Multi-agent system discovers interpretable finance valuation equations
Self-Evolving Multi-Agent Symbolic Discovery for Financial Fundamental Analysis
Multiagent SystemsArtificial IntelligenceMachine Learning
Summary
Figuring out how much a company is worth is tricky because there are many ways to look at it and the market keeps changing. The researchers created MUFASA, a system where many specialized agents each try to find useful formulas from different angles, and another part helps pick the best ones based on market trends and performance. Their system also remembers how well different formulas did over time to keep improving under noisy data. Tests showed that MUFASA performs better than traditional methods and some AI tools, while providing clear and understandable formulas for valuing companies.
What this means in practice
- •For quantitative finance teams: Create adaptable valuation models that produce transparent formulas tuned to changing market conditions.
- •For financial model developers: Integrate multi-agent symbolic discovery techniques to improve reliability and interpretability of valuation tools.
Authors
Kelvin J. L. Koa, Filip Orestav, Shengqiong Wu, Michael J. Wooldridge, Ke-Wei Huang
Abstract
While symbolic regression (SR) has been successfully used in science to discover new equations, its use in financial valuation is hindered by several limitations. Whereas the natural sciences provide objectively correct relationships, financial valuation constitutes a distinct class of symbolic discovery problems, as it admits multiple valid perspectives, operates under non-stationary market conditions, and involves noisy, continuous performance signals. In this work, we propose Multi-Agent Fundamental Analysis with Symbolic Adaptive learning (MUFASA), a hierarchical multi-agent framework for symbolic discovery in finance. MUFASA introduces (1) disentangled equation discovery via specialized agents representing distinct valuation perspectives, (2) a meta-coordinator that performs hierarchical-level reasoning over market context information, and (3) a memory mechanism that reasons over statistical performance summaries (e.g., accuracy, stability, and tail risk) to guide learning under noisy feedback. Experiments across datasets from multiple countries show that MUFASA achieves state-of-the-art performance on the valuation task compared to classical finance methods, financial large language models, and SR approaches, while simultaneously producing interpretable equations, which we share with the community. We also make publicly available the distilled learnings across evolution iterations and context-dependent strategy weights, which might offer useful insights for future research on financial fundamental analysis.