Home › Game Guides › Crash
Crash Gambling Explained: The Math, Odds & House Edge
Crash gambling is one of the defining games of the crypto-casino era: a rising multiplier that can crash at any moment. You bet before the round, watch a number climb from 1.00x upward, and try to cash out before it busts — cash out in time and your stake is multiplied; wait too long and you lose it all. It is fast, simple and brutally mathematical. The whole game runs on one formula, P = RTP ÷ m, and that formula has a consequence most players never realise: every cash-out target you could possibly pick returns the same 0.97 of your stake on average. Aiming for 100x is not "riskier but more rewarding" than aiming for 2x — the long-run maths are identical. The same identity shows up even more starkly in dice, where the player sets the win chance directly. This page shows the numbers, in full.
Crash gambling at a glance
- Bet before the round, cash out before the multiplier crashes — or lose your stake.
- Most crash games run a 97–99% RTP, i.e. a 1–3% house edge; a few run up to 5%.
- Odds of reaching any multiplier m: P = RTP ÷ m (97% RTP → 48.5% chance of 2x).
- The house edge shows up as instant busts: on a 97% game about 3% of rounds crash at 1.00x.
- Every cash-out target has an identical expected return of 0.970 of stake — a flat 3% loss (16Best analysis).
- The median crash point on a 97% game is 1.94x — half of all rounds never reach 2x (16Best analysis).
- A run of 10 losses at a 2x target shows up in the stream about once every 5 hours — but a martingale player, who resets on every win, walks into one about once every 10.5 hours (16Best analysis).
- A $1 Martingale hits a $1,023 cumulative loss by the 10th round in a streak (16Best analysis).
- At $1 a round and 150 rounds an hour, a 97% game costs about $4.50 an hour (16Best analysis).
- Aviator (Spribe) is the most-played crash game, with 77M+ monthly players in 2026.
What is crash gambling?
Crash gambling is a casino game where you bet on a multiplier that rises from 1.00x and can crash at any moment — you win if you cash out first, and lose your whole stake if the round crashes before you do. There are no reels or cards, just a single number climbing an exponential curve. It was popularised by crypto casinos (Stake Crash, Aviator, Bustabit and dozens of clones) and is now one of the most-played formats online: Spribe’s Aviator alone crossed 77 million monthly players in 2026, making it the most-played crash game by a wide margin. If you cash out at 3.00x you win 3× your bet; if the round crashes at 2.40x before you act, you lose everything. That single decision — when to get out — is the entire game, which is exactly why it thrives on crypto casinos: instant, transparent and endlessly repeatable.
How does a single crash round work?
A round has four steps — bet, climb, cash out, crash — and the crash point is fixed before the round even begins. Nothing you do while the number is rising changes where it will bust; the only decision that matters is when you choose to get out.
- Bet. Before the round starts, you place a stake (and optionally set an auto cash-out target).
- Climb. The multiplier starts at 1.00x and rises on an exponential curve — slowly at first, then faster.
- Cash out. At any moment you can lock in the current multiplier × your stake.
- Crash. The round ends at its pre-determined point. Anyone who has not cashed out loses their bet.
Rounds are short — typically 5 to 30 seconds — and run back to back, so players can complete 100 to 160 rounds an hour. That speed is central to how the game separates you from your money, as we quantify below.
What are the odds of hitting each multiplier?
The probability of a round reaching at least a multiplier m is simply the RTP divided by m. On a 97% RTP game the chance of reaching 2.00x is 0.97 ÷ 2 = 48.5%, the chance of 10.00x is 0.97 ÷ 10 = 9.7%, and 100.00x is just 0.97%. The missing ~3% at every level is the house edge, and it surfaces most visibly as an instant bust: on a 97% game roughly 3% of rounds crash at 1.00x before anyone can cash out.
16Best analysis: the formula is P(reach m) = RTP ÷ m. Because the payout on a win is exactly m× your stake, the expected return is P × m = (RTP ÷ m) × m = RTP — the m cancels out. That single line of algebra is why no cash-out target is mathematically better than any other. A game that pays you m× while a fair bet would need to pay m ÷ RTP (the implied fair payout column below) is short-changing you by the same 3% every single time.
Computed from P = RTP divided by m at RTP 0.97. Higher targets pay more but hit far less often. 16Best analysis.
| Cash-out target (m) | P(reach) = 0.97 ÷ m | Implied fair payout (1 ÷ P) | Expected return (P × m) | House take |
|---|---|---|---|---|
| 1.5x | 64.67% | 1.546x | 0.970 | −3.0% |
| 2x | 48.50% | 2.062x | 0.970 | −3.0% |
| 3x | 32.33% | 3.093x | 0.970 | −3.0% |
| 5x | 19.40% | 5.155x | 0.970 | −3.0% |
| 10x | 9.70% | 10.31x | 0.970 | −3.0% |
| 50x | 1.94% | 51.55x | 0.970 | −3.0% |
| 100x | 0.97% | 103.1x | 0.970 | −3.0% |
Table: 16Best analysis, computed from P = RTP ÷ m at 97% RTP. The “implied fair payout” is what a zero-edge game would have to pay for the same probability; the gap versus m is the edge.
Why does every cash-out target have the same expected value?
Because the multiplier you win cancels against the probability of winning it, leaving the RTP unchanged for every target. Look at the fourth column of the table above: whether you cash out at 1.5x, 2x, 10x or 100x, the expected return is exactly 0.970 of your stake — a flat 3% expected loss per round. High multipliers do not improve your odds; they only trade frequency for size, turning many small wins into rare large ones with the same long-run cost.
On a 97% game, every cash-out target returns the same 0.970 of your stake. High multipliers just trade frequency for size.
This is the single most important fact about crash, and it is counter-intuitive: it feels like patiently waiting for a 20x is a different bet from grabbing a quick 1.5x, but to the maths they are the same negative-EV wager. The only thing that changes is variance — how wildly your bankroll swings on the way down. If the RTP-and-edge relationship is fuzzy, our RTP & house edge guide breaks it down, and house edge by game shows how crash compares to blackjack, roulette and slots.
What is the median crash point?
On a 97% RTP game the median crash point is 1.94x — half of all rounds crash before they ever reach 2x. Solve 0.97 ÷ m = 0.5 and you get m = 1.94. That is a useful reality check against the highlight-reel of 100x screenshots players share: the typical round is a short, unremarkable bust below 2x, and a majority of rounds never reward a patient 2x auto cash-out.
The median crash on a 97% game is 1.94x — half of all rounds never reach 2x.
How often will you hit a losing streak?
Far more often than most players expect — a run of 10 losses at a 2x target turns up in the stream about once every 5 hours of steady play. At a 2.00x auto cash-out you win 48.5% of rounds and lose 51.5%, so the chance of losing N rounds in a row is 0.515N. A 5-loss streak is a 1-in-28 event; a 7-loss streak is 1-in-104; a 10-loss streak is 1-in-762. Because you play so many rounds per hour, "rare" events arrive constantly.
One distinction matters here, and most write-ups miss it. The table below counts how often a 10-loss run appears anywhere in a stream of rounds. A martingale player is asking a different question, because they reset the moment they win: their sequences last 1 ÷ 0.485 = 2.06 rounds on average, so 150 rounds an hour is only about 73 fresh attempts, and at 1-in-762 per attempt the blow-up arrives about once every 10.5 hours (16Best analysis). Quoting the 5-hour figure as the martingale number overstates the frequency by roughly two times — the same slip our Mines analysis corrects.
| Consecutive losses at 2x | Probability (0.515N) | Roughly | Expected times per hour (150 rounds) |
|---|---|---|---|
| 3 in a row | 13.66% | 1 in 7 | ~20.5 |
| 5 in a row | 3.62% | 1 in 28 | ~5.4 |
| 7 in a row | 0.96% | 1 in 104 | ~1.4 |
| 10 in a row | 0.131% | 1 in 762 | ~0.20 |
| 13 in a row | 0.018% | 1 in 5,578 | ~0.03 |
Table: 16Best analysis, computed as 0.515N (loss probability at a 2.00x target on a 97% game), with expected occurrences = 150 × probability at 150 rounds/hour.
16Best analysis: at 150 rounds an hour you should expect to see a 7-loss streak more than once every hour and a 5-loss streak about five times an hour. These are not bad luck — they are the ordinary texture of the game. A player who doubles up "just until the next win" is therefore certain to face streaks long enough to matter, which is exactly the trap the Martingale section below quantifies.
Does the Martingale system work on crash?
No — doubling after every loss turns an ordinary streak into a bankroll-ending event, and the math is unforgiving. Start at $1 and double after each loss and the required bet is $2N−1 on the N-th round of a streak, while the cumulative amount lost is $2N − 1. That grows exponentially: by the 10th consecutive loss you are down $1,023 and the next recovery bet is $1,024.
Computed as 2 to the power N minus 1, the sum of a doubling bet sequence starting at 1 dollar. 16Best analysis.
| Loss # in the streak | Bet required (2N−1) | Cumulative lost (2N−1) | Chance of a streak this long (0.515N) |
|---|---|---|---|
| 1 | $1 | $1 | 51.5% |
| 3 | $4 | $7 | 13.7% |
| 5 | $16 | $31 | 3.6% |
| 7 | $64 | $127 | 0.96% |
| 9 | $256 | $511 | 0.25% |
| 10 | $512 | $1,023 | 0.13% |
| 11 | $1,024 | $2,047 | 0.068% |
Table: 16Best analysis, doubling sequence from a $1 base at a 2.00x target (loss probability 0.515 on a 97% game).
16Best analysis: a $1,000 bankroll blows up on the 10th consecutive loss ($1,023 gone) — and a common $500 table maximum blocks the 10th bet ($512) before you even get there. The probability of a 10-loss streak is 0.51510 = 0.13% (about 1 in 762), which sounds safe until you remember the speed: at 150 rounds an hour you meet a 10-loss streak roughly once every 5 hours. Martingale does not lower your expected loss — it concentrates it. You swap many tiny losses for a small chance of one catastrophic one, and the house edge is identical either way. This is the same mechanism behind many large gambling losses.
How much does crash gambling cost per hour?
Expected loss per hour = bet × rounds per hour × house edge — about $4.50 an hour at $1 a round on a 97% game. At 150 rounds an hour, a $1 stake and a 3% edge, that is $1 × 150 × 0.03 = $4.50. The number scales linearly with both your stake and the edge, so a 95% game (5% edge) or a $5 stake multiplies the damage fast.
Computed as bet times rounds per hour times house edge, at 1 dollar per round and 150 rounds per hour. 16Best analysis.
At $1 a round and 150 rounds an hour, a 97% crash game costs about $4.50 an hour — and $22.50 at $5 a round.
The lesson is that crash’s speed, not its edge, is the real cost driver. A 97% RTP looks generous next to a 92% slot, but crash lets you play three or four rounds a minute, so the low edge is applied to a far larger volume of money. Twenty rounds of crash can be done in two minutes; the same twenty spins on a slot might take ten. Same edge, three to five times the exposure. For the slot comparison see our slot machine statistics, and for the industry side of the same arithmetic, how much money casinos make.
What is auto cash-out and does it help?
Auto cash-out exits your bet automatically at a preset multiplier — it removes reaction-time error but does not change your expected value or the house edge. You set a target such as 2.00x and the game cashes you out the instant it is reached, so there is no fumbling to click as the number races upward. That enforces discipline, which is genuinely useful, but the underlying maths are untouched: a low target wins often for a little, a high target wins rarely for a lot, and both return the same 0.970 of stake over time. Auto cash-out changes your experience of the variance, never the long-run result.
Is crash gambling provably fair?
Most reputable crypto crash games are provably fair, meaning the crash point is committed cryptographically before you bet and can be verified afterward — but that proves honesty, not profitability. The casino typically combines a hashed server seed with your client seed and a nonce, then after the round reveals the server seed so you can recompute the exact crash point. Stake and BC.Game, for example, derive the multiplier from a SHA-256 hash via a published formula of the form crash point = (232 ÷ (int + 1)) × (1 − house edge). You can confirm the result was fixed in advance and not altered once your bet was in. What provable fairness does not do is shrink the house edge or make the game beatable — the edge is a deliberate parameter in that very formula. See our deep dive on how provably fair works.
Do any crash strategies actually work?
No strategy beats the house edge. Martingale, fixed low-multiplier auto cash-out, "waiting for a big one", chasing patterns after a run of low crashes — none change the underlying math. Every round is independent (the algorithm has no memory), and every target returns the same expected value, so systems only reshape how you lose — steady small losses versus rare large swings — never whether you lose over time. The one honest edge a player has is stopping: setting a loss limit and a time limit before starting. For the harm side, see our gambling losses and addiction data.
The honest takeaway: crash is entertaining and transparent, but it is a negative-expected-value game like every casino game. Treat any session as paid entertainment, decide your loss limit before you start, and remember that the 1–5% edge grinds your bankroll down the longer you play — the same law of large numbers that keeps casinos profitable and drives the economics of the whole crypto gambling market.
Methodology & limits: what these numbers do and do not tell you
16Best analysis: every figure on this page is expected value — the average over a very large number of rounds — not a prediction of any single session. Two things follow. First, a 0.970 EV does not mean you lose 3% of your stake each round; you either lose 100% of it or win a multiple, and the −3% only emerges across thousands of rounds. Any individual evening can end well up or catastrophically down. Second, our probabilities assume a 97% RTP and independent rounds; real games range from about 97% to 99% RTP (BC.Game and Stake near 99%, Roobet ~3.5% edge, Gamdom ~5%), so check the specific game’s stated RTP — a lower RTP scales every loss figure up proportionally. We also assume 150 rounds/hour; slower play lowers the hourly cost, faster play raises it. These are honest models of a fair game, not a claim about any operator’s integrity, which is what provable fairness exists to check.
Key takeaways
- One formula runs the game: P(reach m) = RTP ÷ m. On a 97% game, that is a 48.5% shot at 2x and a 0.97% shot at 100x.
- Every cash-out target has the same expected value — a flat 0.970 of stake, because the multiplier cancels the probability.
- The median crash is only 1.94x on a 97% game — half of rounds never reach 2x.
- Losing streaks are ordinary: a 7-loss run at 2x happens more than once an hour; a 10-loss run every ~5 hours.
- Martingale concentrates risk, it does not remove it — a $1 base is down $1,023 by the 10th loss and blocked by a $500 table cap.
- Speed is the real cost: ~$4.50/hour at $1 a round, because you play 100–160 rounds an hour.
- Provably fair proves honesty, not profitability — the edge is baked into the crash-point formula itself.
Frequently asked questions
What is crash gambling?
Crash gambling is a casino game where you bet on a multiplier that rises from 1.00x and can crash at any moment. You win if you cash out before the crash; you lose your stake if the round crashes first. It is most associated with crypto casinos and games such as Stake Crash and Aviator.
What are the odds in crash gambling?
The chance of reaching a multiplier m is the RTP divided by m. On a 97% RTP game you have about a 48.5% chance of reaching 2x, a 9.7% chance of reaching 10x, and a 0.97% chance of reaching 100x. The house edge is typically 1 to 3%, and up to 5% on some games.
Why does every cash-out target have the same expected value?
Because the payout on a win is exactly m times your stake, while the probability of winning is RTP divided by m. Multiplying them cancels the m, leaving RTP. On a 97% game that means every target returns 0.970 of your stake on average, a flat 3% expected loss, whether you cash out at 1.5x or 100x.
How much does crash gambling cost per hour?
Expected loss per hour equals your bet times rounds per hour times the house edge. At $1 a round, 150 rounds an hour and a 3% edge, that is about $4.50 an hour. At $5 a round it rises to about $22.50, and a lower-RTP game costs proportionally more.
Does the Martingale strategy work on crash gambling?
No. Doubling after each loss makes your bets grow as 2 to the power of the streak length, so a $1 base is down $1,023 by the 10th consecutive loss and the next bet is $1,024. A 10-loss streak has a 0.13% chance per attempt, and because a martingale sequence lasts about 2.06 rounds, roughly 73 attempts fit into an hour at normal speed — so the blow-up arrives about once every 10.5 hours. Martingale concentrates your losses into rare catastrophic ones without changing the house edge.
Is crash gambling rigged or provably fair?
Reputable crash games are provably fair, meaning the crash point is fixed cryptographically before you bet and can be verified afterward using the revealed seeds. That proves the result was not manipulated, but it does not remove the built-in house edge, which is a deliberate parameter in the crash-point formula.
Is there a winning strategy for crash gambling?
No. Every cash-out target has the same expected value, every round is independent, and the house edge applies to all of them. Systems like Martingale or low-multiplier auto cash-out only change how your losses are distributed, not the negative long-run return.
Sources
- GamblingCalc — Crash Game Strategy: EV, Auto-Cashout & Bankroll Math
- CrashGamesPlay — Crash Game RTP & House Edge: Aviator, Stake, BC.Game (2026)
- CrashGamesPlay — Crash Game Algorithm: Crash Point Formulas & Hash Math (2026)
- Bitcoin.com — Crypto Crash Games Guide
- SoMuchPoker — Best Crash Casino Games 2026: Aviator, JetX & More
- CasinoBeats — How to Play Crash Gambling