Alpha-GPT 2.0: Human-in-the-Loop AI for Quantitative Investment
Hang Yuan, Saizhuo Wang, Jian Guo
TL;DR
Alpha-GPT 2.0 tackles the scalability gap in quantitative investment research by coupling human expertise with LLM-powered agents in a full-pipeline workflow. It introduces a three-layer, human-in-the-loop system—alpha mining, alpha modeling, and alpha analysis—where agents operate with domain memories and Standard Operating Procedures to iteratively discover and validate trading alphas. The work contributes a complete automation framework that extends Alpha-GPT to end-to-end automation, integrates Think-on-Graph-based risk reasoning, and enables real-time, risk-aware decision support. The approach promises improved efficiency, precision, and adaptability in live or semi-live investment settings.
Abstract
Recently, we introduced a new paradigm for alpha mining in the realm of quantitative investment, developing a new interactive alpha mining system framework, Alpha-GPT. This system is centered on iterative Human-AI interaction based on large language models, introducing a Human-in-the-Loop approach to alpha discovery. In this paper, we present the next-generation Alpha-GPT 2.0 \footnote{Draft. Work in progress}, a quantitative investment framework that further encompasses crucial modeling and analysis phases in quantitative investment. This framework emphasizes the iterative, interactive research between humans and AI, embodying a Human-in-the-Loop strategy throughout the entire quantitative investment pipeline. By assimilating the insights of human researchers into the systematic alpha research process, we effectively leverage the Human-in-the-Loop approach, enhancing the efficiency and precision of quantitative investment research.