GrifFinNet: A Graph-Relation Integrated Transformer for Financial Predictions

2025-10-11

AI Summary

Predicting stock returns poses dual challenges: modeling spatiotemporal dynamics and integrating heterogeneous relational information. To address these, we propose a multi-relational graph-enhanced spatiotemporal Transformer. First, we construct a dual-channel stock graph encoding industry membership and institutional ownership. An adaptive gating mechanism then dynamically fuses structural information from these relations. Subsequently, the resulting graph embeddings are jointly encoded with temporal features in a modified Transformer architecture to jointly model spatiotemporal dependencies. This design enhances relational awareness and model interpretability. Experiments on CSI 300 and CSI 500 constituent stocks demonstrate that our model significantly outperforms baselines—including GAT, TimesNet, and Graphormer—in directional accuracy and MSE. Moreover, it quantifies the contribution of each relational edge to predictions, yielding interpretable insights into market behavior.

Technology Category

• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Graph-based Machine Learning

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Predicting stock returns remains a central challenge in quantitative finance, transitioning from traditional statistical methods to contemporary deep learning techniques. However, many current models struggle with effectively capturing spatio-temporal dynamics and integrating multiple relational data sources. This study proposes GrifFinNet, a Graph-Relation Integrated Transformer for Financial Predictions, which combines multi-relational graph modeling with Transformer-based temporal encoding. GrifFinNet constructs inter-stock relation graphs based on industry sectors and institutional ownership, and incorporates an adaptive gating mechanism to dynamically integrate relational data in response to changing market conditions. This approach enables the model to jointly capture spatial dependencies and temporal patterns, offering a comprehensive representation of market dynamics. Extensive experiments on two Chinese A-share indices show that GrifFinNet consistently outperforms several baseline models and provides valuable, interpretable insights into financial market behavior. The code and data are available at: https://www.healthinformaticslab.org/supp/.

Problem

• Predicting stock returns using deep learning techniques
• Capturing spatio-temporal dynamics in financial markets
• Integrating multiple relational data sources for predictions

Innovation

• Combines multi-relational graph modeling with Transformer encoding
• Constructs inter-stock graphs using sector and ownership data
• Uses adaptive gating to dynamically integrate relational information