The number printed on almost every online slot (96%, 97%, 94%) is one of the most misunderstood figures in gambling. Players assume it means something like "for every $100 you put in, you get $96 back." Casinos let that assumption stand because correcting it in plain terms would require explaining something more complicated. This guide corrects it.
RTP, return to player, is a theoretical long-run average. It describes what a game returns across an enormous number of spins under controlled simulation conditions. It says almost nothing about what you will experience in a session, a day, or a month of play. But it is still the most important number attached to any slot game, once you understand what it measures and what it does not.
This article is for players who want to understand the math behind the number, not just the label. You will find out how RTP is calculated, why the same RTP can describe games with wildly different actual experiences, how variance interacts with RTP to determine your realistic outcomes, how to verify whether the RTP you are quoted is real, and what game features do to the headline figure.
Key takeaways
- RTP is a theoretical percentage calculated over billions of simulated spins, not a session guarantee.
- A 96% RTP means the game returns $96 in prizes for every $100 wagered in aggregate, across all players, over millions of rounds. In any individual session the actual return can be anywhere from 0% to several thousand percent.
- The house edge is 100 minus RTP. A 96% RTP game has a 4% house edge per spin.
- Two games with identical RTPs can deliver completely different experiences because of variance (also called volatility). High-variance games pay less often but pay more when they do. Low-variance games pay more often but pay less.
- RTP is certified for specific game versions. A game titled "Starburst" can have different RTPs at different casinos if the casino selected a different variant from the provider.
- Bonus buys, hold-and-spin features and free spin rounds can have different RTPs from the base game. The headline figure is usually for the full game including all features.
- You can verify slot RTPs through provider websites, testing lab certificates and regulatory payout reports.
- Wagering requirements on bonuses effectively reduce your realised RTP below the game's stated figure.
What RTP actually measures
RTP stands for return to player. It is expressed as a percentage and represents the proportion of all money wagered on a game that is returned to players as winnings, calculated over an extremely large number of rounds.
The key word is "all." RTP is not calculated for you. It is calculated across every spin by every player on that game version, summed up over time. When a provider quotes 96% RTP for a slot, they mean: if you add up every pound staked on this game and every pound paid out in prizes across millions of rounds, the ratio of prizes to stakes will approach 96%.
It is a statistical property of the game, not a promise about your experience.
The inverse is the house edge: the proportion of all money wagered that the casino retains. A 96% RTP game has a 4% house edge. A 94% RTP game has a 6% house edge. This is the figure that matters most for comparing games, because it tells you how much you are paying per spin for the entertainment.
Expected loss per spin = stake × (1 − RTP)
So on a 96% RTP game at $1 per spin, your expected loss per spin is $0.04. Over 100 spins that is $4 in expected losses. Over 1,000 spins it is $40. The word "expected" is doing real work here: over small numbers of spins, actual outcomes vary enormously. The expected loss only manifests reliably across a very large number of spins.
How RTP is calculated: the math
RTP is calculated by summing the expected value of every possible outcome weighted by its probability, then expressing this as a fraction of the stake.
The formal expression is:
RTP = Σ (probability of outcome × prize for that outcome) / stake
For a simplified slot with 1,000 possible outcomes at a $1 stake:
| Outcome | Prize ($) | Probability | Expected value ($) |
|---|---|---|---|
| No win | 0 | 0.700 | 0.000 |
| 2× return | 2 | 0.200 | 0.400 |
| 5× return | 5 | 0.070 | 0.350 |
| 20× return | 20 | 0.025 | 0.500 |
| 100× return | 100 | 0.004 | 0.400 |
| 1000× return | 1,000 | 0.001 | 1.000 |
| Total | 1.000 | 2.650 |
RTP = $2.65 / $1.00 = 265%? No. Because the probabilities here are illustrative. In a real game the probabilities are set so that the sum of (probability × prize) equals the target RTP times the stake. The provider's math department builds the paytable to hit the target exactly.
In practice the calculation is run the other way: the provider decides the RTP target (say, 96%), sets the paytable and symbol frequencies, and verifies through simulation that the sum of expected values across all outcomes, divided by the stake, equals 0.96.
For a modern video slot with five reels, multiple paylines, expanding wilds, free spin multipliers and a bonus game, this calculation involves millions of possible combinations. Providers use purpose-built math modelling software. The result is submitted to a certification lab along with the full math model, and the lab runs its own simulation to verify the figure is correct.
Progressive jackpot RTP: the funded vs unfunded split
Progressive jackpots make this calculation more complicated. The jackpot prize grows with every spin because a portion of each stake feeds the prize pool. This means the contribution to RTP from the jackpot changes continuously as the pot grows.
Most providers publish two RTP figures for progressive games:
- Base game RTP (unfunded): the return from fixed prizes only, excluding the jackpot contribution. Typically 88 to 93% for high-value jackpot games.
- Full game RTP (funded at reset): includes the expected value of the jackpot contribution, calculated at the jackpot's seed (starting) value. This is the headline figure.
If a game shows 96% RTP and has a jackpot, the true player-facing RTP is the base game figure, because you almost certainly will not win the jackpot. The funded RTP only approaches the stated figure when averaged across the rare jackpot winner and everyone else. This is disclosed in the math model but often not prominently on the game page.
Why your session return will not match the RTP
There are two reasons your session return will not be 96% even if you play a 96% RTP game, and they are independent of each other.
Reason 1: Sample size. RTP is a long-run average that stabilises over millions of spins. In a session of 200 spins, the outcomes you actually hit are a tiny sample. The variance around the expected value is enormous at small sample sizes. If you flip a fair coin 10 times, you might get 7 heads and 3 tails. That does not mean the coin is biased. You just need more flips for the 50% to emerge. Slots are the same, except the prize distribution is far more skewed.
The number of spins needed for your actual return to be within a few percentage points of the theoretical RTP depends on the game's variance, but for most slots it is in the tens of millions of spins. A human player spinning 5 times per minute for 3 hours a day, 365 days a year, would spin about 328,000 times in a year. That is still nowhere near the sample where RTP stabilises.
Reason 2: The shape of the prize distribution. Even in a long session, a large fraction of your session return can come from a single win. If you have a winning session, it is often because one spin paid 200× your stake. That single event accounts for a huge proportion of your total return for the day. The "average" return of 96% comes from summing millions of these unequal outcomes. Any individual session is just one cluster of those outcomes.
To make this concrete: imagine two games, both with 96% RTP.
Game A: 30% of spins return 2×, 70% return 0. EV = 0.30 × 2 = 0.60. This is actually 60% RTP, which illustrates why you need more outcomes. The point holds at 96%: the distribution of prizes matters as much as the average.
Game B (closer to a real high-variance slot): 98.5% of spins return 0, 1.4% return 10×, 0.099% return 500×, 0.001% return 10,000×. The expected value across all these outcomes sums to 0.96. In 100 spins you will most likely return nothing, occasionally return 10× a few times, and very rarely hit a large win.
Both games return 96% over millions of spins. In 100 spins, Game A will return close to 60% (roughly stable), while Game B will return either much less or much more than 96% in almost any session, because its prizes are far from the mean.
Variance and volatility: the missing half of the story
Variance (used by mathematicians) and volatility (used by the industry) describe the same thing: how far individual outcomes deviate from the expected value. High variance means large deviations, both positive and negative. Low variance means outcomes cluster tightly around the average.
Understanding variance is more useful than knowing the RTP for predicting your actual session experience.
Low variance (low volatility) slots:
- Win frequency is high, often 35 to 45% of spins land some prize.
- Prize sizes are small relative to stake, typically 2× to 20× common.
- Your bankroll drains slowly and gradually.
- You rarely have huge wins, but you also rarely have long losing streaks.
- Suited to players with smaller bankrolls, longer sessions, or lower risk tolerance.
High variance (high volatility) slots:
- Win frequency is low, often 15 to 25% of spins.
- Prize sizes are large when they come, often 100× to 5,000× or more for top prizes.
- Your bankroll can take severe damage before a large win appears.
- Sessions are polarised: many total-loss sessions, occasional big wins.
- Suited to players with large bankrolls relative to stake, or players who are comfortable with infrequent but large prizes.
Medium variance:
- Win frequency around 25 to 35%, mid-range prizes.
- The most common category by game count.
Neither is more or less "fair" in any meaningful sense. Both deliver the same expected return over enough spins. Variance determines the shape of the distribution, not the area under the curve.
Why providers do not publish variance numbers
Most providers classify volatility as low, medium or high without giving a numeric measure. The reason is partly commercial (a volatility index would allow direct comparison between games) and partly that the standard academic measures require the full probability distribution, which providers treat as proprietary.
The measures that do exist include:
- Standard deviation of session return. Not published by providers.
- Win frequency. Sometimes published, usually in the help/paytable screen.
- Max win. Published in almost every game. Higher max wins strongly correlate with higher variance.
- Hit rate. Sometimes quoted in reviews. The proportion of spins that return any prize.
If a game claims a 20,000× max win, it is almost certainly high variance. If the max win is 500×, it is more likely low to medium variance. The relationship is not linear and there are exceptions, but max win is a reasonable proxy when no other information is available.
How RTP and variance interact in practice
The same RTP can describe games with completely different player experiences. The table below illustrates four combinations across a 1,000-spin session at $1/spin ($1,000 total stake, 96% RTP = $960 expected return):
| Scenario | Variance | Probable session range | What actually happens |
|---|---|---|---|
| A | Low | $850–$1,070 | Steady drip of small wins, predictable decline |
| B | Medium | $600–$1,400 | Moderate swings, occasional good wins |
| C | High | $0–$3,500 | Long dry spells, few large wins |
| D | Very high | $0–$8,000+ | Most sessions end near $0, rare sessions end very large |
All four sessions have the same expected return. What differs is the width of the distribution around that expectation. In scenario D, the $960 expected return means very little for any single player: either they lose everything or they win something large. The "average" is only real when you add up hundreds of players.
This is why bankroll management advice for high-variance slots is not just generic caution. It is math. If you are playing a 20,000× max win slot with a $200 bankroll at $2 per spin, you have 100 spins. In 100 spins on a high-variance game, your probability of being up is lower than your probability of being bust before a large win has time to appear. A smaller stake over more spins gives variance more chances to work in your favour.
RTP by game type: slots vs table games vs live
RTP is used across all casino game types but the figure behaves differently because of how the games are structured.
| Game type | Typical RTP range | Notes |
|---|---|---|
| Video slots | 92–98% | Wide range; always check specific game |
| Classic slots (3-reel) | 92–96% | Generally lower than video slots |
| Megaways slots | 94–97% | Usually disclosed; feature-heavy math |
| Progressive jackpot slots | 85–94% (base game) | Headline includes funded jackpot contribution |
| European roulette | 97.3% | Fixed, one mathematical outcome |
| American roulette | 94.7% | Fixed; double zero adds 2.6% house edge |
| Blackjack (basic strategy) | 99.4–99.6% | Depends on specific rule set |
| Baccarat (banker bet) | 98.9% | After commission |
| Video poker (Jacks or Better, 9/6) | 99.5% | Depends on pay table |
| Live roulette | Same as RNG variant | Physical wheel, same math |
| Live blackjack | Same as RNG variant | Depends on rules at that table |
The table illustrates something players often overlook: table games, played correctly, return significantly more than slots. European roulette at 97.3% beats most slots. Blackjack with basic strategy at 99.4% is near the top of any game category. The tradeoff is that table games require decisions that affect the outcome (blackjack), or have slower pace (roulette).
Slots offer something different in return for their lower RTP: entertainment value, theme engagement, the possibility of a 1,000× win on a single spin. That is a product difference, not a value difference in objective terms.
Feature RTP: what bonus rounds do to the headline figure
Modern video slots divide their RTP between the base game and various features. The proportions matter.
A typical structure for a mid-variance slot:
| Component | RTP contribution |
|---|---|
| Base game spins | 55–65% |
| Free spins bonus round | 20–25% |
| Hold-and-spin feature | 8–12% |
| Other bonus games | 2–5% |
| Total | ~96% |
The implication: a large share of a slot's return is locked inside features that trigger relatively rarely. If the free spins round contributes 22% of the total 96% RTP, and it triggers on average once every 150 spins, most of your actual money returned in a session comes from base game wins. But most of the game's theoretical generosity lives in the feature you are waiting for.
Bonus buy (feature buy)
Many slots offer a "bonus buy" option that lets you pay a multiple of your stake (commonly 50× to 100×) to guarantee immediate access to the free spins round. The bonus buy RTP is calculated separately and is usually higher than the base game RTP, sometimes as high as 97 to 98%, because you are buying access to the most valuable part of the game. However, the higher stake and immediate risk of a large outlay change the effective variance dramatically.
In jurisdictions where bonus buys are permitted (they are banned in Great Britain), the RTP for the bonus buy is disclosed separately in the game's help screen.
Free spins promotions from the casino
Casino-issued free spins, common as part of welcome offers, have a stated value (usually $0.10 per spin) and feed the RTP of the specific game they are assigned to. They are not free in any mathematical sense: you spin the game at a fixed stake with the casino's money, any wins are subject to wagering requirements, and your expected value from the free spins is (number of spins × stake × RTP) × (1 / wagering requirement). More on this in the bonuses section.
Provider-selected RTP: why the same game has different percentages at different casinos
This is one of the least-publicised aspects of slot RTP and one of the most important for players to understand.
Many slot providers offer multiple RTP variants of the same game. The operator (casino) selects which variant to deploy. A game titled "Reactoonz" might be available in variants at 96.51%, 95.51% and 94.51%. The game looks identical in all three versions. The gameplay is identical. The RTP is different.
Which variant is live at any given casino is disclosed in the game's information panel (usually a small "i" button or help icon within the game), but not always prominently. The RTP figure on a review site may reflect the highest available variant, not the one the casino you are playing at has deployed.
Why do casinos choose lower RTP variants?
Because a lower RTP increases the house edge and casino profitability per spin without changing anything visible to the player. The difference between a 96.51% and a 94.51% variant doubles the house edge from 3.49% to 5.49%. Over a player population spinning millions of times, this is very significant revenue.
How to find the actual deployed RTP
The reliable method is to open the game at the specific casino you are using and check the in-game help or information panel. Reputable casinos publish the deployed RTP there. Some jurisdictions require this disclosure: the UK Gambling Commission mandates that the RTP for the specific deployed variant is visible to the player in-session.
If the in-game panel is absent or vague, ask the casino's support team to confirm the specific variant deployed. If they cannot answer, that itself is information worth acting on.
How RTP is tested and certified
RTP is not self-reported without verification in licensed jurisdictions. The testing process requires the game provider to submit the full math model, including all symbol frequencies, paytable values, feature trigger probabilities and bonus game logic, to an independent testing laboratory.
The lab does two things:
Mathematical verification. The math model is reviewed to confirm that the stated RTP follows correctly from the paytable and probability weights. If the math says 96%, the algebraic sum of (probability × prize) across all outcomes must equal 0.96 per unit staked.
Simulation. The lab runs the actual game software through a simulation of at minimum tens of millions of spins, more commonly hundreds of millions for complex games. The simulated return is compared to the stated RTP. Small differences within defined tolerance ranges are acceptable (typically ±0.1 to ±0.5%) and are expected at finite sample sizes. Large deviations would indicate a bug or a misleading stated figure and would fail certification.
The tolerance range matters. If a game states 96% RTP and the simulation returns 95.6% across 100 million spins, that may be within tolerance for certification. The game's long-run RTP is still 96% in the math model. The simulation deviation is a statistical artifact of finite sample size, not evidence of inaccuracy.
Certification must be renewed whenever the game software changes, and many regulators require it when the deployed RTP variant changes. The UKGC and MGA are explicit on this point.
How to verify a slot's actual RTP
You have several options, roughly in order of reliability:
1. In-game help or information panel. Open the game at the specific casino and look for an "i" button, "?" icon, "Game Rules" or "Help" link. Most regulated games display the deployed RTP there. This is the most reliable source because it shows the variant currently running at that operator.
2. Provider's website. Most major providers (NetEnt, Pragmatic Play, Play'n GO, Microgaming, Relax Gaming, Push Gaming, Hacksaw Gaming) publish a game library with stated RTPs. This usually shows the highest available variant, not necessarily the one your casino deployed, but it confirms the maximum possible RTP for that game title.
3. Testing lab certificate databases. eCOGRA, iTech Labs and GLI maintain searchable databases of certified games. You can look up a specific game by name and provider to find the certified RTP. Some databases include variant-specific data.
4. eCOGRA payout reports. eCOGRA publishes monthly payout percentage reports for certified casinos. These show actual payout percentages (not theoretical RTPs) across all games played at that casino in that month, broken out by game category. These are real aggregate numbers from real play and give you a sense of what players actually received, though they are at casino level, not individual game level.
5. Regulator-published data. Some national regulators publish RTP data. The UK Gambling Commission publishes industry statistics. The Malta Gaming Authority publishes selected operator data. These are high-level averages but confirm that real payout rates are monitored.
What to do if stated RTPs are vague or missing
If a casino cannot or will not tell you the deployed RTP for a specific game, or if the in-game help panel is missing, treat it as a signal that the casino is either poorly operated or operating without adequate licensing. Reputable casinos using reputable providers in regulated jurisdictions will always be able to show you the in-game RTP.
How bonuses distort your effective RTP
This section gets into territory that most casino marketing material avoids, because it makes the math look less appealing.
When a casino offers a bonus with wagering requirements, the effective RTP of playing through that bonus is not the game's stated RTP. It is significantly lower.
The formula:
Effective RTP from bonus play = game RTP^(wagering requirement rounds)
That is not quite right because it compounds differently, but the intuition is correct: every round of wagering reduces the expected value of the bonus. A more accurate calculation:
Expected value of bonus after wagering = bonus amount × game RTP^wagering requirement multiplier
This overcounts because you keep your stake between rounds. The right formula is:
Expected value retained = bonus amount × (1 − (1 − RTP)^wagering requirement)...
This gets complex fast. The simpler and more useful framing is:
Expected loss to wager through a bonus = bonus amount × wagering requirement × house edge
Example: $100 bonus, 30× wagering requirement, 96% RTP game (4% house edge):
Expected loss = $100 × 30 × 0.04 = $120
You expect to lose $120 while wagering through a $100 bonus on a 96% RTP game. This does not mean you will always lose money from the bonus. Variance means some players will come out ahead. But the expected value of the wagering requirement is negative, and the bonus amount does not offset it unless the bonus is large relative to the expected loss.
The contribution percentage problem
Wagering requirements are complicated further by contribution rules. Most bonuses specify that different game types contribute different percentages of your wager toward clearing the requirement. Slots commonly contribute 100%. Table games often contribute 10 to 20%. Live casino sometimes 5%.
If you play roulette at 10% contribution rate to clear a 30× wagering requirement, you effectively need 300× of actual wagers. That changes the expected loss calculation dramatically.
The only genuinely positive-expectation bonuses
No-deposit bonuses, low-wagering-requirement bonuses (5× or less) on high-RTP games, and cashback offers on losses are the categories that can carry positive expected value under some conditions. Most welcome bonuses are negative expected value after accounting for wagering requirements, regardless of the headline bonus amount.
Common RTP myths, corrected
Myth: A machine that has not paid out in a while is "due" for a win.
No. Every spin on an RNG slot is an independent event. The machine has no memory of previous spins. The probability of winning on spin 1,001 is identical to the probability on spin 1. This is the gambler's fallacy: the belief that independent random events are influenced by previous outcomes. The slot's RNG does not accumulate a debt.
Myth: Casinos can change the RTP remotely during your session.
No. At licensed casinos using certified software, the game RTP is set at the provider level and embedded in the certified game version. The casino operator cannot change it without a software update and recertification. They cannot flip a switch between your spins.
Myth: Playing at max bet improves your RTP.
Usually false, sometimes true. Most video slots have identical RTP regardless of bet size. The exception: some games have a feature that only activates at maximum bet and that feature carries a higher RTP contribution. If this applies to a specific game it is stated in the rules. Check the help panel rather than assuming.
Myth: Playing at night, weekends or off-peak times changes your odds.
No. The RNG runs at the same parameters regardless of what time it is or how many players are on the game simultaneously. RTP does not change based on server load, time of day or jackpot density.
Myth: The 96% means you will end your session with 96% of what you started.
No. As explained above, RTP is a long-run average across millions of spins and all players. In any individual session the return can be anywhere from 0% to several thousand percent. The number tells you about the game's math, not your session.
Myth: High RTP games are "looser" and more likely to pay during your session.
Partially false. A higher RTP does mean a lower house edge, which is good for the player in expected value terms. But a 97% RTP high-variance game can drain your bankroll faster than a 93% RTP low-variance game, because the high-variance game withholds prizes until rare large events. RTP and variance are separate dimensions. Both matter.
Frequently asked questions
What does RTP mean in slots?
RTP stands for return to player. It is the percentage of all money wagered on a slot game that is paid back to players as prizes, calculated over an extremely large number of spins. A 96% RTP means $96 is returned for every $100 wagered in aggregate across millions of rounds. It is a theoretical mathematical property of the game, not a promise about any individual session.
Is a higher RTP always better?
A higher RTP means a lower house edge, which is better for the player in expected value terms over a very large number of spins. However, a high-RTP high-variance game can lose you money faster in a session than a lower-RTP low-variance game, because variance determines how prizes are distributed. For a short session, the combination of RTP and variance together is more useful than either figure alone.
What is a good RTP for an online slot?
Most players treat 96% as a benchmark. Games above 96% (NetEnt's Blood Suckers at 98%, Mega Joker at up to 99%, White Rabbit Megaways at 97.39%) are considered high RTP. Games below 94% are generally considered low. Progressive jackpot games often run 88 to 93% at the base game level. For reference, European roulette is 97.3% and optimal-strategy blackjack is above 99%.
Can I calculate how much I will lose based on RTP?
You can calculate expected loss, which is stake × number of spins × (1 − RTP). At $1 per spin over 300 spins on a 96% game: expected loss = $1 × 300 × 0.04 = $12. This is the expectation, not a guarantee. Due to variance, your actual result will differ, sometimes significantly. The calculation is useful for budgeting, not predicting.
Why does the same slot have different RTPs at different casinos?
Many providers offer multiple RTP variants of the same game, typically a range spanning 2 to 4 percentage points. Operators choose which variant to deploy. Lower RTP variants mean higher house edges and more revenue for the casino. The game looks and plays identically across variants. Always check the in-game help panel at the specific casino to find the deployed RTP.
Does RTP include the jackpot?
For progressive jackpot games, the headline RTP usually includes the expected contribution from the jackpot, calculated at the jackpot's seed (reset) value. The base game RTP, excluding the jackpot, is lower, sometimes by 5 to 10 percentage points. Since most players will never win the jackpot, the base game RTP is the more practically relevant figure.
How do I find the RTP of a specific slot?
Open the game at the casino you are using and look for an in-game information panel ("i" button, "Help" or "Game Rules"). This shows the deployed RTP for that specific variant. Cross-reference with the provider's website for the highest available RTP for that title. If neither is accessible, contact the casino's support and ask them to confirm the RTP of the specific game version they have deployed.
Do wagering requirements affect my effective RTP?
Yes, significantly. Wagering requirements require you to wager the bonus amount (sometimes bonus plus deposit) a set number of times before withdrawing. Each wagering round loses you a fraction of the remaining value proportional to the house edge. The expected loss from clearing a 30× wagering requirement on a 96% RTP game is approximately 30 × 4% = 120% of the bonus value. This makes most standard bonuses negative expected value despite their face value.
Sources
- Cabot A, Hannum R. Practical Casino Math. 2nd ed. Trace Publications, 2005. Chapters on slot machine probability and expected return.
- UK Gambling Commission. Gambling Industry Statistics: Industry Returns, 2023/24. https://www.gamblingcommission.gov.uk/statistics-and-research
- eCOGRA. Monthly Payout Percentage Reports. https://ecogra.org/payout-reports
- Harrigan KA, Dixon M. A software defect in a video lottery terminal leads to an unfair game: Canadian player protection implications. UNLV Gaming Research and Review Journal, 2009, 13(2):36–55. (Historical case study on disclosed vs actual RTP.)
- Griffiths MD. The use of gaming machines: a review of the psychological literature. In Reith G (ed.), Gambling: Who Wins? Who Loses? Prometheus Books, 2003.
- NetEnt AB. Game Library Technical Specifications: Return to Player disclosure methodology. https://netent.com/en/products
- Gambling Commission (UK). Remote Gambling and Software Technical Standards (RTS), Version 3. https://www.gamblingcommission.gov.uk/licensees-and-businesses/guide/rts-technical-standards
This article is for informational purposes. Casino Stats Lab does not operate gambling services. Figures are for educational illustration only and do not guarantee any outcome. Always gamble within your means at licensed operators.
