Stock movement prediction model based on gated orthogonal recurrent units

Abstract

Stock movement prediction has received growing interest in the deep learning community. However, the generalization ability of some existing prediction models is weak due to the highly stochastic property of stock market, and some models suffers from the problem of gradient explosion or gradient vanishing in the training process. To solve the above issues, in this paper, we propose a novel stock movement prediction model based on gated orthogonal recurrent units (GORU) and variational auto-encoder (VAE). Specifically, GORU encodes the text information, and then VAE infers and decodes the market information formed by concatenating encoded text information with normalized historical price information. Meanwhile, orthogonality introduced by GORU can alleviate the problem of gradient explosion or gradient vanishing and enhance the generalization ability of the model. We evaluate the relative contributions of text information and historical prices with respect to prediction accuracy by the results of an ablation study. The experimental results on publicly available datasets show that the proposed model is better than several state-of-the-art models, which indicates that the GORU and VAE can effectively improve the model's generalization ability and accuracy for predicting stock trends.