The Golden Nugget in Atlantic City opened a roulette room in 1978. I was there regularly in the early 1980s. The place smelled like cigarette smoke and possibility. There was a specific regularity to the crowd: older men, mostly, who had notebooks filled with roulette results. They were tracking numbers. The hot numbers (the ones that had hit most frequently in the last hundred or two hundred spins) and the cold numbers (the ones that hadn't hit in a while).
They would sit for hours, just writing down results. Thirty-seven, black. Twenty-four, red. One, red. They weren't playing yet. They were gathering data. The theory was explicit: if a number had been cold for a long time, it was due to hit. If a number had been hot, it would continue. Both versions persisted simultaneously in the same crowd, which should have been a tip-off that at least one was wrong.
The steel pier had views of the ocean. You could smell the salt if you opened the doors facing the boardwalk. The contrast was always striking. Inside: the precise geometry of the roulette wheel, the small mathematics of probability, the sound of chips being collected and rearranged. Outside: the Atlantic, the horizon, the implication that the world is much larger and more random than the roulette room.
I watched a particular player for years. His name was Albert. He had a ledger he'd been maintaining since 1975. Literally a ledger, leather-bound, filled with handwriting. He would sit at the roulette table, place bets on his predetermined cold numbers, and write results in his ledger. He was polite. He wore glasses. He genuinely believed that his data collection had value.
One afternoon I asked him: "How long have you been tracking these numbers?" He said, "Since 1975." I said, "And you've made money?" He said, "Not yet. But the data is clear. The numbers that haven't hit in fifty spins are due." This was his statement of faith. Not that it would work. That the data was clear.
The Mathematics Underneath
Roulette is a discrete random process. Each spin has equal probability (one in thirty-seven for a single number on a European wheel). The outcome of the last hundred spins has zero causal relationship to the next spin. The next spin's probability is identical: one in thirty-seven. This is the gambler's fallacy in textbook form. The belief that past events change the probability of future events in a genuinely random process is false.
But here's where people get confused: if a number has never hit in ten thousand spins, is it really equally likely to come up on the next spin? Mathematically, yes. Actually, maybe not, because that would imply the wheel is biased or the casino is cheating. But if the wheel is fair and the process is genuinely random, then past frequency tells you nothing about future frequency.
"The hot and cold numbers mythology persists because pattern-seeking is hardwired into human cognition. The human brain evolved to find patterns. It's terrible at understanding randomness."
The interesting part was that Atlantic City had actual bias wheels in the early 1980s. Not cheating. Wear and tear. A wheel that gets spun several thousand times daily will wear unevenly. A slight tilt, a slightly loose rotor, and suddenly one section of the wheel comes up more frequently than the mathematics suggests. This was documented. In 1983, a group of physics students at Berkeley used physical observations to predict biased wheels and made a fortune until casinos caught on.
But a biased wheel is a completely different phenomenon than a hot/cold phenomenon. The bias is the exception. The normal case is a fair wheel, which means past history is useless.
Albert passed away in 1997, I heard. I wonder if he ever closed that ledger, ever accepted that fifty years of data had told him nothing predictive. I suspect he didn't. I suspect he believed the pattern was there. The data just hadn't revealed itself yet.
Atlantic City in the 1980s and 1990s was slowly emptying out. The casinos were making money but the city was declining. The Steel Pier where the casinos were located became famous for its slow decay. By the 2000s, Atlantic City was a place people drove through on their way to Philadelphia. The gambling had moved to Las Vegas and eventually online.
The hot and cold mythology moved online too. Modern casino websites have tools that let you view historical results. Players download this data and search for patterns. The computational power is vastly superior to Albert's ledger. But the fundamental error is identical: confusing memory with causation. A roulette wheel has no memory. The fact that seventeen didn't hit for one hundred spins changes its probability zero.
