Skip to content

Seeking Collaboration to Test Automated Research in Finance

Key Ideas

  • Lopez-Lira has built a fully automated research pipeline that takes a one-paragraph idea and produces a submission-ready finance paper in about one day
  • The pipeline handles literature review, model development, mathematical proofs, empirical analysis, paper writing, and simulated peer review with quality gates at every stage
  • The system has access to WRDS (CRSP, Compustat, TAQ), FRED, Ken French data, and other publicly available sources
  • Output quality is described as approximately JFQA level, handling theory, empirical work, or both
  • Two entry paths: (1) run the public idea-evaluation pipeline and score 8/10+, or (2) be faculty at a top-100 finance department or PhD student at a top-50 program
  • The goal is published journal papers, not demos; Lopez-Lira is also studying the process itself for a meta-paper on AI-assisted research

Summary

Alejandro Lopez-Lira announces an open call for collaboration to test his automated research pipeline in finance. The system accepts a one-paragraph research idea and autonomously produces a full submission-ready paper, including literature review, theoretical model development with proofs, empirical analysis, writing, self-critique, and revision until a simulated referee is satisfied. Lopez-Lira oversees the process and ensures quality, offering joint authorship on resulting papers.

The pipeline is designed for standard finance data accessible through WRDS, FRED, and public sources. It explicitly excludes proprietary or exotic datasets. Two pathways are available for collaborators: a public idea-evaluation pipeline on GitHub that scores ideas (threshold of 8/10), and a direct pathway for faculty and PhD students at top programs. The dual goal is both producing publishable research and studying the AI-assisted research process itself.

Relevance to Economics Research

This represents a concrete, production-ready implementation of AI-automated research in finance, moving beyond proof-of-concept demonstrations. The use of standard finance datasets (CRSP, Compustat, FRED) and the explicit goal of journal publication make this directly relevant to the empirical finance research community. The idea-evaluation pipeline and quality gates provide a model for how automated research might be quality-controlled. The meta-paper component adds a self-reflective dimension about the implications of such systems for academic knowledge production.