The published number
What a 97% return actually costs over an evening
Spribe publishes a return to player of 97% for Aviator, which is a statement about a very large number of rounds rather than about the two hundred you will play tonight.
Open Vave Advertising link. 18+.
2.21×
collect here and the round pays 2.21× the stake
seconds
- 1.00× – 9.99×
- 10× – 99×
- 100× and up
| Casino | Named by | Coins | Withdrawal limit | Before ID check | Licence | Action |
|---|---|---|---|---|---|---|
| Vave | 3 / 3 | 6 | no limit | not published | Curaçao Gaming Authority | Open |
What the 97% actually claims
Return to player is a ratio between two totals: everything the game pays out, divided by everything staked into it, measured across an enormous number of rounds. Spribe publishes 97% for Aviator. The complement, 3%, is the house edge — the share of turnover the game is designed to keep.
Two words in that definition carry all the weight. The first is turnover: the divisor is money staked, not money deposited, and the same hundred rupees staked twenty times is two thousand rupees of turnover. The second is enormous: the ratio is a limit that a design converges on over millions of rounds, and it makes no promise about any smaller sample.
The same number, written as money
Take a stake of 100 units per round and hold it constant.
One round stakes 100 and gives back, on average, 97. The expected cost of that round is 3 units. Two hundred rounds — a comfortable hour at ten to thirty seconds each — stake 20,000 units in total, and 3% of 20,000 is 600. Five hundred rounds stake 50,000 and cost 1,500. A thousand rounds cost 3,000, which is thirty times the stake you thought you were risking.
The pattern is the point. The expected cost scales with rounds played, not with the size of the balance you brought, and not with the highest multiplier you happened to catch. A player who doubles the stake doubles the cost; a player who plays twice as fast also doubles it.
A round runs ten to thirty seconds, so an hour of continuous play holds 120 rounds at one end and 360 at the other. At a flat stake of 100 a round, the published 3% edge comes to 360 units in the slow hour and 1,080 in the fast one. Nothing between the two bars changed except how quickly the rounds arrived.
This is also why the second bet panel is not free. Two panels of 100 each is a 200-unit round, so the same 3% applies to twice the turnover. Nothing about the panel is unfair — it simply plays the game twice as hard, which is worth remembering when reading how a round is structured.
Why tonight will not come out at 97%
Almost nobody finishes a session at exactly 97% of what they staked, and that is not a flaw in the published figure. It is what the figure means.
A short session is a small sample drawn from a very wide distribution, and Aviator’s distribution is unusually wide because a single round can pay a multiple in the hundreds. In fun mode on 7 September 2026 we saw a round of 1,095.11× in the history feed above the play area. One such round in a sample changes that sample’s return beyond recognition, and its absence from every other sample is exactly what balances the figure out.
So the honest reading of 97% is: the game keeps 3% of turnover in the long run, and any particular evening lands somewhere on a wide spread around that. Most evenings land below it, because the rare enormous rounds have to come from somewhere, and they come from the many ordinary sessions that never see one.
A winning night proves the spread exists. It does not prove that the 3% stopped applying.
What RTP does not tell you
The published return answers one question well and several adjacent questions not at all. Three of them matter enough to name.
It says nothing about variance. Two games can both return 97% and feel completely different: one paying small amounts constantly, the other paying almost nothing and occasionally paying a fortune. RTP is a single average, and it cannot distinguish them. Aviator sits toward the volatile end, and the same 97% that describes a steady grind also describes a night in which twenty rounds in a row go nowhere.
It says nothing about the maximum multiplier. Spribe does not publish a ceiling on how high the curve can climb, nor a maximum win. The 1,095.11× we saw in the history feed is a thing that happened, not a documented limit, and a number seen once tells you the ceiling is at least that high and nothing more.
It says nothing about the size of bet the game will accept. No upper limit on the stake is published either. In practice the ceiling is set by whichever operator hosts the game, which means the same title can accept very different amounts in two different lobbies, and neither figure comes from the studio.
Those three gaps are worth stating plainly rather than filling in. The maximum multiplier, the maximum win and the maximum bet are all unpublished by Spribe. Any page that quotes a specific figure for one of them is quoting an operator’s configuration, a third-party guess, or nothing at all. We keep a note on how we handle unpublished numbers in our methodology.
The one thing 97% does constrain
There is a useful piece of arithmetic hidden in the number, provided you accept one assumption the studio does not spell out: that the return holds at every cash out target rather than favouring some over others. That is the standard construction for crash games, but it is an assumption, so treat what follows as arithmetic and not as a published fact.
Under it, aiming for a multiplier of m has to hit roughly 97 divided by m of the time. A 2× target lands a little under half the rounds. A 10× target lands under a tenth of them. A 100× target lands about once in a hundred and three rounds, which at fifteen seconds a round is roughly twenty-six minutes of play for one expected hit.
Two inputs, and both are stated on the page: the published 97% return, and the assumption that the return sits evenly across every cash-out point. That second one is standard for crash games and the studio has never published it, so the figure says so on its own face. Under it a 1.20 exit lands about four rounds in five and a 100 exit about once in a hundred and three.
Nothing in that changes the expected cost. Every target sits at the same 97%, which is precisely why no cash out level is a strategy — they trade frequency for size at a fixed price. Products that claim otherwise are examined on our pages about predictor apps and signal groups and bots.
How to use the number
Treat 97% as a cost of admission with a known rate. Decide how many rounds an evening is going to contain, multiply the number of rounds by the stake, take 3% of that, and ask whether that figure is one you would pay to spend the evening this way.
If it is, the maths has done its job. If the answer changes once you see it written down, that is also the maths doing its job.