AlphaSAGE: Structure-Aware Alpha Mining via GFlowNets for Robust Exploration

2025-09-29

AI Summary

Automated alpha signal discovery in quantitative finance faces three key challenges: sparse rewards impeding efficient exploration; sequential modeling of mathematical expressions failing to capture structural semantics; and conventional reinforcement learning (RL) optimizing for a single optimal solution, thereby hindering the generation of uncorrelated alpha combinations. To address these, we propose a structure-aware GFlowNet framework. First, we design a relational graph convolutional network (RGCN)-based structural encoder that explicitly models syntactic and semantic structures of financial formulas. Second, we replace RL policies with generative flow networks (GFlowNets), enabling parallel, multi-path generation of diverse alphas. Third, we introduce a dense, multi-dimensional reward function jointly optimizing predictive power, diversity, and novelty. Experiments demonstrate significant improvements over state-of-the-art baselines in alpha quality, diversity, and pairwise decorrelation—establishing a scalable, interpretable paradigm for automated alpha discovery.

Technology Category

• Machine Learning: Structured Learning
• Search and Optimization: Learning to Search
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

The automated mining of predictive signals, or alphas, is a central challenge in quantitative finance. While Reinforcement Learning (RL) has emerged as a promising paradigm for generating formulaic alphas, existing frameworks are fundamentally hampered by a triad of interconnected issues. First, they suffer from reward sparsity, where meaningful feedback is only available upon the completion of a full formula, leading to inefficient and unstable exploration. Second, they rely on semantically inadequate sequential representations of mathematical expressions, failing to capture the structure that determine an alpha's behavior. Third, the standard RL objective of maximizing expected returns inherently drives policies towards a single optimal mode, directly contradicting the practical need for a diverse portfolio of non-correlated alphas. To overcome these challenges, we introduce AlphaSAGE (Structure-Aware Alpha Mining via Generative Flow Networks for Robust Exploration), a novel framework is built upon three cornerstone innovations: (1) a structure-aware encoder based on Relational Graph Convolutional Network (RGCN); (2) a new framework with Generative Flow Networks (GFlowNets); and (3) a dense, multi-faceted reward structure. Empirical results demonstrate that AlphaSAGE outperforms existing baselines in mining a more diverse, novel, and highly predictive portfolio of alphas, thereby proposing a new paradigm for automated alpha mining. Our code is available at https://github.com/BerkinChen/AlphaSAGE.

Problem

• Addresses reward sparsity in automated alpha mining for quantitative finance
• Overcomes inadequate sequential representations of mathematical expressions
• Solves the lack of diversity in non-correlated alpha portfolios

Innovation

• Structure-aware encoder using Relational Graph Convolutional Network
• Generative Flow Networks framework for automated alpha mining
• Dense multi-faceted reward structure for robust exploration