We Simulated 475 Million Crash Rounds — Every Cash-Out Target Costs the Same 1%

Crash is the emblematic crypto-casino game: a multiplier climbs, you cash out before it crashes, and the entire strategy debate — every Reddit thread, every YouTube “method” — is about where to set the cash-out. So we measured it. 475,353,972 simulated rounds under the standard provably-fair crash distribution with a 1% house edge, one run, fixed seed, every figure below reported as it came out.

The one-line result: the cash-out target does not change what crash costs. It changes everything else.

The distribution we simulated

Major provably-fair crash implementations construct the crash point so that the probability of the round reaching your target x is proportional to 1/x, scaled by the house edge. With a 1% edge:

• P(crash point ≥ x) = 0.99/x for any target x ≥ 1
• rounds that land below the 1.01x cash-out floor are the instant busts — about 2% of all rounds
• expected value of cashing out at any target c: (0.99/c)(c−1) − (1 − 0.99/c) = −1%, independent of c

That last line is the claim every crash strategy implicitly denies, so it is the one we verified hardest.

Measured: every target, same price

Two million rounds per target (twenty million for the two rarest), fresh generator per row:

Auto-cashout	Measured win rate	Theory (0.99/x)	Measured EV per bet	Longest losing streak
1.01x	98.0236%	98.0198%	−0.9962%	3
1.1x	90.0042%	90.0000%	−0.9954%	7
1.5x	66.0062%	66.0000%	−0.9907%	12
2x	49.5052%	49.5000%	−0.9896%	23
3x	33.0755%	33.0000%	−0.7737%	37
5x	19.7944%	19.8000%	−1.0280%	54
10x	9.8864%	9.9000%	−1.1355%	114
20x	4.9694%	4.9500%	−0.6130%	237
50x	1.9706%	1.9800%	−1.4725%	639
100x	0.9880%	0.9900%	−1.1955%	1,331
1000x	0.0979%	0.0990%	−2.0550%	8,940

Win rates track theory to the third decimal. The per-bet EV column scatters around −1% exactly as sampling noise predicts — the scatter widens at rare targets because a single 100x win moves the average, which is why those rows got 20 million rounds. Nothing in 475 million rounds found a target that costs less than the others.

The rightmost column is the real finding. A 2x player’s worst losing streak was 23 rounds. A 100x player’s was 1,331. Same game, same price, completely different experience — and completely different bankroll requirements.

Same 1%, opposite sessions

We then played 50,000 sessions per target: $100 bankroll, $1 flat bets, stop at broke or 2,000 rounds.

Target	Went broke	Median rounds at bust	Still alive at 2,000	Ever doubled the bankroll
1.5x	1.1%	1,777	98.9%	0.01%
2x	6.3%	1,592	93.7%	0.91%
5x	32.8%	1,120	67.2%	20.74%
10x	50.3%	790	49.7%	37.39%
100x	83.0%	200	17.0%	41.39%

Read the two ends of that table together. The 1.5x grinder almost never goes broke in an evening — and almost never gets anywhere: 5 sessions in 50,000 ever saw $200. The 100x hunter doubled the bankroll at some point in 41.39% of sessions — and was broke within 2,000 rounds 83% of the time, typically after about 200. The expected loss is identical in every row. Your target choice is a choice of failure mode, not of profitability.

Martingale, measured again

Our betting-systems study already showed progression systems just repackage the house edge. Crash’s near-coin-flip 2x target is Martingale’s natural habitat, so we gave it its own run: $1 base bet, double after every loss, $100 bankroll, 500-round sessions, 100,000 sessions.

• 85.55% of sessions were wiped out — the bankroll could not cover the next doubled bet.
• The 14.45% that finished 500 rounds held a median of $352.

A $100 bankroll covers six consecutive doubled losses; the seventh needs $64 more than remains. Our 2x streams recorded losing streaks up to 23. Over 500 rounds, meeting a seven-loss streak is not bad luck — it is the arithmetic default, and the occasional $352 finishers are what keeps the system’s reputation alive.

How long until the big one

Median wait for the first appearance of each multiplier (10,000 trials each):

• ≥10x — median 7 rounds, mean 10
• ≥100x — median 69 rounds, mean 100
• ≥1000x — median 710 rounds, mean 1,013

The mean exceeding the median in every row is the signature of skew: half the time the 1000x arrives within ~710 rounds, but droughts run much longer often enough to drag the average past 1,000. “It hasn’t hit in ages, so it’s due” fails here for the same reason it fails everywhere: each round’s distribution is identical and memoryless, which our streak data demonstrates at scale.

The Aviator variant: same run at a 3% edge

Spribe’s Aviator — the most-played crash game in the world — publishes a 97% RTP rather than 99%. We re-ran the entire suite at a 3% edge (451,614,756 rounds, same seed): every cash-out target cost the same ~3% per bet (2x win rate measured 48.5099% against a 0.97/x theory of 48.5%), the streak lengths were essentially unchanged (23 at 2x, 1,331 at 100x), and Martingale’s wipeout rate rose to 88.88%. A tripled edge triples the price and changes nothing about the structure — which is the point.

Methodology

• Distribution: crash point M = 0.99/U, U uniform on (0,1) — the standard provably-fair construction via inverse transform sampling; P(M ≥ x) = 0.99/x; house edge 1%. This models the math the major crypto originals publish (roughly 99% RTP for Crash); individual operators’ exact implementations can differ in details like bet caps and floors.
• Seed: 20260821, Python random.Random, one deterministic run per table row. No reruns, no selection.
• EV rows: 2,000,000 rounds per target; 20,000,000 for 100x and 1000x.
• Sessions: $100 bankroll, $1 flat bets, stop at bankroll < $1 or 2,000 rounds; 50,000 sessions per target.
• Martingale: $1 base, double after loss, reset after win, $100 bankroll, 500-round cap, 100,000 sessions.
• Total simulated rounds: 475,353,972.

Journalists and researchers may reuse any figure on this page with attribution and a link. The comparison across operators that actually run crash games — judged on licensing, fairness documentation and payout terms rather than on the multiplier math, which is the same everywhere — is in the best crash game casinos; the game’s mechanics are in crash games explained; and where crash’s ~1% edge sits against every other casino bet is in house edge statistics.

These are expected values, not predictions of your next session — variance dominates individual outcomes, which is most of what this page measures. Nothing here makes crash profitable; the 1% is the price everywhere. Play only what you can afford to lose, where it is legal for you.