Research On Prediction Of Shanghai Composite Index Based On Decision Tree Algorithm

With the development of China's financial industry,the factors affecting the A-share market are increasing,and the traditional research methods are getting worse and worse in price forecasting.The advent of the artificial intelligence era has enabled us to have more advanced tools to study capital market.This paper uses machine learning related algorithms to explore the operating rules behind the A-share market,and provides suggestions for investors in investment decisions.This paper applies the classical decision tree algorithm in machine learning to the ups and downs of the Shanghai Composite Index.First of all,this paper constructs a large class factor library of explanatory variables from the fundamental,technical and market emotional aspects,and also creatively uses the data of the dragon and tiger to construct the institutional bullish sentiment factor.Secondly,this paper makes a single factor analysis of each subclass factor in the factor library,including the tertile method t statistic test and the single factor strategy back test,and screens out the seven explanatory variables that are most significant for the Shanghai Stock Index's ups and downs.Then,based on the CART algorithm in the decision tree,this paper constructs a trading strategy based on static decision tree and dynamic decision tree by using the above seven explanatory variables,and compares this strategy with the buy-and-hold strategy.Finally,based on the strategy based on dynamic decision tree,this paper analyzes the sensitivity of tree depth.In this paper,three conclusions are drawn: Firstly,the SSE exponential prediction model based on the decision tree algorithm in machine learning has strong predictive ability,and the model can significantly obtain excess returns.Secondly,the dynamic decision tree model is compared with the static decision tree model.There is a stronger predictive ability.Thirdly,as the depth of the decision tree increases,the predictive power first increases and then decreases.