Contents

Contents

Tap any chapter to start reading.

Chapter 1 Alpha Diagnostics & the Information Coefficient

The single most important metric in alpha-model monitoring: rank-based IC, cumulative IC as the model’s equity curve, the Fundamental Law (IR ≈ IC·√BR), P&L attribution, and the early-warning signals that tell you a model is decaying before the wealth curve shows it.

Chapter 2 Model Decay, Drift Detection & Retraining

Why models decay (regime change, feature drift, crowding), the KS test for out-of-distribution monitoring, alpha-decay curves, scheduled vs hybrid retraining policies, and the walk-forward automated alert system that fires before drawdowns hit.

Chapter 3 Multiple Testing

The statistical machinery for honest backtesting: family-wise error rate, Bonferroni and Holm corrections, false discovery rate, Benjamini–Hochberg, the Deflated Sharpe Ratio, and bootstrap null distributions.

Chapter 4 Methodology Snooping

The behavioural side of model risk: the Garden of Forking Paths, researcher degrees of freedom, why train/test alone is insufficient, lessons from a 5.3-year walk-forward study, and the production protocols Renaissance, Two Sigma, DE Shaw, Citadel, and AQR use to keep themselves honest.

Appendix A Solutions to Exercises

Worked solutions for every exercise in Chapters 1–4 — IC and Fundamental Law arithmetic, KS calibration drills, step-by-step Bonferroni / Holm / Benjamini–Hochberg / Benjamini–Yekutieli procedures, deflated-Sharpe practice, and essay answers for the Methodology Snooping reasoning exercises.