Blackjack odds, measured: with basic strategy at six decks, the dealer standing on soft 17 and blackjack paying 3:2, the house edge is 0.44%. A player can expect to lose $0.44 per $100 bet over time.
| Game | House edge | 95% interval | Hands simulated |
|---|---|---|---|
| 6 decks, stands soft 17, 3:2 | 0.44% | 0.43% to 0.44% | 800 million |
| 6 decks, hits soft 17, 3:2 | 0.64% | 0.62% to 0.65% | 200 million |
| 6 decks, stands soft 17, 6:5 | 1.77% | 1.76% to 1.79% | 200 million |
| 6 decks, hits soft 17, 6:5 | 2.00% | 1.98% to 2.01% | 200 million |
| 8 decks, stands soft 17, 3:2 | 0.45% | 0.43% to 0.47% | 200 million |
| 2 decks, stands soft 17, 3:2 | 0.26% | 0.25% to 0.28% | 200 million |
| 1 deck, stands soft 17, 3:2 | 0.01% | -0.01% to 0.03% | 200 million |
How the house edge was measured
We dealt 800 million hands by computer under one rule set. The rules: six decks, dealer stands on soft 17, blackjack pays 3:2, double on any two cards, double after split, split to four hands (aces once, one card each), no surrender, dealer peeks for blackjack, reshuffle after 75% of the shoe. Every hand was played by the basic strategy chart for those rules. The games where the dealer hits soft 17 used the same chart with its three standard changes.
The 95% interval from the simulation is 0.43% to 0.44%. The sample pins the figure to its second decimal place. Each other rule set in the table was dealt 200 million hands.
The 1, 2 and 8 deck games used the six-deck strategy chart. A chart tuned to each deck count can only do as well or better. Those rows can therefore only overstate the edge, never understate it.
At 0.44%, the house edge is the expected result of a bet, and that result is a loss. No betting pattern changes it. The edge is a property of each bet, and a sum of bets with negative expectation is negative.
What the rules at the table do to the edge
The payout on a natural moves the edge most. At 6:5 instead of 3:2, the six-deck game measured 1.77%, which is 1.34 points above the base game.
The reason is one hand. The chance of a natural from a full six-deck shoe is 4.75%, or 1 in 21.1: an ace and a ten-value card in either order. A 6:5 table pays less on exactly that hand. Every other hand is played and paid as before, so the whole 1.34 points comes from the one hand in 21.1.
A dealer who hits soft 17 adds 0.2 points, to 0.64%. Put that rule together with the 6:5 payout and the measured edge is 2.0%.
Fewer decks lower the edge. Two decks measured 0.26%. The single deck measured 0.01%, and its 95% interval runs from -0.01% to 0.03%, so it includes zero: this sample cannot tell that game from break-even. In that run, 206 rounds ran out of cards and drew from a fresh deck. Eight decks measured 0.45%, and the difference from six decks is inside the 95% interval of the eight-deck sample.
Dealer bust rates by up card
| Dealer up card | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | A |
|---|---|---|---|---|---|---|---|---|---|---|
| Dealer busts | 35.3% | 37.4% | 39.5% | 41.8% | 42.2% | 26.2% | 24.4% | 22.9% | 21.3% | 11.5% |
The dealer busts most with a 5 or 6 showing: 41.8% and 42.2% of those hands end above 21. With an ace showing, the rate is lowest, at 11.5%.
These are shares of dealer hands in the base simulation, with every dealer hand played out and dealer blackjacks counted as not bust. The ten-value up card covers four ranks, which is why its row holds 245,902,412 hands.
Your bust rate when you hit
| Your hard total | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
|---|---|---|---|---|---|---|---|---|---|
| Bust if you take one card | 30.8% | 38.5% | 46.2% | 53.8% | 61.5% | 69.2% | 76.9% | 84.6% | 92.3% |
Hard 12 busts on one card 30.8% of the time: 4 of the 13 ranks take it past 21. Hard 16 busts 61.5% of the time, on 8 of 13 ranks. These chances are counted from a full shoe, rank by rank, with no simulation needed.
A hard total of 11 or less cannot bust on one card.
What card counting changes, in numbers
The count measured here is Hi-Lo. Cards 2 to 6 count +1, 7 to 9 count 0, tens and aces count -1. The true count is the running count divided by the decks left, rounded down. The figures come from the same six-deck simulation, with one hand per round and the same bet every hand.
| Hi-Lo true count | Share of hands | Player’s edge |
|---|---|---|
| -4 | 3.90% | -2.18% |
| -3 | 6.74% | -1.66% |
| -2 | 12.41% | -1.10% |
| -1 | 18.91% | -0.65% |
| 0 | 26.65% | -0.19% |
| +1 | 11.79% | +0.38% |
| +2 | 6.48% | +0.83% |
| +3 | 3.67% | +1.25% |
| +4 | 2.09% | +1.76% |
| +5 | 1.18% | +2.09% |
| +6 | 0.64% | +2.43% |
At a true count of 0, the player’s edge measured -0.19%. It turns positive from a true count of +1, at +0.38%, and rises with the count from there.
Those counts are not common. Only 26.6% of hands came at +1 or higher, and 14.8% at +2 or higher. The other 73.4% were played at a true count of 0 or less, where the house keeps its edge. Added up across every count, the same hands give the 0.44% house edge at the top of this page.
This section is a measurement. It is not advice on beating a casino, and it carries no bet-sizing scheme.
Playing within limits
Every rule set in the table measured a house edge above zero except the single deck, whose interval includes zero. Set a budget before you play and a time limit for the session, and stop at either one. Our responsible gambling guide has more on both.
*House edges, bust rates and count figures from a Scores and Stats simulation of basic strategy (800 million hands for the six-deck game, 200 million for each other rule set), November 2026; natural and one-card bust chances computed from a full shoe.*