand back £300 in ten minutes, and neither outcome contradicts the model. The figure is not a schedule of returns, not a prediction for the evening, and not a reason to expect a cold streak to be owed a hot one.

A second number often gets tangled with RTP: hit rate, the share of spins that return anything at all. A game can hit on a third of spins and still lose money slowly, because most hits are smaller than the stake. RTP describes the money; hit rate describes the rhythm. Neither predicts the next spin, but together they explain why two games with identical returns can feel entirely different within an hour.

Against offers, the bands set the terms of comparison. A spins bundle aimed at a middle-band game is priced on a 31 to 40p cost per £10 of play, while the same bundle on a low-band game costs the model 41 to 60p per £10 for identical face value. Where an offer lets you choose the game, the band difference is the cheapest margin available to a reader, and it costs nothing but a minute on the information page.

Practically, read the number before an offer attaches itself to a game. Spins offers usually name their slot, and the game’s information page prints the RTP that applies at the stake the offer uses. Feed that figure, not a marketing round-up figure, into the arithmetic later on this page, because a two-point difference in RTP is a quarter of the expected cost at the middle band.

Volatility and the shape of a session

Volatility describes how a game’s return is distributed, and unlike RTP it rarely gets a printed number. Low volatility means returns land often and stay small: the balance erodes in a gentle slope, sessions run long, and the paytable’s ceiling sits modestly. High volatility means the opposite: long stretches where little lands, punctuated by outcomes that can reach several hundred times the stake. Two games at the same 96 percent RTP can sit at opposite ends of that spectrum, which is why the return figure alone cannot describe an evening.

Session length follows directly. On the model, a £50 bankroll at 20p a spin is 250 spins of raw capacity before any return, and low volatility keeps the actual run close to or beyond that, because returns keep feeding the balance. High volatility can end the same bankroll inside a hundred spins, or triple it, with the outcome decided by whether a feature lands while the balance is still alive. Stake size and volatility together set the odds of a session lasting the hour you had in mind.

A worked stake example anchors the model. Take the £50 bankroll at 20p a spin and a middle-band edge of 4 percent: a hundred spins costs 80p in expectation, so a session of two hundred spins prices at £1.60. Cut the stake to 10p and the same two hundred spins costs 80p; double it to 40p and the session costs £3.20 for the same number of reels. Session length in spins is a choice of stake as much as of game, and the expected cost moves in lockstep with it.

Offer value bends the same way. A wagering requirement clears at the pace of your turnover, and the risk to a bonus balance is busting before the requirement completes. Low-volatility play erodes a bonus smoothly and tends to survive long requirements more often. High-volatility play turns every remaining requirement into a coin-flip with a tail: the balance either spikes past the cap or dies early. That is why the same 35 times requirement feels different on different games, and why the game an offer names matters as much as the multiplier it prints.

Volatility labels on lobbies are a guide, not a measurement: they summarise a game’s payout distribution in three words and compress a great deal. Where a label is absent, the paytable is the better witness. A maximum win in the thousands of stakes, a feature round that carries multipliers, and long gaps between top payouts point one way; frequent small wins across many lines point the other.

The rows below set the British baseline against the offshore end of the market, using category-typical ranges for welcome offers, wagering, minimum deposit, payout windows and payment rails. Nine UKGC-licensed rows and three offshore rows make the comparison concrete; the note underneath says plainly what the figures are and are not.

Figures are category-typical ranges compiled from operator-published terms at the time of writing, not quotes of any single promotion. UKGC-licensed rows are the British baseline shown for comparison; none of them are claimable at the offshore offers modelled on this page. Verify every term on the operator’s own page before you take it.

Read the table by offer shape rather than by size. Spins bundles price a known amount of play on a named game, which makes the arithmetic on this page direct. Match bonuses price flexibility but hand the wagering arithmetic to you. The no-wagering rows trade headline size for the right to keep whatever lands as cash, which the model values precisely: face value minus the game’s edge, with nothing waiting behind a requirement.

The arithmetic of a spins offer

A spins offer looks simple and prices cleanly once you split it into three parts: face value, expected return and clearing cost. Take a worked example with the assumptions in the open: 100 spins at 10p each, on a game publishing 96 percent RTP, winnings credited as bonus funds with a 35 times wagering requirement on winnings, and a £25 cap on what converts to cash.

Face value first. A hundred spins at 10p is £10 of play at the offer’s stake, whether or not you would have chosen that game or that stake. Expected return second: at 96 percent, £10 of turnover returns £9.60 on the model, so the spins’ expected cash-equivalent sits at £9.60 before any requirement touches it, spread across outcomes that range from nothing to several multiples of that.

Clearing cost third, and this is where offers are won or lost on paper. If the £9.60 of winnings lands as bonus funds at 35 times, the requirement is about £336 of wagering. At the same 4 percent edge, clearing it carries £13.44 of expected loss. The balance starts at £9.60, so the linear version of the model puts the offer’s expected value at zero: most balances die inside the requirement before converting anything. The cap only trims the rare tail that clears above £25.

Loosen the requirement and the picture moves fast. At 20 times winnings, the requirement falls to about £192, clearing costs about £7.68 in expected loss, and the modelled leftover after clearing is roughly £1.90. At no wagering at all, the offer is worth its £9.60 expected return and the arithmetic stops there. The difference between those outcomes is entirely in the terms, which is why the wagering line deserves more attention than the spin count.

Caps deserve their own line in any pricing. A conversion cap trims the right tail: every outcome above the cap pays the cap, so the model’s average drops by whatever those tail outcomes contributed. On small offers the effect is modest, because few conversions reach the cap at all; on larger offers, or where a cap sits close to the expected winnings, the cap becomes one of the largest costs in the terms. Read the cap next to the wagering multiple, never instead of it.

Set the same arithmetic against a match bonus to see the shape difference. A £10 match at 35 times the bonus is £350 of wagering, close to the spins example’s £336, but the balance starts at the full £10 rather than at winnings, and the game choice is usually wider. The spins offer prices its game for you; the match bonus prices your discipline instead. Neither is free money on the model; both are discounts on expected cost, and the discount’s size lives in the wagering line.

Two refinements keep the model honest. First, the expected return assumes the spins run at the game’s published RTP; where an offer names a specific game, check that game’s info page rather than assuming the middle band. Second, variance means the typical outcome sits below the mean: a minority of sessions carry the big feature hits, so most spin bundles return less than £9.60 and a few return far more. The model prices the average, and the terms decide how much of it survives.

Bonus buy and what it does to the wagering

A bonus buy, where a game offers one, is a button that pays a multiple of the base stake to enter the feature round directly instead of waiting for it to trigger. Modelled across the market, buys typically price between 50 and 100 times the base stake, so on a 20p game a buy costs £10 to £20 and replaces what might otherwise be a couple of hundred spins of waiting.

The return figure does not change: where a game publishes an RTP, that figure already accounts for feature buys in its model. What changes is concentration. The same expected return compresses into fewer, larger swings, so the session’s variance rises sharply while the edge per pound stays put. Expected loss per hour is the number that moves: ten buys at £10 is £100 of turnover and £4 of expected cost at a 4 percent edge, staked inside perhaps fifteen minutes; the base game at 20p a spin needs the better part of an hour to push the same money through.

Bonus terms treat buys unevenly, which is where the wagering picture shifts. Some offers count feature-buy stakes fully toward a wagering requirement, which clears requirements quickly and at the full expected cost per pound. Some exclude buys or count them at a reduced weight, which makes them useless for clearing. Some bar buys while bonus funds are active at all, with terms stating what happens to winnings if you buy anyway. The offer’s terms decide which of those worlds you are in, so read that line before the feature tempts.

The terms around buys tend to sit in three clauses worth finding: the contribution table that says what counts toward wagering and at what weight, the clause that governs play with bonus funds active, and the clause that lists restricted features or games. Where a buy counts at full weight, it is the fastest way to move a requirement and the fastest way to spend the balance behind it. Where buys are barred with bonus funds, the button is usually disabled rather than dangerous. Either way the clause, not the button, is the governing document.

As a pricing rule, treat a buy as a ticket: same edge, higher variance, faster tempo. If the aim is a long, cheap session, buys work against it, because they raise turnover per hour and shorten the runway a bankroll provides. If the aim is a shot at a large outcome inside a short window, a buy prices that shot explicitly and removes the waiting. Neither aim is wrong; they just cost differently, and the calculator below takes the stake you actually enter.

Reading a paytable before you stake

The paytable is the game’s contract in miniature, and it repays five minutes before the first spin. It shows what each symbol pays at your current stake, how many paylines or ways the game runs, the rules of every feature, the maximum win, the minimum and maximum stake, and usually the RTP itself. Everything the arithmetic on this page needs is on that single screen.

The mechanics line deserves a beat of its own. Fixed-line games pay on defined paths across the reels, ways games pay any left-to-right adjacency and count hundreds of ways, and cluster games pay groups of symbols rather than lines. None of that changes the published RTP at a given stake. What it changes is hit frequency, which is the texture of variance: ways and cluster games land something often and small, top-heavy line games land rarely and large.

Feature rules tell you where the return actually lives. Multipliers, retriggers, sticky symbols and progressive accumulators concentrate the RTP inside the feature round, which is why base-game stretches can feel thin on feature-led games. The maximum win tells you whether the tail exists at all and how far it runs: 500 times stake and 15,000 times stake describe very different games even at the same RTP and the same volatility label.

Stake interacts with mechanics in a way worth noticing: fixed-line games multiply a per-line price across every line, so a 10p line bet across twenty lines stakes £2 a spin, while ways games usually stake the full amount per spin as one figure. Reading which system a game uses prevents the most common pricing mistake, which is comparing a per-line stake on one game with a total stake on another. The calculator takes the total per-spin figure, so read it off the game’s stake box, not off a line-price habit.

Two practical checks close the loop. Confirm the per-spin value a spins offer will actually credit, because offers frequently fix spins at a low stake and the arithmetic changes with it. And confirm the stake range before planning a session, since the minimum and maximum printed there are the range you can actually play, whatever the lobby’s filters suggest. Both figures feed the calculator later on this page without adjustment.

Finding the paytable takes one tap in most lobbies: an information or pays icon on the game’s loading screen opens it. Read it at the stake you plan to play, because several values scale with stake and the maximum win may be quoted at the stake currently set. A paytable read at the wrong stake is the usual source of confusion about what a game paid on a past spin.

Stake limits and the white paper

The government’s gambling white paper, High Stakes: Gambling Reform for the Digital Age, was published in April 2023, and among its proposals were stake limits for online slots products. That proposal is on the record; the size and timetable of any resulting limit are not things the sources on this site carry, so this page prints no current figure and builds no arithmetic on one.

The honest position for a reader is that stake ranges are visible without any regulation at all. Every game prints its own minimum and maximum stake in the stake selector or on its information page, and that printed range is the one that applies when the reels turn. If stake limits tighten in future, the printed ranges will move with them, and the model on this page needs nothing more than the new numbers typed into the calculator.

What the white paper’s direction of travel means for the arithmetic here is simple: expected cost scales linearly with stake. A 20p spin at a 4 percent edge costs 0.8p in expectation; a £2 spin on the same game costs 8p; the game’s label does not change, the multiplier does. Anyone comparing games by session cost should hold stake constant first and volatility second, because stake is the lever that moves the expected cost fastest.

Spins offers complicate the picture in one specific way: the spins arrive at a fixed per-spin value chosen by the offer, not by you. A bundle of 100 spins at 10p stakes the same £10 whatever stake range you would normally play, so the offer’s game and its printed RTP matter more than your usual habits. Price the bundle at its own stake first, then decide whether the game justifies a top-up session at your preferred stake.

Read stake ranges alongside volatility when you plan: a high-volatility game at a low stake and a low-volatility game at a high stake can drain a bankroll at similar speeds, with very different textures. The model treats stake as the input it is and leaves the texture question to the paytable section above.

Where no UK rule set applies, the ranges are whatever the operator prints. That is not a comment on generosity either way; it is a comment on who sets the dial and how visible the setting is. The licence section below takes up that difference in full, because a stake range you can verify is worth more than a larger one you cannot.

What a UK licence obliges, and what its absence leaves open

Providing remote gambling facilities to people in Great Britain requires an operating licence from the Gambling Commission, and the Commission keeps a public register of licence holders that anyone can search by operator name. That register is the first thing a careful reader checks, because every obligation that follows attaches to the licence, not to the marketing.

A UK licence obliges an operator to follow the Licence Conditions and Codes of Practice, including social responsibility requirements on customer interaction and product design. It obliges age and identity verification as a condition of licensing rather than an operator’s choice. It obliges customer funds to be protected at a stated level, not protected, medium or high, with the level declared. It obliges the operator’s own self-exclusion facility on top of the national scheme, and affiliation with an alternative dispute resolution provider so a complaint has a free escalation route. Credit cards are banned for gambling with GB-licensed operators outright.

Enforcement is the part that makes those obligations more than ink. The Commission can fine, suspend or revoke licences, and it publishes the regulatory actions it takes, which means an operator’s record is a public paper trail rather than a claim. For a reader comparing where a session should run, that trail is the practical difference between a promise and a rule.

Checking the register takes a minute and settles most questions. Search the operator’s exact trading name, confirm an active operating licence, and read the conditions the licence carries. Where a site claims an offshore licence instead, the equivalent check runs on the issuing authority’s own register, and a claim that cannot be checked anywhere should be treated as decoration. Neither check tells you how a session will go; both tell you who is answerable if something goes wrong.

Offshore licences work differently, and describing that is not an accusation. The jurisdictions commonly serving UK-facing brands include Curaçao and Anjouan, and Curaçao has been reforming its framework through the National Ordinance on Games of Chance, phasing out its master-licence model. Licence claims from either jurisdiction can be checked on the issuing authority’s register, and this site makes no licence claim about any offshore operator whose register line is not on file. What an offshore licence does not bring is the British bundle: no entry in the Commission’s register, no GamStop coverage, no UK dispute route and none of the declared fund-protection levels. Those gaps do not tell you an offshore site behaves badly. They tell you the routes back are thinner, and that every promise has to be verified rather than assumed.

None of this prices the games themselves, and that is deliberate. The return figure is printed by the game, the protections are set by the licence, and the two travel separately: a well-protected account can host an expensive game, and a cheap game can sit on an unprotected one. Reading both, on the register and on the information page, is the two-minute version of due diligence this page keeps recommending.

Model a session before you sit down

The calculator below runs the session model this page has been circling: it takes a bankroll, a stake per spin, a house edge and a number of planned sessions, and it prices the expected cost of the plan. House edge is 100 minus RTP, so a game publishing 96 percent carries a 4 percent edge, and that single number does most of the work.

The pre-filled example models a £50 bankroll at 20p a spin with a 4 percent edge across four sessions. At that edge, expected cost is eight-tenths of a penny per spin: a hundred-spin session costs 80p in expectation, and four of them cost £3.20. Against £50 of bankroll, that expected cost is small, and the variance around it is not: the same four sessions can plausibly finish anywhere from nothing left to meaningfully ahead, which is the whole point of pricing the edge rather than predicting the outcome.

Use the inputs to compare rather than to predict. Raise the edge input to 6 percent and the same plan reprices at the low band. Halve the stake and the expected cost halves with it. Double the sessions and the cost doubles while the per-session feeling of cheapness stays exactly the same, which is the trap the model exists to expose. The sessions input matters because eighty pence a session sounds negligible, and four sessions a week for a year is a number worth seeing before it arrives.

Treat the output as a floor, not a forecast. Expected cost is the average the model converges toward over very long runs; individual sessions scatter around it, and the scatter is wider than the average at every realistic bankroll. The value of the floor is that it turns a vague sense of cost into a number you can accept or refuse before the first spin.