Blackjack does not have a house edge. It has house edges, one per rule set, and the room you are standing in may contain half a dozen of them at tables that look identical from ten feet away. The useful news is that they are computable, and not by anyone clever: rule effects are close enough to additive that pricing a table is arithmetic a person can do in their head while walking past it.
Our guide to the house edge explains what the number means. This one is about assembling it.
The ledger
Every rule that differs from a baseline game moves the edge by a published amount. Positive numbers favor the house, negative numbers favor you. Per the commonly published tables:
| Rule | Effect on house edge |
|---|---|
| Blackjack pays 6:5 instead of 3:2 | about +1.4 points |
| Eight decks instead of one | roughly +0.5 across the range |
| Dealer hits soft 17 (H17) | about +0.22 |
| European no hole card (ENHC) | about +0.11 |
| Double after split allowed (DAS) | about -0.14 |
| Late surrender offered | about -0.08 |
Two features of this ledger deserve attention before any arithmetic. The first is the enormous gap between the top row and everything else: the payout line is worth about six times the next largest item, which is why it earns its own article and its own ten seconds of your attention. The second is that the player-favorable rules are small. Surrender and DAS are genuinely worth having, and neither of them rescues a bad payout.
Two tables, priced
Take a good baseline: six decks, 3:2 payout, dealer stands on soft 17, double after split allowed, late surrender offered. That combination lands a little under half a percent for a perfect basic strategy player, and it is the game to look for.
Now change one thing at a time. Move to a table that hits soft 17 and drops surrender, and you add roughly 0.22 and 0.08: the price rises by about three tenths of a point, still a decent game. But walk to the single-deck table with the attractive sign and the 6:5 payout, and the ledger reads minus half a point for the deck count, plus 1.4 for the payout, for a net that is roughly a full point worse than the shoe game you left. Our article on deck count covers that specific trap in detail, and the calculator arithmetic is exactly why it is a trap: one prominent good rule financing one quiet expensive one.
21 TrainerThe built-in house edge calculator stacks any rule combination and redraws the strategy chart to match. Free on the App Store.
Get the app
What a calculator adds
You can do the sums by hand, and for the six rules above you probably should, because the point is to internalize the relative sizes. A calculator earns its keep in two other ways.
It handles the long tail. Real tables carry rules the headline list omits: how many times you may resplit, whether aces can be resplit, whether you may double on any two cards or only on 9 through 11, whether early surrender is offered instead of late. Each is worth a small amount, several are easy to overlook, and together they can add up to more than the H17 rule everyone argues about.
And it removes the additive approximation. Rule effects interact slightly, because each published figure was computed against a baseline where the other rules were fixed. Surrender is worth a bit less in a game that already lets you escape trouble other ways; deck count effects shift as the doubling rules change. A calculator computes the combination directly rather than summing estimates. For table selection the difference is immaterial, which is the honest thing to say, but it matters when you are studying the structure rather than picking a seat.
The caveat that outweighs the whole ledger
Every number above assumes perfect basic strategy for that exact rule set. That assumption is doing enormous work.
A player who stands on 16 against a ten, skips doubles on 11, takes insurance when the dealer shows an ace, and splits tens is commonly estimated to give back one to two percentage points. Compare that with the entire ledger above, where the largest single item is 1.4 and most are fractions of a tenth, and the ranking becomes obvious: your own accuracy is a bigger lever than any table in the room. A perfect player at a mediocre table beats a sloppy player at a great one, every time, and it is not close.
This is also why a house edge calculator belongs inside a trainer rather than in a separate tab. In 21 Trainer, changing the rules does two things at once: it recomputes the edge and it redraws the strategy chart, because the correct chart is a function of the same rules. Watching both change together teaches the lesson that a standalone calculator cannot, which is that a rule set is not just a price, it is a set of instructions for how to play.
The ten-second version
At a real table you will not run a calculator, so compress the ledger into a reading order. Payout line first, and if it says 6 to 5, the analysis is over. Soft 17 rule second. Deck count third. Surrender and double-after-split fourth, if the placard mentions them. That sequence spends your attention in proportion to what the numbers are worth, which is the entire practical output of understanding how rules stack.
Frequently asked questions
How do you calculate the house edge in blackjack?
Start from a baseline game whose edge is known, then add or subtract the published effect of each rule that differs. The effects are close enough to additive that the running total lands near the true figure for ordinary rule sets. A house edge calculator automates exactly this: you select the rules, it sums the adjustments and reports the resulting edge.
Which blackjack rules matter most to the house edge?
The blackjack payout dominates everything: 6:5 instead of 3:2 costs roughly 1.4 percentage points, per the commonly published tables. After that comes deck count, worth roughly half a point across the range from one deck to eight, then the dealer hitting soft 17 at about 0.22, then smaller items like double after split and surrender, which help the player by roughly 0.14 and 0.08.
Do blackjack rule effects really just add up?
Close enough to be useful, not perfectly. The published effects are each computed against a baseline, and rules interact slightly: surrender is worth a little less in games where you already escape bad hands other ways, for example. For ordinary rule sets the additive estimate lands within a few hundredths of a percent, which is far more precision than any table-selection decision needs.
What is a good house edge for a blackjack table?
Under half a percent is a good game, and the best commonly available tables sit a little below that with 3:2 payouts, dealer standing on soft 17, double after split and surrender. Above roughly one percent you are usually looking at a 6:5 payout, which no other rule realistically offsets. These figures all assume perfect basic strategy; mistakes add their own cost on top.
Does the house edge assume perfect play?
Yes, and that caveat is easy to forget. Every published rule effect assumes correct basic strategy for that exact rule set. A casual player giving back one to two percentage points through misplayed hands is paying far more than the difference between any two tables in the room, which is why drilling strategy outranks table selection as an improvement.