CLI tool that evaluates trading‑strategy backtest sweeps and decides to ship, iterate or kill
Quant researchers often drown in variant backtest results and need a quick, objective verdict. This tool reads a CSV of parameter sweeps, measures top‑K dispersion, IS/OOS rank stability, and parameter‑plateau structure, then returns a clear CONVERGED, ITERATE or KILL verdict via exit codes. It runs dependency‑free in plain Python, making it easy to integrate into any workflow or Claude Code plugin. Compared to ad‑hoc spreadsheets, it offers reproducible, automated analysis with built‑in statistical checks.
View on GitHub →martianmobile/strategy-evaluation