Applications Of Deep Forest In Stock Index Trend Predicting And Investment Strategy

In recent years,the trend prediction of the stock price index is one of the most challenging problems for investors and researchers in financial time series forecasting.Because machine learning match the characteristics of financial time series analysis and have strong performance on classification and prediction,more and more machine learning methods are applied to the prediction of financial time series.This paper first briefly introduces the theory knowledge of four machine learn-ing models and based on the four models,constructs the stock price index trend prediction model.We select eight technical indicators as the input variables of the model based on previous research experiences,and applies a discretization method for input variables.The four stock price index trend prediction model are applied to predict the trend of CSI300 index.By comparing and analyzing the prediction performance of four models for different prediction frequencies,the va-lidity of the deep forest model is fully verified,based on discretization method for input variable.In the daily and weekly fluctuation forecasting of CSI300 index,the accuracies of the model are 59.78%and 67.11%,respectively,and the values of AUC are 0.5514 and 0.6429,respectively,showing a significant classification performance advantage.Finally,based on the deep forest prediction model,this paper constructs two trend timing strategies.The investment objects of the two strategies are the index fund and stock portfolio of CSI300 constituent stock.Both timing strategies have good performance during the backtest and are better than the market in terms of revenue,risk and risk-adjusted return of capital.The deep forest model based on discretization method for input variable proposed in this paper can obtain very satisfactory results in the stock price index trend prediction and investment strategy construction.It has practical application value and can provide reference and help for the majority of institutional investors and retail investors.