Aviator Game Explained: How the Multiplier Math Works

Pick an Aviator cash-out number by feel and feel puts it somewhere near 2x, because 2x looks like a coin flip and a coin flip looks fair. The arithmetic disagrees. Half of all rounds crash below 1.94x, and the mean sits far above anywhere a person cashes out. This is the Aviator game explained through the distribution underneath it, and every figure below comes out of one division.

How one round of Aviator works

A round opens with a short betting window. The plane leaves the runway at 1.00x and the multiplier climbs until it stops at a point already fixed before anything moved on screen. Cash out before that point and your stake pays at the multiplier you caught. Miss it and the stake is gone.

Two bet panels sit side by side, so you can run two independent bets in a round, banking one early and letting the second run. Auto cashout fires at a preset multiplier without you touching the screen, which takes reaction time out of the round entirely, and where you set the auto cashout trigger changes the shape of your results without changing the return.

About 3% of rounds never climb at all. They stop at 1.00x, roughly one round in 33, and they are the visible face of the house edge.

No studio publishes a round length for Aviator, so every hourly figure on this page runs on one stated assumption: about one round every 20 seconds, roughly 180 an hour. That is four times the pace of a live dealer roulette table, and pace is what multiplies the turnover the edge applies to.

What the 97% model costs, per hour

Spribe publishes no return figure for Aviator. The 97% everyone quotes is the model the market runs on, and every number on this page is computed from it. Taken at face value it makes a 3.00% house edge.

That edge applies to turnover, meaning every lari you stake across a session. Two lari a round at 180 rounds an hour puts 360 GEL through the game every hour, and 3% of that is 10.80 GEL. An hour a day for a month is 10,800 GEL of turnover and an expected cost of 324 GEL. For scale, average monthly nominal earnings in Georgia were 2,363.8 GEL in the first quarter of 2026, so that month of play prices out at 13.7% of one.

Move the stake to five lari a round and the same hour costs 27.00 GEL. Nothing about the game changed. Only the turnover did.

Neither the studio nor the game publishes a bet range. Limits are set by the site you play on, so the only figures that bind you are the ones on the bet panel in front of you.

The real odds behind every multiplier

One division drives the whole game. The chance of a round reaching a multiplier m is 0.97 divided by m, for any m of 1.00 or above. That is a Pareto-shaped tail, which is why the numbers collapse so quickly at the top end.

One row does most of the damage. A 2.00x target, the one that reads as even money, misses 51.5% of the time before you count anything else.

The median crash point is 1.94x and the mean has no fixed value

Almost every odds table gives you the chance of reaching a target. Almost none gives you the centre of the distribution, which is what the question is really about.

The median falls out in one step. Set 0.97 divided by m equal to 0.5, and m is 1.94. Half of all rounds crash below 1.94x, already under the 2.00x that reads as the default.

The mean is a different animal. For a tail shaped like this one it equals 1 plus 0.97 multiplied by the natural log of the ceiling, and the ceiling is the one number nobody publishes. Spribe states no maximum multiplier, and the figures traded between comparison pages run from 100x to unlimited, so the mean follows from whatever ceiling you assume: 7.70x at a 1,000x ceiling, 9.93x at 10,000x, and no finite value at all without one, because a Pareto tail of this shape has no finite mean.

At a 1,000x ceiling 87.4% of rounds crash below the mean, and at 10,000x it is 90.2%. Whichever assumption you make, the mean is dragged upward by a handful of rounds nobody is still holding for, which makes the average multiplier the least useful number in the game and the one quoted most often.

Nobody can capture the mean. Cashing out has to happen strictly before the crash, so the crash point is the number you have to beat, never the number you collect.

Every cash-out target pays the same expected value

Write out the expectation for a stake B at a target m. You win B times m with probability 0.97 divided by m, and lose B the rest of the time. The m cancels, leaving 0.97B minus B, or minus 0.03B.

The target has disappeared from the answer. A 1.20x target and a 100x target both return 97 tetri on the lari.

Two worked lines make that concrete. Ten lari at 1.50x wins 5 GEL 64.67% of the time and loses 10 GEL 35.33% of the time, for minus 0.30 GEL. Ten lari at 10x wins 90 GEL 9.70% of the time and loses 10 GEL 90.30% of the time, for minus 0.30 GEL.

A cash-out target is a variance setting, not a return setting. Every claim about an optimal multiplier is choosing a distribution shape and calling it an edge. The same result shows up from the other direction when you run those targets over long simulated sessions: the return line barely moves while the spread of outcomes changes completely.

Your odds of being ahead after 100, 1,000 and 10,000 rounds

The calculation that follows from those two facts is how often a player is actually in profit. Hold the stake flat, take no bonus, and read the exact binomial on the number of wins needed to break even.

The result runs opposite to what the hit rates suggest. The conservative target converges toward certain loss fastest. At 1.50x you need to win 66.7% of your rounds to break even and you win 64.67%, so a small shortfall compounds with almost no variance to hide behind. At 10x the swings stay wide enough that 15% of players are still ahead after 10,000 rounds.

The 85% who are not ahead at 10x are further down than the 1.50x players, and expected value is identical at minus 3% in every column. Wide variance buys a longer shot at an outlier and a worse typical outcome.

Losing streaks are on the schedule

Rounds are independent, so the wait for a hit at a fixed target follows a geometric distribution, with a mean of 1 divided by p and a variance of 1 minus p, divided by p squared. That single fact predicts every drought you will sit through.

At a 2.00x target, eight misses in a row is 0.50%, or one sequence in 202. Ten in a row is 0.13%, one in 762. Across a 500-round session the longest run averages about nine.

The bottom row is the one worth reading twice. A 43-round drought at a 10x target is the mean longest run across 500 rounds, with a median of 41, and it recurs inside a long session.

Reading a reversal into a run is the oldest error at any table, and it shows up in real play as well as in laboratories. Croson and Sundali measured both the gambler's fallacy and the hot hand in videotaped casino betting.

Why Martingale breaks at the table cap

Doubling after a loss sounds airtight until you put a ceiling on it. Take a 256 GEL maximum bet as a worked example, since every site sets its own. Start at 1 GEL, target 2.00x, double after every miss, and the ladder runs 1, 2, 4, 8, 16, 32, 64, 128 and 256 lari. Nine steps, and then you are out of room.

Nine consecutive misses at 2.00x is 0.515 to the ninth power, or 0.25%. One cycle in 392. By that point you have staked 511 GEL chasing 1 GEL of profit.

The expected result per cycle is minus 0.30 GEL, and the reason is in the turnover. Expected turnover per cycle is 10.16 GEL, and 3% of 10.16 is exactly 0.30. The system shuffled money around the table and left the edge precisely where it started.

A lower ceiling shortens the ladder and raises the collapse rate. The arithmetic keeps its direction and only changes speed.

Provably fair, and the arithmetic that kills every predictor

The crash point is not generated while you watch. In a provably fair crash game the operator commits to a hidden server seed before the round opens, combines it with a client-side value and a round counter, and publishes the seed afterwards, so anyone can recompute a finished round. Whether Aviator works that way at the site you play on is a question for that site's fairness page, and the verification steps are short enough to run by hand.

That mechanism ends the predictor question in two lines.

Mechanically, the number exists before the plane moves, fixed by values no app on your phone can see. Statistically, each round is an independent draw, so the history strip carries no information about the next result. Five 1.00x crashes in a row leave the chance of reaching 2.00x at exactly 48.50%, unchanged.

There is a practical tell too. A working predictor would be worth far more in use than in monthly subscriptions.

Verification of this kind grew up on crypto-first casinos, and Stake still builds its own Originals series, including its house crash title, on provably fair mechanics.

Aviator's 3% next to roulette, blackjack and slots

Aviator lands slightly worse than single-zero roulette and clearly better than the slot floor. The RTP range on most slot titles runs 94% to 96%, which is exactly where the 97% figure gets its marketing power.

The comparison flatters the game, because edge per round is only half of it. At 2 GEL a round Aviator costs 10.80 GEL an hour across 180 rounds, against 2.16 GEL at a 40-spin roulette table on the same stake.

The more useful comparison sits inside the crash category itself, where Aviator and JetX run the same core mechanic on different return settings.

What a crash game does to bonus wagering math

Bonus arithmetic on a 3% edge game is short. A 100 GEL bonus at 35x wagering needs 3,500 GEL of turnover, and 3% of 3,500 is 105 GEL. The expected cost of clearing the bonus is larger than the bonus, and on a 3% edge the break-even line falls at about 33x with full contribution.

If the requirement is calculated on bonus plus deposit, a 100 GEL deposit takes it to 7,000 GEL of turnover and 210 GEL of expected cost.

Then there is the number almost nobody prints on the offer page: game contribution. If a crash title contributes 10% toward wagering instead of 100%, that same 3,500 GEL requirement demands 35,000 GEL of real turnover, and the expected cost jumps to 1,050 GEL. A tenfold change in contribution is a tenfold change in what the offer costs you.

Contribution rate outranks headline size. A large package is only worth more than a small one if the games you actually play count toward it at a decent rate, which is what separates the bonuses worth clearing from the ones that only look big.

Where crash games sit on Georgian sites

Spribe was founded in 2018 and runs offices including Tbilisi, Warsaw and Kyiv, which is as much as the record supports about where Aviator came from.

Coverage differs by site, so the game list is worth reading before the promotion is. Among the operators we score, 1win lists Aviator by name alongside its slots, live casino and sports sections. Crystalbet runs a dedicated crash games section inside a library of more than 4,220 titles. Betlive lists JetX and Aviatrix, Adjarabet and Megapari both carry crash sections, and Megapari runs accounts and bonuses directly in lari, with a casino package advertised up to 4,340 GEL plus 150 free spins.

1win, Crystalbet, Adjarabet, Megapari and Betlive all clear payouts inside 0 to 24 hours, while Stake settles crypto withdrawals instantly, so game coverage and payout speed rank the same ten operators in different orders.

Key takeaways

• The chance of reaching any multiplier m is 0.97 divided by m. That division produces every other number here.
• Half of all rounds crash below 1.94x. The mean crash point depends entirely on a ceiling no studio publishes: 7.70x at 1,000x, 9.93x at 10,000x, no finite value uncapped.
• Expected value is minus 3% at every cash-out target. A target sets your variance, never your return.
• Low targets converge to certain loss fastest: 9.4% are ahead after 1,000 rounds at 1.50x, against 35% at 10x.
• A 43-round drought at 10x is the mean longest run across 500 rounds, with a median of 41.
• Martingale against a 256 GEL maximum bet collapses one cycle in 392, after 511 GEL staked, and leaves the edge untouched.
• On bonuses, a 10% game contribution multiplies the real cost of clearing by ten.