actfap

Crash Auto-Cashout Simulator

Set an auto-cashout target and see what a real session does to your bankroll, not just one round's odds.

This prices one strategy, a fixed target every round at a fixed stake, held for a stated number of rounds from a stated starting bankroll. It says nothing about a player who raises the target after a win, cashes out early on a hunch, or changes stake size mid-session; those are different strategies with their own simulations, not smaller versions of this one.

1% is the generic figure the crypto-native Crash category advertises, not a measurement of any named operator. Enter your own site’s published edge if you know it.

One round, and how often it gets this far

Return to player99%Auto-cashout target2.00xChance a single round reaches your target49.5%

Two different questions, both answered

Most Crash calculators print the streak bound below and call it a bust probability. It is a real number, and it is not the one a session actually runs on.

Chance of missing your cashout on all 200 rounds in a row

0.0000%

the streak bound other Crash calculators print

Chance this session reaches zero within 200 rounds

–

the real session figure, simulated below

The gap is not a rounding difference. A session does not need an unbroken losing streak this long to reach zero; ordinary variance across 200 independent rounds does that on its own, the same way a losing stretch spread across a poker or slots session does. Missing every one of 200 rounds in a row is a different, far rarer event than the bankroll running out.

What a 200-round session actually does to a $300 bankroll

Simulating…

Ruin rate against target

Session ruin rate runs 0% at the bottom to 100% at the top, against an auto-cashout target growing left to right from 1.2x to 10x. The dot marks your 2.00x target.

A higher target does not make the game safer. It wins less often for the same house edge: nothing about the target changes what the game is worth per round, only how that value is distributed across outcomes. The curve above usually climbs as the target rises. At a bankroll deep enough that no target in this range can ruin it within 200 rounds, it goes flat at zero instead, which is the same fact stated differently: there is simply nothing left for variance to spend.

Chance a single round reaches your target
49.5%
House edge on every round
1%
Rounds this session
200

Crash is built around a rising number and a voluntary stop, the mechanical shape of chasing: every round asks whether to take what is on offer or wait for more. There is no target that wins here. The sweep above is what shows that: raising it does not lower the house edge, it only changes how the same edge is distributed across outcomes. If Crash, or any rising-multiplier game, stops feeling like a choice you can walk away from, the tools on our responsible gambling page are a real next step.

Written and reviewed by Alexander Korsager · last reviewed October 6, 2026 · corrections

The idea

What this tool shows

Crash asks one question every round: cash out now, or let the multiplier keep climbing. Set an auto-cashout target and the game answers a single-round question for you, will this round reach that target, but the question a player actually has is different: playing that target for a few hundred rounds in an evening, what does it do to the money they brought.

  1. Every other Crash calculator we found answers the single-round question, or a streak version of it, the chance of missing the cashout on N rounds in a row, and calls that a bust probability. It is a real number, and it is not the number a session actually runs on. Missing an unbroken streak of, say, 200 rounds in a row is astronomically unlikely at almost any target, which is why that figure reads as near zero however the target is set. A real session does not need an unbroken miss streak to end at zero: ordinary variance, a string of losses interrupted by wins too small to recover them, does that on its own. This page runs an actual session, thousands of times, and reports how often that happens.

  2. Set a target, a house edge, a stake, a round count and a bankroll below, and the session simulator does the rest.The streak bound the rest of the category prints is shown too, next to the real figure, with the gap between them stated as a number.

FAQ

Questions players ask

What is the house edge in Crash?

It depends on the site, and this page does not attribute a figure to any named operator because we could not find one publicly cited with a source worth linking. The generic figure the crypto-native Crash category advertises, and the one this page opens on, is 1%. The house edge field is editable because operators differ; if you know the figure your site publishes, use it instead.

Why is the bust probability on this page so different from other Crash calculators?

Because it is answering a different question. The figure most Crash calculators print is the chance of missing your cashout on every one of N consecutive rounds, a genuine number, shown on this page too, and one that reads close to zero for any real session length: at the shipped defaults, a 2x target and 200 rounds, it is effectively 0%. The real session ruin rate this page's simulator reports, about 4.9% at those same defaults, comes from something that number cannot see: a session does not need an unbroken miss streak to reach zero, ordinary variance across 200 independent rounds does it on its own, the same way a losing stretch spread across a poker session or a slot session does. The two are not two estimates of one thing. They are different questions, and the session figure is the one that answers what an evening of play actually risks.

Does a higher auto-cashout target make Crash safer?

No, and the sweep on this page is built to show why in one picture: ruin rate rises as the target climbs, at the same house edge, stake, round count and bankroll. A higher target wins less often, and the house edge stays exactly where it was; nothing about the target changes what the game is worth per round. A cautious-looking 1.5x target and an aggressive 10x one are the same bet at different volatility, not a safe version and a risky one.

Is there a strategy that beats Crash?

No. Every round is an independent draw at a fixed house edge, whatever target is set, and no auto-cashout rule, staking pattern or timing changes that edge. What a target changes is how the same edge is distributed across outcomes, more small wins at a low target, rarer larger ones at a high target, which is what the sweep and the outcome percentiles on this page are for: seeing that trade rather than a promise that one setting is the winning one.

What does the streak bound this page shows actually mean?

It is the probability that a fixed per-round win chance fails on every single one of a stated number of consecutive rounds, the standard formula behind the bust figure most Crash calculators print. It says nothing about a bankroll that survives a miss and keeps playing, which is what an actual session does, so it is a bound on one narrow kind of bad luck rather than an estimate of how a session usually ends. It is shown here so a reader who has seen it elsewhere can see exactly what it does and does not answer, next to a figure that does answer the session question.

Methodology

How this is calculated

Two different quantities, computed two different ways. probabilityOfReaching gives the closed-form chance that one round reaches the target multiplier, (1 - house edge) / target, the standard relationship a fair Crash payout is built from.

The streak bound shown beside the session result is the same per-round chance raised to a power, (1 - winChance)^rounds, the probability of missing every single one of the stated rounds in a row. Neither of those is simulated.

The session result is: this page runs the auto-cashout strategy, the same target every round, for the stated number of rounds, 5,000 independent times, from a fixed seed, so the same inputs always redraw the same 5,000 sessions, on this page load or any other, and reports the share that reached a bankroll of zero along the way, plus the percentile spread of where those sessions finished.

The target sweep further down runs the identical simulator at nine target multipliers, holding house edge, stake, rounds and bankroll fixed, at a reduced trial count sized for a responsive redraw rather than the panel's own precision.

Common misreadings4
  • A low session ruin rate at one target is not a recommendation to play that target. It describes what the stated inputs do to the stated bankroll over the stated number of rounds, and nothing here says that session is worth playing in the first place.
  • The streak bound reading near zero is not evidence that the game is safer than the session figure suggests. It is answering a different question, one an unbroken losing streak this long almost never asks in practice, not a smaller or more precise version of the session ruin rate.
  • A higher target's lower ruin rate at the same bankroll is not a free improvement. The percentile spread widens as the target rises, which the outcome bands on this page show directly: fewer sessions bust, and the ones that do not bust vary more.
  • Ending the streak-bound comparison at a specific round count is not a claim about how long anyone should play. It matches the round count in the session fields so the two figures answer the same session; change the rounds field and both move together.
Assumptions6
  • Every round is an independent draw. The crash point on one round carries no information about the next, which is the standard assumption behind (1 - house edge) / target and is what makes the per-round formula and the session simulator agree on the single-round case.
  • The stake is fixed for the whole session. A player who resizes the bet after a win or a loss is playing a different game, the same distinction risk-of-ruin's own assumptions draw for a fixed-stake model generally.
  • A round that would take the bankroll below zero stops the session there rather than going negative; the reported final bankroll floors at zero, and a session that reaches zero counts as ruined regardless of how many rounds were left to play.
  • 5,000 simulated sessions per redraw, run in your browser from a fixed seed, so the same settings always produce the same picture, on this visit or any other. The trial count is the same order of magnitude the site's other session simulators use, sized to stay responsive on a mid-range phone.
  • The target sweep runs 500 sessions per point rather than 5,000, the same reduced-trial-count trade bankroll-calculator's own preset sweep makes, because nine points at full trials would multiply the per-redraw cost by nine for a curve whose shape does not need the panel's own precision.
  • The streak bound treats the stated round count as the streak length: the probability of missing the cashout on every one of that many consecutive rounds. It is the same formula, and the same question, whatever calculator prints it.
Limits4
  • 5,000 sessions is a sample, not the exact figure. A different seed would land the ruin rate at the shipped defaults a few tenths of a point away from what is shown here; the seed this page uses is fixed, so the same inputs never redraw differently, but the sample itself is still an estimate of the underlying probability, not the probability itself.
  • The streak bound answers a genuinely different, narrower question than the session ruin rate, and the two are never estimates of the same thing. A session can reach zero through an ordinary run of losses spread across many rounds without ever producing the exact unbroken miss streak the bound is asking about, which is why the bound reads close to zero for real session lengths while the session figure does not.
  • Only the fixed-target, fixed-stake strategy is modeled. A martingale-style stake progression layered on top of an auto-cashout target is a different, unmodeled strategy; this site's Martingale simulator prices a doubling progression on its own terms rather than combined with a Crash target.
  • The house edge is whatever is typed into the field. This page does not verify any operator's actual edge and states plainly, in the FAQ, that it could not find a specific operator's figure worth citing; the shipped 1% preset is the crypto-native category's advertised generic figure, not a measurement of a named site.
Evidence

probabilityOfReaching and probabilityOfMissingEveryRound are covered in math.test.ts, which checks probabilityOfReaching against its closed form at hand-checkable points (a 3% house edge at a 2x target gives exactly (1 - 0.03) / 2), checks that both functions strictly decrease as their respective inputs (target, streak length) grow, and pins that feeding one function's output into the other produces a different, independently checked number, the regression guard for the whole reason the module exports two functions with two names rather than one with a mode flag. simulateAutoCashoutSession is checked against the same closed form from the other direction: isolating a single round (rounds = 1) and averaging its simulated net result across ten seeds lands within three pooled standard errors of the exact theoretical mean, winP times the payout multiplier minus 1, derived directly from probabilityOfReaching. sweepAutoCashoutTargets is checked for determinism (the same inputs, targets and seed reproduce the identical sweep) and for the qualitative property the page's own argument depends on, that ruin rate rises as the target climbs at fixed house edge, stake, rounds and bankroll, checked at points far enough apart that the reduced trial count's sampling noise cannot reverse the order.

Gambling can cause serious harm. The National Problem Gambling Helpline is free, confidential and open 24/7, call or text 1-800-MY-RESET (1-800-697-3738), or chat at ncpgambling.org/chat.

Related tools

Try next

We’d like to use privacy-friendly analytics (Google Analytics, Vercel Analytics and Speed Insights) to see which tools and pages are actually useful, so we can improve them. Nothing is sold, and nothing loads unless you say yes. See our privacy policy for details.