Baccarat Terminal · All guides

Baccarat Martingale Strategy: Exact Bust Odds, Bankroll and Real Expected Result

The baccarat martingale is the most popular betting progression at the table because it is easy to state: bet one unit, and after every loss double the bet; after a win, go back to one unit. Each time you win, you are one unit up on the cycle, so the system "always" seems to be ahead until it is not. This page works the numbers out exactly for baccarat rather than for a generic coin flip, because baccarat has two details that change the picture: ties are pushes (they do not move the ladder) and a winning Banker bet pays only 0.95 after the 5% commission. We use the 8-deck probabilities from our odds page (Banker 50.68% and Player 49.32% once ties are removed) and ask four questions: what does each step cost, how often does the ladder run out, how much bankroll is needed, and what is the average result of a session.

How the ladder works

Take a $10 base bet on Banker. A tie returns your stake and the same bet stays on the table; only a Player win counts as a loss. The ladder looks like this:

The doubling formula is simple: step k costs 2^k−1 units and the cycle has staked 2^k−1 units by the time that bet is placed. A seven-step ladder needs 127 units on the table; the eighth bet would be 128 units.

The commission quirk: a Banker win at step 6 is a net loss

In textbook martingale, any win recovers everything plus one unit. At the baccarat table a Banker winner pays 0.95, so the cycle nets 1 − 0.05 × 2^k−1 units. That is +0.95 at step 1, +0.20 at step 5, then −0.60 at step 6 and −2.20 at step 7. In other words, if the first five Banker bets all lose and the sixth one finally wins, you have won the hand and lost money on the cycle. Betting Player removes the quirk (every win nets +1) but Player carries a bigger edge (1.24% versus 1.06%), so it is not a free fix.

How often does the ladder run out?

With Banker only, you lose a decision 49.32% of the time (ties excluded). Running out after n losses in a row therefore has probability 0.4932ⁿ per cycle:

Each extra step halves the bust chance and doubles the loss, which is why the average does not move: you are buying a smaller chance of a bigger disaster. Table limits cap the ladder too. At a table with a $10 minimum and a $1,000 maximum, the seventh bet ($640) is the last one allowed, so a $10 martingale there is really a 7-step system and runs out about once every 141 cycles.

What a realistic target looks like

Suppose you play with a fixed ladder, stop as soon as you are 10 units ahead, and also stop (bust) if the ladder runs out. We solved this exactly for Banker bets:

Read the second column and the system looks wonderful: 91 sessions in 100 reach +10. Read the third column and the catch appears: the average result is negative, because the 8.55% of sessions that bust give back around 127 units each. The more steps you add, the higher the "success rate" and the worse the average. That trade — frequent small wins, rare huge losses — is the whole system.

Why the edge does not move

Every bet still carries the same 1.0579% edge (1.1691% once ties are ignored). A 7-step Banker ladder stakes on average 6.72 units per cycle and loses 0.0786 units per cycle, which is exactly 1.1691% of the money staked. A 10-step ladder stakes 9.41 units per cycle and loses 0.110 units. Martingale makes you bet more money per cycle, so it increases the total expected loss while hiding it in a long series of small wins.

What actually happens in a shoe

In an 80-hand shoe (about 72 decisions), a Banker bettor meets at least 5 straight losses with probability 67.1%, at least 7 with 21.9% and at least 8 with 11.2%. Over ten shoes, at least one 7-loss run happens 92.9% of the time and an 8-loss run 72.4%. Streaks like this are not exotic — they are routine over a few sessions. You can count the real streaks in the shoes you watch with the live baccarat tracker or the free Baccarat Scoreboard app; a patient look at twenty shoes is more convincing than any argument.

What our backtest found

We replayed martingale-family progressions on real shoe data: 3,810 setups combining 13 progression systems, 14 signals, and different units, targets, stop-losses and caps. Tested on 6,000 shoes, 59 setups looked profitable. On the next 10,000 and then 120,825 shoes, zero stayed profitable. Martingale-style plans hit their daily target 80-90% of the time but lost 50-100% of the bankroll on 10-30% of days, and the 30-day median was 0.00. See do baccarat betting systems work? for the wider comparison.

Variations and what they change

• Paroli (reverse martingale): double after wins, reset after losses. Losses are small, but the edge on money wagered is unchanged.
• D'Alembert, Fibonacci, 1-3-2-6: flatter ladders with smaller peaks and smaller wins; the expected loss per unit staked is the same 1.06% on Banker.
• Betting after a streak or "pattern": changes which side you bet, not the odds of the hand — see baccarat patterns.

If you want to compare systems safely, try them in the free practice game at free baccarat with play money, or read the overall baccarat strategy guide.

Frequently asked questions

Figures: 8-deck exact probabilities, ties as pushes, decisions independent; ladder and session results computed by Baccarat Terminal, October 2026. 18+ · for information only, not gambling advice. Please follow the laws where you live.