Higher-order Graph Attention Network for Stock Selection with Joint Analysis

TL;DR

This paper introduces H-GAT, a novel higher-order graph attention network that captures complex stock relations and integrates fundamental and technical analysis factors, improving stock prediction performance.

Contribution

The paper presents H-GAT, the first model to jointly incorporate higher-order graph structures and both fundamental and technical factors for stock selection.

Findings

H-GAT outperforms baseline models in profitability tests.

H-GAT achieves higher Sharpe ratios on NASDAQ and NYSE datasets.

The model effectively captures complex stock relations.

Abstract

Stock selection is important for investors to construct profitable portfolios. Graph neural networks (GNNs) are increasingly attracting researchers for stock prediction due to their strong ability of relation modelling and generalisation. However, the existing GNN methods only focus on simple pairwise stock relation and do not capture complex higher-order structures modelling relations more than two nodes. In addition, they only consider factors of technical analysis and overlook factors of fundamental analysis that can affect the stock trend significantly. Motivated by them, we propose higher-order graph attention network with joint analysis (H-GAT). H-GAT is able to capture higher-order structures and jointly incorporate factors of fundamental analysis with factors of technical analysis. Specifically, the sequential layer of H-GAT take both types of factors as the input of a long-short term memory model. The relation embedding layer of H-GAT constructs a higher-order graph and learn node embedding with GAT. We then predict the ranks of stock return. Extensive experiments demonstrate the superiority of our H-GAT method on the profitability test and Sharp ratio over both NSDAQ and NYSE datasets