PHASE 1.21 — FINDINGS SUMMARY

LAYMAN CONCLUSION
-----------------
Allowing one contrary card made the psychology much more realistic and much more active than the strict consecutive-run test in Phase 1.20.

The clearest result was the high-card continuation belief: after a high-card majority, the player expected another high card and stood early against strong dealer cards. The 2-of-3 model activated 95,753 times and reduced the average ending bankroll from £98.75 under textbook play to £96.58. The independent holdout repeated the direction at £96.42 versus £98.55.

The low-card continuation belief also performed worse. It encouraged extra hits on 15 or 16 against dealer 2–6. In the 2-of-3 model, 57.14% of those extra-hit cases busted. Its primary average ending bankroll was £97.71.

The 3-of-4 windows activated less frequently and generally remained closer to textbook play, but none produced an improvement over the textbook control in both primary and holdout cohorts.

SCIENTIFIC INTERPRETATION
-------------------------
A short-window majority can strongly influence behaviour even when one card contradicts the apparent pattern. However, the observed prediction percentages did not translate into better decision quality or better whole-session results.

Deterministic replay: PASS.
No guide has been created.
