basics
What return to player measures, why it applies to turnover rather than your deposit, how it differs from volatility, why no single session resembles the average, and why the same game can carry different RTP at different casinos.
Return to player is the most quoted number in online casinos and the most often misread. It is a precise statement about a long-run average, and it says almost nothing about what will happen to you this evening.
This guide explains what the number measures, what it is calculated against, and what it cannot tell you. You can watch the distributions behave in the RTP simulator.
Return to player is the proportion of total wagered money a game returns to players across a very large number of rounds. A 96% RTP game returns $96 for every $100 wagered, averaged over millions of rounds.
The complement is the house edge. A 96% RTP game has a 4% house edge. Those are two descriptions of the same quantity.
The figure comes from the game's own mathematics. The paytable, the reel weightings and the bonus frequencies are fixed in the game's design, and the RTP is calculated from them. It is not an observed statistic from real play and it is not a target the game adjusts towards. It is a property of the model, the way the expected value of a die roll is a property of the die.
The critical word is average. Not typical, not likely, not usual. The average is a single point in a distribution that can be extremely wide.
This is the mistake that costs people the most money in understanding.
A 96% RTP does not mean you get 96% of your deposit back. It means you lose 4% of everything you bet. Those are different numbers, because bets recycle.
Deposit $100. Play $1 spins. Every spin costs you 4 cents in expectation, but most spins return something, and everything returned is available to bet again.
Play through your $100 once and you have wagered $100 and expect to be left with $96. Play that $96 through and you expect $92.16. And again, and again. Each pass shaves 4% off what remains.
Played until the balance is gone, total expected turnover is your deposit divided by the house edge.
$100 / 0.04 = $2,500
You wagered $2,500 to lose $100. The 96% figure was accurate at every step. You still lost your whole deposit, because RTP was never a promise about your deposit.
The same arithmetic at other prices:
| RTP | House edge | Expected turnover from $100 played to zero | Expected cost per $1,000 wagered |
|---|---|---|---|
| 99% | 1% | $10,000 | $10 |
| 97% | 3% | $3,333 | $30 |
| 96% | 4% | $2,500 | $40 |
| 94% | 6% | $1,667 | $60 |
Read the right-hand column as a price list. If you intend to put $1,000 through a game, the RTP tells you roughly what that session will cost. That is the useful way to hold the number.
Two games can share an RTP and feel nothing alike.
Consider two fictional games, both exactly 96%.
Game A returns 1.92x your stake on 50% of spins and nothing on the rest. Expected return per $1: 0.5 x $1.92 = $0.96.
Game B returns 960x your stake on 1 spin in 1,000 and nothing on the rest. Expected return per $1: 0.001 x $960 = $0.96.
Identical RTP. Completely different experience.
Play 100 spins of Game A at $1 and you will finish somewhere close to $96, usually between roughly $80 and $110. Play 100 spins of Game B and the probability of never hitting the payout is 0.999 raised to the power 100, which is about 90.5%. Nine sessions in ten end with the full $100 gone. The tenth ends with roughly $960.
The average across all ten sessions is still 96%. The median session returns nothing.
Volatility is the width of that distribution. RTP is its centre. High volatility means the long-run average is delivered by rare large outcomes, so the typical session falls well below it. This is why a player can be told a game returns 96% and reasonably feel that the machine has never paid.
It also explains why bonus cashout caps hurt so much on volatile games. Cutting off the top of the distribution removes exactly the outcomes that carry the average.
The law of large numbers is a statement about large numbers. A few hundred spins is not one.
Take a modest example. 1,000 spins at $1 on a 96% game. Expected loss is $40. But the standard deviation of a typical slot over 1,000 spins is large enough that finishing $200 up or $300 down are both unremarkable outcomes. The expected loss of $40 is a number very few individual sessions land near.
There are three consequences worth holding onto.
First, short-run results carry no information about whether a game is behaving correctly. A dry hour on a 97% game and a dry hour on a 94% game look identical.
Second, the average becomes visible only at turnover volumes no individual reaches. The casino sees the average because it aggregates every player. You see one thin sample.
Third, past rounds do not influence future ones. Each spin is drawn independently. A game that has not paid is not owed, and a game that has just paid is not depleted. No betting pattern, staking system or timing rule alters the expected value of the next bet, because the expected value of every bet is fixed by the paytable.
Here is the detail most players never check. For many third-party studio games, RTP is not a single number. It is a setting.
Studios ship some titles with multiple certified configurations, and the operator selects one. The game keeps the same name, the same artwork and the same feel. The price changes.
Spribe's Aviator is the clearest case. The default is 97%. Operators may configure it at 96% or 94%.
Run those through the price list. On $1,000 of turnover:
The 94% configuration costs twice as much per unit of play as the 97% one, and nothing in the interface announces it. The only reliable source is the game's own information panel, usually reachable from the rules or paytable screen inside the game. Check it per casino, not once.
Some studios publish a fixed figure and do not offer variants. Those are easier to reason about.
| Game | Published RTP | Notes |
|---|---|---|
| Stake Originals | 99% | 1% house edge |
| BGaming Crash | 99% | |
| Spribe Aviator | 97% | operators may configure 96% or 94% |
| Spribe Plinko | 97% | |
| JetX | 96.2% to 98.9% | varies by configuration |
House-built originals sit at the low-cost end of this table. Crash, Plinko, Mines, Dice and Limbo are simple enough that the maths is inspectable, and the edge is applied as a flat deduction rather than buried in a paytable.
In a crash game, for instance, the chance of reaching a multiplier of x is approximately (1 minus the edge) divided by x. At a 1% edge, the chance of reaching 2x is 0.99 / 2, or 49.5%. Cashing out at 2x every round therefore succeeds slightly less than half the time, which is exactly the 1% edge expressed as a probability rather than a percentage of turnover. The same deduction is present whichever multiplier you choose.
RTP is a price comparison, nothing more. Used correctly it does three things.
It ranks games by cost per unit of turnover. A 99% game costs a quarter of what a 96% game costs for the same amount of play. Over a long period that difference is real and compounding.
It sets a realistic budget. Decide how much turnover you want, multiply by the edge, and you have the expected cost of the session. Treat that figure as the ticket price.
It does not predict tonight. Volatility owns the short run entirely, and no amount of RTP knowledge changes what the next spin does.
Check the figure in the game's own info panel rather than trusting a casino's marketing page. Then read house edge explained for the same idea from the other side, and set your limits before you start, using the tools on our responsible gambling page.