You're dealt a pair of 8s. The dealer shows a 9. Split or play it as one hand? Basic strategy answers:
| Split ← | -0.4054 |
| Hit | -0.5093 |
| Stand | -0.5432 |
| Double | -1.0186 |
| You bust on the next card | 61.5% |
| Dealer 9 busts | 22.8% |
Split is worth 0.1039 more per unit than Hit. On a $10 bet that is $1.04 handed over every time you get it wrong. Across the 340 situations in this chart it ranks 320th by expected value — better than 20 of them, worse than 319. Expected value is per unit staked, computed for multi-deck, dealer stands on soft 17, double after split allowed. The model and the code are published as data.
| Option | Win | Push | Lose |
|---|---|---|---|
| Split ← | 33.5% | 11.2% | 55.2% |
| Hit | 21.6% | 5.9% | 72.5% |
| Stand | 22.8% | 0% | 77.2% |
| Double | 21.6% | 5.9% | 72.5% |
Splitting, each of the two hands wins 33.5% of the time, pushes 11.2% and loses 55.2%. Even the right move loses more often than it wins here: the point is to lose less, not to win. Hitting instead: win 21.6%, lose 72.5%. A split plays two hands, one bet each. Same model as the numbers above; after a hit, the rest of the hand is played by the chart. Download the data.
Splitting turns one bad hand into two live ones, at double the stake. Against a dealer busting 22.8% of the time, the extra exposure is weighed against two better starting cards.
Splitting returns -0.4054 per unit, hitting -0.5093. The gap is 0.1039 — play split: large enough that getting it wrong is a real leak.
A pair of 8s is hard 16 — the worst total in blackjack. Splitting escapes the disaster and restarts two hands from 8, which plays far better than 16 against any card, even a 9. It's not about winning big here; it's about losing less, and the math is unambiguous.