Biggest Baccarat Wins in Casino History
The largest documented baccarat wins reveal something interesting: the game's low house edge does not protect players over long sessions. Extraordinary short-term wins are possible precisely because the variance is high enough to overcome the edge.
The deal
Baccarat carries a house edge of 1.06% on banker bets and 1.24% on player bets, making it one of the lowest-edge games in any casino. Those figures assume eight-deck standard rules as certified by most major gaming jurisdictions. The commission on banker bets, standardly set at 5%, is accounted for in the 1.06% figure. This is not a disputed number.
The same mathematics that produces a small house edge also produces high variance in short sessions. A single large bet on a hand with binary outcomes has a standard deviation that swings dramatically relative to expected value. This is why the largest documented wins in baccarat involve very large individual bets, not extended systems.
Round 1: Akio Kashiwagi: The Numbers Behind the Legend
Akio Kashiwagi, a Japanese real estate developer who died in 1992, is the most documented case study in high-stakes baccarat. At Trump Plaza in Atlantic City in 1990, Kashiwagi reportedly sat in a session that saw swings exceeding $10 million in a single extended play. Trump reportedly hired statistician Jess Marcum to calculate the optimal number of shoes required to mathematically recover the house edge against Kashiwagi's $200,000 flat bets.
Marcum's calculation determined that at that bet size, the house needed approximately 500 hours of play to reliably realize its 1.06% edge. Kashiwagi's strategy was to play for fewer hours than the convergence point. The math was not in Trump's favor for short sessions. Trump reportedly lost over $6 million in the encounter.
This is the precise mechanism large baccarat wins exploit: the expected value converges slowly relative to session length when bet sizes are this large.
Round 2: Kerry Packer: $30 Million in a Single Session
The most frequently cited large baccarat win involves Kerry Packer, the Australian media magnate, at the MGM Grand in 1995. Reports from casino sources and later journalism place his single-session winnings at between $20 million and $40 million, with $30 million as the most commonly cited figure. He was betting $250,000 per hand.
At $250,000 per hand and 80 hands per hour, expected loss over a 4-hour session calculates to: $250,000 x 0.0124 x 320 hands = approximately $992,000 expected loss. His actual reported outcome was therefore roughly 30 times his negative expected value. The probability of exceeding $30 million in winnings over 320 hands at $250,000 per hand is astronomically small. It occurred.
This is not a scandal. This is variance operating as variance should operate. The casino is still profitable over all its baccarat sessions. Packer was one data point on the right tail of a distribution that includes many more data points on the left tail, where players lose large amounts quietly.
Round 3: Common Misconceptions About Baccarat Winning Strategies
Several popular misconceptions surface repeatedly in discussions of large baccarat wins.
Myth: Pattern-tracking provides an edge. The Big Road, Bead Plate, and other tracking methods used by many baccarat players have no mathematical basis for predicting future outcomes. Each hand is an independent draw from a fresh shuffle or a running shoe. The fact that banker won the last six hands does not alter the 1.06% edge on the next banker bet. Casinos provide scorecards because pattern-tracking encouraged players to bet more, not because it works.
Myth: Card counting works in baccarat. It technically can, though barely. Edward Thorp demonstrated theoretically that baccarat has a countable edge, but the advantage achievable is estimated at under 0.1% even with perfect tracking, and requires a very deep shoe penetration. No documented casino win of significant size has been attributed to baccarat card counting.
Myth: Tie bets are worth the payout. The 8:1 tie bet carries a house edge of approximately 14.36% in eight-deck games. This is among the worst bets in any table game. The large documented wins avoid tie bets entirely.
Therefore: the historical record of large baccarat wins is a record of high variance operating over small sample sizes with very large bet units. The mathematics of the game did not fail in those sessions. The mathematics produced an expected outcome that happened, with appropriate probability, to favor the player.
