Reproduce “BuySideFlow: A Point-in-Time Benchmark for LLM Agents in Investment Research” 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

Dong Bo, Fanjin Zhang, Ke Cheng, Xinxin Zhao, Xiaowei Su, Yang Wang, Jinghang Xu, Cuiping Li and 2 more

BuySideFlow evaluates historical investment research workflows by checking both answer agreement and whether executed SQL or SQL–Python dependencies follow task-specific point-in-time rules. Across 900 audited runs on 100 tasks, 193 of 380 runs that passed result checks had confirmed A1–A3 violations, a 50.8% pass-side rate. This rate applies only to audited passing runs, not all tasks or outputs' impact.

Summary by alphaXiv.

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