Deviations: when to ignore basic strategy

Basic strategy assumes you know nothing about the cards remaining in the shoe. When you're counting, that assumption is no longer true.

A card counter running the Hi-Lo system (or any balanced counting system) maintains a running count that reflects whether the remaining shoe is ten-rich or ten-poor relative to neutral. When the shoe is ten-rich, some playing decisions that are wrong under basic strategy become correct. When the shoe is ten-poor, the reverse applies. These adjusted decisions are called deviations or index plays, and they contribute meaningfully to the counter's overall edge, particularly for the highest-priority plays. The discipline of identifying which deviations are worth the extra mental effort was formalised by Don Schlesinger in Blackjack Attack, where he introduced what became known as the Illustrious 18.

The Illustrious 18: Priority Index Plays

The Illustrious 18 are ranked by their contribution to player edge. Schlesinger's analysis at Blackjack Forum Online shows that these eighteen plays account for roughly 80% of the total gain from all deviations in a standard six-deck game. Learning all 200-odd possible index plays is not a productive use of effort; learning the top eighteen is.

Insurance is the single most valuable deviation, and it's worth understanding precisely why. In a six-deck shoe, there are 96 ten-value cards in 312 total cards (approximately 30.8% ten-density). The insurance bet pays 2:1, which is correct only if the ten-density is above 33.3%. A Hi-Lo true count of +3 corresponds approximately to that ten-density threshold, which is why the insurance index is +3: take insurance at TC +3 or higher, decline it at lower counts. Over the long run, correctly timed insurance bets contribute more to a counter's edge than almost any other single play change.

The next tier of the Illustrious 18 covers the standing and hitting decisions that shift most dramatically with count. The most frequently encountered are:

Hard 16 vs 10: basic strategy says hit, but at a true count of 0 or higher, standing becomes correct. This is one of the most common deviation opportunities because hard 16 vs dealer 10 is a hand you'll face regularly, and the boundary between hit and stand is very close to neutral count.

Hard 15 vs 10: stand at TC +4 or higher. Below that, hit.

Hard 13 vs 2: basic strategy says stand; at TC -1 or lower (a ten-poor shoe), hit becomes correct.

Pair of 10s vs 5 and 6: basic strategy says never split tens, and the baseline rule remains correct in most situations. But at TC +5 vs a dealer 5, and TC +4 vs a dealer 6, splitting becomes the higher-expectation play. This deviation is controversial because splitting tens is one of the most conspicuous plays a counter can make; it immediately draws pit attention. The gain is real, but the cost in scrutiny can exceed the value in a venue where you intend to play regularly.

Hard 12 vs 3: basic strategy says hit; stand at TC +2 or higher. Hard 12 vs 2: hit under basic strategy; stand at TC +3 or higher.

Negative Count Deviations and the Fab 4

Schlesinger extended the analysis to what he called the Fab 4: four surrender deviations that together account for the bulk of the gain from all surrender-related index plays. These apply when the count drops, meaning the shoe is ten-poor and the dealer is less likely to bust.

Surrender hard 14 vs 10 at TC 0 or higher (basic strategy says hit). Surrender hard 15 vs 9 at TC +2 or higher. Surrender hard 15 vs ace at TC +1 or higher. Surrender hard 16 vs ace at TC -1 or higher. These plays matter most in a game offering late surrender, and they add roughly 0.10% to a counter's edge when executed correctly. Surrender is available at a minority of UK venues, so their practical value varies. Check before you commit them to memory as a priority.

The broader principle of negative count deviations: when the shoe is ten-poor (negative true count), the dealer is less likely to bust on a stiff hand, which means your standing decisions on hard 12-16 become marginally less correct. A deeply negative count argues for hitting slightly more aggressively in some configurations. The gain from negative count deviations is smaller than from positive count deviations because counters typically reduce their bet size in negative counts, which scales down the value of any playing gain proportionally.

When to Abandon Basic Strategy: The Practical Threshold

The honest answer is: not often, and not dramatically. Deviations are a refinement on a strong foundation, not a replacement for it. A player who makes the correct basic strategy play on every hand but gets the index plays wrong will perform significantly better than a player who makes index plays from memory but whose basic strategy is shaky.

At the Hippodrome's main blackjack floor, you're counting in a six-deck shoe with roughly 75% penetration (about 4.5 decks dealt before the cut card). The true count climbs into actionable deviation territory less often than in a two-deck game. The insurance play at TC +3 will present itself several times per shoe. The hard 16 vs 10 at neutral count presents itself frequently. The splitting tens plays present themselves rarely and conspicuously. Prioritise accordingly.

The card counter trainer includes a deviations mode that deals you hands at the specific true count where index plays become relevant. That targeted practice is more efficient than drilling the full deviation table in isolation, because it conditions you to link the count threshold with the play change in a realistic dealing context.

One practical note on the UK game: some venues use continuous shuffle machines (CSMs), which eliminate meaningful penetration and make counting and deviations entirely ineffective. If the shoe never accumulates more than about 20 cards before being reshuffled, the count carries no information. Identify CSMs before you sit. The card counting lesson covers the penetration requirements for a viable count in more detail.

Key numbers

Sources: Schlesinger, Illustrious 18 at Blackjack Forum Online, our calculation, our calculation.