Under standard six deck rules with the dealer standing on soft 17, late surrender is correct on exactly four two card situations: hard 16 against a dealer 9, 10 or Ace, and hard 15 against a dealer 10. Every other hand should be played out. Used correctly, the option reduces the house edge by roughly 0.08%. The derivation below shows why these four hands qualify and why hands that look similar do not.
Rule assumptions
| Rule | Setting |
|---|---|
| Decks | 6 (figures computed on an infinite deck model, see method) |
| Dealer soft 17 | Stands |
| Hole card | Dealer peeks for blackjack with Ace or 10 showing |
| Surrender | Late, first two cards only, after the peek |
| Double | Any first two cards, double after split allowed |
| Split | Once; split Aces receive one card |
| Blackjack pays | 3:2 |
The chart
| Player hand | Dealer 9 | Dealer 10 | Dealer A |
|---|---|---|---|
| Hard 16 (not 8,8) | Surrender | Surrender | Surrender |
| Hard 15 | Hit | Surrender | Hit |
| 8,8 | Split | Split | Split |
Against any dealer upcard from 2 to 8, surrender is never correct under these rules.
The decision rule
Surrender returns half the stake. Its value is therefore fixed at minus 0.5 units. The correct play is to surrender whenever the best alternative, standing, hitting, doubling or splitting, has an expected value below minus 0.5.
This reframes the question. Surrender is not about whether a hand is bad; nearly every stiff hand against a strong upcard is bad. It is about whether the hand is worse than losing exactly half.
Derivation
Expected values per unit staked, conditioned on the dealer not holding blackjack (the peek has already happened when the decision is made).
| Hand vs upcard | Stand | Hit | Best play | Surrender gain |
|---|---|---|---|---|
| 16 vs 9 | minus 0.543 | minus 0.509 | Surrender | 0.009 |
| 16 vs 10 | minus 0.540 | minus 0.540 | Surrender | 0.040 |
| 16 vs A | minus 0.667 | minus 0.517 | Surrender | 0.017 |
| 15 vs 10 | minus 0.540 | minus 0.504 | Surrender | 0.004 |
| 15 vs 9 | minus 0.543 | minus 0.472 | Hit | none |
| 15 vs A | minus 0.667 | minus 0.480 | Hit | none |
| 16 vs 8 | minus 0.511 | minus 0.458 | Hit | none |
Hard 16 against 10
The largest single gain. Standing and hitting are almost exactly equal, both near minus 0.540, which is why this hand is argued about at every table. Surrender sidesteps the argument: at minus 0.5 it beats both by 0.040 units, the biggest margin in the chart.
Hard 16 against Ace
Standing is catastrophic at minus 0.667, because a dealer Ace that has already passed the peek is still likely to finish on 17 to 21. Hitting improves this to minus 0.517, still worse than minus 0.5. Surrender gains 0.017.
Hard 15 against 10
The narrowest surrender in the chart. Hitting a 15 busts less often than hitting a 16, so the hit value, minus 0.504, sits only 0.004 below the surrender line. Against a 9 or an Ace, hitting 15 clears the line comfortably, and surrender is wrong.
Why 8,8 is excluded
A pair of 8s totals 16 but is not played as 16. Splitting against a 10 is worth minus 0.490, marginally better than surrendering at minus 0.5. Against a 9 the split is worth minus 0.405 and against an Ace minus 0.394. The pair is split in every case under these rules.
Why 17 is excluded
Hard 17 against an Ace stands at minus 0.478 when the dealer stands on soft 17. That is better than minus 0.5, so surrender is incorrect.
When the dealer hits soft 17
A dealer who hits soft 17 finishes stronger with an Ace up, and three more hands cross the minus 0.5 line:
| Hand vs Ace, H17 | Best alternative | Correct play |
|---|---|---|
| Hard 15 | Hit, minus 0.507 | Surrender |
| Hard 17 | Stand, minus 0.516 | Surrender |
| 8,8 | Split, minus 0.527 | Surrender |
This is the clearest illustration of why a chart is worthless without its rule set: the same hand flips from split to surrender on a single dealer rule.
Method
Figures are computed on an infinite deck model, in which the probability of each card rank is constant. This removes card removal effects, which in a real six deck shoe shift individual hand values by small amounts. The infinite deck model reproduces the six deck surrender chart for these rules exactly; the edge figures below are infinite deck values and will differ slightly from exact six deck calculations.
What surrender changes, and what it does not
On the infinite deck model, the base game under the listed rules carries a house edge of 0.570%. Adding late surrender with correct use lowers it to 0.485%, a reduction of 0.085%.
Surrender changes one thing: it caps the loss on four specific hands at half a unit instead of a larger expected loss. It does not change card probabilities, the dealer's rules or the long run result. Under perfect play the game retains a house edge of about 0.49% on this model. Surrendering incorrectly, for example on 15 against an Ace or on 8,8, gives back part of the gain on every occurrence.
Responsible gambling
Blackjack at MGA licensed operators is restricted to players aged 18 and over. Surrender is a way to lose less on a bad hand, not a way to come out ahead, and a correct chart still leaves a negative expectation on every session. Deposit limits and self exclusion are available at every licensed operator; the Responsible Gaming Foundation helpline is 1777 and Appoġġ Supportline is 179.