Abstract:

Aiming at the collinear and nonlinear characteristics of stock data, a hybrid forecasting model based on Convolutional Neural Network（CNN） and Gated Recurrent Unit（GRU） neural network is proposed to predict CSI 300 Index, SSE Composite Index and SZSE Component Index. Firstly, this model uses CNN to extract feature vectors and reduce the dimension of original data. Then, it utilizes GRU neural network to learn the dynamic changes of features and predict the stock index. The simulation results show that compared with GRU neural network, Long Short-Term Memory（LSTM） neural network and CNN, this model can mine the information contained in historical data, effectively improve the accuracy of the stock index forecasting, and provide some reference value for the stock index trading.

Key words: stock index forecasting, Convolutional Neural Network（CNN）, gated recurrent unit neural network

摘要：

针对股票数据共线性和非线性的特点，提出了一种基于卷积神经网络（Convolutional Neural Network，CNN）和门控循环单元（Gated Recurrent Unit，GRU）神经网络的混合预测模型，并对沪深300指数、上证综指和深证成指进行了预测。该模型首先采用CNN提取特征向量，对原始数据进行降维，然后利用GRU神经网络学习特征动态变化规律进行股指预测。仿真结果表明，与GRU神经网络、长短时记忆（Long-Short-Term Memory，LSTM）神经网络和CNN相比，该模型能够挖掘历史数据中蕴含的信息，有效提高股指预测的准确率，并可为股指交易提供一些参考。

关键词: 股指预测, 卷积神经网络（CNN）, 门控循环单元神经网络