Charles Wells: The Man Who Broke the Bank at Monte Carlo

Charles Deville Wells is a case study in the difference between outcome and process. In July of eighteen ninety-two, Wells arrived at the Monte Carlo casino with a modest bankroll. Over a period of eleven hours, he won two million francs at the roulette table. The casino's roulette bank was depleted. Wells left with a fortune. The newspapers called him the man who broke the bank at Monte Carlo. The romantic version of the story is that Wells had discovered a system, a method, a way to beat roulette that nobody had thought of before.

The actual version is simpler and less interesting: Wells got lucky. Variance worked in his favor for a sequence of bets. Roulette is a game with a negative expectation. The house edge is approximately 2.7 percent on a single-zero wheel (Monte Carlo uses a single-zero). Over sufficient volume, the house wins. Wells got lucky early in his session. He pressed his bets. He continued to get lucky. After eleven hours and two million francs, probability reasserted itself. Wells stopped. He left. He had caught the wheel at an anomalous moment.

What Wells actually did not understand (or what newspapers did not report, which is the same thing) is that roulette has streaks. A series of red outcomes will occur, despite the true probability being 48.6 percent red and 48.6 percent black on a single-zero wheel. A series of numbers in the 1-18 range will occur. When these streaks align with a player's bets, the player wins large amounts. When they align against, the player loses large amounts. Wells caught the tail of several positive streaks. That is the entire story.

The claim that Wells had discovered a system was commercially useful. Newspapers sold more copies. Wells became a celebrity. He was invited to give lectures on his method. He published an account of his betting progression. The betting progression was not actually a system that beat roulette. It was a retrospective reconstruction of the bets he had placed during his lucky streak, reordered to look like they were part of a coherent strategy. They were not.

The Mathematics of Wells' Actual Performance

Roulette generates a specific expected value per spin. For a single-zero wheel, the EV is -2.7 percent of the bet. Wells' session lasted approximately eleven hours. Assume a moderate pace of thirty spins per hour (fast play). That is three hundred and thirty spins. Assume an average bet of five thousand francs (Wells was adjusting his bet size). That is one point six five million francs in total action. The expected loss on that action would be roughly forty-five thousand francs. Wells won two million francs. The difference between the expected loss and the actual win is approximately two million and forty-five thousand francs. This is the advantage he gained from variance.

Variance of that magnitude is possible but rare. Using the standard deviation of roulette outcomes, a two-million-franc positive swing over three hundred thirty spins with an average five-thousand-franc bet represents approximately a six-sigma event (six standard deviations from the mean). In other words, it would occur roughly once every billion times you repeated the experiment. Wells experienced a one-in-a-billion day. He was lucky.

What followed Wells' win is instructive. He returned to Monte Carlo and lost much of his fortune. He gambled at other venues and continued to lose. He lived the remainder of his life in financial precarity, despite his initial massive win. The reason is that his initial success was not based on skill or a system. It was based on being on the right side of variance. Once variance stopped favoring him (which it inevitably did), he had no edge. He had no system. He had no advantage. He was a man who got lucky once and spent decades unable to replicate it.

The lesson from Wells is that outcome is not process. A good outcome does not prove a good process. A man who wins two million francs at roulette has not proven anything about roulette strategy. He has proven he got lucky. The nineteenth-century newspapers did not understand this distinction. Contemporary gambling media often do not either.