Chapter 10 · §4.7

Weekly reports without revealing the scorer

A feedback loop, not a reverse-engineering contest

Participants should receive weekly automated reports. The purpose is to keep them engaged, reduce confusion, and give them enough signal to debug obvious issues.

The report should stay general. It should say what moved up or down, what looked strong, and what looked weak. It should not reveal exact weights, thresholds, or formulas. The goal is a feedback loop, not a reverse-engineering contest.

VISIBLE TO PARTICIPANT Week 4 — Report Strong week. — Lower turnover than W3. — Sector exposure tilted to financials. — Drawdown remained within band. — Notional capital re-rated upward. No exact weights. No exact thresholds. No formula breakdown. Just enough to debug obvious issues. WALL PRIVATE — NOT EXPOSED Scorer Algorithm w_bench × (R_p − R_b) + … σ_resid penalty: λ × σ_w … turnover_cost = κ × … notional_w+1 = f(s_w, q_w) drawdown_cap … general principles published specific formulas private
Public principles. Private formulas. Enough feedback to debug, not enough to reverse-engineer.
← Prev: Human review All chapters Next: Risks countered →