Blackjack basic strategy, explained

Basic strategy is the mathematically best fixed decision for every combination of your hand and the dealer's upcard, computed for one exact set of rules. That is the whole idea. It is not a betting system, it does not predict the next card, and it does not care about streaks, hunches, or how the table has been running. It also does not beat the house edge — no fixed strategy does — it minimizes it, which over thousands of hands is the difference between losing slowly and losing fast.

Every number on this page is derived from the engine code of our ownblackjack practice simulation, not paraphrased from other strategy sites. The dealer probabilities below come from a small script that exhaustively enumerates every dealer draw sequence from a fresh six-deck shoe, and the chart printed here is the same decision table the simulation's built-in coach uses when it recommends a play. The table deals in practice credits with no cash value, so you can drill every one of these decisions with nothing at stake.

The rules this chart assumes

A basic strategy chart is only correct for the rules it was built for. Change the rules and some cells change with them. This chart — and everything derived on this page — assumes the exact rules our simulation enforces, published in its engine as rules version silver-blackjack-s17-v1:

• Six decks, dealt from a shoe and reshuffled at 75 percent penetration.
• The dealer draws to 16 and stands on all 17s, including soft 17 (S17).
• With an ace or any ten-value card up, the dealer peeks for blackjack. A dealer natural ends the round immediately, and a player natural pushes against it.
• A natural blackjack pays 3:2. Hands made after a split can never count as a natural.
• You may double on any first two cards, including after a split, but never on split aces. Doubling takes exactly one card.
• You may split any two cards of equal value — including two different ten-value cards, so a king and a queen count as a pair — up to four hands total. Split aces receive one card each and stand automatically.
• No surrender, no insurance, no even money. A push returns your stake.

Why does this matter? The dealer standing on soft 17 rather than hitting it changes roughly a half-dozen cells, including 11 against an ace, which is a double under hit-soft-17 rules and a plain hit here. No surrender means the two famous surrender hands, 16 against a 10 and 15 against a 10, become hits. Double-after-split makes several low pairs worth splitting that would otherwise be hits. If you take this chart to a table with different rules, most of it still holds, but the borderline cells will be wrong.

Hard totals: hit, stand, or double

A hard total is any hand without an ace counted as 11. These are the rows you will use most often, so learn them first.

Dealer bust rates by upcard

The logic behind every hard-total row comes down to one fact: the dealer has no choices. The dealer must draw to 16 and stand on every 17, so given an upcard, the probability of every dealer final result can be computed exactly. We did exactly that, enumerating every possible draw sequence from a fresh six-deck shoe, drawing without replacement. For the ten-value and ace columns, the probabilities are conditioned on the dealer not having a natural, because by the time you act, the peek has already told you the round is still live.

Read the bust column and the hard-total chart explains itself. Your totals of 12 through 16 are the bust-or-stand region: hitting them risks breaking your own hand, and standing wins only if the dealer breaks theirs. Against a 2 through 6, the dealer busts between 35.4 and 42.3 percent of the time, often enough that standing on a hopeless-looking 14 or 15 is right — you are not playing to make a hand, you are stepping aside and letting the dealer take the bust risk. Against a 7 through ace, the dealer makes 17 or better between 73.8 and 83.3 percent of the time, so a stiff total that stands is nearly a guaranteed loss. You have to draw and try to improve, even though drawing will sometimes bust you.

The one wrinkle in the region is 12 against a 2 or 3. Those are the weakest of the weak dealer cards — 35.4 and 37.4 percent bust, the lowest rates in the 2-to-6 group — and only a ten-value card can bust your 12, which is roughly 31 percent of the shoe. The math tips to hitting. From 4 through 6, the dealer bust rate climbs to 39.6 percent and above, and standing takes over.

Doubling on hard totals is about attacking with a hand that cannot bust on one card. With 11, no single card breaks you and just under a third of the shoe — about 31 percent — makes you 21, so you double against everything except the ace. With 10, you double against 2 through 9, every upcard your 20-ish outcomes dominate. Hard 9 doubles only against 3 through 6, where the dealer bust rate is high enough to pay for the doubled stake.

Why you hit 16 against a 10

This is the play people hate most, so look at the ten-value row of the dealer table. Even after the peek has ruled out a dealer blackjack, a dealer showing a ten makes 17 or better 77.0 percent of the time, and lands exactly 20 in 36.8 percent of rounds. If you stand on 16, you win only when the dealer busts: 23.0 percent of the time, and you never push. Hitting 16 busts you often — any card above a 5 breaks the hand — but the times you catch a small card and reach 17 to 21 convert enough losses into wins and pushes to beat standing. Both options lose more than they win; hitting simply loses less. Basic strategy is full of decisions like this one, where the job is not to find a good option but to take the least bad one. In a rule set with surrender, 16 against a 10 would be a surrender; here surrender does not exist, so you hit.

Soft totals: the one-card safety net

A soft total contains an ace counted as 11, which means one card cannot bust you — if the draw is too big, the ace drops to 1 and you keep playing. That safety net is why the soft rows are so much more aggressive about doubling than the hard rows.

The pattern: the weaker the dealer, the earlier you double. Against a 5 or 6 — the two highest bust rates in the game at 41.8 and 42.3 percent — even soft 13 is worth a doubled stake, because you cannot bust the draw and the dealer breaks so often that a mediocre finishing total still wins plenty. Against 4 you need soft 15, against 3 soft 17, against 2 you never double a small soft hand. Soft 19 and 20 are simply strong totals: stand everywhere.

Soft 18 deserves its own sentence because it is misplayed constantly. Eighteen sounds like a made hand, but against a 9, 10, or ace it is a loser on average — check the dealer table: those upcards finish on 19 or better 53.3, 52.7, and 45.6 percent of the time. So you hit, protected by the ace. Against 3 through 6 you double, and if the situation does not allow a double, you stand rather than hit. Against 2, 7, and 8 you stand, because 18 beats the 17s a 7 usually makes and ties or beats the typical 8 outcome.

Should you hit soft 17?

Yes — always, in this rule set. Standing on soft 17 is never the play, against any upcard. A 17 can never win against a dealer who finishes a hand; it can only push a dealer 17 or wait for a bust. Since a one-card draw cannot bust soft 17, drawing costs you nothing and can turn the hand into 18 through 21. Against a dealer 3 through 6 you go further and double, collecting the dealer's high bust rate with two units instead of one. It is telling that the dealer in this simulation stands on soft 17 by rule — that rule exists because standing on soft 17 is bad for whoever does it, which is exactly why the S17 rule is player-friendly and why you, with a choice, should never stand on yours.

Splitting pairs

Splitting turns one hand into two, each with its own stake, so a split is really a bet that two hands starting from one card each beat one hand starting from the pair. Sometimes that is obviously true, sometimes obviously false, and a few cells depend on the upcard. Note that any two ten-value cards count as a pair here — the table will offer a split on a king and a queen — but basic strategy says never take it: twenty is the second-best hand in the game, and the coach treats every ten-value pair as a hard 20 and recommends standing.

A few cells worth understanding rather than memorizing. Nines split against 2 through 6 and 8 through 9 but stand against a 7: the dealer table shows a 7 finishing on exactly 17 in 36.9 percent of rounds, and your 18 beats every one of those, so the made hand is worth keeping. Fives never split — a pair of fives is hard 10, one of the best doubling hands in the game, and breaking it into two 5-start hands throws that away. Fours split only against 5 and 6, and only because this rule set allows doubling after the split, which lets each new hand attack the weakest dealer cards. Twos, threes, sixes, and sevens split against the weak half of the dealer range and hit against the strong half.

Why you always split aces and eights

Aces are the easy one: a lone ace is the best starting card in the game, turning about 31 percent of draws into 21. Playing A,A as a hand means starting from a soft 12; splitting means two hands that each start from 11. Even under this simulation's restriction — split aces get exactly one card each and stand, and a resulting 21 pays 1:1 rather than 3:2 because split hands are never naturals — splitting is still clearly right against every upcard.

Eights are the defensive one. A pair of eights is hard 16, the worst total in blackjack: too high to draw safely, too low to win standing. Splitting swaps that one terrible hand for two hands that each start from 8, drawing toward 18. Against a weak dealer card you turn a loser into two contenders; against a 9, 10, or ace you are usually just reducing how much the hand loses. Both are wins for the split compared to the alternative of hitting or standing on 16 twice over.

How much does basic strategy actually save?

Under the exact rules above — six decks, dealer stands on soft 17, double on any two cards including after splits, split to four hands, one card on split aces, dealer peeks, blackjack pays 3:2, no surrender — published combinatorial analyses of this rule set put the house edge for a perfect basic strategy player at roughly 0.4 percent. In plain terms: over a very long run, perfect play loses about 40 credits per 10,000 credits wagered. That figure is not something our dealer-outcome script computes — it requires evaluating every player decision tree, not just the dealer's — so we cite it as consistent with published analysis of these rules rather than as our own derivation.

What is well established is the cost of not using the chart. Typical guessing play — standing on stiff hands out of fear, never doubling soft hands, refusing to split eights against a ten — gives back roughly 1.5 to 2 or more percentage points on top of that baseline, several times the edge of the game played correctly. Basic strategy will not make you a winner; the edge survives perfect play by design. What it does is make every decision cost you the minimum, which is the most any fixed strategy can do. If you want to see the gap for yourself, play a few hundred hands on the practice table with the coach on, then a few hundred ignoring it, and compare the credit totals. The credits have no cash value, which makes it a free experiment — and if you would rather see how a game with no decisions at all behaves, the Two-up rules guide is a good contrast.

There are more practice games and derivation-backed guides in thecasino room and on the guides index.

How we calculated this

Rule set: six decks, dealer draws to 16 and stands on all 17s including soft 17, dealer peeks for blackjack with an ace or ten-value upcard, natural blackjack pays 3:2, double on any first two cards including after splits but never on split aces, doubling takes exactly one card, split any two equal-value cards up to four hands, split aces receive one card each and stand, no surrender, no insurance, no even money, pushes return the stake. This is the rule set the simulation engine enforces, published in its code as rules version silver-blackjack-s17-v1.

The dealer final-result table was computed by a standalone Node script in this repository at apps/games/scripts/derivations/blackjack-dealer-outcomes.mjs. It models ranks as classes (ace, 2 through 9, and a 16-cards-per-deck ten-value class), starts from a fresh six-deck shoe, removes the upcard, and exhaustively recurses over every possible draw sequence without replacement until the dealer reaches 17 or busts, summing exact path probabilities — no simulation or sampling. For ten-value and ace upcards, the hole card is conditioned on not completing a natural, matching the state of play after the engine's peek. Each row was verified to sum to 1 before rounding to one decimal place.

The strategy chart printed above is a direct transcription of the decision table in the simulation's strategy module — the same table its in-game coach and chart panel render — so the guide and the game can never disagree. The roughly-0.4-percent house-edge figure is attributed to published combinatorial analyses of this exact rule combination, as noted above, not to our own script. Everything here describes a practice simulation using credits with no cash value; nothing on this page is a claim that any strategy wins real money.