A Stock Prediction Method based on Futures Data

Stock prediction has always been the focus of attention and difficulty in the field of finance, this thesis selects the stock data of the precious metals sector in the past ten years, and at the same time obtains the futures data related to the precious metals sector, and makes a prediction on it. We used LSTM and various machine learning models. In the regression prediction, basic stock indicators, market sentiment scores, stock technical indicators, and futures indicators are considered, and it is found that the inclusion of futures indicators reduces the MSE by more than 0.0001 compared to the other indicators, and reduces the MSE by more than 0.0003 compared to the other models, and the confidence is significant. This thesis illustrates that the inclusion of futures data in the consideration of stock forecasting can significantly improve the accuracy of stock forecasting.