PHASE 3.05 NEXT DESIGN — THIRD CONTROLLED STABILISATION STEP

Carry forward both validated selective-aggression layers:
1. split-child HARD 20 STANDS;
2. HARD 19 STANDS, ordinary or split-child.

Phase 3.04 evidence shows HARD 18 now dominates the remaining replicated damage ranking in both Primary and Holdout cohorts. Before implementation, treat HARD 18 as the candidate area and select no more than ONE coherent visible-state policy family.

Do not change any other variable. Preserve the six-deck finite shoe, deterministic session-start shuffle, 30-hand horizon, 100-unit start, 1-unit base wager, dealer/payout/split rules, no in-session reshuffle, no surrender, Textbook control, and no-card-history/no-counting information boundary.

Continue observational tracking of positive finish and 2x / 3x / 5x / 10x ending-bankroll thresholds.

Scientific objective: determine whether a third independently justified visible-state correction continues the replicated bankroll improvement without losing one-rule-at-a-time attribution.
