Pair of 3s against a dealer 8

Checked against the textbook basic-strategy chart: 5,5 doubles, 8,8 splits, 9,9 stands against a 7, hard 12 hits against 2 and 3. How

Correct play
Hit
Better than split by
1.25%
Your expected return, once the dealer has checked for blackjack
-21.7%

This is one of the closest decisions in blackjack. Hit is correct, but it beats split by 1.25% of your bet — and of the 350 hand-versus-upcard spots on the chart, only 9 are closer.

Worth being blunt about what that expected return means: at -21.7%, you lose money on this hand in the long run whatever you do. Playing it correctly does not make it a good spot — it makes it a 1.25% less bad one. The choice is between two losing options, which is exactly why it gets argued about.

Every option, ranked

ActionExpected returnCost vs. best
Hitcorrect-21.72%—
Split-22.97%−1.25%
Stand-51.05%−29.33%
Double-100.13%−78.41%

Expected return per unit staked, from an exact recursive engine rather than a copied chart. Figures are for Dealer Stands on Soft 17, DAS, pays 3:2. Actions not shown are not legal here.

The house rules do not change it

For all the argument this hand attracts, the answer is stable: all six rule sets modelled give the same correct play. What changes is how much it costs you, not what to do.

S17 · 3:2, H17 · 3:2, S17 · Surrender, S17 · No DAS, S17 · No Peek, S17 · 6:5Hit-21.7%

What splitting actually buys

Splitting turns one hand into two, each starting from a single 3 against the dealer’s 8, with a second bet on the table. Across both hands it is worth -22.97% of your original bet, against -21.72% for hit. Each of the two hands is worth -11.48% on its own, so splitting doubles the money at stake on a start that loses, and two of them lose more than hitting the pair loses once.

How often it comes up, and what a mistake costs

In a six-deck game you are dealt this exact spot about once every 2,273 hands. Getting it wrong every single time adds 0.00045 percentage points to the house edge. Of the 31 close calls, that makes it the 6th cheapest to get wrong. The mistakes page puts that against the habits that actually cost players money.

What the count does to this

Less than you would think. Shifting the shoe as far as twelve true counts in either direction never makes split overtake hit here — but that result leans on a convention. Index numbers are quoted for a player total, which leaves the two cards in your own hand sitting in the shoe; for a pair that is a fiction, and counting them out instead does give this hand a crossover. Only two of the 31 spots sit on that fence, and both are pairs, which is the tell.

How crossovers like this are found, and what one card of each rank is actually worth, is on the card counting page.

The closest calls either side of this one

Try next

By Gabe · Updated · How every figure is checked

Cookies for analytics and ads; decline and both stop. Privacy