Strategy · the math

Crash gambling strategy: the honest, mathematical version

Almost every 'crash strategy' page sells you martingale, Fibonacci or a magic cashout multiplier. This one proves, with school-level algebra, that none of them changes your expected loss — and shows what strategy genuinely can and can't do.

The uncomfortable claim

Here is the thesis, stated plainly so there is no ambiguity: no crash gambling strategy can produce a long-run profit, because every possible cashout target has exactly the same negative expected value — and that value is minus the house edge. Cashing out at 1.2×, 2×, 10× or 500× makes no difference to your expected return per dollar. The only things that change your actual money are which game you play (its house edge) and how much you wager.

This is not cynicism or a disclaimer. It is a provable fact about how provably-fair crash games are constructed, and you can confirm it yourself with the simulator on our homepage. Let's prove it.

The expected-value proof

Recall how the crash point is generated. For a game with return-to-player r (so r = 0.99 at a 1% edge), the probability that a round reaches at least a target multiplier m is:

For example, at a 1% edge (r = 0.99) and a 2× target, P = 0.99 / 2 = 0.495 — a 49.5% chance. Now compute the expected value of staking $1 with an auto-cashout at m. With probability r/m you win, turning $1 into $m (a profit of m − 1); with probability 1 − r/m you lose your $1:

So EV = r − 1 = −(house edge). At a 1% edge, EV = −0.01 per $1. At 3% (Aviator), EV = −0.03. Every dollar you wager is expected to return the house-edge fraction less than it cost — forever, on average.

Why the multiplier completely cancels

Look at the algebra again: the m in the numerator from the payout and the m in the denominator from the probability annihilate each other. The target multiplier vanishes from the final expression. That is the whole game, mathematically:

• Low targets (e.g. 1.2×): you win often but win little. High win rate, tiny payouts.
• High targets (e.g. 50×): you win rarely but win big. Low win rate, large payouts.
• Both: identical expected value of −(house edge).

Choosing a cashout multiplier is choosing your variance — how bumpy the ride is — not your expected destination. Try it: in the simulator, leave the edge fixed and change the target from 1.5× to 2× to 10×. The "Expected value / $1" readout never moves.

Why martingale fails (with the brutal number)

Martingale is the most-pushed crash "strategy": pick a low target like 2×, and double your bet after every loss so that a single win recovers all prior losses plus one base unit. It feels unbeatable because it produces a steady stream of small wins. The catastrophe is hidden in the tail.

At a 2× target with 97% RTP, your win chance per round is about 48.5%, so your chance of losing is about 51.5%. The probability of a losing streak of length k is roughly 0.515^k. A 10-loss streak happens about once every ~1,300 rounds — which, on a game you can play hundreds of times an hour, is not rare; it's a Tuesday.

And here is what a 10-loss streak demands. To stay in the martingale you must double each time:

To recover a single unit of profit after ten losses you must put 1,024× your base bet at risk. A $1 base means a $1,024 bet — to net $1. Two things then kill you, guaranteed, with finite money: you hit the table maximum, or you run out of bankroll. Either way the streak that was always coming wipes out hundreds of small wins in one stroke. Martingale doesn't beat the edge; it concentrates your inevitable loss into rare, enormous events. Switch the simulator's staking mode to "Martingale" and watch the ruin rate climb.

Fibonacci, D'Alembert, anti-martingale — same fate

Every progression system is a different schedule for sizing bets after wins and losses. None of them touches the per-round EV, because EV is fixed at −(house edge) before any staking rule is applied. The expected value of a sequence of bets is just the sum of each bet's expected value, and each one is negative.

• Fibonacci: a gentler progression than martingale — slower to escalate, but still escalates into ruin on a long enough streak.
• D'Alembert: add one unit after a loss, subtract one after a win. Smoother variance, identical negative expectation.
• Anti-martingale / Paroli: double after wins instead. This caps losing streaks but means one loss erases a winning run. Same EV.

The pattern is total: staking systems redistribute when and how you lose, never whether you lose.

What "strategy" can actually change

Three things are genuinely within your control, and only three:

1. The house edge you play against. This is the single biggest lever. A 1% game (Stake, BC.Game, Bustabit) costs a fifth of a 5% game (Rollbit X Crash) on the same turnover. See the comparison.
2. Rakeback. The one promotion that lowers your effective edge: edge × (1 − rakeback). Gamdom (up to 60%) and Shuffle (5–15%) reduce the bleed — but never to zero.
3. How much you wager and for how long. Your expected loss is turnover × edge. Halve your turnover, halve your expected loss. Set a loss limit and a time limit and stick to them.

Notice what's not on the list: the cashout multiplier, the betting progression, "reading" recent rounds, or any app. Those are variance and noise, not edge.

So what should I cash out at?

Since EV is identical everywhere, the cashout target is purely a variance preference, and the honest answer is: pick the experience you can afford to lose calmly.

• Low multipliers (1.2×–1.5×) give frequent small wins and a slow, steady decline — gentler variance, but you'll bust at 1.00× often enough to sting, and the edge still grinds you down.
• Around 2× is the classic "coin-flip-ish" target — roughly even-money feel at 49–48.5% win rate.
• High multipliers are lottery-style: long droughts punctuated by rare big hits. Maximum variance, same expected loss, far higher chance of ruin before a hit lands.

Frequently asked questions

This guide is educational and based on the standard provably-fair crash model. It is not betting advice and does not suggest any way to profit from gambling — there isn't one. 18+. If gambling is causing harm, see our help resources.