TableWise How casino games actually work

House edge: the price built into every bet

House edge is the share of every bet that a casino keeps, on average. Nearly every payout is set a little below the true odds, and over thousands of bets that gap becomes a steady share of everything wagered. A roulette wheel shows it most plainly, so this page works it out there, one bet at a time.

The same bet on two wheels

The worked example shows the idea: a straight-up bet on a single-zero wheel loses 1/37 of its stake on average. Nobody charges it at the door. It is built into the payout, and the same bet can carry a different price on a different wheel. Here’s the same bet on single- and double-zero wheels:

The same straight-up bet on two kinds of wheel.
Quantity One zero Two zeros
Pockets 37 38
Payout 35 to 1 35 to 1
Chance of a win 1/37 1/38
Expected value of $1 (35 − 36) ÷ 37
= −1/37
(35 − 37) ÷ 38
= −2/38
House edge 2.70% 5.26%

Betting on red feels safer than betting on one number, and in one sense it is: the results swing less. The price per dollar is the same. Red covers 18 of the 37 pockets and pays 1 to 1, so its expected value is (18 − 19) ÷ 37 = −1/37.

Playing longer doesn’t wear the edge down

Every bet adds its own expected loss, so the total grows in step with the amount you bet. Luck grows more slowly. The variance of the result adds up spin by spin, so the typical swing grows only with the square root of the number of spins: four times as many spins means four times the expected loss but only twice the typical swing. The longer you play, the closer your average result per dollar comes to the house edge. Table 2 follows a player who bets $10 on red, spin after spin.

$10 on red every spin, on a single-zero wheel.
Spins Expected loss Chance of being ahead Chance of being behind
1 $0.27 48.65% 51.35%
100 $27.03 35.53% 56.79%
1,000 $270.27 18.76% 79.49%
10,000 $2,702.70 0.33% 99.65%

The rest is the chance of finishing exactly even, which can happen only after an even number of spins: 7.67% after 100, 1.75% after 1,000 and 0.02% after 10,000.

Show the working

Red wins with probability 18/37 and loses with 19/37. With equal stakes you are ahead after n spins when you have won more than half of them, even when you have won exactly half, and behind otherwise. The number of wins follows the binomial distribution:

P(k wins) = C(n, k) × 18^k × 19^(n−k) ÷ 37^n

Adding these terms over every k above n/2 gives the chance of being ahead. TableWise computes the sums in whole numbers when the site is built, rounds only the final percentage, and checks that ahead, even and behind add up to exactly 1. The expected loss is spins × $10 × 1/37. How we research describes the other checks a figure like this goes through.

What this settles about systems

Each bet in a series carries its own price, and averages add up. So over any series of bets the expected loss is the house edge times the total staked, however the bets are sized or ordered. That’s why a run of reds doesn’t make black “due” and why doubling after a loss changes the shape of a session and not its cost. Our page on gambling myths takes the common ones in turn, starting with the gambler’s fallacy.

Where decisions do count

In baccarat, as in roulette, nothing is left to decide once the bet is down. Blackjack and video poker are different, and so are some of the card games played against the house. In each, the choice you make after seeing your cards changes the house edge, and the correct choice can be calculated for every situation: for blackjack, that is basic strategy, which our blackjack trainer lets you practise hand by hand. Correct play makes those games cheaper to play. It doesn’t promise a profit, and TableWise says so wherever strategy comes up.

Rule

If a game gives you no decisions once the bet is placed, nothing you do at the table changes its odds. You can choose a cheaper bet. You can’t make a bet better.

House edge and return to player

Return to player, or RTP, is the same figure seen from the other side: the share of the money staked that comes back, on average. For the single-zero wheel that is 36/37 = 100% − 2.70% = 97.30%. Neither number says anything about one evening; both describe where the average settles over many bets. The roulette simulator shows how far a run of spins strays from them, and our page on slot RTP and volatility works out how widely sessions on one machine spread. What that means for a budget is set out on our responsible gambling page.