Baccarat.
Bet banker, every time.
Baccarat is the lowest-house-edge non-skill game on the floor. There is one decision: which of three bets to place. Math says one. The other two exist to entertain players who want to feel like they are playing.
Baccarat, in 60 seconds
You bet on one of three outcomes: Banker, Player, or Tie. Two hands are dealt, the dealer follows fixed drawing rules (no decisions for anyone at the table), and the hand closest to 9 wins. Tens and face cards count as 0; aces count as 1; double-digit totals drop the tens place (15 becomes 5). That is the entire game.
Banker pays 1:1 minus 5% commission (so 0.95:1 net). Player pays 1:1 flat. Tie pays 8:1 (occasionally 9:1 at premium tables, still bad).
House edge per bet
| Bet | Win prob | Pays | House edge |
|---|---|---|---|
| Banker (with 5% commission) | 45.86% | 0.95:1 | 1.06% |
| Player | 44.62% | 1:1 | 1.24% |
| Tie (8:1) | 9.52% | 8:1 | 14.36% |
| Tie (9:1) | 9.52% | 9:1 | 4.84% |
Banker bet, expected value win : 0.4586 × $0.95 = +$0.4357 lose : 0.4462 × −$1.00 = −$0.4462 push : 0.0952 × $0.00 = $0.0000 EV = −$0.0105 per $1 wagered → 1.06% house edge Player bet, expected value win : 0.4462 × $1.00 = +$0.4462 lose : 0.4586 × −$1.00 = −$0.4586 push : 0.0952 × $0.00 = $0.0000 EV = −$0.0124 per $1 wagered → 1.24% house edge
Banker wins more often than Player because the drawing rules favour it (banker draws second, conditioned on what player drew). The 5% commission was added to push the banker edge above zero. Without commission, banker would be a +1.24% player advantage, which is the actual size of banker’s underlying win-rate edge.
Tie bet, the worst common bet
The tie bet pays 8:1 (sometimes 9:1) but ties only happen ~9.5% of the time. The math:
Tie bet (8:1 paytable, the standard) win : 0.0952 × $8.00 = +$0.7619 lose : 0.9048 × −$1.00 = −$0.9048 EV = −$0.1429 per $1 wagered → 14.36% house edge That is 13× worse than the banker bet, and worse than every common roulette bet, every blackjack mistake, every slots edge. The 9:1 tie paytable cuts it to 4.84%, still 4.6× banker.
The tie bet exists to convert pattern-spotters into bigger losers. It is the “feel like I outsmarted the table” bet. Don’t.
The road, the dragon, and other fan fiction
Live baccarat tables ship with elaborate pattern boards: Big Road, Bead Plate, Big Eye Boy, Small Road, Cockroach Pig. Each tracks streaks and switches in a slightly different visual code. Casinos love these. Players love these. They do nothing.
Baccarat outcomes are conditionally independent. The drawing rules condition each hand on the cards dealt that hand, not on the previous hand. There is no memory across shoes. There is no gambler’s ruin you can read off the road. The boards exist because they make the game feel solvable.
If a streak shows on the road, the next outcome still has the same banker / player / tie probabilities. Past outcomes are noise, not signal.
The closest baccarat gets to a real edge is shoe-edge tracking (counting cards in baccarat), which can flip the EV slightly toward the player on rare occasions deep into a shoe. The math advantage is small, the conditions are rare, and crypto-casino live baccarat uses auto-shufflers that make it impossible. Treat shoe-counting as a curiosity.
Worse than the tie
Premium baccarat tables offer side bets: Dragon Bonus, Perfect Pair, Lucky 6, Either Pair. The house edges compete for last place.
| Bet | House edge | What it pays on |
|---|---|---|
| Dragon Bonus | ~2.65 to 9.4% | Margin of victory plus naturals. Variance product, math product. |
| Perfect Pair | ~13.0% | Banker or Player first two cards form a same-suit pair. |
| Either Pair | ~14.4% | Either side's first two cards form a pair (any suit). |
| Lucky 6 | ~7.6% | Banker wins with a total of 6. |
| Big / Small | ~5.3% / ~4.4% | Total cards dealt across both hands. Asia-pivot bet, still bad. |
The single rule, identical to blackjack’s side-bet rule: don’t take any. They erase the entire reason you came to baccarat in the first place.
The whole strategy in one paragraph
Bet banker every hand. Skip the commission-free tables (they sneak the edge back via pushing on banker-6, which bumps house edge to ~1.46%). Skip every side bet. Skip the tie. Match your unit size to the bankroll guide (1 to 2% of session bankroll for low-volatility games like this). At 1.06% house edge, baccarat extracts the least value per dollar wagered of any non-skill game on the floor. That is the entire pitch.
Pace matters. Live baccarat at 50 hands per hour at $20 a hand costs you $20 × 50 × 1.06% = ~$10.60 per hour in expectation. RNG baccarat at 200 hands per hour at the same stake costs you ~$42.40 per hour. Same edge, same bet, four times the bleed. If you want to play long, go live.
