Splitting Pairs in Blackjack

A split earns its place in one of two ways. The first is escape: some pairs add up to a poor total, and two hands that each start from one card lose less than that total would. A pair of eights makes a hard 16, while a lone 8 is a workable start. The second is pressure: against a dealer 4, 5 or 6, upcards that bust often, a second stake on a decent card collects twice.

Over a session these choices add up. A player who follows every other line of the chart but never splits faces a house edge of 0.97%, where taking every split leaves it at 0.41%.

The key number: Refusing every split hands the house an extra 0.57 percentage points.

Lines hold for the standard six-deck game with a double allowed after a split. A pair that stays together is played as its total, so 5,5 is a hard 10 and 10,10 a hard 20.

Eights are the clearest case of escape. Against a dealer 7, hitting the 16 returns −40.8 per 100 units bet, and two hands that each begin with an 8 return +31.9, a gain of +72.7 that is the largest of any split in the game. Against a 9, a 10 or an ace both choices lose, and the split is simply the smaller loss.

A pair of aces counts as a soft 12, while each ace on its own starts at 11 and makes 21 with any ten-value card. Even against a dealer ace the split is worth +12.7 and a hit −2.2, with each split ace taking the single card most tables allow.

Keeping either pair together is an expensive habit for how seldom it arises. A pair of eights is dealt about 1 in 185 hands, and playing it as a 16 gives up 40.6 per 100 units bet on each occasion, adding 0.22 points to the house edge across the game. Aces left together lose 33.1 per 100 units bet whenever they are dealt, 0.18 points of edge in total. The blackjack tips page ranks these beside the other common mistakes.

Figures are per 100 units bet. A split puts a second stake out, so its figure covers both hands measured against the first stake alone.

Fives and tens are the reverse case, pairs worth more together than apart. Two fives make a hard 10, one of the strongest totals to double, while a split leaves two hands built on a 5. Against a dealer 6, doubling the 10 brings in +59.9 per 100 units bet, the split only +6.1.

Two tens are already a 20, and standing shows a profit against every upcard, from +55.9 against a 10 to +79.0 against an 8. Splitting trades that for two hands rebuilt from a 10, and the trade loses at every upcard, the tempting 5 and 6 included.

The rule: Keep fives and tens together: each pair is worth more as the total it already makes than as two separate starts.

Splitting tens whenever they arrive, about 1 in 11 hands, throws away 65.2 per 100 units bet a hand and lifts the house edge by 5.84 points, the largest price of any single misplay. A split of fives gives away 68.5 per 100 units bet, or 0.371 points over the long run.

A negative gain means the split loses to the play shown, by that many units per 100 bet.

For 2,2, 3,3, 4,4, 6,6, 7,7 and 9,9 the upcard decides. Against a 2 to a 6 the split puts a second stake out while the dealer is most likely to bust; against an 8 or higher the new hands face a dealer who usually finishes on a real total.

The edges of each range are close calls. Splitting 3,3 against a 2 gains only +0.9 per 100 units bet, and 4,4 against a 5 only +2.6, which is why both turn into hits at a table that refuses a double after a split. One upcard past its range, 2,2 against an 8 already gives up 2.0 by splitting.

Nines break the pattern the other pairs follow. A pair of nines is 18, worth keeping only where the dealer is likely to fall short of it. Against a 7 the 18 stands at +40.0 and the split returns +36.4; against an 8 or a 9 an 18 is too often beaten, so the split takes over, and against a 10 or an ace standing loses less again.

Each pair appears at the edge of its range. A negative gain means keeping the pair is worth more.

Small pairs are split against low cards partly for the doubles that can follow. A split 2, 3 or 4 often lands on 9, 10 or 11, a hand worth doubling against a dealer 2 to 6, so a table that allows a double after a split, written DAS on the rule card, makes those splits pay. Without the rule the six-deck game costs 0.55% in place of 0.41%.

The chart answers by giving up 7 splits, each a pair of small cards facing a low upcard, as listed below. The doubles themselves are priced on the doubling down page.

Values for six decks with an S17 dealer, as the average return per 100 units staked on the original pair. Without DAS the hit is worth more in every one of these cells, so the pair is hit, and fours lose their only splits. With DAS the same split moves ahead of the hit. Every other split in the chart is unchanged.

Only the named rule differs from the reference six-deck table in each line, and the chart is refitted to it. The blackjack odds page sets these beside every other rule.

The resplit limit hardly matters, since a third and fourth hand need the same rank to turn up again. Cutting the limit from four hands to three costs 0.01 points, and allowing a single split costs 0.05.

Aces are where split rules carry weight. A split ace normally takes one card and stops, so an ace that draws a 2 through a 7 is left on a soft total another card could have improved. Tables that let split aces be hit give back 0.19 points, the most of any split rule, and resplitting aces gives back 0.07.

Two kinds of table rewrite splitting altogether. Free Bet blackjack backs most splits with a free bet instead of a second stake, and at a no-hole-card table a late blackjack can collect the split hands too, turning 8,8 facing a 10 or an ace and A,A facing an ace into hits; the European blackjack page covers that game.