Blackjack Decision EV Calculator: The Math Behind Every Move

Use this Blackjack Decision EV Calculator to compare hit, stand, double, split and late surrender for an original two-card hand. It shows expected net profit or loss per dollar of your original bet, converts each EV to your stake, and calculates the expected cost of choosing another action.

The model uses an infinite deck with dealer blackjack already excluded, S17 or H17, and your selected split rules. It calculates action values from card probabilities rather than predicting the next card. For a quick conventional 4–8-deck reference, use the rule-aware basic strategy chart; compare model assumptions when the two tools differ.

Default example: 10 + 6 against dealer 7, S17, DAS, up to four split hands and no surrender. Hit has EV −0.414779, compared with −0.475375 for Stand. At a $25 original bet, standing instead of hitting reduces expected value by $1.51. That is an average cost over comparable situations, not an additional charge or a guaranteed loss on this hand.

How to Use the Calculator

1. Enter your original two cards and the dealer upcard. Ten, Jack, Queen and King share one choice because each has value 10. Aces are handled as 1 or 11. Enter pairs as their two cards so that Split is included.
2. Match the blackjack dealer rule. S17 stands on soft 17; H17 draws on soft 17. Both stand on hard 17. Against a dealer Ace or ten-value card, use the results after the dealer checks and does not have blackjack.
3. Set double after split and late surrender. DAS affects the available play on split hands. Surrender loses half the original bet and is offered only as the initial action. It is unavailable after hitting or splitting in this model.
4. Choose the maximum number of split hands. Two hands permits the first split only. Three or four hands allows further splits of non-Ace pairs up to a shared limit. Split aces always get one card each and cannot be split again.
5. Enter the original bet. Use the amount before any double or split. The default is $25. Changing the stake changes the dollar amounts, not the relative ranking of actions.
6. Read every available row. The highest EV is marked Best. Cost vs best compares each alternative with that action. Reset restores the starting example; on a narrow screen, scroll the results table sideways.

A positive blackjack EV is an expected profit; a negative EV is an expected loss. The preferred action can still be negative: −0.40 is better than −0.50 because the expected loss is smaller. An EV of +0.20 does not mean a 20% chance of winning.

Blackjack EV chart: hard 12–17 against 2–A

This static chart shows the best initial action and its expected net return for non-pair hard totals, S17 and no surrender. Each row uses an original two-card hand of 10 plus the remaining value. Cards are drawn from an infinite deck, and dealer blackjack has already been excluded. Double is allowed on the first two cards but is never the best action in these 60 situations.

H = Hit; S = Stand. Hit includes optimal hit-or-stand decisions after the next card. Values are per $1 original bet, rounded to six decimals; 10 includes J, Q and K, and A means Ace. On a narrow screen, scroll the table sideways.

The chart describes hard totals without a split option. For example, its hard 16 row applies to 10 + 6, while 8–8 can have a better Split EV. Enabling late surrender also changes some decisions. Enter the actual cards and rules in the calculator to compare all available actions and their dollar values.

What the blackjack EV model calculates

Every new card is independent: Ace through 9 each has probability 1/13, and the combined ten-value group has probability 4/13. Selecting 10 + 6 does not remove those cards from a finite shoe. The dealer’s closed card is conditioned on the completed blackjack check: a ten-value hole card is excluded behind an Ace, and an Ace is excluded behind a 10.

Stand EV is P(dealer busts) + P(dealer finishes lower) − P(dealer finishes higher). Equal totals push. If the player busts first, the return is −1, regardless of whether the dealer would later bust. Hit EV averages the outcomes of one card followed by the best hit-or-stand continuation.

Double EV is twice the expected return after exactly one additional card and then standing. The factor of two includes both units at risk, so an EV below −1 is possible and meaningful. For example, doubling hard 20 against 7 has EV −1.703704: it is a legal but very poor action under the assumed doubling rule, not missing data.

Split EV combines the net results of every resulting hand, including additional bets and any doubles. The split limit applies to the whole original hand: four hands means at most three split operations, not four hands from each branch. The calculator considers whether an optional resplit improves expected value; choosing an initial split does not force every later resplit.

Example of the resplit convention: after choosing to split 2–2 against 8, optimal later decisions give Split EV −0.174109 under S17/DAS with a four-hand limit. Forcing every available resplit gives −0.176693, matching the published reference-table value. The convention is inferred from reproducing the table numerically. This explains a difference between models; Hit remains the preferred initial action in this example.

A natural blackjack, Ace plus a ten-value card on the original hand, is settled at 3:2. With dealer blackjack already excluded, its net return is +1.5 units. At $25, that is $37.50 profit and $62.50 returned including the bet. Theoretical half-cent amounts are retained; casino chip-rounding rules are not modeled. A 21 made after splitting is an ordinary 21 paid at 1:1 when it wins. Any two ten-value cards are treated as a splittable pair.

The blackjack decision calculator does not model finite-deck depletion, a true count, European no-hole-card/no-peek rules, early surrender, restricted doubling, resplitting aces, hands already hit or split, side bets, or tournament objectives. These are substantive rule differences, not rounding adjustments. The displayed six decimals are numerical precision within this model, not a claim of that accuracy at every casino table.

The Most Expensive Mistakes in Blackjack

The following examples compare specific alternatives under S17, DAS, a four-hand split limit and no surrender. They illustrate the size of a decision error; they are not an exhaustive ranking of every possible mistake. All costs use the original bet as their denominator.

The difference between Split and Hit for 8–8 against 7 is about 0.735821 units, or $18.40 at a $25 bet. That is a large expected-value gap, although either chosen action can win or lose in a particular round. In contrast, the Hit/Stand gap on hard 16 against 10 is only about 0.000604 units, or roughly 1.51 cents at $25.

Insurance uses a different bet and is not one of the action rows above. Before the dealer checks an Ace, the infinite-deck probability of a ten-value hole card is 4/13. At 2:1, insurance EV is 2 × 4/13 − 9/13 = −1/13 per insurance dollar. A $12.50 insurance bet alongside a $25 main bet therefore loses about $0.96 on average. That is 7.69% of the insurance stake, or 3.85% of the original main bet; the denominators must not be mixed.

Anatomy of an EV Decision: Three Case Studies

Case 1: The “Both Options Lose” Scenario — Hard 16 vs 10

With 10 + 6 against dealer 10 under S17, Hit EV is −0.539826 and Stand EV is −0.540430. Both lose on average. When late surrender is available, its −0.500000 return is higher, so surrender is preferred.

At a $25 bet, Hit has expected net loss $13.50, Stand $13.51 and Surrender exactly $12.50. Hitting instead of surrendering costs about $1.00 in expected value. Without surrender, hitting beats standing by roughly $0.0151. The calculator ranks using unrounded values, which matters for close blackjack decisions.

Case 2: The “Defensive Split” — 8,8 vs 10

With S17, DAS and up to four split hands, Split EV is −0.480686. Hit is −0.539826 and Stand is −0.540430. Splitting is preferred even though its combined expectation is negative; with late surrender enabled it also beats −0.500000.

At $25, splitting has expected net loss about $12.02 across all resulting hands. Hitting instead costs about $1.48 in expected value; surrendering instead costs about $0.48. The split EV already includes extra wagers, so do not double it again. The cash required to follow a split can exceed the original bet even when splitting reduces expected loss.

Case 3: The “Soft 18 Trap” — A,7 vs 9

Under S17, Stand EV is −0.183163 and Hit EV is −0.100744. Hitting is preferred, with a gap of 0.082419 units, or $2.06 at a $25 original bet.

A usable Ace prevents an immediate bust on this hit, but the hand can become weaker: A + 7 + 5 is hard 13. The hit calculation includes the later decision from that new total. It does not assume every card improves soft 18, and it does not recommend hitting solely because the next-card bust risk is zero.

How EV Connects to the House Edge

Hand EV and house edge answer different questions. This blackjack hand calculator conditions on a particular starting hand, upcard and completed blackjack check. House edge averages net results across the full distribution of starting hands and dealer outcomes, including naturals and dealer blackjacks. A simple unweighted average of the rows here is not a house-edge calculation.

Cost of error is (best EV − chosen-action EV) × original bet. For repeated decisions, add their expected costs. As an illustration, three errors costing 0.05 units each over 60 equally sized original bets reduce expected return by 0.15/60 = 0.25 percentage points. At $25 per hand, the extra expected loss is $3.75 over those 60 hands. The number and size of errors are assumptions in this example, not a measured frequency for typical players.

Use the table-rules and house-edge tool for rule comparisons, and the session expected-value guide to relate edge, stakes and playing volume. Neither a favorable hand EV nor correct basic strategy guarantees a profitable session.

How H17 vs S17 Changes Your Decisions

The model recalculates the dealer’s full outcome distribution for H17. It then recomputes Hit, Double and Split using those outcomes. A single universal adjustment to S17 cannot capture those changes.

The following results use DAS, a four-hand split limit and late surrender enabled. EVs are per original bet; the action in each column has the highest value under that profile.

For 11 against Ace under H17, Double is +0.103377 and Hit is +0.102702. The double advantage is only about 0.000675 units, or 1.69 cents on a $25 bet. Both actions have lower EV than their S17 counterparts; doubling becomes preferred because their relative order changes, not because H17 makes doubling more profitable in absolute terms.

Under S17, this infinite-deck model also differs from a conventional 4–8-deck chart at A–2 versus 5 and A–4 versus 4: Hit narrowly beats Double. Use a chart or solver for the actual deck model instead of treating every disagreement between blackjack calculators as a bug.

From Decision EV to Advantage Play

This blackjack EV calculator has no remaining-card or count input. Changing the distribution of ranks changes both dealer outcomes and player draw probabilities, so its results cannot be converted into counted-shoe EV by applying one fixed adjustment.

A true count summarizes a shoe; it does not uniquely specify every remaining rank. Different shoes with the same count can have different action values. The Hi-Lo counting guide explains running and true counts. For the separate probability of busting on one additional card, use the next-card bust calculator.

Related Blackjack Tools

• Starting-hand strategy reference — conventional 4–8-deck chart decisions and action fallbacks.
• House Edge Calculator — rule-based estimates of average casino advantage.
• Payout Calculator — profit and total return from settled outcomes.
• Session Variance Simulator — possible short-session swings under its assumptions.
• Bankroll Calculator — requirements based on specified edge, variance and risk.

Frequently Asked Questions (FAQ)

References and model checks

• Wizard of Odds: infinite-deck expected returns — reference values for S17 actions, split limits and the distinction from finite-deck charts. Compare resplit conventions when checking a deliberately unfavorable split.
• BlackjackInfo: dealer outcome probabilities — separate S17/H17 distributions and the conditioning used after a blackjack check.

Set a spending limit before playing. Expected-value analysis explains decisions but cannot make gambling risk-free. If gambling is causing harm, see our warning signs and confidential help resources.