⬡ Pure Mechanics
A multi-factor backtesting framework
A full pipeline from raw data to signals
2026.09.22· 1 min readPythonquantbacktestfactors
- Source
- Python / pandas / numpy
- Duration
- 2026
- Venue
- https://github.com/
This project implements a lightweight multi-factor backtesting framework, aiming to shorten the path from research to backtest.
Pipeline
- Cleaning — halts, adjustments, outliers
- Factor computation — batch factor exposures
- Cross-sectional processing — winsorize, standardize, neutralize by industry and size
- Portfolio construction — constrained optimization
- Attribution — Brinson attribution plus turnover analysis
# factor neutralization
from project.factors import neutralize, standardize
f = standardize(raw_factor)
f = neutralize(f, industry=industry, size=log_mktcap)
The whole framework emphasizes reproducibility: configs, data versions and results are all persisted.