Level 4 · Lesson 14 of 4 · Strategic Thinking
Betting Systems: Martingale, Paroli, Fibonacci, and Why None of Them Work
The principle that governs everything in this lesson
The expected value of a series of bets equals the sum of the expected values of each individual bet.
If every Banker bet carries an expected return of -1.06%, then 100 Banker bets carry a combined expected return of -106 units, regardless of how those bets are sized or sequenced. You cannot improve the expected value of a series of negative-expectation bets by rearranging them. This is the linearity of expectation, and it is not an opinion. It is the foundation of probability theory.
A betting system applied to a negative-expectation game cannot make the game positive-expectation. All it can do is redistribute the losses across different configurations of wins and losses, sometimes in ways that feel better and sometimes in ways that are catastrophically worse. Eliot Jacobson's work on baccarat progressions at apheat.net frames this with rigour: the system changes the shape of the loss distribution, not the expected loss.
The Martingale
Start at a base stake. After every loss, double your bet. After a win, return to base. The logic: a win at any point covers all previous losses and returns a one-unit profit.
At a $10 base:
- Hand 1: bet $10. Lose. Total loss: $10.
- Hand 2: bet $20. Lose. Total loss: $30.
- Hand 3: bet $40. Lose. Total loss: $70.
- Hand 4: bet $80. Lose. Total loss: $150.
- Hand 5: bet $160. Win. Net profit: $10.
The system produces frequent small wins. The problem is the losing streak. A seven-hand losing streak requires a bet of $1,280 on the eighth hand to recover. If the table maximum is $500 (common at London rooms), the progression hits the ceiling before recovery is possible. You take the full loss with no covering win.
The Heliot Salon Prive at the Hippodrome on Leicester Square posts its table maximums. So does every Macau room on the Cotai Strip. Table limits exist precisely to neutralise the Martingale's recovery logic. The system was designed before table limits existed. It does not function as described in a modern casino.
A seven-consecutive-loss sequence on Banker bets happens with a probability of roughly 0.49% per hand sequence. That sounds rare. At 65 hands an hour over a two-hour session, you're playing 130 hands. The probability of hitting at least one seven-loss run in 130 hands is not negligible. The Bellagio Salon Prive in Las Vegas has seen it happen at four-figure base stakes.
The Paroli
The Paroli (the Reverse Martingale) doubles the bet after each win, returning to base after a loss or after three consecutive wins. The logic: ride winning streaks and limit losses to the base stake.
A typical Paroli run: start at $10, win, bet $20, win, bet $40, win, collect $70 profit, return to $10. Or: start at $10, lose, return to $10, net loss $10.
The Paroli is less dangerous than the Martingale because losses are always limited to the base stake. The three-win sequence payoff is satisfying. But the expected value of each hand remains -1.06% on Banker. The system creates a structure where small losses are frequent and occasional triple-win streaks feel like success. Emotionally, this has appeal. Mathematically, it's the same edge.
The Fibonacci
The sequence: 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, each number the sum of the two before it. Applied to betting: after a loss, move one step right along the sequence. After a win, move two steps left. Return to the start when you reach the beginning.
The Fibonacci is a slower progression than Martingale. It reaches large bet sizes after more hands, and recovery after a loss is smoother. The mathematical fact doesn't move: each hand carries a 1.06% edge. A slower sequence of negative-expectation bets is still a sequence of negative-expectation bets.
The D'Alembert
Increase the bet by one unit after a loss, decrease by one unit after a win. Named after the eighteenth-century French mathematician Jean le Rond d'Alembert, who incorrectly believed that after a coin lands heads, tails becomes more likely.
The system embeds the gambler's fallacy in its design. The independence of each hand from the last is not altered by bet sizing. The edge is 1.06% per hand. The D'Alembert spreads losses more gently. It does not recover them.
Why systems persist
Systems persist because they feel like solutions. The Martingale in particular creates a subjective experience of winning: most sessions end with small profits, and the catastrophic losing session is rare. When it arrives, players blame bad luck rather than the expected outcome of a rare event. The emotional appeal is real. Systems reduce decision fatigue. They give a session a structure.
They are emotional management tools. They are not mathematical ones.
The practical upshot
Flat betting Banker is mathematically equivalent in expected value to any progression system. It never runs into table limits. It never creates a situation where one bad streak eliminates the session budget. If you find a system useful for managing your session behaviour, that is a legitimate use of it. Understand that you're buying structure, not edge. The maths doesn't care about your progression.
Key numbers
| System | Bet after loss | Bet after win | Max loss at 7-hand streak ($10 base) | Expected value |
|---|---|---|---|---|
| Flat betting | Same | Same | $70 | -1.06% per hand |
| Martingale | Double | Reset to base | $1,270 | -1.06% per hand |
| Paroli | Same | Double (reset after 3 wins) | $10 | -1.06% per hand |
| Fibonacci | Next in sequence | Two steps back | Sequence-dependent | -1.06% per hand |
| D'Alembert | +1 unit | -1 unit | Moderate | -1.06% per hand |
Sources: Eliot Jacobson on progressions, Hippodrome baccarat, Bellagio casino, Evolution Gaming baccarat.
Welcome to the lesson on betting systems.
I'm Annabel, and I'll be direct about this. Betting systems are not strategies. They are stake sequences. The expected value of your baccarat session is determined by the house edge and your total wagering, not by the pattern of your bets.
Here is the principle that governs everything in this lesson.
The expected value of a series of bets equals the sum of the expected values of each individual bet. If every Banker bet carries an expected return of negative one point zero six percent, then one hundred Banker bets carry a combined expected return of negative one hundred and six units, regardless of how you size or sequence them. You cannot improve the expected value of a series of negative-expectation bets by rearranging them. This is probability theory, not opinion.
The Martingale. Start at a base stake. After every loss, double your bet. After a win, return to base. The logic: a win at any point covers all previous losses and returns a one-unit profit.
At ten dollars base: lose hand one, bet twenty. Lose hand two, bet forty. Lose hand three, bet eighty. Lose hand four, bet one hundred and sixty. Win hand five. Net profit: ten dollars.
The system produces frequent small wins. The problem is the losing streak. A seven-hand losing streak requires a bet of one thousand two hundred and eighty pounds on the eighth hand to recover. If the table maximum is five hundred pounds, which is common at London rooms including the Hippodrome, the progression hits the ceiling before recovery is possible. You take the full loss with no covering win.
Table limits exist precisely to neutralise the Martingale's recovery logic. The Heliot Salon Prive posts its limits. Every room on the Cotai Strip in Macau posts its limits. The system was designed before table limits existed and no longer functions as described.
The Paroli. Double after each win, return to base after a loss or after three consecutive wins. Logic: ride winning streaks, limit losses to one unit.
A Paroli run: bet ten, win, bet twenty, win, bet forty, win, collect seventy dollars profit, return to ten. Or: bet ten, lose, return to ten. Loss: ten dollars.
The Paroli is less dangerous than the Martingale. Losses are capped at the base stake. The three-win sequence payoff is satisfying. But the expected value of each hand is still negative one point zero six percent. The system creates a structure where small losses are frequent and triple-win streaks feel like success. Emotionally, this has appeal. Mathematically, it's the same edge.
The Fibonacci. The sequence: one, one, two, three, five, eight, thirteen, twenty-one, thirty-four, each number the sum of the two before it. After a loss, move one step right along the sequence. After a win, move two steps left. Return to the start when you reach the beginning.
The Fibonacci is a slower progression than Martingale. It reaches large bet sizes after more hands. The mathematical fact doesn't move: each hand carries a one point zero six percent edge. A slower sequence of negative-expectation bets is still a sequence of negative-expectation bets.
The D'Alembert. Increase the bet by one unit after a loss, decrease by one unit after a win. Named after the eighteenth-century French mathematician who incorrectly believed that after a coin lands heads, tails becomes more likely. The system embeds the gambler's fallacy in its design. The edge is one point zero six percent per hand. The D'Alembert spreads losses more gently. It does not recover them.
Why do systems persist? Because they feel like solutions. The Martingale in particular creates a subjective experience of winning: most sessions end with small profits. The catastrophic session is rare, and when it arrives, players blame bad luck rather than the expected outcome of a rare event. Systems reduce decision fatigue. They give the session a shape. These are real benefits.
They are not mathematical benefits. They are emotional ones.
Flat betting Banker is mathematically equivalent in expected value to any progression system. It never hits table limits. It never creates a situation where one bad run wipes the session. If you find a system useful for managing your session discipline, use it for that purpose. Understand what you're buying.
Here is the observation that I think closes this lesson. The Hippodrome's high-limit room has watched Martingale progressions play out at thirty thousand pounds a hand. Not as an unusual event. As a recurring one. Players who wouldn't consider themselves gamblers in any casual sense arrive with a clear system, execute it with discipline, and encounter a losing streak long enough to exhaust the progression. The edge doesn't care about the elegance of the system. Neither does the variance. Both are patient.
The maths doesn't care about your progression.