Research on the selection of stock prediction models

Abstract

Against the backdrop of increasing attention to the integration of machine learning and stock analysis, stock prediction models are a hot topic. The question this paper is studying in this study is which stock prediction model is more accurate in predicting stocks. The method of this study is based on the stock prices of new energy vehicle leader Tesla Motors in the past three years as a data set, using a random forest model and an SVR model to predict the stock prices over the next 10 days. Based on the parameter MSE values of the training models of two stock prediction models, compare their sizes to determine the accuracy and stability of the models. This study found that the stock prediction results of the SVR model are more accurate and stable than those of the random forest model. Therefore, it is believed that the stock prediction model using the SVR method will have more market value and occupy an important position in the integration of machine learning and stock trading analysis.