DETERMINISTIC BLACKJACK DECISION ENGINE RESEARCH
PHASE 1.11 — PACKAGE GUIDE

WHAT CHANGED FROM PHASE 1.10
----------------------------
Phase 1.10 counted how often fingerprints repeat.
Phase 1.11 asks whether a completed previous-hand fingerprint can justify changing only the next wager from one unit to two units.

PLAIN-ENGLISH TEST DESIGN
-------------------------
1. Training shuffles discover possible triggers.
2. Different validation shuffles remove triggers that fail to repeat.
3. The surviving trigger list is frozen.
4. Completely unseen holdout shuffles compare flat one-unit wagering with fingerprint-triggered two-unit wagering on the same hands.

The controller cannot inspect future cards, remaining card composition, shoe position, penetration, seed or shuffle order.

MAIN RESULT
-----------
Exact previous-opening model:
Flat holdout return: -0.2098%
Fingerprint holdout return: -0.1435%
Difference: +198.00 net units and +0.0663 percentage points.

Broader previous-hand model:
Flat holdout return: -0.2098%
Fingerprint holdout return: -0.1802%
Difference: +23.00 net units and +0.0296 percentage points.

Both fingerprint models lost overall. They merely lost less than flat wagering on this holdout cohort. This is not proof that previous cards predict future cards and is not evidence of a guaranteed player advantage.

FILES
-----
DeterministicBlackjackPhase111.java
    Complete Java source code.

Phase_1_11_Results.txt
    Complete console output, including preserved earlier verification.

Phase_1_11_Previous_Fingerprint_Wager_Study.txt
    Main Phase 1.11 evidence with layman explanations, net gains/losses and trigger tables.

Phase_1_11_Replay_Verification.txt
    Independent repeat-run hash comparison.

Phase_1_10_Fingerprint_Frequency.txt
Phase_1_10_Sample_Hands.txt
    Preserved supporting evidence generated by the combined programme.
