Three-Card 16: When the Chart Bends
Key Takeaway: Basic strategy assumes your total is all that matters. It isn't. A 16 built from three or more small cards has already eaten the low cards that could have rescued a hit, which leaves the deck richer in tens. Against a dealer 10, that hand should stand. In single deck, the hit-bust rate climbs from about 59% on a two-card 16 to 62.5% or more on a multi-card 16, and past 64% when the hand is made entirely of sub-6 cards. That flips the decision.
Every player who has put in serious hours has taken a 4, then a 5, then a 7 against a dealer ten, looked at a 16, and reached for the felt to hit because the chart says so. The chart is wrong for that hand. It isn't wrong in general. It is wrong because it was built on an assumption that your specific hand violates. This guide covers exactly where that assumption breaks, how big the error is, and how to fix it at the table without slowing your play or drawing attention.
Total-Dependent vs. Composition-Dependent Strategy
The basic strategy chart you memorized from our learn basic strategy guide is a total-dependent strategy. It maps exactly two inputs, your hand total and the dealer upcard, to one decision. A 10-6 and a 4-5-7 both read as "hard 16 vs 10," so they get the same instruction.
A composition-dependent (CD) strategy adds a third input: the specific cards that make up your hand. Those cards are no longer in the deck. In a finite deck, every card you hold changes the probability of every card you might draw next. It also changes the dealer's hole card and draw sequence. The total-dependent chart averages over all possible compositions of a given total, weighted by how often each one occurs. Since two-card hands vastly outnumber multi-card hands, the chart's 16-vs-10 entry is really optimized for 10-6 and 9-7. Your three-card 16 is a rare composition being forced into a decision tuned for the common one.
The "effect of removal" is the change in expected value for a specific decision when one card of a given rank is taken out of the deck. For a stiff-hand hit/stand decision against a high card, removing low cards (2 through 5) favors standing. Removing tens favors hitting. Mid cards (7, 8, 9) move the needle less. A multi-card stiff is, by definition, a hand that has removed several low cards at once. The removal effects add together, and on a close decision their combined weight can exceed the original margin between hitting and standing.
The Mechanics of Dynamic Card Removal
Start from the infinite-deck baseline, where removal doesn't exist. With a hard 16, any card from 6 through 10 busts you. That is 8 of 13 ranks (6, 7, 8, 9, plus four ten-value ranks), so the hit-bust probability is 8/13 = 61.5%. The five escape cards (A, 2, 3, 4, 5) take you to 17 through 21.
Now make the deck finite and track what your hand has removed. The tool is simple counting, and you should be able to do it yourself:
P(bust on hit) = 8/13 = 0.6154
Approximate EV of standing on final total T vs dealer 10:
EV(17) ≈ -0.42 EV(18) ≈ -0.18 EV(19) ≈ +0.07
EV(20) ≈ +0.55 EV(21) ≈ +0.96
EV(hit 16) = -P(bust) + Σ P(c) · EV(16 + c), c ∈ {A,2,3,4,5}
= -0.6154 + (1/13)(-0.42 - 0.18 + 0.07 + 0.55 + 0.96)
= -0.6154 + 0.0755 ≈ -0.540
EV(stand 16) = 2 · P(dealer busts) - 1 ≈ 2(0.23) - 1 ≈ -0.540
--> Infinite deck: a dead heat to within a fraction of a cent.
SINGLE DECK, first-order approximation (stand EVs held constant)
10-6 vs 10 : 49 unseen | bust cards = 29 | escapes = 4A,4x2,4x3,4x4,4x5
EV(hit) ≈ -29/49 + (4/49)(0.98) ≈ -0.592 + 0.080 = -0.512
4-5-7 vs 10 : 48 unseen | bust cards = 30 | escapes = 4A,4x2,4x3,3x4,3x5
EV(hit) ≈ -30/48 + [4(-0.42)+4(-0.18)+4(0.07)+3(0.55)+3(0.96)]/48
≈ -0.625 + 0.050 = -0.575
Stand EV ≈ -0.54 in both cases (it gets slightly worse for 10-6 and slightly
better for 4-5-7 once dealer draws are recomputed, which widens both gaps).
--> 10-6: HIT by ~3 cents. 4-5-7: STAND by ~3 cents.
A note on rigor: the block above holds the stand-on-total EVs at their infinite-deck values to isolate the player-draw effect. A full CD calculation, the kind you'll find in the composition-dependent strategy tables on BlackjackMath.org, also recomputes the dealer's hole-card and drawing probabilities against the depleted deck. That refinement pushes in the same direction. A ten-rich deck makes a dealer stiff (12 through 16) more likely to bust, which raises the value of standing. The first-order math gives you the correct sign and a reasonable size for the effect.
Case Study: The Multi-Card 16 vs. Dealer 10
The Infinite-Deck Knife Edge
The most important fact about 16 vs 10 is that it is the closest hit/stand decision in the entire game. Both options lose roughly 54 cents on the dollar. In an eight-deck shoe the hit is correct by a hair. When a decision's margin is measured in fractions of a cent, small perturbations can flip it, and in a finite deck a three-card hand is not a small perturbation. That is why this hand is the textbook CD example, and why our stiff hands guide flags it as the one stiff decision where you should be thinking about more than the total.
What a Small-Card 16 Does to the Deck
Look at the single-deck numbers. Each figure below counts the unseen cards after removing the player's hand and the dealer's 10 upcard. The Hi-Lo column is the running count of those visible cards only, so it shows how a counter's tag sum tracks the composition shift.
| Player Hand (vs 10) | Unseen Cards | Bust Cards Remaining | P(Bust on Hit) | Hi-Lo of Visible Cards | CD Play |
|---|---|---|---|---|---|
| Infinite-deck baseline | ∞ | 8/13 ranks | 61.5% | n/a | Hit (razor thin) |
| 10-6 | 49 | 29 | 59.2% | −1 | Hit |
| 9-7 | 49 | 29 | 59.2% | −1 | Hit |
| 3-5-8 | 48 | 30 | 62.5% | +1 | Stand |
| 4-5-7 | 48 | 30 | 62.5% | +1 | Stand |
| 2-4-4-6 | 47 | 30 | 63.8% | +3 | Stand |
| 4-4-4-4 | 47 | 31 | 66.0% | +3 | Stand |
| 2-2-3-4-5 | 46 | 31 | 67.4% | +4 | Stand |
Notice what drives the gap. The two-card 16 is not merely "less bad" than the multi-card version. A 10-6 has removed two tens from the deck (yours and the dealer's), which drops the bust rate below the infinite-deck baseline. That is why the hit is clearly correct in single deck. The multi-card 16 has removed only one ten, the dealer's, while stripping out escape cards, so the bust rate lands above the baseline. The two hands share a total, but their decks are a full three to eight percentage points apart.
The Escape-Card Problem
Bust rate is only half of the story. The other half is the quality of the non-bust outcomes. When you hit 16 vs 10 and survive, drawing an ace gives you a 17, which still loses about 42 cents. A 2 gives you an 18 that still loses money. Only the 4 and the 5 (20 and 21) produce strongly positive results. A 4-5-7 has consumed one of each of the two best rescue cards. That is why the escape-sum term in the math block drops from 0.080 to 0.050. The hand didn't just remove escape cards. It removed the most valuable ones. Composition effects are not uniform across ranks, and any shortcut that treats them as uniform will miss cases like this.
Deck Count Determines How Hard the Chart Bends
Removal effects scale inversely with deck size. One card out of 52 is a 1.9% shift in the deck. One card out of 312 is 0.3%. Run the same two hands through different deck counts and the effect shrinks fast:
| Decks | 10-6 vs 10: P(Bust) | 4-5-7 vs 10: P(Bust) | Spread | Approx. Total Value of Full CD Strategy |
|---|---|---|---|---|
| 1 | 29/49 = 59.2% | 30/48 = 62.5% | 3.3 pts | ~0.04% of action |
| 2 | 61/101 = 60.4% | 62/100 = 62.0% | 1.6 pts | ~0.02% |
| 6 | 189/309 = 61.2% | 190/308 = 61.7% | 0.5 pts | ~0.01% or less |
| ∞ | 61.5% | 61.5% | 0 | 0 |
Two conclusions follow. First, in single and double deck, CD plays are real money. They aren't enormous, but they come free to anyone who learns a handful of rules, and in a game where your total edge may be fractions of a percent, they count. Second, the multi-card 16 vs 10 stand survives even in a six- or eight-deck shoe. The two-card margin there is so thin that a half-point move in bust rate, plus the matching dealer-side shift, is still enough to tip it. Most other CD exceptions disappear in shoes. Our deck count guide covers how to tell, quickly and discreetly, which game you are actually sitting in.
The 12 vs. 4 Composition Pivot
The second classic CD case works in the opposite direction. Basic strategy stands on 12 vs 4. In single deck, however, 10-2 vs 4 is a hit, while 9-3, 8-4, 7-5, and the rest stand. On a 12, only ten-value cards bust you. Here is the single-deck arithmetic after removing your two cards and the dealer's 4:
| Hand vs 4 | Tens Remaining / Unseen | P(Bust on Hit) | Hi-Lo of Visible Cards | Single-Deck CD Play |
|---|---|---|---|---|
| 10-2 | 15/49 | 30.6% | +1 | Hit |
| 9-3 | 16/49 | 32.7% | +2 | Stand |
| 8-4 | 16/49 | 32.7% | +2 | Stand |
| 7-5 | 16/49 | 32.7% | +2 | Stand |
| 6-6 | 16/49 | 32.7% | +3 | Stand |
The 10-2 is the only 12 that removes a ten. That helps the hit in two ways. You have fewer bust cards to draw, and the dealer's 4 now has fewer tens to bust with when you stand, which lowers the value of standing. Both effects point toward hitting. The 7-5 removes a 5, a dealer-helping card, which nudges standing slightly in the other direction. In double deck the 10-2 hit is roughly a coin flip, and in shoes, standing is correct for every 12 vs 4.
How Card Counters Absorb Composition Shifts
If you count, you have been doing a rough version of CD strategy all along, whether you realized it or not. Under Hi-Lo, the index for 16 vs 10 is 0. You stand at any positive true count and hit below it. A disciplined counter includes their own cards and the dealer's upcard in the running count before making the play. Run the 16-vs-10 table again: 10-6 tags to −1 (hit) and 4-5-7 tags to +1 (stand). Off the top of a fresh single deck, the count reproduces the CD decision exactly. In a shoe, the count from earlier rounds dominates, and the hand composition becomes one more input among many, which is the correct mathematical weighting.
The count isn't perfect, though, and the 12 vs 4 case shows where it falls short. Hi-Lo treats a 2 and a 10 as equal and opposite, so 10-2 vs 4 tags to +1. With the Illustrious 18 index of 0 for 12 vs 4, a counter would stand. CD analysis says hit, because for this particular decision the removed ten carries far more weight than the removed 2. A level-one balanced count compresses thirteen ranks into three values. When a decision depends heavily on one rank, that compression loses information. That is the gap CD strategy fills.
Sharp Tip (Marcus Vance): In single deck, update your running count with your own hand and the dealer upcard before you act on any stiff. On 16 vs 10, if the count including your cards is zero or positive, stand. That rule alone captures the multi-card 16 automatically. Then layer exactly one override: 10-2 vs 4 off the top is a hit regardless of the +1 tag. Don't build a mental library of 40 exotic compositions. Memorize the two or three that carry real money and let the count handle the rest. Hesitating at the table costs more in heat than an obscure CD play will ever earn you.
Practical Implementation: CD Rules Worth Memorizing
For a non-counter, or a counter who wants a clean baseline, these are the composition overrides worth learning:
- Hard 16 vs 10, three or more cards: stand. This applies in every game from single deck to eight decks, with the margin largest in single deck. A few knife-edge compositions exist, but stand is the right default.
- Two-card 16 vs 10: surrender if allowed, otherwise hit. Surrender is only available on your first two cards, so it never applies to the multi-card version. Your choice there is strictly hit or stand.
- 8-8 vs 10: split. It is a pair, not a 16. Don't let the CD logic contaminate a pair decision.
- Single deck, 10-2 vs 4: hit. All other 12s vs 4 stand. In shoes, ignore this rule.
- Check the deck count first. Single and double deck justify a fuller CD rule set. In shoes, use only rule 1.
If you want to see these exceptions laid out for your exact rule set (deck count, H17/S17, surrender availability), generate a custom chart in our Strategy Chart Builder. Then drill the overrides until they're as automatic as doubling 11.
Common Errors at the Table
- Over-applying the rule. "Multi-card" isn't the real trigger. "Built from small cards" is. A three-card 16 still removes low cards by construction, which is why the rule works. But don't extend the logic to 15 vs 10, where basic strategy's hit has more cushion and the CD stand doesn't generally hold without a count.
- Ignoring the other cards on the table. In single and double deck, the other players' exposed cards are removed too. If three neighbors just showed a pile of tens and you're holding 4-5-7, the deck is no longer ten-rich. That is precisely why counting is the full solution and CD rules are only the static approximation.
- Trusting the chart over the arithmetic. The chart is an average. When your hand is an outlier composition, the average doesn't describe your deck.
The real lesson of the three-card 16 goes beyond one exception. Blackjack strategy isn't a lookup table. It is a conditional probability problem that the lookup table approximates. Once you understand why the chart says hit on 10-6 and why that reasoning fails on 4-5-7, you have stopped memorizing and started computing. Every serious advantage technique builds on that habit.
