Crash game RTP of 97%: what it means for your stakes

RTP is the single number behind most questions about crash game maths. It also fixes the probability of reaching each multiplier, so the two ideas belong together. This page shows what 97% means in shillings, where the figure comes from and what it leaves out.

Where does the 97% figure come from?

The figure is the provider's own. Spribe's Aviator page lists Return to Player (RTP) as 97%, in the same block that gives the game type and the devices it runs on. Inside the game, the rules panel has a Return to Player section too, which calls the figure the overall theoretical return to player.

Theoretical means the number is worked out from the game's design, not counted from past rounds. It is fixed before any round is played, in the same way a roulette wheel has its edge before the first spin. The round mechanics it rests on are set out in how Aviator-style crash games work.

Kenyan rules put that number on the regulator's file. Regulation 45(4) of the Gambling Control (Conduct of Gambling Operations) Regulations, 2026 says the theoretical return to player percentage for each game shall be disclosed to the Authority. Under regulation 45(2), certification of the random number generator includes a statistical test of its randomness and of the theoretical return to player.

Test labs check such figures against the maths. BMM Testlabs, one such lab, lists the RTP of the base game, the features and the overall game among the maths documents it reviews, and checks the live game configuration against the maths model. Spribe's About page records that Aviator's maker received game certification in November 2019 but does not name the lab.

RTP and house edge: one number, two views

RTP and house edge always add up to 100%. A 97% RTP leaves a 3% house edge, the share of every stake the game keeps on average. In money terms, each KES 100 staked carries an expected cost of KES 3.

Illustrative example: at the KES 20 minimum stake set by section 71(1) of the Gambling Control Act, 2025, one round carries an expected loss of 60 cents. Stake KES 1,000 in total and the expected loss is KES 30.

For comparison, European roulette with a single zero has a house edge of 1/37, about 2.70%, because its 37 pockets pay as if there were 36. The mechanics differ, but the principle matches: the edge is built into the payout rules, not into luck.

Why the 3% applies to every stake, not to your deposit

The edge works on turnover, meaning the total of all stakes, including money that is won and staked again. A deposit can pass through the game many times in one session. Each pass costs 3% on average.

Illustrative example: you deposit KES 1,000 and bet KES 50 a round for 100 rounds, restaking winnings while the balance lasts. Your turnover is KES 5,000, so the expected loss is KES 150. That is 15% of the deposit, even though the game's edge is 3%.

Illustrative example: expected results by session size

The table applies the 97% figure to sessions of different sizes. The expected return is what such a session averages out to if it were repeated many thousands of times. Any one real session will differ.

How far a short session can stray from 97%

Over a handful of rounds, chance dominates. Under the standard crash model, a 2x cash-out target is reached in 48.5% of rounds. Ten rounds at that target can therefore end well ahead or well behind.

Add the rows up and about 34% of such sessions finish ahead, about 25% finish level and about 41% finish behind. The average across all of them is still a loss of KES 30, which is 3% of the KES 1,000 staked.

The more rounds you play, the closer your total loss as a share of stakes tends to come to 3%. That is what "long run" means here: the direction a large number of rounds moves in, not a forecast for tonight. Why no betting method changes that is covered in why you cannot beat a crash game in the long run.

Does the cash-out target change the RTP?

Not under the standard model. A target of m is reached with probability 0.97 divided by m and then pays m times the stake, so the average return is 0.97 of the stake whatever m is. A low target wins often and pays little; a high target wins rarely and pays a lot.

Illustrative example: a 1.5x target pays KES 150 on a KES 100 stake in about 64.7% of rounds, an average of KES 97. A 10x target pays KES 1,000 in 9.7% of rounds, again an average of KES 97. The full working is in the expected value of auto cash-out targets.

Spribe does not publish Aviator's crash-point distribution, so the standard model is an assumption, not a statement from the provider. What Spribe does state is the 97% overall figure.

What the 97% does not tell you

RTP is not a forecast for one session, a refund promise or a sign that a loss is due to be won back. Each round is settled on its own, and the edge applies to the next stake just as it did to the last.

RTP is also measured on stakes placed inside the game. It does not cover what happens to money on the way in or out, such as charges on deposits and withdrawals, which are set outside the game.

If a session is costing more than planned, regulation 84 of the 2026 Conduct Regulations requires licensees to give players deposit, loss and session limits. Where to find them is covered in how limits and self-exclusion work for Kenyan players.