With six decks, the dealer standing on soft 17 and doubling allowed after a split, the correct pair splitting play fits in eight lines. Always split aces and eights. Never split fives or tens. Split nines against 2 to 9 except 7. Split twos, threes and sevens against 2 to 7, sixes against 2 to 6, and fours only against 5 and 6. The rest of this article derives each line and prices the close calls.
Rule assumptions
- Six decks. The dealer checks for blackjack when showing an ace or a ten value card, so every figure below assumes the dealer does not have blackjack.
- Dealer stands on soft 17.
- Doubling is allowed on any first two cards, including after a split (DAS).
- Split aces receive one card each.
- No surrender. Resplitting is left out of the figures, which treat every split as a single split.
- Blackjack pays 3 to 2.
Change any of these and parts of the chart change. The section on DAS shows the biggest of those shifts.
The expected values quoted here were computed for this article with an infinite deck model, in which every card is drawn with fixed probability: 4 in 13 for a ten value card, 1 in 13 for each other rank. For pair decisions that model sits within a few thousandths of a unit of a six deck shoe and does not flip any line of the chart. All values are in units of the original stake. A value of minus 0.489 means an average loss of 48.9% of the first bet across the whole decision, including the second bet that a split adds.
The splitting chart
| Pair | Split against | Otherwise |
|---|---|---|
| A,A | Every upcard | Not applicable |
| 8,8 | Every upcard | Not applicable |
| 9,9 | 2 to 6, 8, 9 | Stand on 18 against 7, 10, ace |
| 7,7 | 2 to 7 | Hit hard 14 |
| 6,6 | 2 to 6 | Hit hard 12 |
| 4,4 | 5, 6 | Hit hard 8 |
| 3,3 | 2 to 7 | Hit hard 6 |
| 2,2 | 2 to 7 | Hit hard 4 |
| 5,5 | Never | Double hard 10 against 2 to 9, otherwise hit |
| 10,10 | Never | Stand on 20 |
Read the chart as a filter that runs before the hard and soft total charts. If the pair line says split, split. If it does not, treat the hand as the total it already is.
Why each line is correct
Aces and eights: two weak starts beat one terrible total
A pair of eights is hard 16, the worst total in the game. Against a dealer 10, standing and hitting both return minus 0.540 in this model. Splitting returns minus 0.489. The gain is 0.051 units per decision, and the split still loses money on average. That is the point many players miss: the chart does not say splitting 8,8 against a 10 wins. It says it loses less than the alternatives.
Against an ace the gap widens. Hitting 16 returns minus 0.517, splitting returns minus 0.394, a gain of 0.123 units. Two hands that each start on 8 have real chances of reaching 18 or better, while 16 has almost none without busting.
Aces are the clearest case. Played as one hand, A,A is soft 12. Against a 10 the best unsplit play (hit) returns minus 0.070. Split, it returns plus 0.180, a swing of 0.250 units even though each ace receives only one card.
Tens and fives: never split, for opposite reasons
A pair of tens is already 20. Against a dealer 6, the weakest upcard, standing returns plus 0.704. Splitting returns plus 0.576. Splitting tens gives up 0.128 units on the single hand where the dealer is most likely to bust, which is why the "the dealer is weak, so split" instinct is wrong here: the existing 20 is worth more than two hands starting on 10.
Fives fail the other way. As a pair they make hard 10, one of the strongest doubling totals, which is why the chart sends 5,5 to the doubling line. Split, each hand starts on 5, a total that cannot be doubled profitably and needs at least two more cards to become competitive.
Nines and the dealer 7 exception
Nines split against every upcard from 2 to 9 except 7. The reason sits in how a dealer 7 finishes. In this model a dealer showing 7 ends on exactly 17 in 36.86% of rounds and busts in 26.23%. A standing 18 beats both outcomes, 63.09% of all finishes, loses to 19, 20 and 21 in 23.13%, and pushes on 18 in the remaining 13.78%. That nets plus 0.400. Splitting returns plus 0.370, so standing is worth 0.030 more.
Against a 10 the dealer finishes on 20 in 37.07% of rounds and 18 stops being competitive enough to split for. Standing returns minus 0.178 and splitting minus 0.297. Against an ace, standing returns minus 0.100 and splitting minus 0.131.
Small pairs: where the DAS rule decides the line
Twos, threes, fours and sixes are the hands that change with the rules. Their split value depends heavily on the chance of catching a card that makes a doubling total on one of the new hands, and that only pays if doubling after a split is allowed.
2,2 against a dealer 2 shows the effect. With DAS, splitting returns minus 0.089 and hitting minus 0.115, so split. Without DAS the split falls to minus 0.152 and hitting becomes correct.
4,4 against a 5 is the closest call on the chart: splitting with DAS returns plus 0.080 against plus 0.071 for hitting hard 8, a margin of 0.009 units. Remove DAS and the split drops to minus 0.025, so the pair is never split.
What changes without double after split
Some tables do not allow doubling after a split. Only four lines move.
| Pair | With DAS | Without DAS |
|---|---|---|
| 2,2 | Split 2 to 7 | Split 4 to 7 |
| 3,3 | Split 2 to 7 | Split 4 to 7 |
| 4,4 | Split 5, 6 | Never split |
| 6,6 | Split 2 to 6 | Split 3 to 6 |
Aces, eights, nines, sevens, fives and tens keep the same lines. Sevens hold because 14 is bad enough that two hands starting on 7 are better even without doubling: against a dealer 2, splitting without DAS returns minus 0.218 against minus 0.293 for standing on 14, the better of the unsplit plays.
Dealer bust rates behind the chart
The chart's shape follows the dealer's bust rate by upcard. Given no blackjack, this model has the dealer busting 35.36% of the time with a 2 showing, 37.39% with a 3, 39.45% with a 4, 41.64% with a 5 and 42.32% with a 6. With a 7 the figure drops to 26.23%, and with a 10 it is 22.98%.
That cliff between 6 and 7 is why so many small pair lines stop at 6 or 7. Against 2 to 6, a split puts a second bet on the table when the dealer is most likely to break. From 8 upward the dealer usually makes a hand, and only aces, eights and nines (against 8 and 9) still justify the extra stake.
What this strategy does not change
Correct splitting reduces the house edge. It does not remove it. Under these rules, a player making every basic strategy decision correctly, splits included, still faces a house edge of a little under half a percent of total stakes. No splitting line, no bet size and no session length turns that into a positive expectation.
The chart is also specific to blackjack. It works because a blackjack decision changes the expected result of the hand. A slot spin offers no decision that does the same, which is why no equivalent chart exists for one.
Responsible gambling
Online blackjack in Malta is open to players aged 18 and over at operators holding an MGA licence. Splitting has a cost that a single hand does not: in one decision, the money on the table doubles, and on hands like 8,8 against a 10 both stakes are on a play that still loses on average. Set the session budget in terms of what a run of split hands could take, not the size of a single bet. If gambling stops feeling like a choice, the Responsible Gaming Foundation offers free, confidential help on its national gambling helpline, 1777.