Baccarat Terminal · All guides

Baccarat Strategy: What Actually Works, With the Math

Search for "baccarat strategy" and you will find pattern charts, progressions and "secret" systems. Almost none of them survive a calculator. Baccarat's outcome is decided by fixed drawing rules and the cards in the shoe, so the house edge on each bet is a fixed number. A real strategy is therefore not about predicting the next hand — it is about the few decisions that do change your results: which bet, which table, how much, and for how long. This guide goes through each one with exact numbers.

1. Bet selection is the whole game

The single most effective baccarat strategy is to bet Banker and ignore the Tie and the side bets. Banker wins more often (45.86% of rounds against 44.62%) because it acts second under the third-card rule; even after the 5% commission it is the cheapest bet on the table. Player is a close second. Everything else costs four to fourteen times more per dollar. Full breakdown: baccarat odds and house edge.

2. Table selection: read the rule sign

• Standard (5% commission): Banker 1.06%, Player 1.24% — bet Banker.
• No-commission (Banker win with 6 pays half): Banker rises to 1.46%, so Player at 1.24% is the better bet at these tables.
• EZ Baccarat (Banker three-card 7 pushes): Banker 1.02% — marginally the best main bet anywhere.
• Tie paying 9 to 1 instead of 8 to 1 cuts the tie edge from 14.36% to 4.84% — still far worse than Banker.

How the variants differ: speed, mini, live, EZ and no-commission baccarat.

3. Speed is a cost

Your expected loss is edge × amount wagered, and the amount wagered grows with the number of hands. A land-based table deals perhaps 60–80 hands an hour; speed baccarat can pass 120. At $10 on Banker, that is roughly $6–8 an hour of expected cost at a normal table and about $13 an hour at a speed table. Playing fewer hands — skipping rounds, taking breaks — is the only "strategy" that reduces what the edge takes from you.

4. Understand variance — why short sessions can win

A Banker bet returns +0.95, −1 or 0 (on a tie). Its expected value is −0.0106 units per hand but its standard deviation is about 0.93 units, almost ninety times larger. In the short run luck dominates; in the long run the edge does. Here is the chance of being ahead after flat Banker bets:

This is the honest answer to "how to win at baccarat": winning sessions are common, winning careers are not. A strategy that respects this sets a session budget, accepts that a +10 unit night and a −15 unit night are both normal, and stops.

5. Betting systems do not change the edge

Martingale (double after a loss), Paroli (double after a win), 1-3-2-6, Fibonacci and D'Alembert only change when you bet big. Every individual bet still carries the same 1.06% edge, so the total expected loss is the same fraction of everything you wager — usually a larger total, because progressions increase average bet size.

Martingale shows the trade-off most clearly. Betting only Banker, a run of 7 losses in a row (ties don't count) happens at least once in 21.9% of shoes and in 92.9% of ten-shoe sessions. Seven losses from a 1-unit start cost 127 units, and the eighth bet must be 128 units — past most table limits. The system wins small amounts often and gives it all back in one run. Worked examples: do baccarat betting systems work?

6. Patterns and "reading the road"

Streaks and chops on the big road look meaningful, but they are what random sequences look like. In a typical 8-deck shoe (about 72 non-tie hands) there is a 69% chance of at least one streak of six or more of the same side, and a 42% chance of seven or more. A long dragon is normal, not a signal. Baccarat's card removal is too weak to make the next Banker/Player result predictable from the scoreboard. Pattern tracking is useful for one thing: measuring, honestly, how often a pattern actually continued — which is exactly what our pattern analysis guide explains, published hit rates included.

7. Card counting in baccarat

It is mathematically possible to count cards in baccarat, but for the main bets the effect of removed cards is so small that a perfect count gains fractions of a cent per hand and only near the end of the shoe. Mathematical analyses going back decades agree it is not practical for Banker or Player. Some side bets (such as pairs or the Dragon 7) react more strongly to the remaining cards, but opportunities are rare and casinos often shuffle early. It is not a reliable strategy for normal play.

8. Bankroll rules that make sense

1. Decide the session budget first — money you can lose — and a unit of 1–2% of it.
2. Flat bet the same unit. It keeps the expected cost proportional to how long you play.
3. Set a stop-loss and a time limit. Chasing losses is how a normal −15 unit session becomes a −60 unit one.
4. Record your sessions. Most players remember wins and forget losses. A written record keeps the numbers honest.

For recording, the free Baccarat Scoreboard app saves every shoe by date with Banker/Player/Tie counts, streaks and side-bet results, and it works offline.

Strategy FAQ

House edges from exhaustive enumeration of an 8-deck shoe; session and streak figures from the exact per-hand distribution (normal approximation for session totals). 18+. Informational only.