Two plays that add to the bet
Most decisions in blackjack concern one hand and one stake. Doubling down and splitting are different, because both increase the amount at risk. Doubling lets you double the bet in exchange for taking exactly one more card. Splitting turns a pair into two separate hands, each with a bet equal to the original.
Because more money is on the table, these are the plays where the maths matters most. The chart recommends them only where the extra stake is expected to earn more than it risks. Where it does not, the chart says no, even when the situation looks promising.
Why doubling can be favourable
Doubling is worth it when your hand is strong at the start, and one more card is likely to leave you in a good position, while the dealer is relatively exposed.
The standard example is a hard 11. Any ten-valued card gives 21, and in a simple model of thirteen card values with four ten-valued, that is 4 of 13, about 31 percent, the most likely single outcome. A 9 or an 8 also gives a strong 20 or 19. Because you cannot bust on one card, the extra stake adds no bust risk. The details of doubling on eleven are covered in their own article.
A hard 10 works similarly, but one card brings fewer top totals. A hard 9 is weaker still and is usually doubled only against the dealer’s weakest cards.
The dealer’s upcard matters because the player’s advantage from a double comes from the combination of a good draw and a dealer likely to struggle. Against a dealer showing a 5 or 6, who busts more often than most, a double on 9, 10 or 11 is commonly recommended. Against a dealer ten or ace, doubling on the same totals is less attractive, and the chart often calls for a simple hit instead.
Soft hands can be doubled too, when the total is low and the dealer is weak, since the ace removes the bust risk from the single card.
When doubling is not favourable
Doubling a hand that is already good does not help much. A hard 17 or 18 gains little from one more card, and often loses value, so the extra stake would be added to a hand that cannot improve. A hard total of 12 to 16 is not doubled, because a single card has a high chance of busting it.
Doubling after a pair split depends on the table rules. Doubling restrictions also matter, since if a table limits which totals you may double, the chart changes. Always check the rules before relying on a chart.
One further point: a double is a bet with a favourable average only relative to the alternatives. It is still a hand against a dealer, and the average gain is modest. The fact that the double wins more often than it loses in a good spot does not mean it wins every time. When it loses, you lose twice the usual stake.
Why splitting can be favourable
Splitting works when two hands starting from the same card are expected to do better, in total, than one hand made of the pair.
Aces. A pair of aces makes only a soft 12 or a hard 2, a poor hand. Split, each ace starts a hand that has a good chance of reaching a strong total, since a ten-valued card (about 31 percent in the simple model) makes 21. The reasoning for splitting aces and eights is set out in a separate piece.
Eights. A pair of eights makes a hard 16, the worst of the usual totals, particularly against a ten. Splitting gives two hands starting from 8, which is a much healthier start. It does not make the situation good, and splitting eights against a strong dealer card is still a loss on average, but it is a smaller loss than the alternative. This is an example of a play that reduces losses rather than creating a favourable bet.
Small and middle pairs. Pairs of 2s, 3s, 6s, 7s and 9s are split against some dealer cards and not others, usually depending on whether the dealer is likely to bust. The details are part of pair splitting rules, and they vary with whether doubling after a split is allowed.
When splitting is not favourable
Tens. A pair of tens is 20, one of the best hands in the game. Splitting it replaces a strong made hand with two hands that start at 10 and usually end below 20. Splitting tens is regarded as a clear error in standard play.
Fives. Two fives are 10, which is a good doubling total, so the chart treats them as a hard 10 rather than splitting into two weak 5s.
Fours. A pair of 4s is 8, which is rarely split, as the chart usually prefers a simple hit.
Limits on splitting also matter. Some tables restrict resplitting aces, or limit the number of hands you can create, so resplit limits are worth reading.
Reading the risk honestly
Both plays raise your exposure. A player who doubles and splits correctly will also see larger swings, because bets are bigger on the hands where these plays occur. A losing double costs two units, and a split followed by two losses costs two units too.
The standard deviation per hand is already large compared with the house edge, and these plays add to it. The mathematics says the plays have a better average than the alternatives. It does not say any individual outcome will be favourable. Results vary widely over a session, and even perfect play is expected to lose slightly over the long term because the house edge remains.
Practical points
Keep doubles and splits within the budget you set before sitting down. A play that is correct on average is not a reason to bet more than you can afford, and a run of losses is not a reason to double up to win it back. Recovering losses by raising stakes is the pattern of chasing losses, which has no mathematical support. If gambling stops feeling like entertainment, a local support service can help.