Home › Casino Games Explained
Martingale, d'Alembert and Fibonacci all rest on the gambler's fallacy and fail to the same arithmetic. The flagship debunk of betting systems, with numbers.
By Anja Wiedenhöft · August 9, 2026 · 6 min read
Betting systems like Martingale and d'Alembert only change how much you stake and when, never the odds of any bet. Because each bet carries a fixed negative expected value and expectations add up, the total stays negative. Table limits and finite bankrolls guarantee escalating progressions eventually collapse. No system beats the house edge.
Betting systems are the most persistent myth in gambling: elegant-sounding staking rules that promise to convert a losing game into a winning one. None of them work, and the reason is not opinion — it is arithmetic. This is the flagship debunk. We will name the famous systems, walk through exactly what each one does, and show with numbers why every single one fails against a game that carries a house edge. Naming these systems is the only way to dismantle them; nothing here endorses any of them.
Start with the principle that decides the whole question. In any casino game the house has a fixed negative expected value for the player on every bet — roughly 2.70% per bet on European roulette, 5.26% on American, around 1.06% on baccarat banker, and so on. Expected value is the average result of a bet, and for the player it is always negative.
Now the key mathematical fact: the expected value of a series of bets is the sum of the expected values of each bet. Because each individual bet has a negative expectation, any sum of them — in any order, at any sizes — is also negative. A betting system only decides how much to stake and when . It never changes the odds of any single bet. So it can rearrange a pile of negative numbers, but it can never make the total positive. That single sentence is the death of every system below.
Almost every betting system smuggles in the same false assumption — the gambler’s fallacy , the belief that independent random events “balance out” in the short run, so that a loss makes a following win more likely.
They do not balance out. On a fair roulette wheel, if red has landed eight times in a row, the probability of red on the next spin is still 18/37 — exactly what it was before the streak. The wheel has no memory. Coins, dice, RNGs and wheels do not keep a ledger and do not owe you a correction. The “law of averages” people invoke is a misreading of the actual law of large numbers, which says results converge to the expected average over a huge number of trials — not that short runs self-correct. Every system that stakes more because a result is “due” is built on this fallacy, and the fallacy is simply false.
The most famous system. After each loss you double your bet, so a single win recovers all prior losses plus one unit of profit. Bet 1, lose; bet 2, lose; bet 4, win — you are up 1 unit. It feels like a guaranteed win. It is not.
Watch the bet sizes: 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1,024. A losing streak of ten — unremarkable over an evening — means your eleventh bet must be 1,024 units, and you have already lost 1,023, all to grind out a one-unit profit. Two things then guarantee failure:
Martingale does not remove risk. It converts many small, frequent, comforting wins into one rare, enormous, catastrophic loss — and because the underlying edge is negative, the trade is a loser on average. You are risking 1,023 to win 1, over and over, until the day the streak arrives.
The reverse — increase stakes on a winning streak, reduce them after a loss — is sometimes sold as safer. It changes the shape of your results (you now risk a lot during good runs) but not the average. Press your bets high enough and one loss surrenders the streak’s gains; the expected value per bet is still negative, so the sum is still negative. Rearranged risk, identical maths.
The d’Alembert system is the gentler cousin of Martingale. Instead of doubling, you increase your stake by one unit after a loss and decrease it by one after a win, on the theory that wins and losses will roughly balance and the staggered stakes will net a profit. This is the gambler’s fallacy in slow motion: it assumes losses will be “made up” by later wins at comparable stakes. They are not guaranteed to be, because the events are independent. The slower progression means slower ruin, not avoided ruin. Each bet still carries the same negative edge, so their sum is still negative — the gentler slope changes how quickly you get there, never the destination.
The Fibonacci system raises stakes along the 1, 1, 2, 3, 5, 8… sequence after losses; the Labouchère (cancellation) system crosses numbers off a list to target a set profit. Both are more elaborate than Martingale, and elaboration is exactly the trick — complexity disguises the same flaw. They rearrange the size and order of negative-expectation bets and run into the same walls: streaks, table limits and finite bankrolls. No amount of sequence design changes the expected value of a single spin, and the sum of negatives stays negative.
Betting systems feel like they work because most sessions are short, and in the short run they usually do produce a small win. Martingale wins a little on the vast majority of nights — that is exactly what makes it dangerous. The rare catastrophic loss that wipes out all those small wins (and more) is spread thin across time, so it is easy to disbelieve until it lands. Survivorship bias does the rest: the winning nights get retold, the busting night gets rationalised as bad luck. The maths, meanwhile, was never in doubt.
Put together, these mean there is no staking pattern — simple or elaborate, aggressive or gentle — that beats a game with a house edge. A system can change your experience of variance. It cannot change your expected result.
The only genuine control a player has is on the demand side: choosing lower-edge games and lower-stakes bets, and deciding a loss limit in advance that you treat as the fixed price of entertainment. That is not a system for winning — there isn’t one — it is a way to keep a form of entertainment affordable and under control.
Martingale, anti-Martingale, d’Alembert, Fibonacci, Labouchère — every betting system rearranges the size and order of your bets while leaving the odds of each bet untouched. Because each bet carries a fixed negative expected value and expectations add up, the total is always negative, and table limits plus a finite bankroll guarantee that escalating progressions eventually collapse. The systems rest on the gambler’s fallacy, which is simply false: random events have no memory and nothing is ever “due.” No system beats the house edge, because the arithmetic does not allow it.
There is no winning system, and gambling is not a way to make money — the maths guarantees a long-run cost to the player. Set a limit you can afford to lose and stop when you reach it. If gambling stops being fun or feels out of control, our responsible gambling resources offer tools and support.
No. Every betting system only decides how much to stake and when; it never changes the odds of a single bet. Because each bet carries a fixed negative expected value and expectations add up, any sum of them is negative in every order and at every stake. Rearranging negative numbers cannot make the total positive.
It is the false belief that independent random events balance out in the short run, so a loss makes a following win more likely. Wheels, dice and RNGs have no memory. If red lands eight times, red's probability on the next spin is unchanged. Nothing is ever due.
Doubling after each loss requires bets of 1, 2, 4, 8 and up. A ten-loss streak needs a 1,024-unit bet to recover, and you have already staked 1,023 to chase one unit. Table limits and finite bankrolls guarantee that a streak eventually stops you, crystallising a large loss. It hides a rare catastrophe behind frequent small wins.
It is gentler, raising the stake by one after a loss and lowering it after a win, but it rests on the same fallacy that losses will be made up by later wins. The slower progression means slower ruin, not avoided ruin. Each bet still carries the same negative edge.
They are more elaborate, but complexity only disguises the same flaw. Both rearrange the size and order of negative-expectation bets and hit the same walls: losing streaks, table limits and finite bankrolls. No sequence design changes the expected value of a single spin.
Only the demand side: choosing lower-edge games and lower stakes, and setting a loss limit in advance that you treat as the fixed price of entertainment. That is not a way to win, because there is not one; it is a way to keep gambling affordable and under control.