Application of LSTM-CNN model to forecast the rise and fall of CSI 300

Stock price prediction has always been a focal issue in the field of quantitative finance. By analyzing historical data related to stock prices, researchers attempt to infer the patterns of stock price changes in order to achieve excess returns. However, due to the influence of multiple factors, traditional linear forecasting models often struggle to provide effective predictions. In recent years, non-linear predictive models using machine learning and deep learning methods have been widely used in quantitative investment. Compared to linear models, these models possess stronger predictive performance, learning ability, and adaptability. This paper conducts an empirical analysis from the perspectives of individual models (CNN model, LSTM model) and the combined CNN+LSTM model. It aims to predict the change of the stocks in the CSI 300 Index and compares the accuracy of their predictions. The results demonstrate that the combined CNN+LSTM model outperforms the individual models in predicting the directional movements of stocks in the CSI 300 Index.