Forecasting Stock Prices with Machine Learning: A Practice in China A-Share Market

A supervised machine learning method, Support Vector Machine (SVM) has been employed in this study to predict the stock prices of Kweichow Moutai (Maotai) and Contemporary Amperex Technology (CATL) in China by using daily trading data, macroeconomic indicators and events, seasonal impact as well. For each stock, three models were developed based on different features included: (1) stock-specific features only, (2) adding macroeconomic factors, and (3) including seasonal indicators and global events. Evaluation metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Direction Accuracy were used to assess model performance. Results indicate that including macroeconomic factors improved predictive accuracy for CATL only.