What does RTP mean? Return to player explained with real numbers

RTP means return to player: the share of all money bet on a game that the game is designed to pay back over the long run. A game with a 96% RTP returns $96 for every $100 wagered on average, and keeps $4. That missing part is the house edge. This guide explains what the number measures, what it does not, how regulators set and display it, and how you can check it yourself on a provably fair game.

What RTP means in plain numbers

RTP is a long-run average, worked out from the game's rules and pay table. It is always quoted against the total amount wagered, not against your deposit.

The second column is simply 100% minus the RTP. The third is the amount wagered times the edge. Notice that $5,000 can be wagered from a much smaller deposit, because winnings are bet again. A player who deposits $100 and plays $10 spins until the money is gone has usually wagered far more than $100 by then.

The UK Gambling Commission's remote technical standards describe the figure the same way. RTS 3C lists the information a player must be able to find before they gamble, and one option is "the return to player (RTP) percentage". The guidance adds that it should be "the average theoretical return to player percentage." Theoretical is the key word: it is a property of the game, not a promise about a session.

Where the number comes from

For a table game or a simple crypto original, RTP comes straight from the payouts.

• European roulette: 37 pockets, a single number pays 35 to 1, so each unit bet returns 36/37 on average, or 97.30%. theScore Bet's rules page for Evolution's Euro Roulette lists the same 97.30%. Our roulette payouts and odds guide does the sum for every bet.
• Dice at 99%: a roll over 50.50 wins about 49.5% of the time and pays 2x. 2 x 49.5% = 99%.
• Slots: the RTP comes from the reel strips, symbol weights and pay table, worked out by the studio and checked by a test lab. You cannot calculate it from the screen, which is why labs matter. See are online slots rigged for how labs certify it.

Where a game involves choices, like blackjack, the published RTP assumes the best play. RTS 3C says the figure should be calculated "using either the auto-play strategy or a standard/published strategy". Poor decisions lower your real return below the quoted number.

RTP is not what you get back in one session

The quoted RTP describes millions of bets. Over a few hundred, almost anything can happen. We computed the exact spread for a 2x dice bet at 99% RTP:

"Typical swing" is one standard deviation of the realised return. After 100 bets a result anywhere from about 89% to 109% is ordinary. As the number of bets grows, the swing shrinks and your results drift towards 99%, which is why the chance of being ahead falls the longer you play.

Two games with the same RTP can feel completely different. A 99x target multiplier on a 99% limbo game wins only 1% of the time. In 100 bets there is a 36.6% chance of not hitting once, while 2x dice almost never goes 100 bets without a win. That difference is volatility: how the same average return is spread out. RTP tells you the average cost, volatility tells you the shape of the ride.

What regulators say about RTP

There is no single worldwide minimum. A few examples from regulators' own texts:

These are floors. Most online slots sit well above them, but a legal floor of 75% or 85% shows how wide the range can be.

Real results back this up. The Nevada Gaming Control Board figures compiled by UNLV for calendar 2024 show slot machines statewide won 7.20% of the amount played, which means players got back roughly 92.8% on average. Penny slots won 9.36%, so they returned about 90.6%. Those are land-based machines, but the same logic applies online: the RTP is a design choice within the rules, not a fixed law of nature.

RTP versions: the same game, different numbers

Many slot studios release one game in several RTP versions and let the casino choose. The game looks identical; only the long-run return changes. A slot offered at 96% at one casino may run at 94% or lower at another. The only reliable check is the game's own info or help screen, opened at the casino where you play, which normally states the RTP of that copy. Our guide on are online slots rigged shows how to find it and how labs certify each version. Table games do the same through pay tables: our guides to Three Card Poker odds and Let It Ride odds show one side bet under the same name returning anywhere from about 87% to 98%.

RTP on crypto originals

Crypto casinos that make their own games often publish RTP per game. Some examples from their own pages:

Gamdom's page also adds a fair warning: "RTP is theoretical; it's based on thousands of bets and may vary in short sessions."

The difference between 97% and 99% looks small but triples the cost: $1,000 wagered costs $30 on average at 97% and $10 at 99%. We checked the edges of 156 originals ourselves in our house edge audit.

How to check RTP on a provably fair game

On a provably fair game you can check that each result came from seeds fixed before you bet, and you can count your own results against the stated RTP. Here is a worked example using made-up seeds and the Stake-style dice scheme in our verifier.

1. SHA-256 of the server seed matches the hash, so the seed was fixed before the first roll.
2. Nonce 0 rolls 41.45, a loss. Nonces 1 to 4 roll 20.53, 83.30, 27.39 and 49.08.
3. Over all 1,000 nonces, 513 rolls land above 50.50. 513 wins at 2x return 1,026 units for 1,000 staked.
4. Realised return: 102.6%, against a theoretical 99%.

That is a winning run, and well inside the normal swing in the table above. You can open nonce 0 in our verifier and step through the others. How to verify a provably fair bet explains the steps with your own seeds.

What this proves, and what it does not: verification shows results were not changed after you bet. The RTP itself comes from the payout rules, so check the multiplier against the win chance. If a dice game pays 1.96x at 49.5%, its RTP is 97%, however fair each roll is.

How to use RTP when choosing a game

• Compare like with like. RTP tells you the average cost per unit wagered. Pick the higher number between two games you enjoy equally.
• Count the turnover, not the deposit. Expected cost = amount wagered x house edge. Fast games burn through more wagering per hour.
• Expect swings. A session result says little about the RTP. Hundreds of bets can land 10 points either side.
• Check the copy you play. For slots, read the info screen at your casino. For originals, compare the payout to the win chance.
• Set a limit first. A deposit limit controls your real cost better than any RTP figure. Our responsible gambling page lists the tools, and risk of ruin in gambling shows what the edge does to a bankroll over time.

Sources

• UK Gambling Commission: RTS 3, rules, game descriptions and the likelihood of winning
• Malta Gaming Authority: RTP percentage streamlined, 28 May 2021
• Nevada Gaming Commission: Regulation 14 (section 14.040)
• Nevada Gaming Win 2024, from the Nevada Gaming Control Board revenue report (UNLV Center for Gaming Research copy)
• theScore Bet: Euro Roulette rules (Evolution)
• Gamdom: Gamdom Originals
• MetaWin: Originals hub
• NIST FIPS 180-4, Secure Hash Standard
• Session spreads, limbo odds and the 1,000-roll example computed by us (Python and our verifier's pf.js)