House Edge: What It Costs Per Hour, Why Hold Differs and Why No Betting System Moves It
A percentage is not a price until you know the decisions per hour behind it.
The house edge is quoted as a percentage, and a percentage on its own is not a price. Roulette’s 2.70% and a slot’s 4% describe the rate at which each game takes money, and a rate only becomes a cost once you know how many times an hour it is applied.
This is the part that comparison tables leave out. The missing variable is decisions per hour. Ranked by edge alone, American roulette at 5.26% looks like the worst bet in this section; ranked by what it actually costs to sit and play for an hour, it is not.
The edge multiplied by the pace: what it costs per hour
Three things set the cost of an hour: the house edge, the number of decisions per hour, and the average stake per decision. Multiply them.
Expected cost of one hour's play. Decision rates are typical rather than fixed; a crowded table is slower and a solo one faster
| Game | House edge | Decisions per hour | Average stake | Expected cost |
|---|---|---|---|---|
| Blackjack, basic strategy | 0.50% | 80 | 10 | 4.00 |
| Craps, pass line only | 1.41% | 30 | 10 | 4.23 |
| Baccarat, banker bet | 1.06% | 60 | 10 | 6.36 |
| Roulette, European | 2.70% | 40 | 10 | 10.80 |
| Roulette, American | 5.26% | 40 | 10 | 21.04 |
| Online slots | 4.00% | 600 | 1 | 24.00 |
| Category | Value |
|---|---|
| Blackjack, 80 hands at 10 | 4 |
| Craps pass line, 30 at 10 | 4.23 |
| Baccarat banker, 60 at 10 | 6.36 |
| European roulette, 40 at 10 | 10.8 |
| American roulette, 40 at 10 | 21.04 |
| Online slots, 600 spins at 1 | 24 |
Expected hourly cost from the table above, at the stated pace and stake. The slot has a lower house edge than American roulette and is the more expensive hour, because it is played fifteen times faster. Source: Edge figures are the game manufacturers’ published values; decision rates are typical observed table and online pace, stated per row rather than assumed. Cost equals edge multiplied by decisions multiplied by stake.
The two highlighted bars are the point of the plate. The slot carries a house edge a full 1.26 percentage points better than the American wheel and costs more per hour, because six hundred spins an hour is fifteen times the pace of a roulette table. A one-unit slot spin looks harmless next to a ten-unit roulette chip, and the arithmetic does not care what it looks like. Running the same sum the other way, holding the pace fixed and improving the return instead, is what the highest RTP slots are worth, and it is a larger saving than any promotion.
Speed is the variable nobody quotes and the one the player controls most easily. Halve the spin rate and you halve the cost of the hour, with no change to the game, the edge or the stake. Nothing else available to a player moves the number that far.
The edge is not uniform inside a game
Quoting one figure per game hides spreads that are larger than the differences between games.
Craps is the clearest case. The pass line runs 1.41%, which is among the better bets on any floor. The one-roll any-seven bet on the same felt, placed by reaching six inches further across the layout, runs 16.67%. Both are craps. Both are available to the same player in the same second.
Roulette is the opposite case and unusually honest for it: on a European wheel every standard bet carries exactly 2.70%, whatever combination you cover, which is why the roulette odds are the cleanest example in the building. American wheels break that uniformity in one place. Every bet is 5.26% except the five-number basket covering zero, double zero, one, two and three, which pays 6:1 and runs 7.89%.
Blackjack is the exception in the other direction, because the player’s own decisions move the number rather than the layout — the basic-strategy matrix is what that movement looks like written down.
Blackjack’s spread comes from the side bets bolted onto it. The main game at 0.5% sits next to optional wagers several points worse, offered on the same hand, resolved by the same cards. A player making every correct decision in the main game and one side bet per hand is not playing a 0.5% game.
House edge is not the same as hold
Casino accounting uses a different number, and the two get confused constantly.
House edge is theoretical and measured against total stake. Hold is actual, and measured against the money a player brought: win divided by drop. Hold is always far higher than the edge, often by a factor of five or more, and the reason is recycling. A player who buys in for 100 and plays it through six times has staked 600, and the edge has had six opportunities at the same money.
This is why the published payout rates of an entire casino look generous next to what a player recognises from their own account, and it is covered in more detail under online casino payout rates. The percentage is honest. The denominator is the thing to read.
Why a betting system cannot move it
Every betting system that has ever been sold works on the same principle: change the size of the next stake according to what happened last. The Martingale doubles after a loss, the Paroli doubles after a win, the d’Alembert steps up and down by one unit.
All of them redistribute outcomes and none of them change the expectation. The Martingale is the instructive one. It converts a symmetrical distribution into a lopsided one: many small wins, and a rare catastrophic loss when a long losing run meets either the table maximum or the end of the bankroll. The expected value per unit staked after that transformation is identical to before, because each round is independent and carries the same edge regardless of what preceded it.
A roulette wheel keeps no record. That is not a limitation of any particular system; it is the property that makes all of them equivalent.
What the edge is genuinely useful for
It tells you the direction of the long run and nothing about tonight. Over a session the variance dominates: a 2.70% edge is invisible inside the swing of forty spins, which is precisely why the game is playable and why the edge takes so long to become obvious.
Used as a comparison tool it works, provided the pace is included. Used as a prediction of an evening it is worthless. The one operational conclusion in this section is the plate above: at a given stake, the cost of the hour is set more by how fast you play than by which game you chose.
Where the return is verified externally rather than derived from the game’s own geometry, the equivalent figure is a payout percentage. That is return to player, and it is the only number in a casino somebody outside it has to sign.