Crash Game Odds and RTP

The RTP printed on a crash game's rules screen is the share of all money staked that the game pays back over a very long run. On a crash curve it fixes everything else, because the chance of a round reaching any multiplier equals the RTP divided by that multiplier. A game set at 97% reaches 2.00x in 48.50% of rounds and 10.00x in 9.70%, and in each case the payout multiplied by its chance comes back to the same 97%.

Studios commonly sell one title in several return versions and leave the casino to pick which one it runs. The same game can therefore carry a different figure from one lobby to the next, and a listing that quotes a single number for a title may describe a version a given casino does not offer. The figure that applies is the one on the rules screen of the game as it opens where you play.

What the game keeps shows most plainly at the bottom of the curve. At 97%, 3.96% of rounds bust before the first payable step, about one in 25. Of that share, 3.00% is the return the game withholds, and the remaining 0.96% are rounds that would stop below the first step even in a game that kept nothing.

A stated maximum deserves a second look, because three different limits travel under that name. A round cap ends every flight at a set multiplier and pays anyone still in, leaving every lower target intact. A multiplier payout cap stops the payment growing while the curve carries on. A money win cap limits what one bet can collect, a different multiplier ceiling at every stake. The last two pull the return down, as worked out further below.

Each cell is the share of rounds whose bust point gets as far as the target, which is the chance that a cash-out set there is paid. The three columns are illustrative return levels, not the figures of any particular title; a game set elsewhere sits between or beside them, and a higher return lifts every row in the same proportion.

Reading across a row shows how little the return level moves the middle of the curve: the median shifts by a few hundredths between columns, while the share of rounds that bust at once falls by roughly a point for every point of return added.

Ask for the average bust point of an uncapped crash game and there is no figure to give. The odds of getting to a multiplier shrink only in proportion to its size, so every further doubling of the range, from 10x to 20x and then on to twice that again without end, adds about the same amount to the running total. A sum that keeps growing by a steady step never settles, which is why these pages describe the middle of the curve by its median rather than its mean.

Once every round has to stop somewhere, an average finally exists, but it depends almost entirely on where the ceiling sits, since the rare rounds at the far end each count for a great deal.

None of this alters what a cash-out returns. A bet is paid at its own target, not at the bust point, so the far tail of the curve concerns nobody whose target sits well short of it.

Each round counts at its bust point, or at the ceiling when it gets that far. Every step down the table multiplies the ceiling many times over yet adds only a little to the average, the mark of a quantity with no limit once the ceiling is removed.

Figures for an uncapped game at 97%. Half of all runs of a given length beat the median column. The top result grows almost in step with the number of rounds, so the largest figure a busy lobby displays reflects how long it has run far more than how generous the game is.

A crash game takes its share before the multiplier has climbed anywhere. Once a round has passed a given point, the chance of it reaching a higher one is simply the lower multiplier divided by the higher, with no RTP in the formula: a round that has reached 2x goes on to 4x 50.00% of the time, at any return level.

Every stretch of the flight after the start is therefore priced as a fair bet, whatever the round has done so far: a player still in at 10x has a 10.00% chance of seeing 100x. The cost of the game lives entirely in the rounds that end early.

The key point: The game collects its share at the start, in the rounds that bust early. After that, staying in until the multiplier doubles is an even chance from any point in the flight.

These chances describe a round still in flight, so they hold at every return level: the return has already been settled by the time the first of these points is passed. The odds of reaching a target from the start of a round are the ones in the reach table.

On a clean curve every cash-out returns the stated RTP. Three settlement rules can break that; each is something to look for on the rules screen, not to assume of any title.

A money win cap. If one bet can collect no more than a fixed amount, dividing that amount by the stake gives the cap counted in stakes. A target above it is paid only the cap, so its return falls to the RTP times the cap divided by the target. At 97%, with a cap worth 1,000 stakes, a 5,000x target returns 19.400% and a 10,000x target 9.700%. The same cash limit bites lower as the stake rises, since doubling the stake halves the cap counted in stakes.

Payouts rounded down to the cent. If a game settles this way, a small stake loses a little on every target whose exact payout falls between two cents. A 1.37x cash-out on a stake of 0.1 works out at 0.137 and pays 0.13. At that stake the targets from 1.01x to 10.00x average 95.889% instead of 97%, with the lowest, 88.991%, at 1.09x. From a stake of 1 upward the rounding costs nothing.

The bust point rounded to the nearest step. If a game rounds the bust point to the nearest hundredth instead of down, the player gains slightly, and most at the lowest targets: on a game set at 97%, a 1.01x cash-out then returns 97.483%.

The rule: Both the cash limit and cent rounding cut the return only for some stakes and targets: the cap at high targets on large stakes, the rounding at low targets on small ones.

Returns at 97%, with the cap counted in stakes: the money limit on a single bet divided by the amount staked. Targets at or below the cap keep the full return, and every target above it loses in proportion to how far it sits past the cap.

These rows apply only to a game that floors each payout to the cent, at 97%, with stakes in currency units. The loss is the fraction of a cent cut off each win, so it fades as the stake grows.

Most studio crash games take each bust point from a random number generator the studio runs, normally tested by a laboratory before release. Some also let players check a finished round against a value committed before it began. Either way the curve stays the same: the return comes from the version the casino runs, and a round that checks out says nothing about which version that is or what cap applies. How a seed-based round is locked before betting opens and recomputed by the player is shown step by step on the provably fair page.

Not every game filed under crash runs on a climbing multiplier. In step games a chicken crosses lanes or a climber works up a tower, and the player may cash out between steps. Each step is priced from how many spaces end the walk at the chosen difficulty, so the chance of surviving several steps is a product of per-step odds rather than the RTP divided by a multiplier, and none of the tables above describes them. Target games are the opposite case: there the multiplier is picked before the draw, and the reach table applies to them exactly.