Reproduce “FinForesight: A Diagnostic Backtesting Benchmark for Deep Research Systems in Financial Forecasting” 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

Fengyuan Liu, Mukai Li, Jieshun You, Yilin Zheng, Chenjia Min, Yizhou Zhang, Harper Chen, Qi Liu

The authors introduce FinForesight, a benchmark for evaluating forecasts across three periods. It restricts retrieved evidence to information public by the forecast date and compares post-cutoff forecasts with live results. In one-week US tests, point-in-time controls plus a leakage filter reduce mean absolute score gaps from 0.1203 to 0.0285, but do not show T2 universally replaces live validation.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.