Strategy Deviations — True Count Index Plays

Basic strategy answers: What is the best play if I know the rules but do not know anything useful about the undealt card composition?

A strategy deviation answers a different question: Has the remaining shoe changed enough that another play now has the higher expected value?

That second question belongs to card counting. A deviation is not intuition, a “dealer tell,” a streak response, or an excuse to improvise. It is a precomputed index decision tied to a count system and a true count.

For non-counting play, stop at Blackjack Basic Strategy. This page is for readers who already understand the Hi-Lo system and true-count conversion. For off-table checking of the conversion step, use the True Count Calculator.

Three layers must be correct before an index matters

A valid deviation requires three separate skills:

1. Running count: cards must be tagged and accumulated accurately.
2. Deck estimation: the player must estimate how many decks remain undealt.
3. True count: the running count must be normalized for the remaining shoe.

For Hi-Lo:

[ TC = \frac{RC}{D} ]

where:

• (TC) = true count;
• (RC) = running count;
• (D) = estimated decks remaining.

If the running count is +8 with about two decks remaining:

[ TC = +8/2 = +4 ]

If the same running count occurs with four decks remaining:

[ TC = +8/4 = +2 ]

The running count is identical, but the composition signal is not. That is why shoe-game index play based on running count alone is structurally wrong.

What an index number means

An index is a threshold where the expected values of two actions cross.

Suppose the basic-strategy action for a hand is Hit. A count system may assign an index of +2 for standing. The interpretation is not “+2 guarantees a win.” It is:

• below the chosen +2 threshold, Hit has the higher expected value;
• at or above the threshold, Stand has the higher expected value under the index system’s assumptions.

The two actions are often close near the crossing point. That is why small errors in true-count conversion can erase the value of a deviation.

Index numbers also depend on conventions. Some systems floor, truncate, or round the true count differently. Some index sets assume S17, others H17; some include surrender; some are optimized for different deck counts.

Memorizing a number without memorizing the conditions attached to it is not advanced strategy.

High-value Hi-Lo examples

A commonly cited multi-deck Hi-Lo set, the “Illustrious 18,” prioritizes a small group of deviations that capture much of the practical value of index play.

Selected examples include:

These are examples from a specific Hi-Lo index framework, not universal blackjack commandments. Rule assumptions and index methodology matter.

The value comes from changing only the close decisions that the composition signal has actually moved across a threshold.

Insurance is different from most playing deviations

Insurance is often the highest-value index deviation because its payoff depends directly on the density of ten-value cards behind the dealer’s Ace.

The bet pays 2:1 and breaks even at a one-third ten-value probability. A sufficiently high count is evidence that the undealt cards are rich in tens, which can move insurance from negative to positive expected value.

That is a cleaner relationship than many hit/stand deviations, which depend on how the altered composition changes several possible dealer and player outcomes at once.

The When to Take Insurance page derives the 33.33% threshold. Do not treat the +3 Hi-Lo index as a shortcut around that reasoning.

Stiff-hand deviations are close decisions, not “safe stands”

Hard 16 against a dealer 10 is famous because both Hit and Stand are unattractive. Basic strategy selects the action with the smaller expected loss under a neutral composition.

When the count rises, the remaining shoe is richer in high cards. That changes both the danger of drawing to a stiff hand and the dealer’s outcome distribution. At the relevant index, the balance between Hit and Stand reverses.

The important word is reverses, not “becomes good.” Standing on 16 against 10 at the correct index can still have negative expected value and can still lose immediately. The deviation is valuable because it loses less on average than the alternative at that composition.

This distinction is essential for interpreting all index play.

Double-down deviations change both decision and exposure

A high count can make a borderline double more attractive because high cards improve the value of strong player starting totals. But doubling also increases the amount at risk on the hand.

A player who has learned an index but cannot comfortably fund the extra wager has a bankroll problem, not a strategy problem. The correct response is a smaller base unit, not refusing profitable doubles because the session bankroll is too thin.

The same principle applies to split deviations. Strategy and bankroll sizing have to agree with each other.

Split deviations can be mathematically correct and operationally conspicuous

Some high-count deviations involve splitting tens. Ordinary players almost never split a strong 20, so the play is visually unusual.

That does not make the deviation mathematically wrong. It does make it easy for experienced staff to notice when it coincides with larger bets and a strong count.

This page will not provide camouflage tactics. From an operations perspective, unusual high-value decisions are simply part of the observable evidence a casino may review when assessing advantage play.

The distinction matters: a strategy deviation is a mathematical decision; how a casino responds to advantage play is an operational and jurisdictional matter.

Count accuracy is worth more than a longer index list

Suppose a player has memorized 50 index numbers but estimates 1.5 decks remaining when there are really 2.0. With a running count of +6:

Estimated:

[ TC = 6/1.5 = +4 ]

Actual using 2 decks:

[ TC = 6/2 = +3 ]

That one deck-estimation error can move the player across an index threshold and reverse a decision.

Now compare that with a player who knows only a handful of high-value indices but keeps an accurate running count and reliable deck estimate. The shorter list may produce better decisions because the inputs are trustworthy.

A sensible learning order is therefore:

• perfect basic strategy;
• counting accuracy;
• deck estimation;
• true-count conversion;
• a small high-value index set;
• broader index coverage only after the first five are automatic.

Bet ramp and playing deviations are separate systems

Card counting has at least two distinct uses:

• betting correlation: changing the amount wagered when the estimated player advantage changes;
• playing efficiency: changing a hit/stand/double/split/insurance decision when an index is crossed.

A player can understand one and make mistakes in the other. A large betting spread does not make bad index decisions correct, and perfect index play does not compensate for a bet schedule that is too aggressive for the bankroll.

The Blackjack Risk of Ruin page explains why variable betting requires its own bankroll model.

Negative indices matter too

Not every deviation waits for a strongly positive count.

Some basic-strategy stands become hits when the count is sufficiently negative because a low-card-rich remaining shoe changes the balance. Selected Hi-Lo examples from the same well-known index set include:

• 13 vs 2: Stand at true count -1 or higher; otherwise Hit;
• 12 vs 4: Stand at 0 or higher; otherwise Hit;
• 12 vs 5: Stand at -2 or higher; otherwise Hit;
• 12 vs 6: Stand at -1 or higher; otherwise Hit;
• 13 vs 3: Stand at -2 or higher; otherwise Hit.

These examples are useful because they correct a common misconception that “higher count means use deviations, lower count means use basic strategy.” Index play works in both directions around the threshold.

Again, the exact numbers belong to a particular Hi-Lo framework. They should not be transferred automatically to another count system.

Surrender has its own compact deviation set

When late surrender is available, count information can move some borderline hands across the -0.5-unit surrender threshold.

A commonly cited “Fab 4” group for Hi-Lo focuses on surrender decisions such as hard 14 or 15 against strong dealer upcards. The logic is the same as every other index: surrender when the true count reaches the threshold where surrender has the higher expected value than the best continuation.

This is another reason the table rules must be known first. A surrender index has no use if the table does not offer the relevant surrender rule.

Read Blackjack Surrender before adding surrender indices; otherwise it is easy to memorize a count number without understanding what the underlying -0.5 comparison means.

True-count rounding can reverse borderline plays

Different counting references use different conventions:

• floor toward negative infinity;
• truncate toward zero;
• round to nearest integer;
• half-deck or quarter-deck estimation with correspondingly finer true counts.

Suppose an index is +2 and the calculated true count is +1.7.

• nearest-integer rounding may call that +2;
• truncation calls it +1.

Those methods can produce different decisions at the same physical shoe. The solution is not to argue about which integer “looks right.” Use the rounding convention that belongs to the index set being studied.

This is why copying isolated index numbers from forum posts is risky. The number may be correct under a convention that the reader is not using.

Composition-dependent basic strategy is not the same thing as card-count deviation

A single-deck player can sometimes improve on total-dependent basic strategy by considering the exact ranks that make up the hand. For example, two different three-card 16s against a dealer 10 may have different optimal actions because the specific removed cards slightly change the remaining deck.

That is composition-dependent strategy, not necessarily card counting.

A card-count index compresses many observed cards into a count statistic and uses a threshold. Composition-dependent strategy can make a decision from the exact cards in a small-deck situation even when no running count has been maintained.

The concepts overlap because both exploit card removal, but they should not be taught as synonyms.

Index value is concentrated, not evenly distributed

The practical reason people learn a short index set first is that deviations do not contribute equal amounts of value.

Insurance has a direct relationship with ten-value density. Hard 16 vs 10 occurs frequently enough and sits close enough to the decision boundary to matter. An obscure index on a rare hand may add only a tiny fraction of the value.

That means study time has diminishing returns. Going from zero deviations to a few high-value ones can matter much more than going from 50 memorized indices to 100.

For a recreational reader, this also sets a boundary: if basic strategy itself is not automatic, studying low-value deviations is solving the wrong problem.

An index does not tell you how much to bet

A play index says which action has higher expected value for the hand. It does not tell you the optimal wager size.

Bet sizing depends on estimated advantage, bankroll, variance, risk tolerance, table limits, and the player’s chosen risk-of-ruin framework. A +4 decision index is therefore not a command to bet four units, four times the minimum, or any other amount.

Confusing play indices with bet sizes is a category error.

Deviation gains can be erased by rule selection

A counter can execute index plays perfectly and still choose a poor game.

Examples:

• 6:5 blackjack removes substantial value from naturals;
• poor penetration reduces the time spent at useful counts;
• no DAS or restrictive doubles reduce player options;
• unfavorable no-hole-card settlement can expose added wagers.

Advanced decision accuracy should come after game selection. A small gain from index play cannot be assumed to compensate for a much worse base rule package.

The House Edge by Rules page is therefore relevant even to readers focused on counting.

Operational review looks for patterns, not one strange hand

A single unusual play does not prove anything. Recreational players split tens, stand on 15, take insurance, and change bets for many reasons.

What becomes informative operationally is a repeated relationship among:

• wager size;
• count-favorable moments;
• index-like decisions;
• shoe depth;
• multiple sessions.

This is why serious game protection relies on pattern review rather than superstition about one “counter move.” It is also why this article avoids turning casino response into a cat-and-mouse manual. The mathematical topic is the deviation itself.

The Fab 4 illustrates why surrender deviations need a rule first

A compact Hi-Lo surrender set often taught alongside the Illustrious 18 uses four threshold plays. One common version includes:

The interpretation is the same: surrender at or above the specified true count when the applicable surrender rule is available; otherwise use the normal continuation action.

These indices are not useful at a table with no surrender. They also should not be mixed casually with a different counting system or rule set.

The broader lesson is that advanced strategy has dependencies. An index is the last step in a chain, not a self-contained fact.

Why a short list captures so much value

High-value index sets are intentionally selective. They concentrate on hands that meet some combination of three properties:

• the hand occurs often enough to matter;
• the basic-strategy alternatives are relatively close in expected value;
• the count strongly changes the relevant card probabilities.

A source table may contain dozens or hundreds of possible indices, but their marginal value declines. The well-known Illustrious 18 was designed around this concentration of value rather than completeness.

This is a useful study principle: learn decisions in order of expected contribution, not in numerical or alphabetical order.

Errors near zero matter more than they look

Indices near zero are encountered more frequently than extreme positive counts because the true count spends a great deal of time near neutral.

That makes plays such as 16 vs 10 at index 0 operationally significant. A player who inconsistently treats +0.3, -0.2, and exactly 0 because of rounding confusion can give back value repeatedly.

Before memorizing more indices, define exactly how your count system treats fractional true counts. Consistency matters more than pretending the estimate has perfect precision.

Index generation is an expected-value crossing problem

Conceptually, an index can be generated by calculating the expected value of two actions across different deck compositions and identifying where their ranking changes.

For a hit-versus-stand decision, define:

[ \Delta EV(TC) = EV_{stand}(TC)-EV_{hit}(TC) ]

The index is near the true count where:

[ \Delta EV(TC)=0 ]

Below that point, the sign favors one action; above it, the sign favors the other.

Real index-generation software models the deck composition, rules, rounding and simulation method in much more detail. The equation is useful because it explains what the index means: a crossover in expected value.

Deviations are not predictions about the next card

A true count of +4 does not mean the next card will be a ten or Ace. It means the remaining shoe is estimated to contain a higher concentration of high cards than a neutral shoe.

An individual next card can still be low. The count changes probabilities across the remaining population; it does not identify card order.

This is why a correct deviation should be judged by whether the input and index were correct—not by whether the hand won.

A practical audit after a session

For training purposes, an advanced player can record a small set of decisions without recording every hand:

• hand and dealer upcard;
• running count before the decision;
• decks remaining estimate;
• calculated true count;
• index used;
• action taken.

Reviewing those five fields can reveal whether mistakes come from counting, division, memory, or action execution.

Do not use the log to search for “lucky indices.” Short-term wins and losses are noisy. The purpose is process quality, not outcome superstition.

The published Hi-Lo reference

The Wizard of Odds High-Low reference lists the Illustrious 18 and gives example index numbers, including Insurance +3, 16 vs 10 at 0, and 15 vs 10 at +4. It also makes clear that index decisions use the true count.

Use one consistent index source. Mixing numbers from different rule sets, counting systems, or rounding conventions is an easy way to create contradictions that look like “advanced play” but are only bookkeeping errors.

How to practice deviations without turning the session into a memory test

A useful drill has four columns:

The point is not the table itself. The point is to practice the threshold concept until the basic-strategy action and deviation action do not blur together.

If a player has to reconstruct the running count, divide by decks remaining, remember the index, and then recall basic strategy from scratch while the dealer waits, too many skills are still competing for attention.

What deviations can and cannot do

Deviations can improve expected value relative to basic strategy when the count contains real information.

They cannot:

• predict the next card;
• prevent losing streaks;
• guarantee a winning session;
• repair an inaccurate count;
• make a weak bankroll adequate;
• turn a 6:5 game into a desirable counting game merely through clever play;
• justify chasing after a losing shoe.

The gain from an index is statistical. A correct deviation can lose, and a wrong play can win. Outcome feedback from one hand is therefore a poor teacher.

Responsible play and practical limits

Card counting and index play can make blackjack feel more controllable than it is. Even a player with a genuine positive expectation can experience large drawdowns because variance remains.

Do not use advanced strategy as a reason to borrow, raise stakes beyond a planned bankroll, or continue after the session limit. A mathematically interesting edge does not make the money immune to short-term loss.

Author / Editorial Note

This page is written from a land-based casino operations perspective. The goal is not to sell card counting as easy money. The goal is to explain why real deviations require count accuracy, rule awareness, bankroll discipline, and the emotional control to make small edges matter.

Bottom line

An advanced blackjack deviation is a true-count threshold decision. Basic strategy is the baseline, the count supplies new information, and the index identifies where another action becomes better.

If the running count, deck estimate, or true-count convention is unreliable, the deviation is unreliable too. Master fewer high-value indices with clean inputs before adding more.