PHASE 3.05 SECOND INTERVENTION RESULTS

PRIMARY
Average ending bankroll Textbook/Selective: 99.956 / 86.922
Maximum bankroll observed Textbook/Selective: 126.0 at hand 29 (session 1777) / 127.0 at hand 30 (session 1543)
Positive sessions Textbook/Selective: 48.14% / 10.18%
Paired winner Textbook/Selective/tie: 86.34% / 12.50% / 1.16%
Doubles Textbook/Selective: 15716 / 134059; splits Textbook/Selective: 3808 / 21731
Mean cards consumed Textbook/Selective: 165.674 / 175.828
Ending bankroll >=2x (200) Textbook/Selective: 0.00% / 0.00%
Ending bankroll >=3x (300) Textbook/Selective: 0.00% / 0.00%
Ending bankroll >=5x (500) Textbook/Selective: 0.00% / 0.00%
Ending bankroll >=10x (1000) Textbook/Selective: 0.00% / 0.00%

HOLDOUT
Average ending bankroll Textbook/Selective: 99.972 / 86.706
Maximum bankroll observed Textbook/Selective: 122.5 at hand 24 (session 1273) / 129.0 at hand 18 (session 3473)
Positive sessions Textbook/Selective: 47.42% / 9.80%
Paired winner Textbook/Selective/tie: 87.34% / 11.68% / 0.98%
Doubles Textbook/Selective: 15412 / 133685; splits Textbook/Selective: 3747 / 21757
Mean cards consumed Textbook/Selective: 165.647 / 175.694
Ending bankroll >=2x (200) Textbook/Selective: 0.00% / 0.00%
Ending bankroll >=3x (300) Textbook/Selective: 0.00% / 0.00%
Ending bankroll >=5x (500) Textbook/Selective: 0.00% / 0.00%
Ending bankroll >=10x (1000) Textbook/Selective: 0.00% / 0.00%


PHASE 3.05 — SELECTIVE-AGGRESSION INSTRUMENTATION SUMMARY

PRIMARY resolved selective-aggression hands: 171731; mean net/hand -0.3808; mean committed stake 1.7806; bust rate 25.47%; third-card bust rate when a third card was taken 32.63%.
Textbook would have chosen on those same visible first-decision states: HIT 70311, STAND 68112, DOUBLE 15768.

HOLDOUT resolved selective-aggression hands: 171757; mean net/hand -0.3870; mean committed stake 1.7783; bust rate 25.53%; third-card bust rate when a third card was taken 32.80%.
Textbook would have chosen on those same visible first-decision states: HIT 70405, STAND 67880, DOUBLE 15740.

INTERPRETATION
Phase 3.05 carries forward split-child HARD 20 STAND and the HARD 19 STAND family, then adds one new family only: split-child HARD 18 now STANDS instead of automatically DOUBLING. Ordinary HARD 18 remains unchanged. All other Maximum Aggression behaviour remains frozen.

PRIMARY — WORST SELECTIVE-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 4 vs 11 [split-child] | textbook=HIT     |      6 |  -2.000 | 100.00% |   0.00% |   0.00% | 2.000
 3 | SOFT 16 vs 3 [split-child] | textbook=HIT     |     21 |  -1.619 |  85.71% |   0.00% |   0.00% | 2.000
 4 | HARD 4 vs 9 [split-child] | textbook=HIT      |     10 |  -1.600 |  90.00% |   0.00% |   0.00% | 2.000
 5 | HARD 4 vs 8 [split-child] | textbook=HIT      |      9 |  -1.556 |  88.89% |   0.00% |   0.00% | 2.000
 6 | HARD 5 vs 10 [split-child] | textbook=HIT     |     90 |  -1.511 |  87.78% |   0.00% |   0.00% | 2.000
 7 | HARD 4 vs 7 [split-child] | textbook=HIT      |      8 |  -1.500 |  87.50% |   0.00% |   0.00% | 2.000
 8 | HARD 5 vs 9 [split-child] | textbook=HIT      |     21 |  -1.429 |  85.71% |   0.00% |   0.00% | 2.000
 9 | HARD 5 vs 9 | textbook=HIT                    |    131 |  -1.359 |  83.97% |   0.00% |   0.00% | 2.000
10 | HARD 6 vs 11 | textbook=HIT                   |     99 |  -1.333 |  82.83% |   0.00% |   0.00% | 2.000
11 | HARD 7 vs 9 [split-child] | textbook=HIT      |     32 |  -1.313 |  78.13% |   0.00% |   0.00% | 2.000
12 | HARD 18 vs 4 | textbook=STAND                 |    583 |  -1.310 |  81.82% |  79.76% |  79.76% | 2.000
13 | HARD 6 vs 9 [split-child] | textbook=HIT      |     23 |  -1.304 |  82.61% |   0.00% |   0.00% | 2.000
14 | HARD 18 vs 10 | textbook=STAND                |   2047 |  -1.301 |  80.17% |  76.89% |  76.89% | 2.000
15 | SOFT 14 vs 9 [split-child] | textbook=HIT     |     17 |  -1.294 |  82.35% |   0.00% |   0.00% | 2.000
16 | HARD 5 vs 11 | textbook=HIT                   |     87 |  -1.264 |  81.61% |   0.00% |   0.00% | 2.000
17 | HARD 18 vs 9 | textbook=STAND                 |    512 |  -1.250 |  79.30% |  76.37% |  76.37% | 2.000
18 | HARD 7 vs 8 [split-child] | textbook=HIT      |     42 |  -1.238 |  76.19% |   0.00% |   0.00% | 2.000
19 | HARD 18 vs 2 | textbook=STAND                 |    609 |  -1.225 |  78.98% |  75.04% |  75.04% | 2.000
20 | HARD 18 vs 5 | textbook=STAND                 |    564 |  -1.223 |  79.26% |  76.95% |  76.95% | 2.000
21 | HARD 18 vs 8 | textbook=STAND                 |    473 |  -1.218 |  79.28% |  78.22% |  78.22% | 2.000
22 | HARD 7 vs 11 [split-child] | textbook=HIT     |     33 |  -1.212 |  78.79% |   0.00% |   0.00% | 2.000
23 | HARD 18 vs 3 | textbook=STAND                 |    519 |  -1.206 |  79.00% |  77.26% |  77.26% | 2.000
24 | HARD 14 vs 11 [split-child] | textbook=HIT    |    176 |  -1.205 |  76.70% |  50.57% |  50.57% | 2.000
25 | HARD 5 vs 7 [split-child] | textbook=HIT      |     15 |  -1.200 |  80.00% |   0.00% |   0.00% | 2.000

HOLDOUT — WORST SELECTIVE-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 2 [split-child] | textbook=HIT      |      6 |  -2.000 | 100.00% |   0.00% |   0.00% | 2.000
 2 | HARD 4 vs 5 [split-child] | textbook=HIT      |      3 |  -2.000 | 100.00% |   0.00% |   0.00% | 2.000
 3 | SOFT 15 vs 11 [split-child] | textbook=HIT    |     13 |  -1.846 |  92.31% |   0.00% |   0.00% | 2.000
 4 | HARD 7 vs 11 [split-child] | textbook=HIT     |     29 |  -1.724 |  93.10% |   0.00% |   0.00% | 2.000
 5 | HARD 4 vs 10 [split-child] | textbook=HIT     |     34 |  -1.647 |  91.18% |   0.00% |   0.00% | 2.000
 6 | HARD 6 vs 8 [split-child] | textbook=HIT      |     32 |  -1.625 |  90.63% |   0.00% |   0.00% | 2.000
 7 | HARD 4 vs 8 [split-child] | textbook=HIT      |      9 |  -1.556 |  88.89% |   0.00% |   0.00% | 2.000
 8 | HARD 5 vs 11 [split-child] | textbook=HIT     |      8 |  -1.500 |  87.50% |   0.00% |   0.00% | 2.000
 9 | HARD 6 vs 9 [split-child] | textbook=HIT      |     39 |  -1.487 |  87.18% |   0.00% |   0.00% | 2.000
10 | SOFT 17 vs 11 [split-child] | textbook=HIT    |     18 |  -1.444 |  83.33% |   0.00% |   0.00% | 2.000
11 | HARD 5 vs 8 [split-child] | textbook=HIT      |     24 |  -1.333 |  83.33% |   0.00% |   0.00% | 2.000
12 | HARD 18 vs 2 | textbook=STAND                 |    554 |  -1.325 |  81.77% |  78.88% |  78.88% | 2.000
13 | HARD 6 vs 11 | textbook=HIT                   |     99 |  -1.293 |  81.82% |   0.00% |   0.00% | 2.000
14 | HARD 17 vs 10 [split-child] | textbook=STAND  |    847 |  -1.292 |  79.46% |  70.01% |  70.01% | 2.000
15 | HARD 7 vs 9 [split-child] | textbook=HIT      |     45 |  -1.289 |  80.00% |   0.00% |   0.00% | 2.000
16 | HARD 18 vs 10 | textbook=STAND                |   1993 |  -1.276 |  80.08% |  76.37% |  76.37% | 2.000
17 | HARD 4 vs 11 [split-child] | textbook=HIT     |     11 |  -1.273 |  81.82% |   0.00% |   0.00% | 2.000
18 | HARD 18 vs 4 | textbook=STAND                 |    551 |  -1.267 |  80.76% |  78.77% |  78.77% | 2.000
19 | HARD 6 vs 11 [split-child] | textbook=HIT     |     21 |  -1.238 |  80.95% |   0.00% |   0.00% | 2.000
20 | HARD 18 vs 7 | textbook=STAND                 |    512 |  -1.227 |  80.08% |  78.13% |  78.13% | 2.000
21 | HARD 18 vs 8 | textbook=STAND                 |    538 |  -1.223 |  79.18% |  77.32% |  77.32% | 2.000
22 | HARD 7 vs 11 | textbook=HIT                   |    200 |  -1.220 |  78.50% |   0.00% |   0.00% | 2.000
23 | HARD 5 vs 11 | textbook=HIT                   |     97 |  -1.216 |  80.41% |   0.00% |   0.00% | 2.000
24 | HARD 16 vs 11 [split-child] | textbook=HIT    |    181 |  -1.215 |  77.35% |  60.22% |  60.22% | 2.000
25 | HARD 17 vs 11 [split-child] | textbook=STAND  |    158 |  -1.215 |  76.58% |  68.99% |  68.99% | 2.000


PHASE 3.05 — REMAINING REPLICATED DAMAGE STATES AFTER SPLIT-CHILD HARD 18

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

Rank | State | Primary n / net | Holdout n / net | P bust% / H bust% | Textbook first action
 1 | HARD 18 vs 10                      |  2047 / -1.301 |  1993 / -1.276 |  76.89% /  76.37% | STAND
 2 | HARD 18 vs 4                       |   583 / -1.310 |   551 / -1.267 |  79.76% /  78.77% | STAND
 3 | HARD 18 vs 2                       |   609 / -1.225 |   554 / -1.325 |  75.04% /  78.88% | STAND
 4 | HARD 18 vs 9                       |   512 / -1.250 |   555 / -1.196 |  76.37% /  75.14% | STAND
 5 | HARD 18 vs 8                       |   473 / -1.218 |   538 / -1.223 |  78.22% /  77.32% | STAND
 6 | HARD 18 vs 5                       |   564 / -1.223 |   567 / -1.210 |  76.95% /  75.66% | STAND
 7 | HARD 17 vs 10 [split-child]        |   784 / -1.120 |   847 / -1.292 |  66.33% /  70.01% | STAND
 8 | HARD 16 vs 11 [split-child]        |   166 / -1.193 |   181 / -1.215 |  60.24% /  60.22% | HIT
 9 | HARD 5 vs 9                        |   131 / -1.359 |   149 / -1.034 |   0.00% /   0.00% | HIT
10 | HARD 18 vs 11                      |   387 / -1.189 |   413 / -1.196 |  75.19% /  75.30% | STAND
11 | HARD 18 vs 7                       |   530 / -1.151 |   512 / -1.227 |  75.09% /  78.13% | STAND
12 | HARD 18 vs 3                       |   519 / -1.206 |   561 / -1.166 |  77.26% /  76.11% | STAND
13 | HARD 17 vs 11 [split-child]        |   151 / -1.152 |   158 / -1.215 |  69.54% /  68.99% | STAND
14 | HARD 17 vs 10                      |  2530 / -1.180 |  2534 / -1.174 |  69.13% /  67.36% | STAND
15 | HARD 17 vs 9 [split-child]         |   244 / -1.197 |   212 / -1.132 |  74.59% /  68.40% | STAND
16 | HARD 17 vs 6 [split-child]         |   235 / -1.149 |   229 / -1.179 |  70.64% /  73.36% | STAND
17 | HARD 18 vs 6                       |   560 / -1.179 |   527 / -1.131 |  75.36% /  74.95% | STAND
18 | HARD 7 vs 11                       |   202 / -1.089 |   200 / -1.220 |   0.00% /   0.00% | HIT
19 | HARD 5 vs 10                       |   517 / -1.118 |   517 / -1.172 |   0.00% /   0.00% | HIT
20 | HARD 17 vs 2 [split-child]         |   233 / -1.176 |   225 / -1.102 |  70.39% /  68.44% | STAND

DESIGN USE
Phase 3.05 carries forward the validated HARD 20 split-child and HARD 19 protections and removes one additional automatic DOUBLE family: split-child HARD 18. Ordinary HARD 18 remains unchanged. Remaining replicated damage states are retained for later one-rule-at-a-time interventions.

PHASE 3.05 — RUNNING TXT AUDIT RECORD

OBJECTIVE
Make the third controlled stabilisation intervention while preserving the frozen architecture. Carry forward split-child HARD 20 STAND and HARD 19 STAND, then add one new policy family: split-child HARD 18 STANDS rather than taking Maximum Aggression's automatic DOUBLE.

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 3.05 deliberately uses no previous-card-history, running count, remaining-rank inventory, hidden card or future-card feature. The accumulated policy uses only current visible states: split-child HARD 20, HARD 19, and split-child HARD 18. No card-history or composition feature is introduced.

PHASE 3.05 DECISION RULE
Carry forward split-child HARD 20 STAND and the HARD 19 STAND family. ONE additional Phase 3.05 intervention: if a split-child first-decision hand is hard (not soft) and totals 18, Selective Aggression STANDS. Ordinary HARD 18 remains unchanged. Otherwise the frozen Maximum Aggression policy is unchanged. Textbook remains the control.

PHASE 3.05 REPLAY FINGERPRINT: aa09cba3babf71a4c178492a68c85b69339a17f0548e563d02802e423533c3fa
