Betting Systems in Roulette: Martingale, Fibonacci and the Maths That Doesn’t Change

Roulette betting systems — Martingale, Fibonacci, D’Alembert and dozens of variations — promise to turn a losing game into a winning one by changing how much you stake after each spin. The short answer is that they can’t: every system only reshapes when you win and lose and how big those swings are, while the long-run expected value stays exactly the same as betting a flat amount every spin. What these systems genuinely change is the shape of your variance and the risk to your bankroll — a Martingale can feel brilliant for an hour and then wipe out a session in one losing streak. This guide walks through how each popular system works, what it actually does to your risk, and the mathematical reason no staking plan can beat the house edge.

What a betting system actually changes (and what it can’t)

Before any system, you need the one number that matters: the house edge. On European roulette, with a single zero, a bet on red or black pays 1:1 but wins 18 times out of 37 possible outcomes. Your true odds of winning are 18/37, or about 48.65%, yet you’re paid as if the odds were 50/50. That gap — the difference between true odds and the payout odds — is the house edge, and on European roulette it is 2.70%. On American roulette, with both a zero and a double zero, it rises to 5.26%. This edge applies to every single spin, on every single bet type, and it is completely independent of how much you stake.

A betting system is a rule for changing your stake between spins. Some raise bets after losses, some after wins, some follow a sequence of numbers. What every system has in common is that it cannot alter the probability of any individual spin — a roulette wheel has no memory, and each spin’s outcome is independent of the last. No staking plan can make the ball land on your colour more often, and no staking plan can change the payout odds the casino offers you. Both of those are fixed by the rules of the game.

So what does a system change? It changes the distribution of your results. A flat bettor who bets £10 on red for 100 spins will have a fairly predictable range of outcomes. A Martingale player who doubles after every loss will experience long strings of small wins punctuated by occasional catastrophic losses. The average over many sessions is the same — the house edge grinds both players down at the same rate in the long run — but the path is completely different. One player’s bankroll moves smoothly; the other’s is a rollercoaster. Understanding that distinction — variance versus expected value — is the entire story of betting systems.

This matters practically because of a concept called risk of ruin. A system that produces frequent small wins and rare large losses can feel like it’s working, right up until the rare loss arrives and takes your whole session bankroll in one go. The longer you play, the more likely you are to hit that rare event. Systems don’t fail because they’re badly designed — they fail because they can’t change the underlying probabilities, and the variance they create eventually catches up with you.

Martingale: doubling after every loss

The Martingale is the most famous betting system in casino history, and the simplest to explain: after every loss, double your previous bet. The logic is seductive. You bet £5 on red. If you lose, you bet £10. Lose again, bet £20. Lose again, bet £40. The moment you win, you recover all your previous losses plus a profit equal to your original stake. A single win ends the losing streak and puts you £5 ahead. Because the bet pays even money, one win always covers everything you’ve lost so far and leaves you with your starting stake as profit.

Here’s a worked example of how the bets escalate:

In this example, after three losses the player is down £35. The fourth bet of £40 wins, paying £40 profit, which covers the £35 lost and leaves £5 ahead. The system “works” exactly as advertised — if you can keep doubling until you win, and if you have the bankroll to survive the streak.

The problem is that those two conditions are precisely where the system breaks. First, the bets escalate exponentially. A losing streak of 10 spins doesn’t mean you’re down 10 times your stake — it means your next bet is 512 times your original stake. Starting at £5, ten consecutive losses put your next bet at £2,560, and you’re already down £5,115. Starting at £10, the same streak needs a £5,120 bet and you’re £10,230 in the hole. The table below shows how quickly stakes grow:

Second, every real roulette table has a maximum bet. Casinos set table limits precisely to make the Martingale unplayable in its pure form. If the table limit is £500 and you’re down £635 needing to bet £640, you simply can’t place the bet that would recover your losses. The system collapses not because the maths of doubling fails, but because the casino’s limit cuts the sequence off before you reach your win. And even without a table limit, your own bankroll is a limit: a long losing streak requires exponentially more money than most players carry.

How likely is a long losing streak? On European roulette, betting red, your chance of losing a single spin is 19/37, about 51.35%. The chance of losing ten in a row is (19/37) raised to the tenth power — roughly 1 in 784. That sounds rare, but it means over the course of 10,000 spins, you’d expect that specific streak about 12 times. The streaks are rarer than short losing runs, but they are not rare enough to avoid, and when one arrives, it erases many hours of small £5 wins in a single sequence. That’s the real trade-off of the Martingale: frequent small profits in exchange for the certainty that, eventually, one streak will cost you more than all of them combined.

Fibonacci and D’Alembert: gentler progressions

The Fibonacci system borrows its name from the famous number sequence where each number is the sum of the two before it: 1, 1, 2, 3, 5, 8, 13, 21, 34, and so on. The system works like this: after a loss, move one step forward in the sequence; after a win, move two steps back. Your bet is the current number in the sequence, usually multiplied by a base unit. So with a £5 base unit, a losing streak takes your bets through £5, £5, £10, £15, £25, £40, £65, £105, £170 and so on.

The idea is that a win at any point in the sequence will recover your recent losses, because each number is larger than the sum of the two previous ones — loosely, one win cancels out the two losses that preceded it. Fibonacci is called a “gentler” Martingale because the stakes grow more slowly. Where the Martingale doubles each loss, Fibonacci’s growth is roughly exponential but at a much slower rate, so your bankroll lasts longer and the bets feel less terrifying. The table below compares the two systems over a losing streak:

What Fibonacci gains in gentleness, it loses in efficiency. Because the stakes rise more slowly, each win recovers less of your total losses, so you need a higher proportion of wins to climb back to profit. A losing streak of nine costs you far less than under the Martingale, but a single win doesn’t restore you to even — you need several wins in a row, and the sequence’s “two steps back after a win” rule means you oscillate around a level for a long time. The system is slower and more survivable, but it has the same fundamental flaw: it cannot change the house edge, and a long losing streak still pushes your required bets toward your bankroll ceiling.

D’Alembert is gentler still. It’s based on the (incorrect) idea of “equilibrium” — that wins and losses should balance out, so after a loss you increase your bet by one unit, and after a win you decrease it by one unit. With a £5 unit, a losing streak takes you through £5, £10, £15, £20, £25 and so on — arithmetic growth rather than exponential. This keeps stakes the lowest of any popular system and makes sessions last much longer. But it also means each win only recovers one unit of loss, so a losing streak of ten requires ten subsequent wins just to get back to even, and the house edge means you don’t get ten wins without ten losses mixed in. D’Alembert is the least dangerous of the three in terms of bankroll swings, and the least effective at recovering losses — which is the honest trade-off.

All three systems share the same structural truth: they are rules for managing the size and timing of your bets, and none of them touches the probability of any individual outcome. The wheel doesn’t know or care what sequence you’re following. What these systems genuinely offer is a structure that can feel more disciplined than betting randomly — some players find the routine keeps them from chasing losses with reckless one-off bets — but the structure is a psychological aid, not a mathematical edge.

Flat betting vs progressive betting

Flat betting means staking the same amount on every spin, regardless of what happened before. If you bet £10 on red every time, win or lose, your next bet is still £10. Progressive betting is the umbrella term for every system that changes your stake based on prior results — the Martingale, Fibonacci and D’Alembert are all progressive systems, and there are positive progressions too, where you raise your bet after a win rather than a loss.

Flat betting has two important properties. First, it produces the most predictable bankroll path. Your swings are bounded by your stake size, and the house edge grinds you down at a steady, predictable rate. Second, and this is the key point, flat betting has exactly the same expected value as any progressive system. Over a large number of spins, a flat bettor and a Martingale player lose the same percentage of their total turnover to the house edge. The difference is purely in the distribution of results: the flat bettor’s bankroll moves in a narrow band, while the progressive player’s bankroll swings wildly but averages out to the same destination.

This is a genuinely counterintuitive result, so it’s worth spelling out. The house edge is a percentage of the total amount wagered, not of your bankroll. A progressive system bets more in total over a session than a flat system does, because losing streaks force larger bets. So over a fixed number of spins, the progressive player’s total turnover is higher, which means the expected total loss is also higher — not because the edge is bigger, but because more money is passing through it. The Martingale player who bets £5, £10, £20, £40 across four spins has turned over £75; the flat bettor has turned over £20. The house edge applies to all £75, not just the £20.

Positive progressions — raising your bet after a win and dropping it after a loss — are the mirror image. They’re marketed as “let your winners run” strategies, and they produce a bankroll path that’s the reverse of the Martingale: long, slow declines punctuated by occasional big wins when a hot streak arrives. But the same maths applies. The house edge is unchanged, and the positive progression’s total turnover is still higher than flat betting over the same number of spins, so its expected loss is higher too. The direction of the progression changes the shape of your risk, not its average.

The practical takeaway is that flat betting is the honest baseline. It’s the strategy that makes the house edge easiest to see and to plan around, and it never puts you in a position where a single losing streak forces you to bet more than you’re comfortable with. If a player insists on a system for the structure and discipline it provides, the milder progressions are less dangerous than the Martingale — but no progression is ever a way to turn a negative-expectation game into a positive one.

Why no system beats the house edge, mathematically

The mathematical reason no betting system can work is a theorem, not an opinion: the expected value of a bet is the sum of each outcome’s probability multiplied by its payout, and a betting system only changes the sequence of stake sizes, not the probabilities or the payouts. Each individual spin has a fixed negative expected value equal to the house edge. A system strings these spins together in a particular order, but the expected value of a string of negative-expectation bets is just the sum of those negative expectations — it cannot be positive.

There’s a formal version of this idea called the optional stopping theorem, which in plain language says that you cannot use a rule for deciding when to stop betting to turn a losing game into a winning one, as long as your bankroll is finite and the house edge is negative on every bet. The Martingale looks like it exploits the fact that a win is “guaranteed” eventually — and it’s true that if you can keep doubling forever, a win is certain. But you cannot keep doubling forever. Your bankroll is finite, the table limit is finite, and at the moment you run out of one of those, the losing streak ends with you losing everything you’ve put in. The system’s apparent certainty of a win is bought with the certainty of eventual ruin.

Another way to see it is through the gambler’s fallacy — the belief that a string of losses makes a win more likely. It doesn’t. A roulette wheel has no memory. If red has lost nine times in a row, the probability of red on the tenth spin is still exactly 18/37, the same as it was on the first. The Martingale’s whole logic depends on the idea that a win is “due” after a losing streak, and that idea is false. The losing streak doesn’t make a win more likely; it just means you’ve already lost money, and your next bet is now large enough that a win merely returns you to where you started.

There is one genuinely mathematical edge in roulette, and it’s worth naming so you can distinguish it from systems: the en prison rule on some European tables, where an even-money bet that lands on zero is not lost but held for another spin. This rule cuts the house edge on even-money bets roughly in half, to about 1.35%. That’s a real change to the mathematics of the game, but it’s a change to the game’s rules, not to your betting strategy. No staking plan can replicate it, and it’s a property of the table you choose, not of how you stake. If you’re choosing between roulette tables, the one with en prison is genuinely better for even-money bets — that’s a real, defensible difference, unlike every betting system.

The honest summary is this: roulette is a game designed so the house has a fixed, small edge on every bet. Betting systems are elaborate ways of rearranging the losses that edge produces. They can make a session feel more controlled, they can make losses come in bigger chunks less often, and they can make a short session look like a win — but over the long run, the house edge is a constant, and no rearrangement of stakes changes it. The only real choices a player has are which table to play (European beats American), whether to play even-money bets (lowest variance) or inside bets (higher variance, same edge), and how much to budget for the session. Those choices are all about managing risk and entertainment value — none of them is about beating the house.

If you’re playing roulette for fun, play a system you enjoy, bet only what you can afford to lose, and understand that the system is entertainment, not strategy. If you’re playing to win, the only honest advice is that the game is not set up for you to win in the long run — set a budget, set a time limit, and treat any money you put on the table as the cost of the entertainment, not as an investment. The maths doesn’t change, and no system, however cleverly named, will make it change.

Frequently Asked Questions