The Tran Organization: false shuffles, 42 guilty pleas, and the slug of cards that came back in order

Ninety minutes in East Chicago

In October 2005, at the Resorts East Chicago casino in Indiana, a group of players at a mini-baccarat table won $868,000 in about ninety minutes. Nothing about the hands looked wrong on tape: the dealer dealt, the players bet, the cards fell as cards do. What the tape could not show was that the players knew the order of the next several dozen cards before they left the shoe, because the dealer had been paid not to shuffle them.

The Department of Justice describes the mechanism in two sentences that are worth reading slowly. Members of the organisation bribed casino card dealers and supervisors to perform false shuffles, creating “slugs”, groups of unshuffled cards. After tracking the order of cards dealt, a member signalled the dealer to perform the false shuffle, and the group bet on the known order when the slug came back to the table. The tracking was not done in anyone’s head: the indictments describe hidden transmitter devices and specially created software that predicted the order in which the cards would reappear.

Sources: DOJ releases of 2 April 2008, 22 January 2009, 15 March 2010 and 28 September 2012; FBI San Diego release of 13 July 2009; the Associated Press report of 25 May 2007 for the $868,000 session and the $3.3 million indictment figure.

The crew was not small and not subtle about scale. The first indictment, unsealed in San Diego on 24 May 2007, named 19 people and listed 16 casinos; by the time Van Thu Tran was sentenced in 2012 the admitted list ran to 29 casinos from Foxwoods and Mohegan Sun in Connecticut to Casino Rama in Ontario, and 42 defendants had pleaded guilty. A pit boss at a Washington tribal casino took $5,000 to introduce the crew to dealers willing to be bribed. The dealer who went to trial, Mike Waseleski, was convicted of helping take about $1.5 million from a single property.

What a slug does to the edge

Mini-baccarat is a good game to cheat precisely because it is a bad game to beat. The player has no decisions; the third-card rules are fixed; the only choice is which side to back. On a random shoe the Banker bet wins about 45.8% of hands, the Player bet 44.6%, and 9.5% tie; Banker pays 19 to 20, Player even money. That prices the Banker bet at a house edge of about 1.06% and the Player bet at about 1.24%, which is why baccarat has the reputation of being nearly fair and why nobody beats it by skill.

Now suppose you know the next six cards. You know which side wins the next hand, or that it ties. You back the winner for the table maximum and sit out the ties. Your return on the hand is +100% on a Player win and +95% on a Banker win, every time, with no variance. The “edge” has not been reduced from 1.06% to something smaller. It has been replaced by a different number with the opposite sign, roughly +97% per hand across a slug, and the only limit on the take is the table maximum multiplied by the number of hands the slug covers and the number of seats the crew fills.

Infinite-shoe probabilities from the baccarat pricing article (45.84 / 44.61 / 9.54). Random-shoe expectations: 0.4584 × 0.95 − 0.4461 = −1.06%; 0.4461 − 0.4584 = −1.23%. A known slug of cards fixes each hand’s outcome, so the bettor’s win rate is 100% on the hands they choose to play. The $868,000 session is consistent with a few seats at a high table limit over a slug that covered a few dozen hands.

Two things had to be true for this to work, and neither is about cards. First, the order of a physical shoe is state that persists between rounds: whatever is not shuffled stays in sequence, so a bribed dealer who holds back a block has handed the crew the future. Second, the casino’s defence against that state was a person. The shuffle is a procedure performed by an employee, watched by a supervisor who was also for sale, under cameras that record a shuffle but cannot tell a real one from a good imitation. The cheat lived in the gap between what the camera saw and what the cards were.

A shoe that cannot be false-shuffled

It took the FBI, the IRS, the California Bureau of Gambling Control and the Ontario Provincial Police five years of indictments to put the crew away, and the first sealed charge in Seattle in February 2006 came almost four years after the first false shuffle. Surveillance did not catch it as it happened. Nothing in the hands was statistically suspicious hand by hand; a player who wins six baccarat bets in a row at the maximum is lucky until the pattern is matched against a dealer’s shift roster and a slug that keeps reappearing.

Compare the thing that was attacked, the order of the shoe, with how a provably fair table produces its cards. There is no shoe. Each card in a round is a fresh draw from a number derived from a committed server seed, the player’s own seed and a round counter, so the sequence of one round carries nothing into the next; the engine deals as if from an infinite deck, each rank one in thirteen on every card. A dealer cannot hold a block still because there is no block, and a supervisor cannot look away from a shuffle that does not exist. The outcome is fixed at the moment the round is dealt, by a function neither the house nor the player controls alone, and it is not stored anywhere a person could preserve it.

That is not a claim that an online table is safer than a casino floor in every respect. Custody, payouts and the honesty of the committed seed are separate questions, and the Stake hack is the reminder that provable fairness covers the result, not the vault. It is the narrower claim this case invites: the Tran Organization’s entire method was to make a random process remember, and a process that is recomputed from a commitment for every round has nothing to remember. The 1.06% stays 1.06% on the first hand and the ten-thousandth, and there is no one at the table to bribe into making it otherwise.