Title: A GRU-based hybrid global stock price index forecasting model with group decision-making

Authors: Jincheng Hu; Qingqing Chang; Siyu Yan

Addresses: School of Information Management, Shanghai Lixin University of Accounting and Finance, Shanghai 201209, China ' School of Information Management, Shanghai Lixin University of Accounting and Finance, Shanghai 201209, China ' School of Information Management and Engineering, Shanghai University of Finance and Economics, Shanghai 200433, China

Abstract: To predict the global stock price index daily more effectively, this study develops a new filtering gate recurrent unit group-based decision-making (FiGRU_G) model that combines GRU group network and group decision-making strategy. This proposed FiGRU_G model can effectively overcome the shortcoming of traditional neural network algorithms that the random initialisation of network weights may cause worse performance to some extent. The experimental results indicate visually the proposed FiGRU_G framework outperforms other competing methods in terms of prediction accuracy and robustness for both Chinese and international stock markets. In the short-term prediction scenario, the FiGRU_G framework achieves 20% and 19% performance improvements on evaluation criteria MAPE and SDAPE respectively compared with the GRU model in the Chinese stock market. For the international markets, this FiGRU_G framework also achieves 23% and 22% performance improvements respectively compared with the GRU model.

Keywords: stock closing price prediction; deep learning; GRU model; group decision-making.

DOI: 10.1504/IJCSE.2023.129153

International Journal of Computational Science and Engineering, 2023 Vol.26 No.1, pp.12 - 19

Received: 14 Dec 2021
Accepted: 31 Dec 2021
Published online: 23 Feb 2023 *