Baccarat Rules

Key Takeaways

• Baccarat (punto banco) is a no-decision game: after you bet on Player, Banker or Tie, every card is dealt by fixed rules — there is no strategy inside the hand, only in bet selection.
• Card values: ace = 1, pips = face value, tens and faces = 0; only the last digit of a hand's total counts, so 7+8 = 15 = 5.
• Standard 8-deck probabilities: Banker wins 45.86%, Player 44.62%, Tie 9.52%. House edges: Banker 1.06% (after 5% commission), Player 1.24%, Tie 14.36% at 8:1.
• The third-card tableau looks arcane but you never need to apply it — the dealer (or software) executes it identically everywhere.
• Baccarat's low edge and zero decision load make it one of the cheapest table games to play correctly — because "correctly" is just betting Banker and avoiding side bets.

How Baccarat Works

Setup and card values

Punto banco — the version in virtually every casino and online lobby — deals from 6–8 shuffled decks (8 is standard, and all figures here assume it). Two hands are dealt regardless of how many people bet: Player (Punto) and Banker (Banco). These are betting positions, not people; you can back either.

Totals use baccarat counting:

Sum the cards and keep only the last digit: 9+7 = 16 → 6; K+5 = 5; 8+8 = 16 → 6. No hand can exceed 9, and nothing "busts" — the worst total is 0 (baccarat, the name's origin).

The deal

1. Bets down: Player, Banker, Tie (plus optional side bets).
2. Two cards each to Player and Banker.
3. Natural check: if either hand totals 8 or 9, the hand ends immediately — no third cards.
4. Otherwise the tableau (fixed drawing rules) runs:

Player's rule — simple: draw on 0–5, stand on 6–7.

Banker's rule — responds to the Player's third card:

(If Player stood with 6–7, Banker simply draws on 0–5 and stands on 6–7.)

1. Closest to 9 wins. Player win pays 1:1; Banker win pays 1:1 minus 5% commission (i.e. 0.95:1); Tie pays 8:1 (sometimes 9:1) and pushes Player/Banker bets.

The tableau's asymmetry is deliberate: Banker acts with information — the Player's third card — and its rules are optimized to exploit it. That is the entire source of Banker's statistical advantage, and the 5% commission is the casino's mechanism for pricing it back.

The Math: what each bet costs

The fixed rules make baccarat exactly solvable. Standard 8-deck results:

Bet economics per unit staked:

Banker: EV = 0.4586 × 0.95 − 0.4462 = −0.0106  → 1.06% edge
Player: EV = 0.4462 − 0.4586          = −0.0124  → 1.24% edge
Tie:    EV = 0.0952 × 8 − 0.9048      = −0.1436  → 14.36% edge (8:1)

(Tie at 9:1 improves to a still-poor 4.84%. Banker and Player bets push on ties, which is already reflected above.)

What follows from these numbers:

• Banker remains the best bet despite the commission — the tableau advantage is worth more than 5% of Banker's win frequency takes away.
• Ignoring ties, Banker wins 50.68% of resolved hands. This is why every scoreboard shows more red (Banker) than blue over time, and why "follow the shoe" feels validated — the lean is real but already fully priced.
• No-commission variants are not a gift. Paying Banker 6 wins at 50% raises the Banker edge to ≈1.46%. The rebranding sells convenience, not value.
• Side bets are expensive. Pairs, Perfect Pair, Dragon/Panda bonus wagers run mid-single to double-digit edges — an order of magnitude above the main bets they sit beside.

Squeezes, roads and rituals

Baccarat's culture — the slow card squeeze, the Big Road / Bead Plate / derived-road scoreboards tracking patterns — is presentation over an unchanging distribution. Roads have zero predictive power. Shoes hold no exploitable memory at practical scale: card-counting analyses of baccarat find edges too small and rare to act on, a well-established result in the gambling-math literature. Casinos happily print pattern cards because patterns don't matter. Enjoy the theater; audit the math.

Table formats: big, midi, mini, speed

The rules never change; the ritual does, and with it the economics of an hour:

• Big table baccarat — the traditional 12–14-seat VIP format: players take turns dealing from the shoe, winners may squeeze the cards, and hands run slow (~40/hour). The ceremony is the product; minimums are high.
• Midi baccarat — 9 seats, one dealer handles cards but the biggest bettor squeezes. The mainstream Asian-floor format, ~60 hands/hour.
• Mini baccarat — blackjack-sized table, dealer does everything, no squeeze, 70–80+ hands/hour at mass-market minimums. Identical math, roughly double the hourly expected cost of the big table at the same stake, purely through speed.
• Speed/online baccarat — live-dealer studios push past 80 hands/hour; RNG versions run as fast as you click. Online tables also popularized side-bet menus and no-commission rules as default rather than option.

One more variant deserves its line: EZ Baccarat replaces the commission with a rule that a winning three-card Banker 7 pushes ("Dragon 7" bar). Banker's edge lands at 1.02% — actually the best mainstream main bet — while the associated Dragon 7 side bet (40:1) carries a 7.6% edge; the pairing is a masterclass in how the industry gives back a rule and sells a side bet.

A worked session

Concrete cost of an hour, flat-betting $25 on Banker:

Variance towers over these numbers in any single hour — the standard deviation on 75 near-even-money hands at $25 is roughly $215, an order of magnitude above the $20 expected cost. That gap is the game's honest description: cheap on average, swingy in the moment, and utterly indifferent to which scoreboard pattern preceded your bet.

Baccarat vs the other low-edge tables

Baccarat's niche is precise: near-blackjack economics with zero skill risk. A basic-strategy mistake rate of even 1% of decisions erases blackjack's edge advantage; baccarat has no mistakes to make on the main bets.

For Players

• Bet Banker by default. 1.06% vs 1.24% is small but free; over $10,000 of turnover it is $18 of expected value — and there is no countervailing argument.
• Never make the Tie your plan. At 14.36% it costs more per dollar than almost anything else on the casino floor with a "table game" label.
• Streak systems change nothing. Following or fading the shoe, doubling after losses, waiting for "Banker runs" — the next hand's distribution is identical regardless. Systems reshape variance, never expectation.
• Watch the commission bookkeeping (live tables). Commissions accrue per Banker win and settle per house procedure; online it is automatic. In no-commission rooms, remember the Banker-6 half-pay rule is the price.
• Budget by turnover, not buy-in. At 1.06% and ~40–70 hands/hour flat-betting $25, expected cost is roughly $11–$19/hour — among the cheapest premium-feel gambling available, if you stay off the side bets.

For the Industry

• Baccarat is a volume business on thin, stable margins. The game's 1%-class hold percentage is compensated by enormous handle, high bet ceilings and VIP concentration — revenue models should treat volatility of win (not edge) as the dominant planning variable, especially in junket/VIP segments where theoretical vs actual hold diverges wildly short-term.
• Side bets are the margin lever. Main-bet edges are fixed by tradition; monetization innovation happens entirely in side-bet design (edges 5–15%) and in no-commission repackaging (+0.4pp on Banker). Both trade player value for hold and should be disclosed accordingly.
• Speed formats multiply exposure. Speed baccarat's hand rate raises turnover per seat-hour proportionally; RG analytics should track hands/hour as a risk variable, since the per-hand cost illusion ("only 1%") compounds fast at 80+ hands/hour.
• Live-dealer baccarat is a studio product. Card handling, squeeze mechanics and road displays are regulated presentation; the SlotCodex catalog models live and RNG baccarat as one game type with distinct delivery formats, since their math is identical while their certification paths differ.