Editorial photograph of a printed house-edge comparison chart tacked to a corkboard, a fountain pen and a small leather notebook resting on a shelf below the board.

Strategy

Strategy: where the betting systems stop working

The fixed edge is the foundation. Betting systems change the shape of a session; they do not change the edge. This chapter reads the martingale family, the named systems, flat staking and the only decisions a player actually controls: how much to bring, when to leave and how to behave while the wheel turns.

Start with the fixed edge Read the responsible-play page

Chapter 01

The fixed edge is the foundation

European

2.70 percent on every bet. The cleanest reference point.

American

5.26 percent on every bet. The extra green pocket is the cause.

French (la share)

1.35 percent on even-money bets when la partage or en prison applies.

Every strategy discussion begins with the fixed edge because no strategy discussion is honest without it. On a European wheel the house edge is 2.70 percent on every bet; on an American wheel it is 5.26 percent; on a French wheel with la partage or en prison applied to even-money bets, the edge on those bets falls to 1.35 percent. The edge is a property of the wheel and the payouts, both of which are public. Nothing a player does at the table changes the edge on the next spin.

The fixed edge is also the answer to most strategy questions. A betting system can change the shape of a session — the size of the bets, the order in which they are placed, the duration of a winning or losing run — but it cannot change the per-spin expectation. Any claim that a system "beats" roulette is a claim about expectations, not about a single session. The honest reading of every published roulette strategy is the same: it manages variance, it does not remove the edge.

Chapter 02

Martingale and cousins: how doubling collapses

The martingale family of systems is the most popular negative-progression system sold in books and on websites. The basic version asks the player to double the stake after each loss, then return to the starting stake on a win. A single win at any point in the sequence recovers all prior losses and returns a profit equal to the starting stake. The maths of the family are tidy; the practical flaw is equally tidy.

The flaw is geometric. A one-unit starting stake becomes 2 after one loss, 4 after two, 8 after three, 16 after four and 1024 after ten. A 1024-unit stake on a 1-unit starting bankroll is a stake the bankroll cannot cover. The table maximum exists precisely to interrupt this growth; the practical floor on the bet size is the point at which the next double is unavailable. No published martingale removes that floor.

The anti-martingale

The anti-martingale asks the player to double the stake after each win and reset after each loss. It is the symmetrical cousin of the basic martingale. The mathematics are the mirror image: a long winning streak at the start of a session produces a very large stake, but a single loss at any point returns the player to the starting unit. The anti-martingale rewards concentration of the bankroll during winning runs but does not change the per-spin expectation.

The d'Alembert and the Fibonacci

The d'Alembert progression asks the player to add one unit after a loss and subtract one after a win. The Fibonacci progression asks the player to step forward in the Fibonacci sequence after a loss and step back two places after a win. Both systems reduce the rate of stake growth compared to the martingale, but neither removes the edge. The smaller growth rate produces smaller reward per recovery, which means more recoveries to clear the same loss. Over a long sample the per-spin expectation is unchanged.

The honest summary

No negative progression reverses the house edge. The systems change the shape of a session — the size of the bets, the duration of losing runs, the size of winning runs — but the average return per spin is fixed by the wheel and the payouts. An adult reader who treats these systems as entertainment budgets rather than as edge-beaters will be reading them correctly.

Editorial photograph of a notebook page showing neat handwritten columns representing a flat bet progression, a fountain pen resting across the page.

Chapter 03

Flat staking as the baseline

Flat staking is the simplest system and the easiest one to defend. The player bets the same stake on the same bet every spin, until the session ends. The variance is the natural variance of the bet; the edge is unchanged. There is no doubling, no progression, no recovery and no special handling of losing runs beyond the natural.

For an adult player who plans to read roulette as a small entertainment expense, flat staking is the easiest discipline to keep. The session has a single variable — the outcome of each spin — and the player does not have to track a sequence of bets. The bankroll declines at the expected rate of the house edge, with the natural variance of the chosen bet layered on top.

The flat stake is also the right unit for measuring session cost. A player who bets one unit per spin across 200 spins has wagered 200 units. The expected loss is 200 × 2.70% = 5.4 units on a European wheel. The actual loss can be much smaller or much larger; the expected loss is the right starting point for budgeting the session.

Chapter 04

Stop-loss and stop-win as the only real controls

The only decisions a player actually controls are how much to bring to the table, when to leave and how to behave while the wheel turns. A stop-loss is the largest acceptable loss for a session; a stop-win is a softer target that gives a session a clean ending. Both are more reliable than any system sold with a name.

Choosing a stop-loss

The right stop-loss is the largest amount of money the player can afford to lose in a single sitting without affecting the rest of the week's budget. The number is personal, not mathematical; it should be written down before the session begins and treated as a hard line. A session that hits the stop-loss is over, regardless of the player's mood.

Choosing a stop-win

The right stop-win is the largest amount of money the player is willing to leave with. It is a softer target than the stop-loss because winning sessions tend to feel less urgent than losing ones; the discipline of stopping is harder when the wheel is being kind. A common rule is to set the stop-win at half the stop-loss; a session that ends at the stop-win has recovered half of what it could have lost. That is a reasonable target.

Editorial photograph of a small printed reference card listing roulette payouts and odds, the card resting on a felt surface beside a cream chip.

Chapter 05

Variance: why one session is not the average

Variance is the reason a session that loses its whole stack in twenty spins is not a refutation of the fixed edge, and a session that doubles its bankroll in twenty spins is not a confirmation of a secret system. The fixed edge describes the long-run behaviour of a wheel over thousands of spins; a twenty-spin session is a small sample of that distribution.

The simplest reading of variance is in streaks of one outcome. On a European wheel the probability of red appearing six times in a row is roughly 1.4 percent, which means a session of one hundred spins will see such a streak roughly once or twice. A streak is not a prediction and not a refutation; it is the natural behaviour of a fair coin with a small bias.

The bankroll and the variance

The relationship between bankroll and variance is the practical reason a session ends. A small bankroll on a high-variance bet (a straight-up number) will end quickly, regardless of the fixed edge. A large bankroll on a low-variance bet (red/black) will last longer, because the per-spin variance is smaller. The choice of bet is a choice of session length as well as a choice of payout.

Chapter 06

Named systems: what the labels actually do

The roulette literature contains dozens of named systems. Some are progressive (the bet size changes with the outcome); some are pattern-based (the bet size or position changes with a perceived sequence); none are edge-beating. The honest summary is short: a named system is a stake-management pattern with a marketing name. Some patterns suit some players; no pattern suits every player.

Common names and what they describe

The Labouchere system asks the player to write a sequence of numbers and bet the sum of the first and last. The sequence shrinks with each win and grows with each loss; the system ends when the sequence is cleared. The Oscar's Grind system asks the player to increase the stake by one unit after each win and reset after each loss. The 1-3-2-6 system asks the player to step through a four-bet sequence that resets on a loss.

None of these systems removes the house edge. The 1-3-2-6 sequence has the cleanest recovery profile, the Labouchere has the most variable stake growth, the Oscar's Grind has the slowest progression. A player who understands the stake shapes can choose the system that matches their variance tolerance. A player who believes any of these systems beats the wheel is mistaken.

Chapter 07

An honest summary

The strategy section of a serious roulette handbook ends with a short summary because the rest of the section has done its work. The fixed edge is a property of the wheel and the payouts. Betting systems change the shape of a session but not the edge. Flat staking is the simplest discipline and the easiest to keep. Stop-loss and stop-win are the only decisions a player actually controls.

The rest of the chapter lives in the responsible-play section of this handbook. The decisions a player makes about how much to bring, how long to play and when to walk away are the decisions that shape the experience. The systems sold with names are entertainment, not edge. The reading an adult player should leave this chapter with is the difference.

Read the platform reviews Responsible play