Stake Crash RTP & House Edge: The Math Behind the Multiplier
By LiveCasinoData editorial team · Updated 2026-08-28 · 14 min read
Stake Crash pays a published RTP of 99%, a 1% house edge, and that edge is identical no matter what multiplier you plan to cash out at. The crash point follows P(crash ≥ x) = 0.99/x, and across 2,794 rounds tracked so far the average realized return sat at 99.55%. The distribution has a heavy tail: most rounds crash low, a small number crash enormous, and no fixed target changes your long-run expectation.
Key takeaways
- Stake Crash's published RTP is 99.00%, a 1% house edge, the lowest of any game we track.
- Every fixed auto-cashout target carries the exact same -1% expected edge, from 1.50x to 100x, because the crash formula P(crash ≥ x) = 0.99/x cancels out at every x.
- Across 2,794 rounds tracked since 2026-08-27, the average realized round RTP was 99.55%, close to the 99% theoretical figure.
- The crash-point distribution has a heavy tail: median multiplier is 2.02x, but 3 of 2,794 rounds so far crashed above 1,000x, which is why the 11.02x mean is not a useful 'typical' number.
- Bucket shares tracked so far (under2 49.43%, 2to5 30.28%, 5to10 9.41%, 10to50 8.41%, over50 2.47%) sit within about a point of the theoretical model.
Stake Crash pays back 99% of everything wagered, a 1% house edge that is fixed by a single mathematical formula rather than a wheel layout or paytable. Unlike a segmented game show, there is no menu of bets with different returns: every player is exposed to the same curve, and our live tracker has logged 2,794 rounds so far with an average realized return of 99.55%.
What is the RTP of Stake Crash?
Stake Crash's theoretical RTP is 99.00%, giving a house edge of 1.00%, among the lowest of any game type we track. That number comes from lib/games.ts as expectedRTP and matches what Stake states in the client's game rules for this title.
Unlike a wheel game where different bet spots carry different RTPs, Stake Crash has only one RTP figure, because there is only one underlying random process: a multiplier curve that climbs from 1.00x until it crashes at a random point. A bet wins only if the player cashes out before that happens. There is no separate "bet type" menu the way Crazy Time has number bets and bonus bets; the entire game is one continuous mechanic, and the 99% figure applies to it in full.
Stake is a different kind of provider than the live-dealer studios behind most games on this site. It is one of Stake's in-house "Stake Originals," a provably fair, server-driven game with no dealer, no camera, and no physical wheel. Curacao eGaming licenses the platform. The how to play guide covers the client interface in more depth; here the focus is purely the math.
How is the 99% RTP actually calculated?
The crash point follows a survival function: the probability that a round reaches at least multiplier x is P(crash ≥ x) = 0.99 / x, for any x at or above 1.00x. That single line generates the entire game.
Plug in a few values and the shape becomes clear. The probability of reaching 2x or higher is 0.99/2, or 49.5%. Reaching 5x or higher is 19.8%. Reaching 10x is 9.9%, 50x is 1.98%, 100x is 0.99%, and 1,000x is 0.099%, or about one round in a thousand. Each step down in probability matches a proportional step up in payout, which is exactly what a fixed house edge requires.
| Target multiplier | P(crash ≥ x) | Meaning |
|---|---|---|
| 2x | 49.5% | Roughly a coin flip |
| 5x | 19.8% | About one round in five |
| 10x | 9.9% | About one round in ten |
| 50x | 1.98% | About one round in 50 |
| 100x | 0.99% | About one round in 101 |
| 1,000x | 0.099% | About one round in 1,010 |
This formula is also why the 1% edge is baked in rather than set by a paytable. Multiply the payout at any target x by the probability of hitting it, subtract the probability of missing, and the expected value always resolves to -1%, which the next section works through directly.
Does the RTP change based on your cashout target?
No. Every fixed cashout target carries exactly the same -1% expected value, whether it is set at 1.50x or 100x, because the 0.99/x formula cancels itself out algebraically.
The expected value of a fixed target x is: EV = P(≥x)·(x-1) minus P(<x)·1, the stake lost on a bust. Substitute P(≥x) = 0.99/x and the x terms cancel, leaving EV = 0.99 - 1 = -1% every time, regardless of what x you chose. A player who always cashes out at 1.5x has the same theoretical long-run edge against them as a player who always waits for 100x. Neither approach is mathematically "safer" in terms of expected return; they differ only in how the losses and wins are distributed across sessions.
This is a genuine point of confusion for players coming from wheel games, where different bet spots really do carry different RTPs (see how Crazy Time's bet table ranges from 94.33% to 96.08% depending on the spot). Stake Crash has no such spread. There is one edge, applied uniformly, no matter what number you type into the auto-cashout field. Our strategy page covers what actually differs between low and high targets: bankroll behavior, not expected return.
What are the crash-point probabilities and payouts by range?
Because Stake Crash has no fixed betting spots, we bucket every round's crash point into five ranges for the site's wheel-style widgets, and each bucket has both a theoretical share and an observed one from our tracker.
| Bucket | Crash range | Theoretical share | Observed share (2,794 rounds) | Rounds observed |
|---|---|---|---|---|
| under2 | below 2.00x | 50.5% | 49.43% | 1,381 |
| 2to5 | 2.00x to 4.99x | 29.7% | 30.28% | 846 |
| 5to10 | 5.00x to 9.99x | 9.9% | 9.41% | 263 |
| 10to50 | 10.00x to 49.99x | 7.92% | 8.41% | 235 |
| over50 | 50.00x and above | 1.98% | 2.47% | 69 |
The 10to50 and over50 buckets are flagged in our data as bonus_triggered, our tracker's way of marking the "big win" rounds since there is no separate bonus round for the tracker to log. Every bucket sits within roughly a point of its theoretical share; over50 runs the furthest off at +0.5 points, unsurprising with only 69 hits recorded and a data set that is barely a day old. As more rounds accumulate, expect these gaps to narrow rather than widen if the model holds.
Within each bucket, the average and median crash points diverge sharply the higher you go, a preview of the tail problem discussed next.
| Bucket | Average multiplier | Median multiplier |
|---|---|---|
| under2 | 1.37x | 1.32x |
| 2to5 | 3.00x | 2.78x |
| 5to10 | 6.80x | 6.50x |
| 10to50 | 20.35x | 16.87x |
| over50 | 286.89x | 94.32x |
Why is "average crash multiplier" a misleading number?
Because the crash-point distribution has a heavy tail, and its overall mean is dragged around by a handful of enormous rounds rather than representing anything typical. Across all 2,794 rounds tracked so far, the mean multiplier is 11.02x. The median is 2.02x. That is not a rounding gap; it is the defining feature of the game's math.
Since P(crash ≥ x) is proportional to 1/x, the theoretical distribution has no stable, finite average in the way a coin flip or dice roll does. In practice, on any real batch of rounds, one or two outsized crashes end up carrying a wildly disproportionate share of the "total multiplier" if you summed every round. Our data shows this directly: 3 of 2,794 rounds, about 0.11%, crashed above 1,000x, including the current high of 3,234.05x on 2026-08-27. Remove those 3 rounds and the average would collapse by more than half.
This matters practically because "average crash multiplier" sounds like it describes a typical round, and it does not. The median, 2.02x, is the far better description of what usually happens: half of all rounds crash below roughly 2x, and half above. If you want to reason about a "normal" outcome, use the median or the bucket shares above; treat any quoted average multiplier as a tail-weighted artifact, not a target to plan around. This is the same reason a casino's average jackpot payout tells you little about your own likely night.
How volatile is a low cashout target versus a high one?
Volatility rises sharply as the target multiplier increases, even though the expected return stays fixed at -1% throughout. A target near 2x behaves like a near coin-flip bet: it wins close to half the time (49.5% theoretical, 49.43% under the under2 bucket observed so far, though that specific figure describes crashes below 2x rather than cashouts exactly at 2x). A target above 50x behaves like a lottery ticket: it wins on roughly 1 in 50 rounds, but the payout when it hits is large enough to offset the long runs of losses in between.
Our observed gap data shows this rhythm concretely. The average gap between over50 (50x-plus) hits so far is 40.0 rounds, with the longest stretch recorded at 193 rounds without one. The average gap between 10x-plus hits is much shorter, 9.2 rounds, with a longest gap of 54 rounds. Because tracking only began on 2026-08-27, treat these as early figures likely to shift as more history builds, not settled long-run averages.
| Cashout style | Approx. win rate | Typical payout when it hits | Character |
|---|---|---|---|
| Near 2x | ~49.5% | Small, near-even | Low volatility, frequent small results |
| Near 5x-10x | ~10-20% | Moderate | Medium volatility |
| Near 10x-50x | ~2-8% | Larger, less frequent | High volatility |
| Above 50x | ~2% | Large, rare | Very high volatility, long droughts possible |
None of these rows differ in expected value; they only differ in how the same -1% edge is distributed across wins and losses. A player chasing 50x-plus targets should expect long stretches with no payout at all, interrupted occasionally by a result large enough to swing an entire session. The cashout strategy guide walks through bankroll sizing for each style in more depth.
What is the maximum multiplier in Stake Crash?
The platform ceiling configured in our system is 1,000,000x, though that has never come close to appearing in the rounds we have tracked. The highest crash point recorded so far is 3,234.05x, on 2026-08-27 at 15:41 server time, followed by 3,116.66x earlier the same day and 2,593.79x that evening.
| Crash point | Date recorded |
|---|---|
| 3,234.05x | 2026-08-27 15:41 |
| 3,116.66x | 2026-08-27 12:04 |
| 2,593.79x | 2026-08-27 19:21 |
| 993.39x | 2026-08-27 14:54 |
| 953.70x | 2026-08-27 17:44 |
Because the crash formula has no hard mathematical ceiling below the platform's configured cap, in theory a round could crash at any multiplier above 1.00x, with the probability of very large values shrinking continuously rather than stopping at a fixed line the way a wheel game's segment count does. With just over a day of tracked data, these figures represent an early snapshot; a longer sample would likely surface additional multi-thousand-x rounds and push the recorded maximum higher, since 3 of 2,794 rounds already cleared 1,000x.
What are the betting limits?
Stake Crash bets are placed in cryptocurrency, including BTC, ETH, USDT, LTC, and other currencies Stake supports, rather than in fixed fiat units. There is no published fixed minimum or maximum bet size the way a live table might advertise; limits are effectively set by currency denomination and account status rather than a table rule, and different players in the same round can be staking very different amounts simultaneously.
Rounds are multiplayer: many players bet into the same climb at once, and the raw feed tracks fields like bettorsCount and totalBetAmount for the round as a whole, alongside each individual winner's own cashout point. Our database stores only the round-level crash point, not each player's personal stake or cashout, so per-player betting patterns are outside what this tracker can show. What every player in a given round shares is the same crash point and, by extension, the same underlying 99% RTP.
Does the observed data match the theoretical RTP?
Yes, closely, though the sample is still young. Since tracking began on 2026-08-27, our tracker has logged 2,794 Stake Crash rounds, all within roughly the first day and a half. The average realized round RTP, drawn from the round_rtp field in the raw feed and averaged across every round, is 99.55%, close to the 99.00% theoretical figure and a useful independent check because it comes from the real-money cash-return side of the feed rather than just the crash-point distribution.
| Measure | Theoretical | Observed (2,794 rounds) |
|---|---|---|
| RTP | 99.00% | 99.55% (average round_rtp) |
| under2 share | 50.5% | 49.43% |
| 2to5 share | 29.7% | 30.28% |
| 5to10 share | 9.9% | 9.41% |
| 10to50 share | 7.92% | 8.41% |
| over50 share | 1.98% | 2.47% |
Rounds reaching at least 10x totaled 304, or 10.9% of all rounds tracked, close to the 9.9%-plus theoretical figure once the 10to50 and over50 buckets are combined. Rounds reaching 100x or higher totaled 33, about 1.2%, again in the expected neighborhood. These are the kinds of numbers a fair 0.99/x curve should produce over a few thousand independent rounds, and as of 2026-08-28 nothing in the sample looks inconsistent with the published model. Because this is roughly one day of history rather than the 90-day windows we use for older games, treat every figure here as an early read that will firm up, not a final verdict.
How do you read our Stake Crash RTP tracker?
The live stats page presents Stake Crash through the same widget set used for wheel games, adapted to a single continuous multiplier rather than fixed segments, and a few things are worth understanding before you use them.
The bucket panel compares observed under2/2to5/5to10/10to50/over50 shares against the theoretical 0.99/x model, the same numbers tabulated above. Expect short windows to wander further from theory than the full 2,794-round sample does; that is normal sampling noise, not a sign the game is misbehaving.
The gap or drought tracker counts rounds since a bucket, most usefully over50 or 10x-plus, last appeared. A long gap does not make a large multiplier "due." Each round's crash point is generated independently through the provably fair seed process, with no memory of what happened before it, so the probability of a 50x-plus round stays 1.98% regardless of how many rounds have passed since the last one. We show the gap because a 193-round drought in real data is the clearest cure for the gambler's fallacy that exists.
The probability tool answers a genuinely useful planning question: given a target multiplier, how many rounds do you need to watch before there is a good chance of seeing at least one hit. That calculation follows directly from the 0.99/x survival function and is legitimate math, unlike reading a drought as a signal.
Is Stake Crash's RTP the same as other crash-style games?
Not necessarily, and the RTP itself is specific to how each provider tunes its curve, even though the general mechanic (a rising multiplier that can crash at any moment) is shared across the category. Stake Slide, another Stake Originals title, uses a related multiplier structure but its own published figures. Cash or Crash borrows the crash theme for an Evolution game show format with a wheel and a host, a genuinely different mechanic behind a similar name, worth checking if you assumed the two worked the same way. Lightning Dice is a different RNG multiplier game entirely, dice-based rather than curve-based.
What is specific to Stake Crash is the 0.99/x formula and its resulting 99% RTP, both stated in Stake's own game rules and confirmed independently by our 99.55% observed round RTP. Our sitewide RTP monitor tracks this figure alongside every other game on the platform, and it is worth checking there periodically as more history accumulates behind the number.
No fixed cashout target beats the underlying 1% edge, and no amount of watching gaps between big multipliers changes the odds of the next round. If you play Stake Crash or any multiplier game, treat it as entertainment with a fixed statistical cost, set a loss limit before you start, and use resources like our responsible gambling page if a session stops feeling optional.
Frequently asked questions
What is the RTP of Stake Crash?
Stake Crash has a published RTP of 99.00%, meaning a 1% house edge built into the crash-point formula. That figure is the same for every player and every cashout target; it is not adjustable by the operator the way bonus terms are. Our tracker's average realized round RTP, 99.55% across 2,794 rounds so far, sits close to that published number.
Does the cashout multiplier you pick change the RTP?
No. Because the crash formula is P(crash ≥ x) = 0.99/x, the expected value of any fixed auto-cashout target works out to the same -1% edge, whether you target 1.5x or 100x. What changes is volatility, not expected return: low targets win often for small amounts, high targets win rarely for large amounts, but both average to a 1% loss per unit staked over the long run.
What is the highest multiplier possible in Stake Crash?
Our game configuration lists a platform ceiling of 1,000,000x, though that has never been observed on our tracker. The highest crash point recorded so far is 3,234.05x on 2026-08-27. The crash-point formula theoretically allows any value above 1.00x, with the probability of reaching very large multipliers shrinking as 0.99 divided by that multiplier.
Is Stake Crash provably fair?
Stake publishes a hashed server seed before each round and reveals it afterward, combined with a client seed and nonce, so a result can be independently verified once the round ends. This is the actual trust mechanism for a game with no wheel or dealer to watch; there is no pattern-based way to predict a crash point in advance, since each round is generated independently of every prior one.
How volatile is Stake Crash compared to a wheel game?
It depends entirely on your cashout target rather than a fixed bet type. Cashing out near 2x wins close to half the time for a small gain, similar to an even-money wheel bet. Waiting for 50x or higher wins only about 2% of rounds tracked so far, but the payout is proportionally larger, a volatility profile closer to a jackpot bet than a standard casino wager.