No roulette strategy changes the price: every system on a standard European wheel loses about 2.70 kr per 100 kr staked. Our simulation of ten systems shows what changes instead: the shape of the results and how much you stake.
As seen on...
No roulette strategy changes the price: every system on a standard European wheel loses about 2.70 kr per 100 kr staked. Our simulation of ten systems shows what changes instead: the shape of the results and how much you stake.

No roulette strategy changes the price of the game. On a standard European wheel, every system loses about 2.70 kr per 100 kr you stake. A strategy only changes the shape of your results and how much you end up staking. And the table you pick (la partage) matters more than the system.
Below is our simulation of ten systems on the same 1,000 kr, plus the rules that stop systems in practice.
Last updated: 29 October 2026
The tiles below come from our live casino category. The tag means our review found a live casino at that site. It says nothing about betting systems, and the order is not a ranking.
It can change how your results are spread out and how much money you push across the table. It can’t change what each kroner of stake costs you. That price comes from the zero on the wheel. No order of bets gets around it.
Here’s the whole idea in one line: your expected loss is the house edge multiplied by the total amount you staked. On a European wheel the edge on a red bet is 1/37, or 2.70 kr per 100 kr. Stake 2,000 kr over a session and you lose about 54 kr on average, however you split the bets.
So a system has two levers. Stake more, and the average loss grows. Stake less, and it shrinks. The Kelly criterion, which sizes a stake from a positive edge, returns zero on a bet like this: try it in our stake calculator. Our house edge guide explains why the percentage stays put, and our roulette odds guide prices each bet on the layout.
What the systems do sell you is a feeling: a high chance of a small win. Is that worth having? Let’s see what it costs.
We ran every system through the same session, so you can compare them head to head. It’s a simulation, not a visit to a casino. Here’s the setup:
The same rules apply to all ten systems. The wheel type changes only the zeros: one zero (European), a French table with la partage, or two zeros (American).
Most systems end ahead more often than flat betting, and most also lose more on average. Here’s how they did on a European wheel, with one zero:
| System | Ends ahead | Goes broke | Average result | Average staked |
|---|---|---|---|---|
| Flat 10 kr | 32.32% | 0% | -54.5 kr | 2,000 kr |
| Martingale | 60.05% | 38.01% | -87.1 kr | 3,222 kr |
| Paroli (bank after 3 wins) | 35.90% | 0.02% | -88.7 kr | 3,336 kr |
| D’Alembert | 54.59% | 44.71% | -171.5 kr | 6,338 kr |
| Fibonacci | 62.75% | 31.88% | -116.6 kr | 4,325 kr |
| Labouchere 1-2-3-4 | 60.28% | 39.72% | -94.8 kr | 3,501 kr |
| Oscar’s Grind | 61.57% | 30.46% | -141.2 kr | 5,238 kr |
| James Bond (200 kr over 25 numbers) | 52.39% | 47.59% (under 200 kr left) | -144.9 kr | 5,380 kr |
| Two dozens (10 + 10 kr) | 29.01% | 0% | -108.4 kr | 3,998 kr |
| Bold play | 64.86% | 35.14% | -27.1 kr | 1,010 kr |

Look at the “ends ahead” column first. Almost every system beats flat betting there. Now check “goes broke” and “average staked”. Different picture, isn’t it?
Martingale doubles your stake after every loss, so one win wipes out the whole losing streak and leaves you 10 kr up. It reached the +500 kr target in 59.52% of sessions. But it went broke in 38.01%, and the average result was -87.1 kr.
That’s the deal: a small, frequent win against a rare, huge loss. Our Martingale guide walks through the doubling ladder, so we won’t repeat it here. The same idea turns up in crash games, and our guide to Aviator and crash games shows where doubling goes wrong there.
Fibonacci ended ahead 62.75% of the time, yet it hit the +500 kr target in only 5.76% of sessions. It goes up a step after a loss and down two steps after a win, so wins are slow to pile up. It went broke in 31.88% and averaged -116.6 kr.
Labouchere with the 1-2-3-4 list behaved much like Martingale: 60.28% ahead, 39.72% broke, -94.8 kr. You cross numbers off a list when you win and add one when you lose. More paperwork, roughly the same outcome.
Both are gentle systems, and gentle is expensive. D’Alembert adds one unit after a loss and removes one after a win. It staked 6,338 kr on average and lost 171.5 kr, the worst average in the table. It went broke in 44.71% of sessions, too.
Oscar’s Grind aims for one unit of profit per cycle. It ended ahead 61.57% of the time, but it staked 5,238 kr and lost 141.2 kr on average. Slow systems mean more spins, and every spin costs the same 2.70 kr per 100 kr.
Paroli is Martingale turned upside down: you raise your stake after wins and bank the profit after three. It went broke in just 0.02% of sessions, the best safety figure of any system that changes its stake. Only flat betting and two dozens did better, at 0%. The catch: it ended ahead only 35.90% of the time and lost 88.7 kr on average, about 60% more than flat betting, because it staked 3,336 kr instead of 2,000 kr.
You can’t lose your bankroll fast, because you never chase a loss. You also rarely get anywhere. That’s the trade.
James Bond spreads 200 kr over 25 numbers, and two dozens covers 24 numbers with 10 + 10 kr. Both win often on a single spin and lose big on the rest.
James Bond ended ahead 52.39% of the time and fell under 200 kr in 47.59% of sessions. Two dozens ended ahead just 29.01% of the time but went broke in none of 200,000 sessions. The averages were -144.9 kr and -108.4 kr, because both stake a lot of money for the same 2.70 kr per 100 kr.
Flat betting is the control. You bet 10 kr on red, 200 times. You’d end ahead in 32.32% of sessions, never go broke, and average -54.5 kr on 2,000 kr staked.
It’s the dullest line in the table, and close to the cheapest. Keep that in mind.
Because the zero takes the same bite from each bet, whatever its size or timing. Add up all the kroner you stake and multiply by 2.70%: that is your average loss. Our measured cost per 100 kr staked came out between 2.66 and 2.73 kr for every system, which is simulation noise around 2.70.
Check it with flat betting. 200 spins at 10 kr, times 1/37, gives 54.05 kr. Our simulation said 54.5 kr. The gap is random variation in the simulation.
So when a system loses more, it didn’t get a worse price. It just staked more. That’s all. D’Alembert’s 171.5 kr loss comes from 6,338 kr staked, not from a bad formula. The only way to lose less is to stake less, or to play a table with a lower edge.
That’s also why “I’ve found a system that beats the house” is a sentence you can ignore. Does a system still feel good when it wins? Of course it does. The maths just doesn’t care.
If your only aim is to reach a fixed target, bold play gives you the best chance of getting there in an unfair game. It’s also the most boring answer. You stake what you need to reach the target, or everything you have if that is less, and repeat until you reach it or lose everything. In our runs that took 2.0 spins on average.
Mathematician Jason Schweinsberg describes the result like this: “Dubins and Savage showed that the optimal strategy, which they called ‘bold play’, is always to stake min{f, 1-f}, where f is the gambler’s current fortune”. That holds when your chance of winning each bet is below 1/2, which it is on roulette, and when you are free to choose any stake. “Optimal” here means the highest chance of reaching the target, not the smallest loss.
In plain words: stake everything you need to reach the target, or everything you have, whichever is smaller. From 1,000 kr to 1,500 kr, the chance of getting there is p / (1 – p q), with p = 18/37 and q = 19/37. That works out to 64.85%, and our simulation gave 64.86%. The expected result is -27.3 kr, against -27.1 kr in the simulation. On an American wheel the chance falls to 63.10%, and the simulation gave 63.07%.
Bold play wins most often and loses least because it stakes least: 1,010 kr on average. There’s a catch, though.
So bold play is the best way to chase a target, not a way to beat the wheel. You still lose 27 kr on average. It’s simply the cheapest way to take that shot.
Yes. The wheel and its rules change the price of every bet, while a system never does. Pick the right table and you cut the cost without learning a thing.
On a French table with la partage, an even-money bet loses only half when the ball lands on zero. The even-money cost falls to 1.35 kr per 100 kr, and all seven even-money systems we ran there came out between 1.30 and 1.37. Flat betting averaged -26.9 kr, half of the 54.5 kr on the standard European wheel. The exact figure is 27.03 kr. Martingale on a French table ended ahead 63.03% of the time and broke 35.10%, at -40.7 kr.

The American wheel does the opposite. With 0 and 00 the cost is 5.26 kr per 100 kr, and every system came out between 5.22 and 5.35. Flat betting averaged -105.2 kr, against an exact 105.26 kr. D’Alembert averaged -344.0 kr there, and 56.10% of its sessions went broke.
Rules are public. Monte-Carlo SBM’s rules for European roulette say even-money bets on zero are barred and lose half of their value; the player may split the bet or leave it barred for the next spin, when it is freed if it wins. Evolution’s live roulette page says “if you make an even-money bet and the ball lands on zero, you will lose only half your bet”.
Danske Spil prints both returns on its winning-chances page: “Live Roulette 97,3%” and “Live French Roulette 98,65%”. The 98.65% figure applies to even-money bets. By our arithmetic, the other bets on the French table still return 97.30%.
Our roulette odds guide lists every bet and the la partage rule in full.
Rules, not luck, usually stop a system first. Three kinds are worth knowing: stake caps, bonus clauses and speed limits.
Norway caps the table games in Norsk Tipping’s KongKasino. Section 31 of the Norwegian gambling regulations says that for table games in KongKasino the total stake can be at most 500 kr, and the first stake no more than 100 kr (Lovdata).
We ran a cap variant, with total stake per spin limited to 500 kr on a European wheel. Martingale from 10 kr hits the cap in 60.53% of sessions, because the seventh stake in the ladder, 640 kr, is over it. It still ended ahead 59.29% of the time, with an average of -96.3 kr. Fibonacci was capped in 42.25% of sessions and Labouchere in 12.01%.
Bold play was unchanged in practice, at -26.0 kr: its biggest stake is 500 kr, so the cap never binds and the gap to -27.1 kr is noise.
Two cautions. The text doesn’t define the total stake per spin, so we treat it as the total per spin. And a 500 kr even-money stake needs the round total, not the first stake. We don’t claim KongKasino’s interface allows it. See our Norway licence page for the wider rules.
Some bonus terms ban hedging. Can they really? Yes. Take Wildz’s bonus terms, version 1.2, last updated 31 July 2026. Clause 10 says: “Betting strategies intended to eliminate or significantly reduce risk, including but not limited to placing opposing bets on the same outcome, do not contribute towards wagering requirements or promotional progress. Players found to be using such strategies may have their Bonus Money, promotional rewards and any associated winnings removed.”
Red and black at once is the clear case. Where does that leave a system that covers most of the table? Whether covering most of the table, like James Bond or two dozens, counts as risk hedging is the operator’s call.
The same terms add more limits. Clause 7: “All Table Games and Live Casino games contribute 50%.” The maximum bet while bonus money is active is “€5 per transaction” (clause 6). And clause 4 says “A maximum of €5 in Bonus Money per wager will count towards the wagering requirement.” A doubling ladder from €0.20 meets the €5 ceiling after five losses.
Our guide to non-sticky bonuses explains how bonus money and cash balance interact.
Sweden sets a minimum round time. SIFS 2022:3, chapter 15, section 2 says that every game round must last at least three seconds, and that this also applies to an autoplay function. Read it at lagen.nu or on the Spelinspektionen site.
In English: every round must last at least three seconds, autoplay included. That matters for slow systems. Our Oscar’s Grind sessions average 173 spins, which takes a while, and you pay the same 2.70% price on every one. See our Sweden licence page.
The best roulette strategy is the one that costs least for the fun you get. That usually means staking less, on a better table. Here’s your checklist:
Roulette is entertainment with a fee, not an income. Our responsible gambling page lists tools and helplines if it stops feeling like fun.
No system changes the house edge. Every one we tested lost about 2.70 kr per 100 kr staked on a European wheel. Some raise your chance of ending ahead, at the cost of a bigger loss when it goes wrong.
For reaching a target, bold play. It ended ahead 64.86% of the time and lost 27.1 kr on average. For lowest cost, a French table with la partage.
It ended ahead in 60.05% of our sessions and went broke in 38.01%. The average result was -87.1 kr. It’s a small win with a large risk.
D’Alembert, at -171.5 kr on average. It staked 6,338 kr, and that’s why. The price per kroner staked was the same as for the others.
Yes. It halves the loss on even-money bets when the ball lands on zero. The cost falls from 2.70 to 1.35 kr per 100 kr staked.
Yes. With 0 and 00 the cost is 5.26 kr per 100 kr on nearly every bet, about double the European wheel.
Some do. Wildz’s terms say players using strategies meant to eliminate or significantly reduce risk, such as opposing bets on the same outcome, may have their bonus money and winnings removed. A progression like Martingale is not named. Read the terms before you play.
Yes, for KongKasino table games. The total stake can be at most 500 kr, and the first stake at most 100 kr.
The numbers come from our simulation of 200,000 sessions per system and wheel, with a seeded random generator. We checked the flat-bet loss, the la partage loss and the bold-play success chance with exact calculations, and they match the simulation to within rounding. We read the rules and terms listed below on 29 October 2026.
What we did not verify: we didn’t read Dubins and Savage’s 1965 book, so we cite the result through Schweinsberg’s paper. We found no roulette return figure for Norsk Tipping’s KongKasino or for Veikkaus. We can’t say whether covering most of the table counts as risk hedging under a bonus clause. We also don’t claim that KongKasino’s interface allows a 500 kr even-money stake.
We opened no account and did no hands-on test. This page rests on our simulation, published rules and terms. How we work is in how we rate, and our casino list shows the operators we review.
New reviews, guides and bonus news by email. Unsubscribe at any time.