Research on Stock Volatility Prediction Using Attention Mechanism Enhanced GRU Ensemble Model
Mingjuan Li, Huimei Chen, Bingda Yan, Qingzhen Xu, Qiang Chen · 2024
This paper proposes an ensemble model integrating attention mechanism with GRU (Gated Recurrent Unit) to improve the accuracy of stock volatility predictions. The study initially utilizes the "Tushare" platform to obtain stock market data and calculates its volatility time series. Furthermore, by incorporating the attention mechanism and a bidirectional processing strategy, this paper constructs an enhanced GRU model. Experimental results indicate that this model surpasses some existing prediction methods in terms of prediction accuracy. The main contribution of this paper is to enhance the GRU model equipped with an attention mechanism by integrating various network structures to achieve more accurate and stable predictions of stock volatility.