PHASE 3.13 — NEXT DESIGN NOTE

Phase 3.13 followed the 3.12 instruction to compare mechanisms before intervening.

PRE-INTERVENTION EVIDENCE
HARD 12–13 vs dealer 7–Ace:
Primary n=18,075, mean net=-0.7296; Holdout n=18,245, mean net=-0.7604.
Residual HARD 5–8 vs dealer 2–6:
Primary n=4,887, mean net=-0.2202; Holdout n=4,737, mean net=-0.2681.

This justified selecting HARD 12–13 vs 7–Ace -> HIT.

RESULT
Primary average bankroll: 95.203 -> 96.617 (+1.414).
Holdout: 95.065 -> 96.547 (+1.482).
Positive sessions: +4.72 pp Primary / +4.22 pp Holdout.

PHASE 3.14 DESIGN QUESTION
The remaining ranking now contains at least two distinct coherent candidates:
A. HARD 8 against strong dealers -> Textbook HIT, zero immediate bust mechanism.
B. HARD 13 against weak dealer cards such as dealer 2 -> Textbook STAND, substantial bust mechanism.

Aggregate both families across Primary/Holdout before choosing the next intervention.
Do not modify both in the same experiment.

All frozen architecture and information controls remain unchanged.
