Good GamblingSites

Free tool

Betting System Simulator: Martingale, Fibonacci and D’Alembert tested

Pick a system, a bankroll and a table limit, then run up to 20,000 simulated roulette sessions. The results show how often each system wins a session, how often it goes bust, and what it costs on average.

Updated 5 min read By Richard Smith

Wheel

This simulator needs JavaScript. How each system works and what the long run looks like is explained below the tool.

Expected loss per session from the house edge
—
Average result per session in this run
—
Sessions that ended ahead
—
Sessions that lost the whole bankroll
—
Worst session
—
How the sessions ended
0
  • Sessions that ended behind
  • Sessions that ended level or ahead
  • Bar height = number of sessions

Even-money bets (red/black). Each run uses fresh random spins, so figures move a little between runs; the pattern does not.

The three systems

  • Martingale doubles the stake after every loss and returns to one unit after a win. One win recovers all earlier losses plus one unit.
  • Fibonacci moves one step along the sequence 1, 1, 2, 3, 5, 8… after a loss and two steps back after a win. It climbs more slowly than Martingale.
  • D’Alembert adds one unit after a loss and takes one away after a win, the gentlest of the three.

Flat betting, the same stake every time, is included as the baseline.

What the simulation shows

With the default settings — a 200 bankroll, 2-unit bets, a 500 table limit and a target of +40 — progressive systems end most sessions ahead, typically three to four sessions in five, while flat betting ends ahead in roughly one in three. The other side of the same picture is the bust rate: around one session in five loses the entire bankroll under the progressions, against almost none under flat betting. The histogram makes the shape visible: a tall cluster of small wins and a thin tail of total losses.

How to read the results

The first figure is the one that does not depend on luck: the expected loss per session from the house edge, which is the average amount wagered per session multiplied by 1/37 on a European wheel or 2/38 on an American one. The second figure, the average result of this particular run, lands close to it once a few thousand sessions are simulated. Only after those two come the share of sessions that ended ahead and the share that lost the whole bankroll, because a high win rate says nothing about cost on its own.

In a 20,000-session run with the defaults, Martingale ended about 80 sessions in 100 ahead and lost the bankroll in about 20; its average result was roughly −8 per session on about 300 wagered. Flat betting ended only about 31 sessions in 100 ahead, never went bust, and averaged about −15 on about 570 wagered. The loss per unit wagered was the same 2.7% in both cases.

The histogram shows how the sessions ended. Pink bars are sessions that finished behind and blue bars sessions that finished level or ahead; the height of each bar is the number of sessions in that band. A progression produces a tall blue spike just above zero and a pink cluster at minus the whole bankroll, with little in between.

Table limits and the size of the bankroll

A doubling system needs 2n − 1 units to survive n losses in a row: 255 units for eight, 1,023 for ten. With a 2-unit base and a 500 table limit, the ninth stake in a losing run would be 512, above the limit, so the system cannot recover its losses after eight straight losses even with a larger bankroll. The default 200 bankroll gives out sooner still: after six losses in a row (2 + 4 + … + 64 = 126 lost) the next doubled stake of 128 is more than the 74 left, so one more loss ends the session. The streak odds calculator shows that six losses in a row turn up in most 300-spin sessions.

Why no progression beats the house edge

Every spin is independent and pays 1 to 1 on a bet that wins 18 times in 37. Whatever stake sequence a system uses, each unit wagered loses 1/37 on average — about 2.7% on a European wheel, 5.3% on an American one. A system can rearrange when the losses arrive, turning many small losses into rare large ones, but it cannot change their total. The roulette simulator shows the same edge spin by spin, and the guide on how to bet on roulette explains the odds of each bet.

How the simulation runs

Each session starts with the bankroll you set and places even-money bets until it reaches the target, runs out of spins, or can no longer cover the next bet. Bets above the table limit are capped at the limit. Spin outcomes come from a pseudo-random generator with the true probability of red on the chosen wheel. Because each run draws fresh spins, the percentages move slightly from run to run; with 5,000 or more sessions the pattern is stable.

Portrait of Richard Smith

About the author

Richard Smith

Founder & editor, GoodGamblingSites.com

Started the site to show where online gambling is legal. Every list here begins with the licence check.

  • 12Countries & US states
  • Licence firstRule for every list

FAQ 5 questions

Frequently asked questions

Why do progressions win most sessions and still lose money?

They win small amounts often and lose large amounts occasionally. Most sessions end slightly ahead, but the rare losing streak that hits the table limit or empties the bankroll wipes out many small wins. Over many sessions the result follows the house edge on the money wagered, as it does for flat betting.

Is Fibonacci safer than Martingale?

It climbs more slowly, so a losing run takes longer to reach the table limit, but it also needs more wins to recover. With the default settings it ends ahead slightly less often than Martingale and goes bust slightly more often. The house edge on everything wagered is the same.

What does the table maximum change?

A table maximum stops a progression from doubling indefinitely. Once the next stake would exceed it, the system can no longer recover its losses with one win. Lowering the maximum in the simulator shows how quickly that turns a Martingale session from a likely small win into a large loss.

Why does Martingale sometimes show a smaller average loss than flat betting?

Because sessions stop as soon as the target is reached, Martingale sessions often end early and wager less money in total. The loss per unit wagered is the same 2.7% on a European wheel; less money wagered simply means a smaller total loss on average.

Is there any betting system that changes the odds?

No system based on changing stake sizes can change the probability of each spin or the payout. On a fair-paying game nothing would be lost on average; on a casino game with an edge, every system loses that edge on everything it wagers.