The House Edge, Game by Game

By Alex Thornton · author of the MIT-licensed teachersbettextbook library · Last reviewed 24 July 2026

Every casino game is a bet with a price. Almost every game rigs the price against you; exactly two let skill flip it. “The house always wins” is a theorem, not a slogan.

How to read any casino game

One tool reads them all: expected value. Multiply each payoff by its probability; add:

The house edge is that, negated, per dollar staked. When EV is $-0.05$, the edge is 5%. The casino keeps a nickel of every dollar, forever. It never needs to win every hand: it pays winners less than true odds and lets the law of large numbers grind.

The two you can beat

Two games break the pattern with skill, not a system. Blackjack’s optimum strategy appeared in a statistics journal in 1956.^[1] Card counting flipped it positive in 1962.^[2] Poker’s skill premium was measured at the World Series of Poker itself.^[3] The edges are real. They are also far harder to earn than the movies let on.

Blackjack: Basic Strategy & Card CountingThe one house game with memory. Basic strategy cuts the edge to ~0.5%; counting the shoe can flip it positive, and get you barred.Player edge is possible →

Poker: Why Skill Wins (and the Rake)You play the table, not the house (minus the rake). Pot odds, position, and why the better player wins.A game of skill →

The house’s games

Everything below is negative expected value by construction. The only variable is how fast it bleeds you.

Roulette: 5.26% a SpinTwo green pockets are the whole edge. Every American-wheel bet pays the same 5.26% tax; Europe’s single zero halves it.−5.26% →

Craps: Good Bets and Sucker BetsThe pass line, −1.41%, is the house’s cheapest price. Ten feet away, “Any 7” is −16.67%. Same dice.−1.41% to −16.67% →

Slots: The Edge You Can’t SeeThe edge hides in the machine’s return-to-player. No skill, no count, no timing: a fixed tax, spun 600 times an hour.−3% to −12% →

Baccarat: Banker, Player, TieBanker, at −1.06%, is among the floor’s lowest. Still negative, still skill-free. The Tie bet is a −14% trap.−1.06% to −14% →

Keno: The Worst on the FloorA lottery paying pennies on the true odds. The edge runs 20–35%, worst in the building.−20% to −35% →

Read this part: the danger in “beatable”

Counting blackjack and winning poker are mathematically real. For almost everyone they are still a bad idea. A ~1% edge means variance dwarfs the signal: surviving to the long run takes a large bankroll, iron discipline, and Kelly criterion sizing. Overbetting a small edge is how skilled players go broke. The casino fights back: early shuffles, deep cuts, lowered limits, the door.

The rule of this whole site: know the price. For every game but two, the price is a wall. Arithmetic, not bad luck.

The same math runs the sports side. See the free +EV tool, read how expected value finds an edge, or open it in the live app.

Where these numbers come from

Every edge above is an exact, decades-old derivation from the game’s own rules.^[4] The reading:

1. Baldwin, R. R., Cantey, W. E., Maisel, H. & McDermott, J. P. (1956). “The Optimum Strategy in Blackjack.” Journal of the American Statistical Association 51(275), 429–439. The first complete basic strategy.
2. Thorp, E. O. (1962). Beat the Dealer: A Winning Strategy for the Game of Twenty-One. Random House; rev. ed. 1966. Card counting: how the game with memory flips positive.
3. Levitt, S. D. & Miles, T. J. (2014). “The Role of Skill Versus Luck in Poker: Evidence From the World Series of Poker.” Journal of Sports Economics 15(1), 31–44. Poker skill, measured: pre-rated skilled players returned over +30% at the 2010 WSOP; everyone else about −15%.
4. Ethier, S. N. (2010). The Doctrine of Chances: Probabilistic Aspects of Gambling. Springer. The canonical exact house-edge derivations for every game on the scoreboard.