Analysis Of Public Influencing Factors Of Sock Return Based On Approxinate Factor Model

In modern market economy conditions,the financial sector is one of the most risky and competitive sectors.As a display of the national economy,the stock market is closely related to the development of the macro economy.With the improvement of people’s income level and the strengthening of national financial management awareness,more and more people are involved in stock investment.As a highly profitable investment tool,stocks have attracted great attention from investors.Due to the rapidly changing stock market,precise analysis is difficult.Therefore,it is of great research value and practical significance to find the public factors that affect stock returns,reduce the risks faced by investors in the stock market,and enable investors to obtain higher investment returns on the basis of minimizing risks.Since the stock return data is a matrix time series observed over time,in order to separate the dependencies between each time series in the matrix observations and the matrix factors,this paper uses standard vector time series data,making full use of Based on existing information,we explore the relationships between a set of time series driven by common factors,and use these common factor information for forecasting to grasp future trends.This paper selects the return of common stocks in A shares of Shanghai Stock Exchange from January 2003 to May 2021 as the research object,analyzes the common factors that affect the stock return by establishing an approximate factor model.The empirical analysis is divided into three parts: First,the number of factors is preliminarily observed through the eigenvalue rubble plot,a total of 8 types of PC criteria,RE criteria,ER criteria,GR criteria,TCR criteria,RRE criteria,ED criteria and BIC criteria are selected.The estimated number of factors is obtained by the estimation method.Considering the actual situation of this paper,the number of factors is selected as 3,and the cumulative variance contribution rate is 83%.Then,the load matrix and common factor vector of the approximate factor model are estimated,and the trend between common factors and stock returns is analyzed.The second is to establish a vector autoregression model to further analyze the relationship between public factors and stock returns.Through the model,it is found that three public factors have a negative correlation with stock returns.Finally,the stock return is predicted by the linear regression model,and the data of the average stock return is divided into 70%training set and 30% test set,using the training set to learn the linear regression model and obtain the relevant parameters of the model,then carry out the test set.Prediction,after obtaining the predicted value,calculate the accuracy rate and the average error rate.The real sequence and fitting sequence of stock returns of two listed companies are drawn,and the results show that the model fitting effect is better,which can help investors make better decisions in the investment process.