FinMamba: Market-Aware Graph Enhanced Multi-Level Mamba for Stock Movement Prediction

2025-02-10

AI Summary

Financial time series exhibit low signal-to-noise ratios and high dynamics, rendering existing methods incapable of adaptively modeling evolving inter-stock topological relationships driven by market conditions, while suffering from inefficient long-range dependency capture and excessive memory consumption. To address these challenges, we propose a novel framework integrating dynamic graph pruning with multi-level selective Mamba architectures. First, we introduce a market-aware dynamic graph pruning mechanism that adapts the stock interaction topology in real time. Second, we design a Mamba variant featuring multi-granularity selective state resetting and cross-scale pattern retrieval, enhancing long-horizon modeling capability while preserving linear-time inference complexity. Evaluated on U.S. and Chinese equity datasets, our method achieves state-of-the-art forecasting accuracy and demonstrates statistically significant profit improvement in live backtesting. Moreover, it reduces GPU memory usage by 42%, striking an optimal balance among prediction accuracy, computational efficiency, and deployment feasibility.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Recently, combining stock features with inter-stock correlations has become a common and effective approach for stock movement prediction. However, financial data presents significant challenges due to its low signal-to-noise ratio and the dynamic complexity of the market, which give rise to two key limitations in existing methods. First, the relationships between stocks are highly influenced by multifaceted factors including macroeconomic market dynamics, and current models fail to adaptively capture these evolving interactions under specific market conditions. Second, for the accuracy and timeliness required by real-world trading, existing financial data mining methods struggle to extract beneficial pattern-oriented dependencies from long historical data while maintaining high efficiency and low memory consumption. To address the limitations, we propose FinMamba, a Mamba-GNN-based framework for market-aware and multi-level hybrid stock movement prediction. Specifically, we devise a dynamic graph to learn the changing representations of inter-stock relationships by integrating a pruning module that adapts to market trends. Afterward, with a selective mechanism, the multi-level Mamba discards irrelevant information and resets states to skillfully recall historical patterns across multiple time scales with linear time costs, which are then jointly optimized for reliable prediction. Extensive experiments on U.S. and Chinese stock markets demonstrate the effectiveness of our proposed FinMamba, achieving state-of-the-art prediction accuracy and trading profitability, while maintaining low computational complexity. The code is available at https://github.com/TROUBADOUR000/FinMamba.

Problem

• Adaptively capture evolving stock interactions
• Extract patterns efficiently from historical data
• Maintain low computational complexity in predictions

Innovation

• Dynamic graph learns inter-stock relationships
• Multi-level Mamba discards irrelevant information
• Linear time costs for historical pattern recall