Comparison of Future Price Prediction in Context of Machine Learning Approach

Abstract

Contemporarily, futures markets have become prosper on account of its intrinsic risk hedging function as well as double-side trading advantages compared to Chinese stock market. In this case, finding the suitable and appropriate way to predict future price changes will be a crucial thing for investors and traders of commodity. In context of theoretical analysis in terms of summary of previous models, this article firstly selects features and market information of all trading days from 2016 to 2021 as input. The research sample is the CSI300 stock index futures, and 1-daylow-frequency data is selected. Three price prediction models were built in terms of the data, i.e., linear regression, random forest, and support vector machine algorithms. Accordingly, the random forest algorithm has the lowest error, high accuracy and stability in the prediction of stock index futures price. These results shed light on guiding further exploration of future price prediction based on specific machine learning approach.