Crash Game Simulator
Two tools on one page. The first shows the exact odds of a crash round reaching any multiplier, and why every cash-out target carries the same negative expectation. The second runs a seeded Monte Carlo backtest so you can watch martingale, paroli and flat betting fail in different shapes. Everything runs in your browser — no money, no account, no data sent anywhere.
♠ Part 1 — What are the odds of reaching x?
A Stake-style crash round draws a 32-bit integer and turns it into a multiplier withcrash = max(1, (2^32 / (int + 1)) × (1 − e)). That gives a clean survival function: the chance a round reaches at least x is(1 − e) / x.
| Cash out at | Reaches it | Busts first | 1 in | EV / $1 |
|---|
Read the last column. It never moves. Chasing 100× is not "riskier but better value" than grinding 1.5× — it is the same expected loss per dollar, delivered in rarer and larger pieces. The only thing your target multiplier chooses is the variance of your session, never its expectation.
♦ Probability of reaching the target
Green bar = rounds that reach the multiplier. Red remainder = rounds that bust first.
Where the edge lives: when the drawn integer lands in the top of its range the formula clamps to 1.00× — an instant bust before any cash-out can fire. That clamp is the house edge; the rest of the distribution is a fair 1/x curve.
♥ Part 2 — Strategy backtest
Monte Carlo over the same distribution. Every run is a fresh session starting from your bankroll; the summary aggregates all of them. The generator is seeded, so the same seed and inputs always reproduce the same numbers — change the seed to draw a different sample.
♣ Session Settings
Hits about 49.5% of rounds.
After this many consecutive doublings the bet resets to base — the same cap a real table maximum imposes. 8 doublings means the last bet is 256× your base bet.
Bank the streak and drop back to the base bet after this many wins in a row.
Any 32-bit integer. Same seed plus same inputs = identical results, every time.
♦ Results
Equity curve — last session
One session out of the batch, plotted round by round. The dashed line is your starting bankroll.
How a round resolves: you stake the current bet, a crash point is drawn from the formula above, and you win stake × (target − 1)if the crash point reaches your target, otherwise you lose the stake. If the bankroll cannot cover the next bet the remainder is staked instead; a bankroll under one cent counts as bust and ends the session.
This tool does not find you an edge — there isn't one
No betting pattern changes the −e expected value of a crash bet. Martingale, paroli, D'Alembert, Fibonacci, "wait for ten reds", auto-bet scripts with clever stop conditions — all of them are sums of individually negative-expectation bets, and a sum of negative expectations is negative. What they change is variance: how often you finish up, how bad the worst session is, and how quickly a bad streak ends the session. That is the only thing this page measures. If a simulation run here shows a profit, you have found a lucky seed, not a system. Play with money you can afford to lose, and read ourresponsible gambling pageif any of this feels like a plan rather than a game.
♣ Where the model comes from
The crash point distribution used here is Stake's published one, taken from the Bitcointalk pre-commitment post where the crash chain was seeded: a 32-bit integer is read from the round's hash and converted withmax(1, (2^32 / (int + 1)) × (1 − e)), then floored to two decimals. Ourprovably fair verifieruses exactly this function to recompute real Stake crash rounds against their published game hashes, and this page reuses that maths unchanged.
Two caveats worth stating plainly. First, other operators' crash games may differ — house edge, instant-bust rules and even the shape of the distribution vary between sites, so these odds describe Stake's game and games that copy it, not crash in general. Second, the simulator draws crash points from a seeded pseudo-random generator sampling that distribution; it does not replay real hash chains. It is a model of the odds, not a record of past rounds.
Last updated: August 2026.
Keep reading
Crash Odds, Explained
Does any crash strategy beat the house edge?
No. Cashing out at multiplier x wins with probability (1 − e) / x and pays x times your stake, so the expected value of every single round is x × (1 − e) / x − 1 = −e. The multiplier cancels out. Whether you cash at 1.5× or at 100×, you lose the same 1% of every dollar wagered on average. Betting patterns only change how that loss is distributed across time — they cannot change its sign.
Why is the expected value identical at 1.5× and at 100×?
Because the crash distribution is built to make it identical. The payout multiplier and the hit probability are exact reciprocals of each other, scaled by (1 − e). A 1.5× target hits about 66% of the time and pays 1.5×; a 100× target hits about 0.99% of the time and pays 100×. Multiply each pair out and you get 0.99 either way. That is what a house edge is: a constant multiplicative haircut applied to a fair game.
Does martingale work on crash games?
Martingale converts a small, frequent loss into a rare, catastrophic one. Doubling after every loss means each losing streak costs 2^n − 1 base bets, so the bet size grows faster than any realistic bankroll. In the backtest below, martingale usually shows a higher share of sessions ending in profit than flat betting does — and a much higher bust rate. That trade is the entire trick: it hides the loss in the tail. Table maximums and the casino bet cap make it strictly worse in practice.
What are the odds of a long losing streak at a 2× target?
At a 1% house edge a 2× cash-out misses about 50.5% of rounds. Ten misses in a row is 0.505^10, roughly 1 in 1,090 — so in a 5,000-round session you should expect several of them. Fifteen in a row is roughly 1 in 34,000. Streaks that feel impossible are routine at volume, which is exactly why doubling systems fail: the streak that ruins you is not unlikely, it is scheduled.
Is this simulator using real casino numbers?
The crash point distribution modelled here is the one Stake published when it seeded its crash chain — crash = max(1, (2^32 / (int + 1)) × (1 − e)) with int read as a 32-bit integer. It is the same formula our provably fair verifier uses to check real Stake crash rounds. Other operators run crash games with different house edges, different instant-bust rules and sometimes an entirely different distribution, so treat these numbers as a model of Stake's game rather than a universal one. Checked August 2026.
Why does the simulator show a seed, and why can I change it?
So you can reproduce a result. The simulation uses a seeded mulberry32 pseudo-random generator, not your browser’s entropy, which means the same seed plus the same inputs always produces exactly the same bankroll curve, bust rate and drawdown. Change the seed and you get a different sample of the same distribution. If a strategy only looks good on one seed, it does not look good.
Can a simulator prove a strategy is bad?
It can show you the shape of the risk, not prove anything. The proof is the algebra in the first table: every target has the same negative expectation, and no sequence of bet sizes changes the expectation of a sum of negative-expectation bets. The Monte Carlo section exists to make the variance visible — the drawdowns, the bust rate, the gap between the median session and the mean.