The casino's business model fits on one line

1. The house edge: paying less than the true odds

A bet is a list of outcomes, each with a probability p_i and a payout x_i per unit bet (+35 for a winning roulette number, −1 for a loss). Its expected value is the probability-weighted average:

The edge on the main bet of every game

2. Loss speed: a small edge, multiplied by a lot of bets

b

bet per round

n

rounds per hour

E

house edge

A 1% edge sounds harmless. But at 72 hands an hour, $50 a hand, you bet $3,600 an hour, and 1% of that is $36 an hour, every hour, for as long as you sit there. Slots look cheap per spin and are the opposite: hundreds of spins an hour at 10%+.

Every game at the same bet

Expected loss per hour at a round, fastest-losing first.

3. Your chance of being ahead shrinks every hour

Each round's result has a mean of −E and a standard deviation σ (per unit bet). Over n rounds the means add up in proportion to n, but the luck only grows with √n. By the central limit theorem:

Φ

the standard normal distribution function

σ

volatility per round, in bets

Early on, luck is bigger than the edge and anything can happen; that's the part people remember. Past n*, the line only goes one way.

4. Gambler's ruin: "I'll stop when I've doubled it"

You start with B betting units, bet one unit at a time on an even-money bet you win with probability p (q = 1 − p), and stop at T units or at zero. The classic result:

5. Churn: you lose a share of what you bet, not of what you bring

If you keep playing until the money is gone, Wald's identity turns around: the amount you'll have bet in total is your bankroll divided by the edge.

This is why people say "I only lost $500" after betting $5,000. The money goes round the table again and again, and the casino keeps a slice each time. Wins don't make you richer; they put money back in the machine.

6. Betting systems: changing the shape, never the sign

Martingale says: bet 1, and after every loss double the bet, so the first win recovers everything plus 1. With k doublings your bankroll can afford, the series wins 1 unit unless you lose k times in a row:

With any edge, q > ½, so 2q > 1 and the expectation is negative for every k. You win small, often, and lose big, rarely, and the rare loss is larger than all the small wins put together. Table limits make sure k stays small. The same goes for D'Alembert, Fibonacci, Labouchère, Paroli and every "system" sold online: by the optional stopping theorem, no rule for sizing bets or choosing when to stop can turn a game with negative expectation into a positive one.

7. 1,000 players walk in with the same money

Same game, same bet, same bankroll, different luck. The bands show where the middle 50% and 90% of players are after each hour; the dashed line is the expected value. Players who hit zero stay at zero.

8. Psychology: why people keep playing a game they know they lose

Nobody thinks they're beating the house edge. People keep playing because these games are designed, very carefully, around how our brains handle reward, uncertainty and loss. Knowing the tricks makes them easier to see.

9. How to stop

10. Free, confidential help