PHASE 3.02 ? TEN-VALUE SPLIT PRECEDENCE AUDIT

Stored-card rule: 10, J, Q and K are all represented as integer value 10.
Maximum Aggression precedence: SPLIT eligible initial pair first; only otherwise consider first-decision DOUBLE.

10+J  -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
10+Q  -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
10+K  -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
J+Q   -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
J+K   -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
Q+K   -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
10+10 -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
J+J   -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
Q+Q   -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT
K+K   -> stored 10+10 -> split eligible = true -> first aggressive action = SPLIT

RESULT: PASS ? every two-card ten-value combination is routed to SPLIT before DOUBLE.

PHASE 3.02 BASELINE REPLICATION

PRIMARY
Average ending bankroll T/A: 99.956 / 78.978
Positive sessions T/A: 48.14% / 2.48%
Paired winner T/A/tie: 95.72% / 3.88% / 0.40%
Doubles T/A: 15716 / 154310; splits T/A: 3808 / 21838
Mean cards consumed T/A: 165.674 / 177.840

HOLDOUT
Average ending bankroll T/A: 99.972 / 78.895
Positive sessions T/A: 47.42% / 2.42%
Paired winner T/A/tie: 95.86% / 3.86% / 0.28%
Doubles T/A: 15412 / 154170; splits T/A: 3747 / 21789
Mean cards consumed T/A: 165.647 / 177.749


PHASE 3.02 ? BASELINE INSTRUMENTATION SUMMARY

PRIMARY resolved aggressive hands: 171838; mean net/hand -0.6117; mean committed stake 1.8980; bust rate 35.76%; third-card bust rate when a third card was taken 39.83%.
Textbook would have chosen on those same visible first-decision states: HIT 70163, STAND 68221, DOUBLE 15926.

HOLDOUT resolved aggressive hands: 171789; mean net/hand -0.6143; mean committed stake 1.8974; bust rate 35.72%; third-card bust rate when a third card was taken 39.80%.
Textbook would have chosen on those same visible first-decision states: HIT 70374, STAND 68038, DOUBLE 15758.

INTERPRETATION
Phase 3.02 does not optimise the player. It first locks and verifies ten-value split precedence so the optimisation programme cannot misclassify K+J, Q+10 or any other two-card ten-value combination as a non-pair DOUBLE state. State evidence remains available for the next optimisation phase.

PRIMARY ? WORST MAXIMUM-AGGRESSION VISIBLE STATES BY MEAN NET/HAND
Rank | Visible state | Hands | Mean net | Loss% | Bust% | 3rd-card bust% of 3rd-card draws | Mean stake
 1 | HARD 5 vs 11 [split-child] | textbook=HIT     |     11 |  -2.000 | 100.00% |   0.00% |   0.00% | 2.000
 2 | HARD 20 vs 5 [split-child] | textbook=STAND   |    672 |  -1.750 |  93.30% |  93.30% |  93.30% | 2.000
 3 | HARD 20 vs 2 [split-child] | textbook=STAND   |    642 |  -1.748 |  93.30% |  93.30% |  93.30% | 2.000
 4 | HARD 20 vs 6 [split-child] | textbook=STAND   |    660 |  -1.742 |  93.18% |  93.18% |  93.18% | 2.000
 5 | HARD 20 vs 4 [split-child] | textbook=STAND   |    643 |  -1.742 |  93.31% |  93.31% |  93.31% | 2.000
 6 | HARD 20 vs 7 [split-child] | textbook=STAND   |    664 |  -1.723 |  92.77% |  92.77% |  92.77% | 2.000
 7 | HARD 20 vs 9 [split-child] | textbook=STAND   |    659 |  -1.718 |  92.87% |  92.87% |  92.87% | 2.000
 8 | HARD 20 vs 3 [split-child] | textbook=STAND   |    704 |  -1.702 |  92.19% |  92.19% |  92.19% | 2.000
 9 | HARD 20 vs 11 [split-child] | textbook=STAND  |    446 |  -1.691 |  92.15% |  92.15% |  92.15% | 2.000
10 | HARD 20 vs 10 [split-child] | textbook=STAND  |   2385 |  -1.652 |  91.11% |  91.11% |  91.11% | 2.000
11 | HARD 20 vs 8 [split-child] | textbook=STAND   |    681 |  -1.648 |  91.19% |  91.19% |  91.19% | 2.000
12 | HARD 19 vs 4 [split-child] | textbook=STAND   |    186 |  -1.624 |  89.78% |  89.78% |  89.78% | 2.000
13 | HARD 19 vs 6 [split-child] | textbook=STAND   |    222 |  -1.559 |  87.84% |  86.94% |  86.94% | 2.000
14 | HARD 19 vs 7 [split-child] | textbook=STAND   |    220 |  -1.555 |  88.18% |  88.18% |  88.18% | 2.000
15 | HARD 19 vs 10 | textbook=STAND                |   1994 |  -1.532 |  86.91% |  86.56% |  86.56% | 2.000
16 | HARD 19 vs 5 | textbook=STAND                 |    569 |  -1.515 |  86.99% |  86.64% |  86.64% | 2.000
17 | HARD 19 vs 10 [split-child] | textbook=STAND  |    770 |  -1.504 |  86.10% |  85.84% |  85.84% | 2.000
18 | HARD 4 vs 7 [split-child] | textbook=HIT      |      8 |  -1.500 |  87.50% |   0.00% |   0.00% | 2.000
19 | HARD 19 vs 11 [split-child] | textbook=STAND  |    158 |  -1.494 |  86.08% |  86.08% |  86.08% | 2.000
20 | HARD 19 vs 2 | textbook=STAND                 |    579 |  -1.475 |  86.01% |  84.46% |  84.46% | 2.000
21 | HARD 6 vs 11 [split-child] | textbook=HIT     |     22 |  -1.455 |  86.36% |   0.00% |   0.00% | 2.000
22 | HARD 19 vs 8 | textbook=STAND                 |    516 |  -1.453 |  85.47% |  85.08% |  85.08% | 2.000
23 | HARD 19 vs 11 | textbook=STAND                |    368 |  -1.446 |  85.33% |  84.51% |  84.51% | 2.000
24 | HARD 19 vs 3 | textbook=STAND                 |    553 |  -1.443 |  85.17% |  84.63% |  84.63% | 2.000
25 | HARD 19 vs 9 [split-child] | textbook=STAND   |    197 |  -1.431 |  85.79% |  85.28% |  85.28% | 2.000

HOLDOUT ? WORST MAXIMUM-AGGRESSION VISIBLE STATES BY MEAN NET/HAND
Rank | Visible state | Hands | Mean net | Loss% | Bust% | 3rd-card bust% of 3rd-card draws | Mean stake
 1 | HARD 4 vs 5 [split-child] | textbook=HIT      |      3 |  -2.000 | 100.00% |   0.00% |   0.00% | 2.000
 2 | HARD 20 vs 2 [split-child] | textbook=STAND   |    628 |  -1.768 |  93.47% |  93.47% |  93.47% | 2.000
 3 | HARD 20 vs 6 [split-child] | textbook=STAND   |    667 |  -1.739 |  93.40% |  93.40% |  93.40% | 2.000
 4 | HARD 20 vs 9 [split-child] | textbook=STAND   |    639 |  -1.734 |  93.11% |  93.11% |  93.11% | 2.000
 5 | HARD 20 vs 5 [split-child] | textbook=STAND   |    690 |  -1.730 |  92.61% |  92.61% |  92.61% | 2.000
 6 | HARD 19 vs 9 [split-child] | textbook=STAND   |    166 |  -1.723 |  92.17% |  91.57% |  91.57% | 2.000
 7 | HARD 20 vs 7 [split-child] | textbook=STAND   |    725 |  -1.705 |  92.41% |  92.41% |  92.41% | 2.000
 8 | HARD 20 vs 10 [split-child] | textbook=STAND  |   2422 |  -1.699 |  92.20% |  92.20% |  92.20% | 2.000
 9 | HARD 20 vs 4 [split-child] | textbook=STAND   |    690 |  -1.699 |  91.88% |  91.88% |  91.88% | 2.000
10 | HARD 19 vs 8 [split-child] | textbook=STAND   |    178 |  -1.685 |  92.13% |  91.01% |  91.01% | 2.000
11 | HARD 20 vs 3 [split-child] | textbook=STAND   |    661 |  -1.676 |  91.83% |  91.83% |  91.83% | 2.000
12 | HARD 20 vs 11 [split-child] | textbook=STAND  |    474 |  -1.667 |  90.93% |  90.93% |  90.93% | 2.000
13 | HARD 20 vs 8 [split-child] | textbook=STAND   |    626 |  -1.658 |  90.73% |  90.73% |  90.73% | 2.000
14 | HARD 4 vs 11 [split-child] | textbook=HIT     |     10 |  -1.600 |  90.00% |   0.00% |   0.00% | 2.000
15 | SOFT 15 vs 11 [split-child] | textbook=HIT    |     10 |  -1.600 |  80.00% |   0.00% |   0.00% | 2.000
16 | HARD 19 vs 3 [split-child] | textbook=STAND   |    200 |  -1.540 |  88.00% |  86.50% |  86.50% | 2.000
17 | HARD 4 vs 9 [split-child] | textbook=HIT      |     16 |  -1.500 |  87.50% |   0.00% |   0.00% | 2.000
18 | HARD 4 vs 7 [split-child] | textbook=HIT      |      8 |  -1.500 |  87.50% |   0.00% |   0.00% | 2.000
19 | HARD 19 vs 2 | textbook=STAND                 |    558 |  -1.491 |  86.38% |  85.30% |  85.30% | 2.000
20 | HARD 19 vs 5 | textbook=STAND                 |    544 |  -1.489 |  86.76% |  86.03% |  86.03% | 2.000
21 | HARD 4 vs 10 [split-child] | textbook=HIT     |     39 |  -1.487 |  87.18% |   0.00% |   0.00% | 2.000
22 | HARD 19 vs 3 | textbook=STAND                 |    539 |  -1.466 |  85.53% |  85.34% |  85.34% | 2.000
23 | HARD 7 vs 9 [split-child] | textbook=HIT      |     41 |  -1.463 |  85.37% |   0.00% |   0.00% | 2.000
24 | HARD 19 vs 7 | textbook=STAND                 |    561 |  -1.462 |  86.10% |  85.20% |  85.20% | 2.000
25 | HARD 19 vs 10 | textbook=STAND                |   1972 |  -1.459 |  84.58% |  84.18% |  84.18% | 2.000


PHASE 3.02 ? REPLICATED CANDIDATE DAMAGE STATES

Filter: same visible state appears >=100 times in BOTH primary and holdout, and Maximum Aggression has negative mean net/hand in BOTH cohorts. This avoids choosing Phase 3.02 from tiny-sample extremes.

Rank | State | Primary n / net | Holdout n / net | P bust% / H bust% | Textbook first action
 1 | HARD 20 vs 2 [split-child]         |   642 / -1.748 |   628 / -1.768 |  93.30% /  93.47% | STAND
 2 | HARD 20 vs 6 [split-child]         |   660 / -1.742 |   667 / -1.739 |  93.18% /  93.40% | STAND
 3 | HARD 20 vs 5 [split-child]         |   672 / -1.750 |   690 / -1.730 |  93.30% /  92.61% | STAND
 4 | HARD 20 vs 9 [split-child]         |   659 / -1.718 |   639 / -1.734 |  92.87% /  93.11% | STAND
 5 | HARD 20 vs 4 [split-child]         |   643 / -1.742 |   690 / -1.699 |  93.31% /  91.88% | STAND
 6 | HARD 20 vs 7 [split-child]         |   664 / -1.723 |   725 / -1.705 |  92.77% /  92.41% | STAND
 7 | HARD 20 vs 3 [split-child]         |   704 / -1.702 |   661 / -1.676 |  92.19% /  91.83% | STAND
 8 | HARD 20 vs 11 [split-child]        |   446 / -1.691 |   474 / -1.667 |  92.15% /  90.93% | STAND
 9 | HARD 20 vs 10 [split-child]        |  2385 / -1.652 |  2422 / -1.699 |  91.11% /  92.20% | STAND
10 | HARD 20 vs 8 [split-child]         |   681 / -1.648 |   626 / -1.658 |  91.19% /  90.73% | STAND
11 | HARD 19 vs 9 [split-child]         |   197 / -1.431 |   166 / -1.723 |  85.28% /  91.57% | STAND
12 | HARD 19 vs 4 [split-child]         |   186 / -1.624 |   217 / -1.447 |  89.78% /  83.41% | STAND
13 | HARD 19 vs 5                       |   569 / -1.515 |   544 / -1.489 |  86.64% /  86.03% | STAND
14 | HARD 19 vs 10                      |  1994 / -1.532 |  1972 / -1.459 |  86.56% /  84.18% | STAND
15 | HARD 19 vs 6 [split-child]         |   222 / -1.559 |   221 / -1.412 |  86.94% /  84.16% | STAND
16 | HARD 19 vs 2                       |   579 / -1.475 |   558 / -1.491 |  84.46% /  85.30% | STAND
17 | HARD 19 vs 8 [split-child]         |   227 / -1.278 |   178 / -1.685 |  81.50% /  91.01% | STAND
18 | HARD 19 vs 7 [split-child]         |   220 / -1.555 |   229 / -1.380 |  88.18% /  83.84% | STAND
19 | HARD 19 vs 10 [split-child]        |   770 / -1.504 |   729 / -1.429 |  85.84% /  83.81% | STAND
20 | HARD 19 vs 3                       |   553 / -1.443 |   539 / -1.466 |  84.63% /  85.34% | STAND

DESIGN USE
The next optimisation phase should modify ONE policy rule from a replicated, adequately sampled state family. Initial ten-value pairs are now explicitly excluded from non-pair DOUBLE interpretation because SPLIT has precedence.

PHASE 3.02 ? RUNNING TXT AUDIT RECORD

OBJECTIVE
Preserve the finite six-deck baseline while explicitly verifying the action-precedence rule for all ten-valued two-card combinations before policy optimisation begins.

EVIDENCE HIERARCHY
1. Primary design authority: finite six-deck Phase 2.xx evidence and new Phase 3.xx data.
2. Earlier non-depleting/infinite-generator behaviour-profile data: historical corroboration and hypothesis generation only.
3. Old behaviour-profile evidence must not directly define a new policy unless finite-shoe evidence independently supports the same conclusion.

FROZEN VARIABLES
Six decks, deterministic session shuffle, 30 hands, 100-unit start, 1-unit base wager, dealer rules, blackjack payout, split limit, no in-session reshuffle, textbook control and Maximum Aggression policy are unchanged from Phase 2.02.

POLICY INFORMATION BOUNDARY
The future optimiser may use player hand composition/total, pair status, dealer upcard and legal actions. Phase 3.02 deliberately records no usable previous-card-history, running count, remaining-rank inventory, hidden card or future-card feature.

PHASE 3.02 DECISION RULE
No optimisation is performed. Maximum Aggression remains the extreme baseline. The next experiment is selected only after primary and holdout state-level damage patterns are compared.

PHASE 3.02 REPLAY FINGERPRINT: 83321a80f1d40d3a35158ddd8e3d850f813b1fc3c8b9895c4fcd62d6d2c25045
