Index Tracking Problems And Empirical Analysis Based On Regression Models

Index tracking is a method of tracking a market index trend with a small number of stocks,which is a kind of passive portfolio management strategies and rapidly develop in our country recently years.Now many fund companies and investors make use of index tracking to invest.So there are more and more attention paid by people in how to effectively track the market index,and many of them are scholars interesting it.Scholars hope to find a proper method from the historical data of the market index and stocks,and make the tracking performance more better,then provide more choices for fund companies and investors by this.This paper establishes the linear regression model and the quantile regression model of the index tracking,and then obtains three sample space through the three stock selection methods,finally make empirical analysis for the model in the three sample space.Because the linear regression model has many conditions,and the parameters are easily affected by the extreme values.In order to avoid these problems,we propose a quantile regression model.Specifically,this paper selects the SSE 50 index in the first half of 2016 as the target index,and constructs two index tracking model and gives the constraints,then respectively select the stocks on the space of stock,finally gets the better tracking effect among those tracking portfolios by fitting the real data,and forecasts the gains of the next 30 days,then compares with the SSE50 return.The main conclusions are got as follows:(1)Among the three approaches,the maximum correlation coefficient method has the best stock tracking effect and the K value of it is as small as possible;(2)Comparing the two models,the linear regression model is better than on the aspect of tracking effect,but the quantile regression model can get more returns than the linear regression model;(3)The changes of K value has different effect on the models in the different stock selection method.