Ask a table of regulars for the best roulette strategy and you will hear about doubling after a loss, a sequence scribbled on a beer mat, or a bet that covers most of the layout at once. Each roulette strategy on that list changes how your stakes move. None of them changes what the wheel keeps. On a single-zero wheel that is 2.70% of every dollar you put on the felt, in whatever order the dollars go down, and on a double-zero wheel it is 5.26%.
That sounds bleak, but it clears the ground. Once you accept that no system changes the edge, the decisions that do move the price stand out plainly: which wheel you sit at, what happens to your even-money bets when zero lands, how big each bet is next to your budget, and when you get up. This hub covers both halves. First the betting systems people search for, each with its own worked page, then the choices that really change what an evening of roulette costs.
What a roulette strategy is up against
A European wheel carries 37 pockets: the numbers 1 to 36 and one green zero. The payouts are set as if that zero did not exist. A straight-up bet wins once in 37 spins on average and pays 35 to 1. Bet C$1 on one number for 37 spins and the expected return is C$36 (one win: your C$1 back plus C$35 of winnings) against C$37 staked. The missing dollar is the house edge, 1/37 or 2.70%.
The same sum works for every bet on the layout. Red wins on 18 of the 37 pockets and pays 1 to 1. A dozen wins on 12 and pays 2 to 1. A corner wins on 4 and pays 8 to 1. Work any of them through and you land on minus 1/37 of the stake each time. The American wheel adds a double zero, which pushes the edge to 2/38, or 5.26%, and its five-number bet (0, 00, 1, 2, 3) is the worst bet in the building at 7.89%.
Every system runs into the same fact: spins are independent. the ball has no memory, and a well-kept wheel or a tested random number generator has none either. So the expected value of a bet is fixed the moment it is placed. A betting system is only a rule for sizing the next stake. Add up all the stakes in a session, multiply by minus 2.70%, and you have the expected result of that session, whichever system produced the stakes.
Order and size change two things: the total amount you put through, and the shape of the results around that average. A progression makes many short sessions end a little ahead and a few end far behind, yet the average result is still a loss of 2.70% of the total staked, and the share of sessions ahead shrinks with every extra spin. A flat bet makes results spread more evenly. Neither touches the 2.70%.
If you would like to check any figure on this page yourself, the roulette odds calculator gives the expected value of any bet at any stake.
Betting systems compared in one table
The table follows each system through the same bad patch: seven even-money losses in a row, starting from a C$5 unit on a European wheel. Seven straight losses happen about once in every 106 attempts, so a regular player meets them.
| System | Stake rule | Lost after 7 losses | Next stake | How it fails |
|---|---|---|---|---|
| Martingale | Double after a loss, reset after a win | C$635 | C$640 | Hits a C$500 table maximum |
| Fibonacci | Next number in the sequence after a loss, two steps back after a win | C$165 | C$105 | Long slow runs drag the line upward |
| D’Alembert | One unit up after a loss, one down after a win | C$140 | C$40 | Needs more wins than the wheel supplies |
| Labouchère (1-2-3-4) | Bet first plus last number, add the stake to the line after a loss | C$280 | C$60 | The line grows faster than it shrinks |
| James Bond | Flat C$200 spread over 19-36, a six line and zero | C$1,400 | C$200 | 12 numbers lose everything, every spin |
Every row has the same expected cost per dollar staked. What differs is how quickly the stake climbs and how much you need on hand to keep following the rule. The Martingale is the extreme case: it needs C$635 behind a C$5 bet to survive seven losses, and the eighth step is not even allowed at a table that caps bets at C$500.
Martingale: the one everyone tries first
Doubling after every loss means the first win covers all the earlier losses in the run and leaves one unit of profit. It feels airtight until you do the sums. With a C$5 start and a C$500 maximum, the Martingale system runs out of room after seven losses. That streak has a probability of (19/37)7, about 0.94%, each time you start a new cycle, and roughly a 36% chance of turning up somewhere in a session of 100 spins. When it arrives, C$635 is gone for the sake of five-dollar wins. Our worked page shows the full table of streak probabilities and why the expected value per spin stays where it was.
Fibonacci, D’Alembert and Labouchère: slower ladders
These three climb more gently. The Fibonacci system follows 1, 1, 2, 3, 5, 8 and so on, stepping back two places after a win. The D’Alembert system adds or removes one unit at a time and leans on the idea that wins and losses balance out. The Labouchère system writes a target as a line of numbers and cancels them as you win. A gentler climb means a bad run takes longer to hurt, and when it does, the damage is smaller per spin but often spread over more of them. Each of these pages works through a full sequence in Canadian dollars.
James Bond: coverage instead of progression
The James Bond strategy does not change stakes at all. It splits C$200 across high numbers, a six line and zero so that 25 of the 37 pockets pay something. That makes winning spins common, about two in three, and the other third costs the whole C$200. Coverage changes how often you win, never what each dollar is expected to lose.
The best roulette strategy starts with the wheel
If one choice deserves the label of best roulette strategy, it is refusing double-zero tables. Put C$500 of total bets through a European wheel and the expected cost is C$13.51. Put the same C$500 through an American wheel and it is C$26.32. Nearly double, for the same evening and the same bets. No staking plan comes close to that saving.
Online, the wheel type is usually in the game name, and live tables show the layout on screen before you bet. Evolution, whose live tables most of the casinos we list carry, runs single-zero wheels on titles such as Lightning Roulette and Fireball Roulette, and keeps a separate American Roulette table with 0 and 00. At land-based casinos you may need to look at the wheel itself. If you see two green pockets, the table costs twice as much. The variants guide lists what each version does to the numbers.
La Partage and En Prison: the 1.35% bets
French roulette uses the same single-zero wheel, with one extra rule for even-money bets (red or black, odd or even, 1-18 or 19-36). Under La Partage, when zero lands you lose only half of an even-money stake. Under En Prison, the stake is held for the next spin and returned if it wins then. Either way, the expected loss on those bets halves, from 1/37 to 1/74, or 1.35%.
That is the lowest house edge on any standard roulette bet. The catch is availability. Not every casino offers a French table, and the rule only covers even-money bets: a straight-up bet on a French table still costs 2.70%. Our French roulette page explains how each rule settles, and the house edge breakdown shows the working for all three wheels.
Bet choice changes the ride, not the price
On a European wheel, every bet type carries the same 2.70% price. So does bet choice matter? Very much, for how a session feels and how long it lasts. The useful measure is the standard deviation of a single spin, the typical size of a swing around the average.
Swipe the table
| Bet | Pays | Wins out of 37 | Swing per spin |
|---|---|---|---|
| Straight up | 35:1 | 1 | 5.84 × stake |
| Split | 17:1 | 2 | 4.07 × stake |
| Corner | 8:1 | 4 | 2.80 × stake |
| Dozen or column | 2:1 | 12 | 1.40 × stake |
| Even money | 1:1 | 18 | 1.00 × stake |
A C$5 straight-up bet swings almost six times as hard as a C$5 bet on black. With a fixed budget, a string of straight-up bets usually burns through it quickly, with the occasional big hit. Even-money bets stretch the same budget over many more spins. Our view, put plainly: for most players, flat even-money bets on a single-zero wheel, or a French table where one exists, make the most sensible default. Inside bets are fine if you want the chance of a large single payout and accept a shorter session.
Session limits: the part of strategy you control
The last piece has nothing to do with the layout. Set a money budget and a time budget before you start, at an amount you are comfortable losing in full, and stop when either one runs out. A system that asks you to keep raising stakes after a loss is pulling the other way, which is one more reason to be wary of progressions.
At C$5 per even-money spin on a single-zero wheel, the expected cost is about 14 cents a spin. On Evolution’s Auto Roulette, which runs 60 to 80 games an hour, that adds up to roughly C$8 to C$11 an hour at that stake, while a typical swing over the same hour is around C$40 either way, and C$80 is not unusual. Short-term results are mostly noise. Over many sessions the price is collected.
Our bankroll and session limits guide turns this into budgets you can actually keep, and the responsible gambling page lists the limit tools and support lines for your province. If you want to try a system first without spending anything, the free practice wheel uses play chips only.
Where the systems got their names
A brief detour, because the names explain a lot about why these ideas survive. The Martingale comes from 18th-century France, where the word was already used for a doubling bet. D’Alembert borrows the name of Jean le Rond d’Alembert, the French mathematician, who wrote on probability and doubted that coin tosses were truly independent, which is exactly the belief his namesake system rests on. The Fibonacci sequence is from Leonardo of Pisa’s Liber Abaci of 1202, a book about arithmetic and trade with no roulette in it at all. Labouchère is named after Henry Labouchère, the 19th-century British journalist and politician whose name stuck to the cancellation method. The James Bond split borrows Ian Fleming’s spy for glamour.
None of the five was built on the mathematics of an independent wheel, and the most famous demonstration of that came at the Casino de Monte-Carlo on 18 August 1913, when black came up 26 times in a row. Players kept piling onto red as the run went on, sure that the wheel owed them. The chance of 26 blacks from a standing start is about 1 in 137 million. The chance of red on the 27th spin was still 18 in 37.