The D'Alembert betting system is a way of sizing bets on an even-money wager, something like red or black in roulette. Raise your bet by one unit after every loss. Lower it by one unit after every win. That is the entire system. It is named after a real 18th century mathematician who, it turns out, got a basic probability question wrong in print, which is a strange thing to build a betting strategy around and an even stranger thing to name it after.
What is the D'Alembert betting system?
It is a staking pattern, not a way to change your odds: you pick a unit size, and you move your bet up or down by exactly one unit depending on whether your last bet won or lost.
Say your unit is $10, and you are betting on red at roulette.
| Bet # | Bet size | Result | Next bet |
|---|---|---|---|
| 1 | $10 | Loss | $20 |
| 2 | $20 | Loss | $30 |
| 3 | $30 | Win | $20 |
| 4 | $20 | Win | $10 |
| 5 | $10 | Loss | $20 |
Lose, and the bet climbs by one unit. Win, and it drops by one unit, down to a floor of one unit (you never bet less than your starting size). The idea behind it is that a losing streak and a winning streak should roughly cancel out, so raising the bet slightly while you are down and trimming it while you are up should, in theory, land you back near even. It is built for even-money bets specifically, wagers that pay close to 1 to 1, because the whole system depends on a win recovering roughly what the last loss cost.
Compared to other progression systems like the Martingale system, D'Alembert looks almost cautious. That reputation is fair, and it is also the whole reason the name on the label is worth a second look.
Who was D'Alembert, and why is his name on a betting system?
He was Jean le Rond d'Alembert, a real French mathematician and physicist who lived from 1717 to 1783, and the system is named after him because he made a documented mistake in exactly this kind of coin-toss math. He was not a gambler and never published a betting strategy. He was one of the sharpest scientific minds of the French Enlightenment: he co-edited Diderot's Encyclopedie, one of the defining intellectual projects of the 1700s, and there is still a real law of motion named after him, d'Alembert's principle, taught in mechanics courses today.
In 1754 he wrote the Encyclopedie's own article on coin tossing, titled "Croix ou Pile," an old French term for heads or tails. In it, he worked through a simple question: toss a fair coin twice, what is the chance of getting heads at least once? The correct answer is 3 in 4. There are four equally likely results (heads-heads, heads-tails, tails-heads, tails-tails), and three of the four include at least one heads.
D'Alembert got 2 in 3. His reasoning was that once the first toss lands heads, nobody in real life bothers with a second toss, since the question is already answered. So he counted only three outcomes, heads on the first flip, tails then heads, and tails then tails, and treated all three as equally likely. They are not: the first of those three outcomes is really two outcomes (heads-heads and heads-tails) collapsed into one. That shortcut, letting what already happened quietly change how you count what comes next, sits right next to the mistake gamblers still make at the table: treating a coin, a wheel, or a shoe of cards as though it owes the next result a correction for the last one.
Later mathematicians were not gentle about it. The French mathematician Joseph Bertrand wrote in 1889 that "D'Alembert's astute mind slips completely" the moment probability enters the picture. The statistician Karl Pearson went further in his own history of the field, writing that d'Alembert's actual contribution to probability was, in his words, "absolutely nothing."
So the name on this betting system is not a professional gambler and not a casino insider. It is a genuinely brilliant scientist who, on this one specific problem, got the math wrong in print, and whose name stuck to a system that leans on the same instinct that tripped him up: the belief that outcomes owe each other a balance.
How does D'Alembert compare to Martingale in practice?
It compares well on risk, because it grows so much more slowly. Martingale doubles your bet after every loss, so both your bet size and the bankroll needed to keep playing grow exponentially. D'Alembert only adds one unit at a time, so it grows in a straight line instead of a curve that shoots upward.
Starting from a $10 bet, after 8 losses in a row, Martingale's next bet is already $2,560, and getting to that point means you already had to risk $2,550 across the first 8 bets. D'Alembert's next bet after the same 8 losses is $90, with $360 risked so far. Same losing streak, two very different nights.
Compare the bankroll each system needs for the same losing streak
Set a starting bet and a number of losses in a row. See what each system needs you to have risked to survive that exact streak, and what the next bet looks like.
Martingale, bankroll needed so far
$2,550
D'Alembert, bankroll needed so far
$360
Martingale, next bet required
$2,560
D'Alembert, next bet required
$90
For 8 losses in a row, Martingale needs 7.1x more bankroll than D'Alembert to survive the exact same streak.
Illustrative math only, based on a fixed unit bet and consecutive losses. Not a prediction of any real session, and not gambling advice.
Does a smaller bankroll mean better odds?
No. D'Alembert changes how fast your bet grows when you are losing. It does not change your chance of winning any individual bet, and it does not touch the house edge sitting underneath the game you are playing.
On a single-zero roulette wheel, a bet on red wins about 48.6% of the time and loses about 51.4% of the time, spin after spin, no matter what happened on the last ten spins and no matter what betting system is riding on top of it. Every spin is independent. The wheel has no memory, and neither does a shoe of cards between hands where nothing has been removed to change the count. Betting systems only ever resize the bet. They cannot resize the odds.
Run any staking pattern out over enough bets, flat betting, Martingale, D'Alembert, and the ratio of money lost to money wagered settles toward the same number: the game's house edge. That is what the underlying expected value of a negative-EV game means. No sequence of bet sizes changes the sign of that number, only how bumpy the ride to it feels along the way. D'Alembert's real advantage is a smoother ride and a smaller worst case, which is a legitimate reason to prefer it over Martingale. It is not a reason to expect a different result over time, and it is not a substitute for a real plan around managing a bankroll in the first place.
The bottom line
D'Alembert is the gentler cousin of Martingale: it raises your bet by one unit instead of doubling it, so it needs a fraction of the bankroll to survive the same losing streak, and it is named, honestly, after a real mathematician who got a coin-toss problem wrong in 1754. Neither the man nor the system he lent his name to changed the actual math of a coin flip or a roulette spin. The odds on each bet stay exactly what the game says they are, streak or no streak.
If you are building the game floor rather than betting on it, none of this changes what matters on your side of the table: a correctly configured house edge holds regardless of which staking pattern a player brings to it. Whitelabels.com ships a casino games library with published RTP on every title, so the numbers underneath every bet are never a mystery. For the system this one gets compared to most often, see what the Martingale system does, and for the math that explains why no staking pattern beats a negative-EV game, see what house edge means.
Key takeaways
- The D'Alembert system raises your bet by one unit after a loss and lowers it by one unit after a win, on an even-money bet like red or black in roulette.
- It is named after Jean le Rond d'Alembert (1717 to 1783), a real French mathematician and co-editor of Diderot's Encyclopedie, not a professional gambler.
- In 1754, d'Alembert's own Encyclopedie article on coin tossing, Croix ou Pile, got a basic probability question wrong: he calculated a 2 in 3 chance of at least one heads in two tosses, when the correct answer is 3 in 4.
- For the same losing streak, D'Alembert needs far less bankroll than Martingale. After 10 losses in a row from a $10 bet, Martingale needs $10,230 while D'Alembert needs $550.
- D'Alembert's slower bet growth lowers the risk of one catastrophic loss compared to Martingale, but it does not improve the odds of winning any single bet or shrink the house edge.
- Run for long enough, every betting system, D'Alembert included, loses money at a rate that converges on the game's house edge.
