Prediction Models Of Stock Return Rate Based On Gradient Lifting Regression Tree And Its Applications

Financial market plays a pivotal role in modern society.Its good operation has positive significance to the stable growth of national economy.And financial market is widely favored by the public as an investment channel.However,there are many factors that affect its change,such as macro factors(policies,GNP,etc.)and micro factors(company operation,public opinion,etc.).Therefore,it is very necessary to use artificial intelligence method to build a scientific model for related research.This thesis selected the trading index(the highest and the lowest price,closing price,trading volume and turnover rate,etc.)and technical indicators(moving average,exponential moving average momentum,relative strength index,index,index,etc.)as the explained variable,and selected short(15),medium(30,60 days)and long-term(90)yields as explanatory variables;Secondly,two different models were constructed: the fusion model(clustering and gradient lifting regression tree model)and the single model(gradient lifting regression tree model based on moving window).Finally,two different models are used to predict the stock market and simulate investment.It is found that the prediction effect of the model is good in the short and medium term.The gradient lifting regression tree model based on moving window(MW-GBRT)is superior to the clustering and gradient lifting regression tree model(K-Means-GBRT).The prediction accuracy is more than 70% and the is less than 0.005.And the simulated investment under two different investment strategies has obtained good returns.