Latest revision as of 08:06, 10 December 2024

Abstract

Stock price fluctuations reflect market expectations for the economic situation and company profits. Accurately predicting stock prices has become a hot topic in academia. With the rapid development of artificial intelligence, many researchers are starting to use machine learning algorithms to predict stock prices. In this paper, a new time series prediction model, the combination of the convolutional neural network and long shortterm memory neural network with additive attention mechanism (CNNLSTM-AAM), is proposed for stock price prediction. It can combine the advantages of the convolutional neural network (CNN), the long shortterm memory (LSTM) neural network, and the additive attention mechanism (AAM), and better capture nonlinear features of time series data. In the simulation analysis, we select sample data of three stocks (Vanke A, Shanghai International Port Group, and China Merchants Bank) and three stock price indexes (China Securities 500 Index, Shanghai Stock Exchange 50 Index, and Growth Enterprise Index) in the Chinese stock market for comparative analysis, and use mean square error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) as the evaluation indexes. The CNN-LSTM-AAM model has the best prediction ability relative to the CNN and CNN-LSTM models. In addition, we also find that the prediction ability of the CNN-LSTM-AAM model for different stock data sets is different under the existing parameter conditions. For a specific dataset, the parameter conditions of the CNN-LSTM-AAM model need to be further adjusted to achieve the best prediction effect. Based on the above findings, the CNN-LSTM-AAM model has better performance and higher accuracy, and can provide credible decision-making basis and research methods for investors, financial institutions, and regulators.

Full document