Most betting systems tell you to double your bet after a loss. The Fibonacci system is different. Instead of doubling, you add your last two bets together to get the next one. That slower climb is the whole point.
Fibonacci is a negative progression system. You increase bets when you lose and pull them back when you win. The bet sizes follow a number sequence that has been kicking around mathematics for centuries: 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144. Each number is just the sum of the two before it.
The sequence is named after Leonardo of Pisa, an Italian mathematician who wrote about it in 1202 in his book Liber Abaci. He did not invent it. Indian mathematicians had described similar patterns centuries earlier, and the sequence appears throughout nature in everything from sunflower spirals to seashell curves. Whether it has any business being applied to a roulette table is another question entirely.
One thing to get out of the way early: this system does not change the house edge. Not by a fraction of a percent. The casino's mathematical advantage is baked into every single bet you place, regardless of how you size your wagers. What Fibonacci gives you is a structure for managing those bets. It does not give you an edge over the game. No betting system does.
You need three things. The Fibonacci sequence (written down or memorized), a base unit (whatever amount you want to start betting), and a game with even-money bets.
Start at the first number in the sequence: 1 unit. After a loss, move one step forward and bet the next number. After a win, move two steps back. If you are already near the beginning when you win, just stay at 1 unit. When you work your way back to the start with a profit, the cycle is done. Start fresh.
Lose, step forward. Win, step back two. That is the entire system.
Why two steps back? Because of how the sequence is built. Each Fibonacci number is the sum of the previous two. So a win at any position pays enough to cover the losses from the two positions before it. One win wipes out two losses. Not all your losses. Just the last two.
That is the core trade-off, and it is worth understanding clearly. Martingale recovers everything in a single win, but it doubles your bet every time you lose, which gets dangerous fast. Fibonacci recovers piece by piece across multiple wins, but the bets grow at a fraction of the speed. You are giving up recovery speed in exchange for a longer leash on your bankroll.
Theory is one thing. Watching it play out with real numbers is another. Here is a session using a $10 base unit on European roulette, betting on red.
Eleven rounds. Six losses, five wins. The player ended up $20 ahead despite losing more often than winning. That is the system doing its job. It absorbed a four-loss streak at the start, clawed back over three wins, completed the cycle, then repeated a shorter one.
Now here is the session nobody puts in their guide.
Nine losses, three wins. The seven-loss streak pushed the bet all the way to $210. Winning at that level only pulled the player back to position 6. After twelve rounds, they are still $40 in the hole, still sitting deep in the sequence, with no clean exit in sight.
Both of these sessions are realistic. The first one is what Fibonacci looks like when the wins and losses mix together. The second is what it looks like when they cluster. Any honest guide should show you both.
The practical question with any negative progression: how much money do you need to survive a bad streak? Here is Fibonacci side by side with Martingale.
After 10 straight losses, Fibonacci has you betting 55 units with 143 in the hole. Martingale has you at 512 units with 1,023 gone. That is roughly a 5x difference.
But look at the recovery side. One Martingale win at position 10 erases every prior loss and nets 1 unit profit. Clean reset. One Fibonacci win at position 10 only steps you back to position 8. You are still buried and you need more wins to dig out.
A player using a $10 base unit who wants to survive 10 consecutive losses needs about $1,430 in bankroll. That is a lot of runway for a system that bets $10 on the first round. But compared to Martingale needing over $10,000 for the same streak, it looks modest.
This is the part that makes people uncomfortable.
In a perfectly fair coin flip, the probability of losing 8 times in a row is 0.39%. That sounds like it will basically never happen. But probabilities stack up over time. Play 1,000 rounds and the chance of experiencing at least one 8-loss streak is uncomfortably high. Run the numbers over a full year of regular play, and it stops being a question of "if" and becomes "when."
And roulette is not a coin flip. Even-money bets on a European wheel win 48.65% of the time, not 50%. On an American wheel, 47.37%. That gap is small, but it tilts every streak slightly in the house's favor. Losing runs happen a touch more often. Recovery takes a touch longer.
This does not mean your next session will blow up. Most sessions are short enough that the long streaks never appear. But the system assumes you can always keep betting the next number in the sequence. When a streak finally arrives that outpaces your bankroll or the table limit, the cycle breaks. That is the structural weakness of every negative progression system, Fibonacci included.
Fibonacci was built for even-money bets with a win probability near 50%. The closer the game gets to a true coin flip, the better the system performs.
Roulette is the natural fit. Red/Black, Odd/Even, High/Low on a European wheel (2.70% house edge). If you can find French roulette with La Partage, even better. That rule cuts the even-money house edge to 1.35%, which is the friendliest table you will find for any progression system. American roulette works mechanically, but the 5.26% house edge chews through your bankroll noticeably faster.
Baccarat fits well. The Player bet (1.24% house edge, no commission) is the cleanest option. The Banker bet has a lower edge (1.06%) but the 5% commission on wins makes unit tracking messy. Most Fibonacci players who use it on baccarat stick with the Player side.
Craps pass line and don't pass line bets work. House edges of 1.41% and 1.36% respectively. Straightforward even-money payouts.
Blackjack is a poor fit, and this is where a lot of people get it wrong. Blackjack payouts are not uniform. A natural pays 3:2. Doubles and splits change how much you have on the table mid-hand. The system assumes every bet either wins or loses one unit. When a single hand can cost you 2x or 4x your planned bet because you split and doubled, the progression math falls apart.
Slots do not work at all. Variable payouts, no even-money structure, and house edges typically running 3% to 10% or more.
This is the matchup most people are actually looking for. Both systems increase bets after losses. The question is how they do it and what that means for your money.
Martingale is a chainsaw. It cuts through losing streaks in one swing, but if the wood is too thick, it kicks back hard. Fibonacci is a handsaw. Slower, more controlled, less dramatic. It gets through the same log, just in more strokes.
Neither one changes the house edge. They reshape how your wins and losses are distributed across a session. Martingale packs the risk into rare but catastrophic events. Fibonacci spreads the pain across longer, grinding recovery periods.
If you have a large bankroll and strong nerves, Martingale recovers faster. If you prefer to stay alive longer and absorb hits gradually, Fibonacci is the less painful path. Your call.
D'Alembert is what you get when you take negative progression and make it as gentle as possible. Instead of adding the previous two bets (Fibonacci) or doubling (Martingale), D'Alembert increases by just one flat unit after each loss. The bet sequence goes 1, 2, 3, 4, 5, 6. Straight line. No acceleration.
For the first 5 or 6 losses, Fibonacci and D'Alembert feel almost identical. After that, Fibonacci starts pulling ahead. By loss 10, Fibonacci has you at 55 units. D'Alembert has you at 10. That is a massive gap. But D'Alembert's recovery is even slower because each win only drops the bet by one unit.
The choice between them is a question of patience versus pace. D'Alembert is the most conservative negative progression out there. You will survive longer but recover slower. Fibonacci sits in the middle ground between D'Alembert's caution and Martingale's recklessness.
Flip the whole thing around. Instead of increasing bets after losses, increase them after wins. A loss sends you two steps back (or all the way to 1 unit). A win moves you one step forward in the sequence.
The goal is to press winning streaks. When you are running hot, the bets grow along the Fibonacci sequence, stacking profits. When a loss comes, you scale down fast and limit the damage.
The risk is the mirror image of the standard system. A single loss after five or six consecutive wins erases most of what you accumulated, because your last bet was the biggest one. Standard Fibonacci protects you from big one-off losses. Reverse Fibonacci protects you from grinding drawdowns but leaves you exposed to a single well-timed loss.
It suits a different kind of player. If chasing losses makes you uneasy but pressing wins feels natural, the Reverse version fits that instinct.
Depends entirely on what you are expecting.
If you want a structured way to manage your bets that produces frequent small wins under normal conditions, yes. When wins and losses shuffle together in the typical way, Fibonacci regularly completes cycles at a modest profit. Most sessions feel like you are making progress.
If you want a system that beats the casino over time, no. Nothing does.
The house edge applies to every bet you place. A $10 bet on red carries the same 2.70% disadvantage whether it follows three losses or three wins. The sequence does not know or care about probability. It just tells you how much to bet next. Over enough rounds, the casino's edge chews through any system.
What Fibonacci actually does is redistribute your outcomes. Instead of a flat line of small wins and small losses, you get a pattern: lots of short profitable cycles punctuated by occasional ugly streaks that eat the profits. The wins are frequent and feel earned. The losses are rare and feel unfair. Over the long run, the rare losses consume the accumulated wins. That is the fingerprint of every negative progression system.
So should you use it? If you value structure, if you like having a clear plan for every outcome, and if you understand that the plan manages your bets but not the odds, Fibonacci is genuinely one of the better systems available. It is less reckless than Martingale, more dynamic than flat betting, and well suited to even-money wagers at lower-edge tables.
Just understand what you are paying for. Structure and discipline, not an edge.
Use it on the lowest-edge even-money bets you can find. French roulette with La Partage (1.35%) is the gold standard. European roulette (2.70%) and baccarat Player (1.24%) are both solid. The lower the house edge, the more runway your bankroll gets.
Set a ceiling before you start playing. Decide how deep into the sequence you are willing to go. Position 8 (21 units) or position 10 (55 units) are reasonable cutoffs. When you hit that number, stop. Walk away. Coming back tomorrow with a full bankroll is always better than pushing further today.
Set a profit target too. Something like +10 to +20 units per session. When you reach it, cash out. The system does not protect you from giving back profits if you keep playing past your target.
Check the table limits. If your base unit is $10 and the maximum even-money bet is $500, you can sustain 12 consecutive losses. The 13th Fibonacci bet would be 144 units ($1,440), which exceeds the cap. Know where that wall is before you start climbing.
Write down your position in the sequence. This sounds trivial, but losing track after a bathroom break or a conversation with the person next to you means betting the wrong amount. A wrong bet defeats the purpose of using a system in the first place. A note on your phone takes two seconds.
And keep the system where it belongs. Even-money bets on low-edge games. Do not try to force it onto blackjack, slots, or anything with variable payouts. The math only holds up when every win returns exactly one unit of profit on the amount wagered.
