No baccarat strategy beats the house. The only meaningful choice is which bet to place, and betting Banker every time gives the lowest house edge in the game, about 1.06% with the usual 5% commission. Roadmaps, streak-chasing and betting progressions change how your money moves around, but not how much you lose per unit staked on average.
The rest of this page goes through the common systems one at a time, with the maths behind each verdict.
Why baccarat has no strategy to learn
In blackjack you decide whether to hit, stand, double or split, and some decisions are better than others. In baccarat the third-card rules are fixed (see the draw chart). After you place a bet there is nothing to decide. The game is closer to a coin flip with a built-in tax than to a skill game.
That leaves two decisions: which bet and how much. For the first, the Wizard of Odds lists these house edges for an 8-deck game:
| Bet | Pays | House edge |
|---|---|---|
| Banker | 19:20 (even money less a 5% commission) | 1.06% |
| Player | 1:1 | 1.24% |
| Tie | 8:1 | 14.36% |
The approximate win rates behind them, for 8 decks, are Banker 45.86%, Player 44.62% and Tie 9.52%. Banker wins more often than Player but pays less than even money because of the commission, and the net effect is still negative.
That is the whole mathematical case for “banker only”. It does not make you a winner. It minimises the cost.
Banker only: the right default, not a winning system
Betting Banker every hand is the cheapest way to play. At $25 a hand and 60 hands an hour, you wager $1,500 an hour and the average loss is about $16 at 1.06%. Betting Player costs about $19. Betting Tie costs about $215.
| Bet at $25 per hand, 60 hands/hour | Expected loss per hour |
|---|---|
| Banker | about $16 |
| Player | about $19 |
| Tie | about $215 |
These are averages. In a single session the result will be dominated by chance. The point is that the Tie bet’s headline 8:1 payout hides a much higher price.
Mixing Banker and Player does not help or hurt in any systematic way. Each hand is a fresh bet at the odds above, so the average cost is just a weighted mix of 1.06% and 1.24%.
Roadmaps: what they are and what they cannot do
Baccarat tables often display roadmaps. The Bead Plate shows each result in order, in a grid. The Big Road starts a new column each time the result switches between Banker and Player, so streaks run downward. The smaller derived roads (Big Eye Boy, Small Road, Cockroach Pig) are built from the Big Road using pattern rules.
All of them are records of what has already happened. They are useful for one thing: seeing how the session has gone. They do not predict anything, because each hand is dealt from a shoe of 416 cards (8 decks) that does not remember patterns, and the rules produce the same probabilities each hand regardless of what the last hands were.
There is one honest caveat. Cards leave the shoe as hands are played, so the probabilities do shift very slightly as a shoe is dealt. That effect is real, but it is tiny, and a roadmap does not measure it. Nothing about a “choppy” or “streaky” road tells you the remaining cards are favourable.
So when someone says a road pattern “shows” that Player is due, they are describing a feeling. The same is true if they say Banker “is hot”. The road is a diary, not a forecast.
The gambler’s fallacy, in baccarat form
If Banker has won five times in a row, the chance of Banker on the next hand is still about the same as it was before the streak, not lower. The inverse mistake is equally common: assuming that a streak will continue because it has gone on so far. Both come from the same belief, that independent events are linked.
Betting progressions: change the pattern, not the price
A progression changes how much you bet after a win or a loss. The idea is that clever sizing turns a losing bet into a winning plan. It cannot, because every bet has the same negative expected value, and the total expected loss is the sum of the edges on each bet you place.
In other words, the expected loss equals your total amount wagered times the house edge. A progression changes the pattern of results and often increases the amount wagered, but it cannot make the edge positive.
Doubling after a loss at the baccarat table
Double the bet after each loss, return to the base bet after a win. Each winning round nets one base unit.
The trouble is the losing streak. Say you bet Banker at a $10 base bet. Losing a bet on a resolved hand (ignoring ties, which push) happens about 49.3% of the time, so seven losses in a row occurs on roughly 0.7% of rounds, about once in 140 rounds. After seven losses you are down $1,270 (10 + 20 + 40 + 80 + 160 + 320 + 640), and the next bet you would need is $1,280, which may exceed the table maximum.
| Consecutive losses | Total lost so far (base $10) | Next bet needed |
|---|---|---|
| 3 | $70 | $80 |
| 5 | $310 | $320 |
| 7 | $1,270 | $1,280 |
| 9 | $5,110 | $5,120 |
Martingale produces many small wins and occasional large losses. Its expected result per unit wagered is the same as flat betting.
Paroli (pressing a win streak)
Double after a win, up to a set number of wins (often three), then reset. It risks small amounts in a losing run and tries to ride a winning one. The chance of three wins in a row on Player is about 12% per attempt, so most rounds end with a loss of the base bet. Again, the edge on each bet is unchanged.
Fibonacci and others
Fibonacci uses the sequence 1, 1, 2, 3, 5, 8, 13 and moves along it. It grows more slowly than Martingale, so the downside is slower to build, but a long losing streak still costs more than the typical win. D’Alembert, Labouchère and similar plans work the same way: they rearrange when you win and lose, and the expected loss per unit wagered stays the same.
If you want to see the effect rather than take it on trust, use the bankroll simulator on this site. Run a flat bet and a Martingale for the same number of hands and compare how often each one hits the table limit or runs out of funds.
Dragons and streaks
A long run on the Big Road, seven or more of the same result, is called a “dragon”. The probability of a run is simple multiplication: for Banker winning seven resolved hands in a row (ignoring ties), it is about 0.507 to the power of 7, a little under 1% from any given starting point. Over a long session, runs like that will appear.
Do they mean anything? No. People bet on dragons continuing (“follow the dragon”) or ending (“cut the dragon”), and both approaches run at the same house edge as any other hand.
Verdict table
| Idea | Verdict | Why |
|---|---|---|
| Bet Banker every hand | Lowest cost, not a winner | Edge about 1.06% |
| Bet Player | Slightly worse | Edge about 1.24% |
| Bet Tie | Avoid | Edge about 14.36% |
| Read roadmaps | Entertainment only | Hands are effectively independent |
| Follow or cut the dragon | No edge | Streaks happen at random |
| Martingale | Variance trade, not an edge | Same expected loss per unit wagered |
| Paroli, Fibonacci | Variance trade, not an edge | Same |
| Counting in baccarat | Impractical | Tiny, rarely usable effect |
Common mistakes
- Believing a system “wins most sessions” means it wins. A system that wins small amounts often and loses a lot rarely will feel good until the big loss arrives.
- Raising bets to chase losses. That is exactly what Martingale automates.
- Placing Tie bets on a hunch. The payout looks generous, but the Tie bet has the highest cost on the table.
- Hunting for patterns in the road. The mind finds shapes in random sequences.
- Ignoring table limits. They are what makes a doubling system fail in practice.
Try it yourself
Open the bankroll simulator, set a starting bankroll, a bet size and a number of hands, and run the same session with flat betting and a Martingale. Look at the distribution of outcomes rather than a single run. You will see that the average result is negative in both, with different shapes.
Baccarat is cheap to play at Banker or Player, but the casino is still ahead on average, so decide what you are willing to spend before you sit down.
Related guides
- Baccarat draw rules for Player and Banker: when each side draws, with worked examples.
- Casino games ranked from best to worst odds: about 30 bets in order, with cost per hour.