A counter sits at a 6-deck H17 game with a $25 table minimum and a true count of +4. The dealer's hole card is an Ace, hers is a 10. She has hard 16. The basic-strategy chart on her wallet card says: hit. She stands instead. The dealer flips a 9 — hard 19, dealer wins all the 17s and 18s but loses to her 16-she-stood-on by zero margin. She lost the hand. She would have lost it harder by hitting and drawing a 6 to bust. Across 1,000 hands of hard 16 vs 10 at moderate-to-high counts, the deviation 'stand instead of hit' saves about 0.6% per hand on the bet at risk. Over a counter's career, it's the second-most-valuable single decision she will ever make at the felt.
The Illustrious 18 — named by Don Schlesinger in Blackjack Attack — is the curated short list of the 18 highest-EV deviations from basic strategy based on the true count. There are hundreds of statistically correct deviations across the full BS chart, but most are vanishingly rare or carry tiny EV. The Illustrious 18 covers roughly 95% of the deviation EV a Hi-Lo counter can capture with a fraction of the memorization burden of the full chart. This post walks through why these 18 specifically, the three that matter most, and the full table with thresholds — sourced directly from TableSharp's ILLUSTRIOUS_18 constant.
Why these 18 and not the full list
Card counting opens up two profit channels: bet sizing and play deviations. Bet sizing — putting more money out when the count is positive — is the larger of the two by far, typically 75-80% of a counter's edge. Deviations are the remaining 20-25%, and they don't all carry equal weight.
Schlesinger's analysis ranked every possible deviation by EV per occurrence × frequency of occurrence. The full chart of statistically correct deviations is over 100 entries deep. The top 18 — the Illustrious 18 — capture roughly 95% of the cumulative EV. Adding deviations 19-30 (sometimes called the Fab 4 or Catch 22 depending on the author) picks up another 3%. Deviations 31+ collectively contribute less than 2%.
For a working Hi-Lo counter, the math is clear: memorize 18 deviations, capture 95% of the deviation EV. Memorize 30, capture 98%. Memorize 100, capture 100%. The diminishing returns past 18 are real, and the cognitive load of running 100 indices under casino conditions is what trips up players who try to over-engineer.
The top 3 (and why they carry most of the value)
Three deviations carry roughly 70% of the total Illustrious 18 EV: insurance, hard 16 vs 10, and hard 15 vs 10. If you only have time to drill three deviations before your next session, drill these three.
Rank 1: Insurance at TC ≥ +3
The single most profitable index play in blackjack. Basic strategy: never take insurance. Deviation: at true count +3 or higher, always take insurance.
Why it pays so much: insurance is a yes/no decision, the threshold is one number, and the EV swing per occurrence is large. In a fresh six-deck shoe, the dealer's hole card is a 10-value about 30.87% of the time — at 2:1 payout, that's a -7.4% bet. By TC +3, the shoe is rich enough in 10s that the probability crosses 1/3 (33.33%) and the bet flips to +EV. By TC +6, the EV is +10% or higher per dollar bet. Counters who get nothing else right but nail insurance every time recoup a meaningful chunk of the basic-strategy house edge in counted shoes.
The same rule applies to 'even money' — when you have a blackjack and the dealer shows an Ace, the dealer will offer you 1:1 immediately instead of waiting to see if she also has BJ. Even money is mathematically identical to taking insurance for half your bet on your BJ. Take it at TC ≥ +3; decline below.
Rank 2: Hard 16 vs 10 at TC ≥ 0
Basic strategy says hit hard 16 vs 10 — it's the lesser-evil decision in a hand where both alternatives are bad. Deviation: at true count 0 or higher, stand instead.
The intuition: at neutral or positive count, the shoe is rich enough in 10s that hitting hard 16 busts more often than the dealer's likely outcome warrants. Standing lets you win the dealer-bust hands; hitting wastes the 'free' wins by busting before the dealer plays. The deviation triggers at the very edge of the count range — TC 0 is essentially every hand from neutral on up — which means it fires frequently and consistently. Hard 16 vs 10 is the second-most-valuable single deviation behind insurance.
Worth noting: 8,8 vs 10 is split (basic) at any count; the deviation only applies to non-pair hard 16 compositions like 7+9, 6+10, 5+J.
Rank 3: Hard 15 vs 10 at TC ≥ +4
Basic strategy says hit hard 15 vs 10. Deviation: at true count +4 or higher, stand.
The threshold is higher than for hard 16 because hard 15 is a marginally better hitting hand — you need a 6 or lower to stay alive on the draw, which is more cards than hard 16. At neutral count, hitting wins. By TC +4, the shoe is rich enough in 10s that the hit-bust rate climbs and standing wins more in expectation. The cell triggers less often than hard 16 vs 10 (you have to be in a count strong enough), but the EV per occurrence when it does trigger is substantial.
The full Illustrious 18 table
The full list, ranked by EV contribution, sourced from TableSharp's ILLUSTRIOUS_18 constant. Each row gives the hand, the dealer upcard, the basic-strategy play, the deviation threshold (TC ≥ for stand/double/split deviations, TC ≤ for hit deviations), and the deviation action. The 'at_or_below' direction means the deviation fires below the threshold; at the threshold itself, basic strategy still plays.
- Insurance vs Ace — basic: No insurance — at TC ≥ +3, take insurance
- Hard 16 vs 10 — basic: Hit — at TC ≥ 0, stand
- Hard 15 vs 10 — basic: Hit — at TC ≥ +4, stand
- Pair 10s vs 5 — basic: Stand — at TC ≥ +5, split
- Pair 10s vs 6 — basic: Stand — at TC ≥ +4, split
- Hard 10 vs 10 — basic: Hit — at TC ≥ +4, double
- Hard 12 vs 3 — basic: Hit — at TC ≥ +2, stand
- Hard 12 vs 2 — basic: Hit — at TC ≥ +3, stand
- Hard 11 vs Ace — basic: Hit (H17) / Double (S17) — at TC ≥ +1, double
- Hard 9 vs 2 — basic: Hit — at TC ≥ +1, double
- Hard 10 vs Ace — basic: Hit — at TC ≥ +4, double
- Hard 9 vs 7 — basic: Hit — at TC ≥ +3, double
- Hard 16 vs 9 — basic: Hit — at TC ≥ +5, stand
- Hard 13 vs 2 — basic: Stand — below TC -1, hit
- Hard 12 vs 4 — basic: Stand — below TC 0, hit
- Hard 12 vs 5 — basic: Stand — below TC -2, hit
- Hard 12 vs 6 — basic: Stand — below TC -1, hit
- Hard 13 vs 3 — basic: Stand — below TC -2, hit
Notes on the table:
- Ranks 1-13 are 'at_or_above' deviations — fire at TC = index and higher. Ranks 14-18 are 'at_or_below' — fire strictly below the index, so at TC = index the basic-strategy answer still applies.
- The Pair 10s splits (ranks 4 and 5) are 'never split 10s' deviations — useful for aggressive counters but visible to surveillance. Cover-conscious counters often skip these even at threshold.
- Hard 11 vs Ace varies by ruleset — basic strategy already says double on S17 tables, so the deviation only adds value on H17 tables where basic says hit.
- The negative-count deviations (ranks 14-18) are rarely triggered in practice because TC -2 and lower is usually a Wong-out situation — you've left the table.
The EV math: what each deviation actually buys
Per-occurrence EV varies widely. Insurance at TC +3 saves about $0.05 per dollar bet over not insuring. Hard 16 vs 10 standing instead of hitting saves about $0.02-$0.06 per dollar bet depending on the exact count. The deeper-index deviations (rank 13+) save closer to $0.01-$0.02 per occurrence.
Multiply by frequency. Insurance vs Ace upcard comes up roughly 1 in every 8 hands (Ace is 1 in 13 cards × frequency of being upcard). Of those, the TC is ≥ +3 in maybe 1 in 4 cases at typical penetration. Net: insurance deviation fires roughly 1 in 30 hands and saves about $0.05 per dollar bet on those hands.
Hard 16 vs 10 fires more often — it triggers at TC ≥ 0, which is roughly half of all hands once you're playing through positive-count shoes. The combined EV across all 18 deviations is the ~0.10-0.15% lift on top of the bet-ramp gain.
How to drill the Illustrious 18
Two layers of drill, in order:
- Memorization layer — the 18 hand/threshold pairs as flashcards. Drill until you can call the threshold for any hand in under 2 seconds. The TableSharp counting trainer has a dedicated I18 flashcard mode for this.
- Application layer — running the count, converting to TC, AND recognizing deviation cells as they appear on the felt. This is the harder layer because the cognitive load is real. The trainer's 'Live Pace' drill mode at /train/counting simulates this — random hands at TC values near and across deviation thresholds, with a clock.
A reasonable timeline: 1-2 weeks of pure flashcard memorization, then 4-6 weeks of integrated drill with the count + TC + deviations all simultaneous. Expect to make errors on the rare deviations (ranks 12-18) for months before they become reflexive. Drill the top 5 most often; they account for most of the EV anyway.
Beyond the I18
Once the Illustrious 18 is locked in, the next layer is usually the Fab 4 surrender deviations (specific surrender hands that are correct only at certain TC thresholds). After that, the Catch 22 expands the deviation set further. Each subsequent expansion adds smaller EV: I18 captures ~95%, Fab 4 adds another ~2%, Catch 22 adds another ~1%. Past that point, the EV gain per memorized deviation is essentially noise.
Working counters generally stop at the I18 plus Fab 4 surrenders. The cognitive load of the next layer outweighs the EV gain at any realistic playing volume. The exception is full-time professional counters playing thousands of hours a year — for them, the marginal 1% is meaningful money.
The deviations that matter for cover
Some I18 plays are 'cover-friendly' (look like basic-strategy mistakes a recreational player might make) and others are obvious counter-tells:
- Cover-friendly: hard 16 vs 10 stand (rec players often stand on stiff 16s out of fear), hard 12 vs 3 stand (rec players often misremember 12 as a stand vs 2-6).
- Counter-tells: insurance taken at any count (rec players almost never insure), pair of 10s split (rec players almost never split 10s), doubling hard 10 vs 10 (almost no rec player does this), doubling hard 9 vs 7 (also rare).
Cover-conscious counters sometimes skip the loudest deviations (insurance, pair splits) at threshold and accept the EV loss for the heat reduction. The trade-off is property-dependent — at a casino that aggressively backs off counters, the cover skip is worth it; at a casino with weak counter-detection, taking the full I18 EV is fine.
Bottom line
The Illustrious 18 is the highest-leverage memorization any Hi-Lo counter can do beyond basic strategy. Three deviations (insurance, hard 16 vs 10, hard 15 vs 10) carry most of the value; the remaining 15 add up to the rest of the ~0.10-0.15% EV lift. The deviations don't replace the bet ramp — they layer on top of it — but they're the difference between a Hi-Lo player and a Hi-Lo counter who actually wins at the rate the math predicts.