The Mathematics of Gambling: Why the House Usually Wins
Bet $10 on red at an American roulette table. You'll win 47.4% of the time, so almost half your bets pay off. The casino's edge on each spin is barely more than 5 cents on the dollar.
That sounds like a game you could beat with a bit of luck. Over 100 spins, you have about a 30% chance of being ahead. Over 1,000 spins, under 5%. Over 10,000 spins, effectively zero. The house doesn't need to win big. It needs a small edge and time.
Casinos Welcome Winners
Casinos aren't afraid of people who win. Winners are good advertising, and short-term wins are mathematically guaranteed to happen to some players. What casinos rely on is that variance shrinks relative to the edge as the number of bets grows.
Here's what happens to a player betting $10 on red, spin after spin, on a double-zero wheel:
| Spins | Expected result | Chance of being ahead |
|---|---|---|
| 100 | −$52.63 | ≈ 30% |
| 1,000 | −$526.32 | ≈ 5% |
| 10,000 | −$5,263.16 | ≈ 0% |
The expected loss grows in proportion to n. The typical swing (standard deviation) grows only with √n. Eventually the steady loss overwhelms the random swing.
House Edge: The Only Number That Matters
The house edge is the expected loss per unit bet, expressed as a percentage. It comes from expected value:
EV = Σ (probability × payout)
On American roulette there are 38 pockets: 18 red, 18 black, and 2 green (0 and 00). A $1 bet on red:
EV = (18/38) × (+1) + (20/38) × (−1) = −2/38 ≈ −5.26%
A single-number bet pays 35 to 1:
EV = (1/38) × 35 + (37/38) × (−1) = −2/38 ≈ −5.26%
The same edge! Roulette's payouts are set as if there were 36 pockets. The extra green pockets are pure profit. On a European single-zero wheel, the edge falls to 1/37 ≈ 2.70%.
| Game | Typical house edge |
|---|---|
| Blackjack (basic strategy) | ≈ 0.5% |
| Baccarat (banker bet) | ≈ 1.06% |
| European roulette | 2.70% |
| American roulette | 5.26% |
| Slot machines | ≈ 2%–15% |
| Keno | ≈ 25%+ |
Variance: Why Anyone Wins at All
If every bet lost 5.26 cents on the dollar with certainty, nobody would play. Gambling is fun because of variance, the spread of outcomes around the expected value.
For an even-money bet of $10, the standard deviation per spin is almost exactly $10. Over n spins:
σ_total = 10 × √n
After 100 spins, the standard deviation is about $100, twice the expected loss of $52.63. Luck dominates. After 10,000 spins, it's about $1,000 against an expected loss of $5,263. The edge dominates. You can check these probabilities with a z-score calculator.
The Gambler's Fallacy
After ten reds in a row, is black "due"? No. Each spin is independent, so the probability of black is still 18/38.
The most famous example happened at the Monte Carlo Casino on August 18, 1913, when black came up 26 times in a row. Gamblers lost millions of francs betting on red, convinced the streak had to end. The probability of 26 blacks in a row on a single-zero wheel is about (18/37)²⁶ ≈ 1 in 136.8 million, but any particular sequence of 26 spins is just as unlikely.
Betting Systems Don't Beat Math
The Martingale system says: double your bet after every loss, and one win recovers everything plus one unit. It sounds foolproof. It isn't, for two reasons:
- Losing streaks grow exponentially. After 10 straight losses starting at $10, your next bet is $10 × 2¹⁰ = $10,240.
- Table limits and finite bankrolls cap how far you can double.
Mathematically, no betting system can change the expected value of a sequence of negative-EV bets. That result is formalized in the optional stopping theorem for martingales.
An Insider Reference: The Man Who Beat Blackjack
In 1962, MIT mathematics professor Edward O. Thorp published Beat the Dealer. He showed that blackjack, unlike roulette, has memory: cards that have been dealt don't return until the shoe is reshuffled. When the remaining deck is rich in tens and aces, the player can gain an edge.
Thorp's card counting worked so well that casinos changed their rules: more decks, earlier reshuffles, and banning counters. Thorp later took the same probability ideas to Wall Street and became a pioneer of quantitative hedge funds.
Two Concepts Worth Knowing
Gambler's Ruin
If you play a negative-EV game against an opponent with far more money (the casino), the probability that you eventually go broke approaches 100%. Even a fair game ends in ruin against an effectively infinite bankroll.
The Kelly Criterion
When you do have an edge, the Kelly criterion (John L. Kelly Jr., Bell Labs, 1956) says what fraction of your bankroll to bet to maximize long-run growth. For an even-money bet with win probability p, it's f = 2p − 1. When p is below 0.5, Kelly says bet zero.
Quick Answer: Why Does the House Always Win?
Every casino game pays slightly less than the true odds, which creates a negative expected value for the player called the house edge. Over a few bets, variance lets players win. Over thousands of bets, the law of large numbers makes the casino's profit close to the house edge times the total amount wagered.
Try Them Yourself
- Random Number Generator: simulate a roulette wheel with numbers 1–38
- Z-Score Calculator: how likely is a winning session?
- Standard Normal Table: turn z-scores into probabilities
- Statistics Formulas: expected value, variance and standard deviation
- Algebra Formulas: exponents behind the Martingale's doubling
- Winning With Numbers: where probability works in your favor
Simulate 100 spins with the random number generator and track your balance. Then imagine doing it 100 times over. That's the casino's view of the game.