An editorial encyclopedia of casino table games · Vol. III · MMXXVI
Front Page / The Canon / Baccarat

Baccarat

/ˈbækərɑː/ · Punto Banco · Chemin de Fer
Baccarat casino table with cards spread — the canonical game of nine
Image: Pixabay Content License.

I. Origin — From 15th Century Italy to the Macau VIP Room

The earliest documented baccarat ancestor is the 15th-century Italian game baccara (meaning "zero" — the value of all face cards and tens in the game's scoring system). It crossed the Alps to France around 1490, where it became chemin de fer ("railway") — so named for the slow movement of the shoe from player to player. Two formal variants emerged in 19th-century French casinos: Baccarat Banque (a permanent banker) and Chemin de Fer (rotating banker).

The American version, Punto Banco, was standardized at Havana's Capri Hotel in 1958 by Tommy Renzoni and exported to Las Vegas's Sands in 1959. Punto Banco eliminated player banker rotation — both hands are dealt by the casino, the player only chooses which hand to bet on. This sterile version traveled to Macau in the 1960s and became the dominant Asian table game by 1980. Today over 95% of baccarat played worldwide is Punto Banco.

II. The Rules — Brutally Simple

  1. Two hands are dealt: Player and Banker. Bettors wager on Player win, Banker win, or Tie.
  2. Card values: 2-9 face value, 10/J/Q/K = 0, Ace = 1.
  3. Hand totals are computed modulo 10 (only the units digit counts). A hand of 7 + 8 = 15 → 5.
  4. If either initial two-card hand totals 8 or 9, that's a "natural" — no further cards dealt, hand wins immediately.
  5. Otherwise, the third-card rule applies (full table below).
  6. Hand closest to 9 wins. Banker pays 19:20 (1:1 minus 5% commission). Player pays 1:1. Tie pays 8:1 (or 9:1 in some venues).

The Third-Card Rule (Player)

  • Player total 0-5: Draw
  • Player total 6-7: Stand
  • Player total 8-9: Natural (already won; Banker doesn't draw)

The Third-Card Rule (Banker — depends on Player's third card)

Banker TotalDraws if Player's 3rd card is...
0-2Always draws
3Anything except 8
42-7
54-7
66-7
7Always stands

III. The Mathematics — Why Banker Wins 50.68% of Resolved Hands

Casino dealer handling baccarat cards — the mechanical execution of the third-card rule
Image: Pixabay Content License.

Exact probabilities per 8-deck shoe (the standard configuration):

OutcomeProbabilityPayoutHouse Edge
Banker wins45.86%1:1 minus 5% commission1.06%
Player wins44.62%1:11.24%
Tie9.52%8:114.36%
(Tie alt. payout)9.52%9:14.85%

Banker's mathematical advantage comes from positional asymmetry: the Banker's third-card decision is made after seeing Player's third card. Banker can adapt; Player cannot. The 5% commission was calibrated in 1959 to leave Banker very slightly less attractive than Player after commission — but the calibration was inexact. Banker's net 1.06% remains marginally better than Player's 1.24%.

IV. The 5% Commission — and Why It Hurts Cash Flow

Standard Banker bet wins pay 1:1 nominally, but 5% of the winnings is owed back to the casino. On a $100 winning Banker bet, the casino pays $95 and tracks the $5 commission in a side compartment. The commission is collected at the end of the shoe, not after each hand — meaning a $100/hand bettor on a six-decision shoe with three Banker wins ends with a $15 commission tab.

Variants:

  • EZ Baccarat / No-Commission Baccarat: pays 1:1 on Banker except a winning Banker total of 7 with three cards, which pushes. House edge 1.018% on Banker (slightly better than standard 1.06%), but adds the "Dragon 7" side bet (Banker wins with three-card 7), which carries ~7.6% house edge — the side bet is where the casino recovers its commission income.
  • Lucky Six / 6 Trick: Banker wins with a 6 pays 0.5:1. House edge on Banker rises to 1.46%.

The 5% commission is universal in Macau, Las Vegas, Atlantic City. No-commission tables exist mostly as a marketing variant in California Indian casinos and parts of Asia.

V. Why Baccarat Dominates Macau

Macau casino skyline at night — the baccarat capital of the world
Image: Pixabay Content License.

According to Macau's Direcção de Inspecção e Coordenação de Jogos (DICJ), baccarat (combined Punto Banco + minor variants) accounts for:

  • 88% of Macau's total HK$207 billion table-game revenue in 2023
  • 96%+ of VIP-room revenue (high-stakes rooms exclusively offer baccarat)
  • Average VIP single-bet size: HK$25,000-HK$200,000
  • Top single-bet record (verified): HK$10 million per hand at Wynn Macau, 2015

The cultural-mathematical-structural triad — game of pure chance, high-bet ceiling, junket-room dead-chip economics — created a flywheel that no Macau operator has been able to break. Sands China's repeated attempts to push craps or roulette to Asian whales have all failed; the whales return to baccarat tables within minutes.

VI. The "Derived Roads" — Sophisticated Superstition

Macau-style baccarat displays six "roads" (路 / lù) on overhead screens, tracking shoe history:

  • 珠路 (Bead Plate) — raw results in a 6-row grid, red = Banker, blue = Player, green = Tie
  • 大路 (Big Road) — collapsed display showing streaks of same outcome stacked vertically
  • 大眼路 (Big Eye Road) — derived from Big Road, indicates whether the current shoe is "regular" or "choppy"
  • 小路 (Small Road) — alternative derivation with different shift parameter
  • 曱甴路 (Cockroach Pig) — third derived road
  • Tie Marks — overlay of ties on whichever road is displayed

The mathematics: each hand is independent (the shoe penetration that would allow card-counting analysis is too shallow — typically 75% before reshuffle). The roads describe pattern, not probability. Treating them as predictive is the cluster illusion — the human tendency to perceive patterns in random data.

And yet the roads remain. Why? Because they keep players betting longer (Macau studies show 20-30% session-length increase when roads are visible), which is exactly what the casino wants.

VII. Edge Sorting — The Phil Ivey Case

In 2012, professional gambler Phil Ivey won £7.7 million at Crockfords Casino (London) playing punto banco. His method, executed with partner Cheung Yin Sun: identifying a manufacturing defect in Gemaco brand cards where the diamond pattern on the back was slightly asymmetric. By asking the dealer to rotate certain high-value cards (8s and 9s) 180° before reinserting them in the shoe — under the pretext of "superstition" — Ivey could later identify those cards by their orientation, gaining a 6.7% edge on Player bets.

Crockfords refused to pay. Ivey sued. The UK High Court ruled in Crockfords' favor (2014), upheld by Court of Appeal (2016) and Supreme Court (2017): edge sorting constituted "cheating" under the Gambling Act 2005 even though Ivey did not handle the cards. Lesson: the legal boundary of "advantage play" is jurisdiction-specific and ill-defined. Ivey separately lost a US case at Borgata for $10.1 million using the same technique.

VIII. Online Baccarat — Live Dealer vs RNG

Two formats dominate online:

  • RNG baccarat: Each hand uses fresh shuffle; pace 200-400 hands/hour; house edge identical to physical (Banker 1.06%). Common at NetEnt, Evolution, Microgaming.
  • Live-dealer baccarat: streamed from Manila / Bucharest / Riga; pace 70-90 hands/hour; identical rules. Evolution Gaming dominates with 70%+ Asian market share. Bet ranges from $1 to $500,000+ on dedicated VIP tables.

The 2020-2024 pandemic accelerated online baccarat — Macau visitor numbers fell from 39 million (2019) to 5.8 million (2020); meanwhile online baccarat GGR (Asia-licensed) tripled. Recovery has been partial: 2023 Macau visitor count was 28 million; combined online baccarat from Asian players still exceeds 2019 levels.

IX. Common Misconceptions

  • "Following the trend boards is profitable." Patterns in random sequences are not predictive. The boards exist to extend session length, not to help you win.
  • "I should switch from Banker to Player after a Banker streak." Each hand is independent. The next hand's Banker probability remains 45.86% regardless of prior outcomes.
  • "Tie pays 8:1, so it's worth taking occasionally." 14.36% house edge says no.
  • "Counting cards works in baccarat." Theoretically yes — practically no. The optimal counting system gains under 0.005 cents per dollar bet, far below table minimum and worth less than the time spent.

X. FAQ · Sources · Responsibility

Why does Banker have a lower house edge than Player?
Because the third-card rule favors the Banker — the Banker's third-card decision depends on what the Player drew, while Player's decision is fixed in advance. This asymmetric information lets Banker win slightly more often (about 45.86% vs Player's 44.62%, with 9.52% ties). The casino offsets this advantage by charging a 5% commission on winning Banker bets, leaving a net house edge of 1.06%. Player bet pays even money with no commission — house edge 1.24%. Net: Banker is still mathematically the best baccarat bet despite the commission.
Is the 'Tie' bet ever worth taking?
Almost never. Tie pays 8:1 (sometimes 9:1) but ties occur only ~9.52% of the time. House edge on 8:1 Tie is 14.36% — the worst bet on the table by a wide margin. Even at 9:1 the house edge is still 4.85%. Compare with Banker at 1.06% and Player at 1.24%. Tie is statistically the same as donating money. The only marginal exception: card-counted shoes with extreme imbalances of 8s or 9s, but the swing required is rare and the casino reshuffles before it matters.
Does the third-card rule create any opportunity for skill?
No. The third-card rule is mechanical — every action is determined by the cards already dealt, applied identically every hand. There is no decision, no betting structure, no skill component for the player. Baccarat is a pure coin-flip in casino clothing. The only 'skill' is bankroll management (knowing when to stop) and bet selection (Banker > Player > Tie). Edge-sorting, card counting, and shuffle tracking are theoretical but require specific manufacturing defects (the famous Phil Ivey / Crockfords case turned on edge-sorting Gemaco cards in 2012).
Why is baccarat 88% of Macau's table revenue?
Three reasons: ① Cultural fit — Chinese gamblers historically favor games of pure luck over skill (since 'fate' is more aligned with traditional fortune concepts than mathematical optimization); ② Speed and stakes — baccarat hands take 30-60 seconds, with VIP rooms permitting bets of HK$1M+ per hand, far exceeding what blackjack or roulette tables allow; ③ Network effects — Macau's junket-room ecosystem evolved around baccarat-specific commission and Dead Chip ('泥码') systems, creating high switching costs to other games. DICJ 2023 data: baccarat = HK$182 billion of HK$207 billion total Macau table revenue.
What are 'derived roads' and trend boards in baccarat?
Macau-style baccarat tables track shoe results on six 'roads' (蛇形 / Big Road, 大眼仔 / Big Eye Boy, 小路 / Small Road, 曱甴路 / Cockroach Pig, plus two derived). These are sophisticated visualizations of trend patterns that supposedly predict future outcomes. They are pure superstition. Each hand is independent — the roads describe past patterns but provide zero predictive power. Casinos display them prominently because they keep players betting longer (the 'illusion of control' effect documented in Journal of Gambling Studies). Treat them as decoration, not strategy.
Mini-baccarat, midi, and full table — what's the difference?
Same game, different theater: ① Full table baccarat — 12-14 players, players take turns acting as Banker (passing the shoe), pace ~40 hands/hour, high minimums ($25-$100), traditional Macau / Monte Carlo VIP setting; ② Midi-baccarat — 9 players, dealer always handles cards, faster pace, common in US high-limit rooms; ③ Mini-baccarat — 6-7 players, dealer-handled cards, fast pace ~100 hands/hour, $5-$25 minimums. Mini-baccarat is the casino's preferred format because the high hand count amplifies house edge into faster revenue. House edge percentages are identical across formats — only the speed of accumulation changes.

Sources

  1. Macau DICJ, Gross Gaming Revenue Quarterly Report Q4 2023
  2. Edward O. Thorp & W.E. Walden (1973), The Optimal Strategy for Casino Baccarat, IJGT
  3. Ivey v. Genting Casinos UK Ltd [2017] UKSC 67, UK Supreme Court
  4. UNLV Center for Gaming Research, Baccarat & the Asian High-Roller Market
  5. Tommy Renzoni (1973), Renzoni on Baccarat, Lyle Stuart

Responsible play: This article is historical and mathematical analysis, not gambling advice. Baccarat carries low theoretical house edges, but the pace (60-100 hands/hour at mini-baccarat) means hourly expected loss can rival much higher-edge games. If gambling stops being entertainment, contact the National Council on Problem Gambling: 1-800-522-4700.