This repository contains the FinSearchComp dataset, a benchmark for evaluating financial search and reasoning capabilities of LLM-based agents, as presented in the paper FinSearchComp: Towards a Realistic, Expert-Level Evaluation of Financial Search and Reasoning.

Project Page: https://randomtutu.github.io/FinSearchComp/

FinSearchComp is the first fully open-source agent benchmark designed for realistic, open-domain financial search and reasoning. It comprises three tasks that closely reproduce real-world financial analyst workflows:

• Time-Sensitive Data Fetching
• Simple Historical Lookup
• Complex Historical Investigation

The benchmark includes 635 questions spanning global and Greater China markets, meticulously annotated by 70 professional financial experts.