Anti-Martingale (Paroli) in Crash Games: Why It Cannot Beat the House Edge

It is sold as the safe, disciplined way to ride a winning streak. The maths is far less flattering. Like every staking system, it reshapes how your losses arrive without changing how much you lose.

🎮 How the Anti-Martingale works

The rule is simple: start at a base unit, increase the stake after a win, and drop back to base after a loss. That is the precise opposite of the Martingale, which doubles after losses. Because the doubling logic assumes a roughly even-money bet, the natural target in a crash game is 2.00x, where a win returns twice your stake.

The names are interchangeable. “Anti-Martingale”, “Paroli” and “Reverse Martingale” all describe the same positive progression. The one distinction worth knowing is whether the system is capped, and that gives you three practical variants.

🔢 Why no staking order can change the edge

No betting system can turn a negative-edge game positive. This is a proven result, not an opinion. Your expected loss over a session equals the house edge multiplied by the total amount you stake, and a staking system only changes how that total is distributed across rounds.

The mathematics is settled. Epstein’s theorem states that any and all betting systems lead to the same expectation per unit wagered; Ethier calls the same result the conservation of fairness, the principle that a sequence of unfavourable bets cannot be rearranged into a favourable one. Doob’s optional stopping theorem is the formal probability-theory version, and the one-line intuition is linearity of expectation: the expected value of a sum of bets is the sum of their expected values, and every term here is negative. We derive the full result on our crash gambling maths page; here we just work the crash-specific arithmetic.

The crux for crash games is that the bet is never a coin flip. At a 2.00x target with 97% RTP, the chance of reaching it before the crash is 0.485, not 0.5, with a 0.515 chance of crashing first. More generally the probability of reaching any multiplier m is 0.97 divided by m, so the expected value of a round is 0.97 minus 1, which is minus 3p per pound at every target. The multiplier cancels out completely.

📊 The 100-round proof

Put numbers on it. Take 100 rounds at the same 2.00x target and a one-pound base stake, and compare flat betting against the two progressions. The proportional loss is identical in every case; only the amount churned changes.

The Anti-Martingale figure is an illustrative model: because completed streaks are rare, the player sits at the base stake most of the time, so the average stake works out at roughly 1.45 pound a round and the total churn at about 145 pound. The absolute loss is larger than flat betting only because the system pushes more money through the game, not because the rate is any worse.

The appeal is the streak, so price it honestly. Three straight wins at a 0.485 win probability happens about 11.4% of the time. That run stakes one pound, then two, then four across three rounds, and the final four-pound stake returns eight, for a tidy +7 pound cycle. But the cycle is just three independent minus-3% rounds: the rare 7-pound windfall is exactly offset over the long run by the far more common early failures. The streak is high variance, not positive expectation.

🧠 Why it feels safer than the Martingale

It feels safer because the downside is linear rather than exponential. The Martingale escalates during losses: ten losses in a row turn a one-pound base into a 1,024-pound stake, with 1,023 already gone, and it breaks the moment it hits a table cap or your bankroll. The Anti-Martingale escalates only during wins, so its worst case is simply losing the base bet over and over, a slow bleed instead of a cliff.

The comfort rests on documented biases. The hot-hand fallacy, the belief that a winning streak will continue, was established by Gilovich, Vallone and Tversky in 1985. Loss aversion, from Kahneman and Tversky, means a loss hurts roughly twice as much as an equivalent gain feels good, so a string of small base-bet losses is far easier to stomach than one big doubled-up loss even when the expected cost is identical. Xu and Harvey, writing in Cognition in 2014, even found that winners kept winning because they chose safer odds after a win, which manufactures the appearance of a streak.

🏆 How it compares to the other staking systems

Set the Anti-Martingale against every rival staking plan and the verdict never changes. The systems differ only in the shape of their variance and how much they churn; the expected loss is the same minus-3% of everything staked.

The Martingale is the mirror image of this system, and we debunk it in full in our guide to the Martingale in crash games. The honest alternative is the dullest line in the table: flat betting carries the same expected cost with the lowest variance and the smallest drawdowns, which is why we treat it as the rational baseline.

⚙️ Setting it up in the auto-bet panel, and why it still loses

Every major crash game automates the Anti-Martingale in two clicks. On Stake Crash you open the Auto tab, set On Win to “increase by 100%” and On Loss to “reset”, and add a stop-on-profit and a stop-on-loss. Aviator, JetX, Lucky Jet and Spaceman all expose the same increase-on-win and reset-on-loss controls, often with their own stop thresholds. Setting “increase by 100%” doubles the stake; any smaller percentage gives you a gentler progression variant.