Reproduce “FactorBench: A Portfolio-Aware Benchmark for Automated Factor Mining” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhuohan Wang, Carmine Ventre
FactorBench compares nine automated factor-mining methods across five equity markets, testing stock-ranking signals from discovery through portfolios. It preserves each method’s search process while standardizing data, evaluation, and execution. On test-period composites, R&D-Agent’s raw IC is 0.0253 versus residual IC of 0.0011; AlphaQCM’s are 0.0191 and 0.0175. After-cost Sharpe ratios are similar, and no paradigm consistently leads among the tested implementations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.