The Splitting Mistakes That Cost the Most, Measured

The Five Most Expensive Pair Errors

Every figure below is the expected value per unit staked, computed for a multi-deck game where the dealer stands on soft 17 and double after split is allowed. The gap column is what choosing the second-best action costs you on that hand.

Notice the pattern. Three of the five most expensive errors are splitting tens when you should stand, and the other two are failing to split when you should. Both mistakes come from the same place: treating the pair as a total rather than as two hands.

Eights Against a Ten: Losing Less

This is the hand that makes people distrust the chart, so it is worth going through slowly.

You hold 8,8 — sixteen, the worst total in blackjack. The dealer shows a ten. The options:

Every option loses. Splitting loses least, by about five hundredths of a unit — so splitting is correct, and it is correct while guaranteeing that you will usually lose more money than if you had hit, because you now have two bets on the table.

Against a nine and against an ace the gap is larger still — 0.10 and 0.12 units respectively — and in both cases splitting still shows a loss. The hand is never good. It is only ever less bad.

The Tens Trap

Twenty is the second-best hand in blackjack. Splitting it turns one hand worth roughly +0.55 to +0.79 into two hands starting from a ten, and the arithmetic is brutal: against a nine, standing is worth +0.76 and splitting +0.23. You give up half a unit per hand for the chance to win two bets instead of one.

The temptation is strongest exactly where the loss is largest — against a weak dealer card, where splitting tens looks like maximising a good spot. It is the same instinct that makes small bets feel safer at roulette, and it is wrong for the same reason: the shape of the outcome is being confused with its value. The bold play article works through that confusion in a different game.

There is one exception, and it is not for you: card counters split tens at very high true counts, when the shoe is dense enough in tens to make two hands starting from ten better than one twenty. It requires a count, and it is considerably harder to do than to describe.

Aces, and Why the Rule Is Absolute

Split aces against every dealer card, without exception. The gap against a 6 is 0.48 units, against a 5 it is 0.46, against a 4 it is 0.44 — the rule is never close.

The reason is structural. A pair of aces played as one hand is a soft twelve, which is close to the worst soft total there is. Played as two hands, each starts from eleven with a ten as the single most likely next card. Most casinos give you exactly one card per split ace, and even with that restriction the two hands are worth far more than the twelve.

The same logic is why eights are always split against low cards: sixteen is the worst hard total, and two hands starting from eight are better than one hand of sixteen, even though eight is not itself a good start.

When the Rules Change the Answer

Everything above assumes double after split is allowed. Without it, several pair decisions flip — 2,2 and 3,3 against a 2 or a 3, and 4,4 against a 5 or a 6 — because the value of splitting depended on being able to double the resulting hands. The rule is worth about 0.14% overall, and the pairs are where it does its work.

Resplitting matters less. Being unable to resplit aces costs about 0.08%, and it never changes whether you split in the first place.