House edge, explained
Every casino game is designed so that, over time, the casino keeps a small share of the money wagered. That share is the house edge. It's the single most useful number for comparing bets — and one of the most misunderstood.
The definition
House edge = the average amount you lose per dollar wagered, over a very large number of bets.
It comes from paying winners slightly less than the true odds. A single roulette number, for example, should pay 37 to 1 on an American wheel (one way to win, 37 ways to lose). It pays 35 to 1. That gap is a 5.26% edge.
Common bets, compared
| Bet | House edge | Expected cost per $100 wagered |
|---|---|---|
| Craps odds (behind the line) | 0% | $0 |
| Blackjack, basic strategy (6 decks, H17) | ≈0.55% | $0.55 |
| Craps Don't Pass | 1.36% | $1.36 |
| Craps Pass Line / Come | 1.41% | $1.41 |
| Craps Place 6 or 8 | 1.52% | $1.52 |
| Baccarat Banker | 1.06% | $1.06 |
| European roulette | 2.70% | $2.70 |
| American roulette | 5.26% | $5.26 |
| Craps Hard 6 / Hard 8 | 9.09% | $9.09 |
| Craps Any 7 | 16.67% | $16.67 |
Myth 1: "House edge is my chance of losing"
It isn't. Look at how often these bets lose versus what they cost:
| Bet | Chance a single bet loses | House edge |
|---|---|---|
| Craps Pass Line | 50.7% | 1.41% |
| Blackjack hand (basic strategy) | ≈49% (8.5% are pushes) | ≈0.5% |
| Roulette red (American) | 52.6% | 5.26% |
| Roulette single number (American) | 97.4% | 5.26% |
| Craps Any 7 | 83.3% | 16.67% |
A single roulette number loses almost every time, yet costs exactly the same per dollar as red, because when it wins it pays 35 times your bet. The edge combines how often you win with how much you win.
Myth 2: "Expected value is what will happen tonight"
The house edge tells you the average. Individual sessions scatter widely around it.
Take blackjack at $25 a hand with an edge of about 0.55% and a typical swing (standard deviation) of about 1.14 bets per hand:
| Hands played | Expected result | Typical swing (1 SD) | Chance you're ahead* |
|---|---|---|---|
| 100 | −$14 | ± $285 | ≈ 48% |
| 1,000 | −$138 | ± $900 | ≈ 44% |
| 10,000 | −$1,375 | ± $2,850 | ≈ 32% |
*Normal approximation.
Over one night, luck completely swamps the edge. Over a lifetime of play, the edge wins. That's why a winning weekend proves nothing about a system — and a losing one doesn't mean you played badly.
Per bet or per roll?
Craps edges are sometimes quoted two ways:
- Per bet resolved: the cost each time the bet wins or loses. Place 6 = 1.52%.
- Per roll: the same cost spread over every roll, including rolls that don't affect the bet. Place 6 resolves on only 11 of 36 rolls, so it's 0.46% per roll.
Per-bet figures are the fair way to compare bets; per-roll figures help estimate cost per hour.
The formula that actually matters
Expected loss = house edge × total amount wagered.
The total amount wagered is usually much larger than the money you bring, because you keep betting winnings. At $10 a spin of American roulette for 40 spins, that's $400 of action and an expected cost of about $21 — even if you only ever had $100 in front of you. (Pace assumptions vary; the multiplication doesn't.)
So there are only three ways to lower what you expect to lose: choose bets with a lower edge, play correctly (in blackjack), and bet less in total. No betting system changes the edge — see why.
How Table School checks its numbers
We treat mathematical accuracy as the product's foundation:
- Every craps and roulette probability and house edge is computed exactly from the dice combinations and wheel pockets, and compared with published figures (Wizard of Odds).
- The blackjack strategy charts were transcribed twice and compared cell by cell — all 1,920 cells across six published charts match — and every recommendation was checked against an independent expected-value calculation.
- Every craps bet settlement was compared with a separately written reference implementation across all dice outcomes and table states (over 230,000 checks, zero mismatches).
- The full games were simulated for millions of rounds, and results land within normal statistical error of the theory.
If you ever find a number that looks wrong, tell us — accuracy reports are fixed first.
See the numbers move
The Math Lab simulates thousands of sessions so you can see the expected loss and how wide the swings are.
Frequently asked questions
Does a 1.41% house edge mean I lose 1.41% of my bets?
No. You lose about 50.7% of Pass Line bets. The 1.41% is the average cost per dollar over many bets — wins and losses included.
If the house edge is small, why do people lose so much?
Because the edge applies to every dollar you bet, including money you've already won and are betting again. $100 bet 300 times is $30,000 of action; at 1.41%, that's an expected $423 — more than four times the original $100.
Which casino game has the lowest house edge?
Among common table games, blackjack played with correct basic strategy (roughly 0.3–0.6% depending on rules) and craps line bets with odds (under 0.5% on total action) are the lowest. Craps odds on their own have no house edge at all.