A Catalogue of Systematic Trading Models

Preface

Each model in this repository is documented as a standalone working paper. The papers share a common evaluation protocol — lookahead-safe signal construction, one-bar execution lag, explicit transaction costs, and a fixed battery of risk-adjusted performance metrics — so that results across families are read on the same footing.

Two of the hypotheses fail outright, and the two options papers are deliberate demonstrators. They are written up with the same rigour as the winners, because a catalogue that reports only winners is, statistically, one that has been overfit. Negative results here are evidence, not omissions.

Conventions. Returns are simple unless noted; the risk-free rate is taken as zero; the Sharpe ratio is annualized by √(periods/year) with the period count stated per paper. See methodology.md and glossary.md for shared definitions.