Font Size: a A A

A Comparative Study On The Effectiveness Of Investor Emotion Based On Text Analysis And Principal Component Analysis

Posted on:2020-07-08Degree:MasterType:Thesis
Country:ChinaCandidate:Z P WangFull Text:PDF
GTID:2429330572466702Subject:Finance
Abstract/Summary:PDF Full Text Request
Investors' irrational investment behavior is an important reason for the deviation of market price from value and the increasing deviation,which leads to false prosperity in the period of inflation and excessive panic in the period of collapse.Studying the effect of investor sentiment on the stock market can prevent the risk of the stock market,which has great practical significance.This paper reviews the background of the emergence and development of investor sentiment theory,the research on the relationship between investor sentiment and stock market,the evolution of measuring methods of investor sentiment,and the testing methods of the effectiveness of investor sentiment indicators.It is found that in the current research on investor sentiment,we use Baker's and Wurgler's methods for reference and adopt principal component analysis.Analytical method is still the mainstream method to construct investor sentiment indicators,and more and more researchers use network text to construct investor sentiment indicators directly measured.In the past,most of the research focuses on the relationship between investor sentiment and stock market,but there are few studies on the effectiveness of investor sentiment index itself.Firstly,this paper introduces how to construct the direct measure of investor sentiment index based on text analysis and indirect measure of investor sentiment index based on principal component analysis.When constructing investor sentiment based on text analysis method,the source of text acquisition,the method of text processing and the steps of emotional analysis to construct sentiment indicators are introduced in detail.When constructing investor sentiment based on principal component analysis,this paper introduces in detail the selection of original sentiment indicators,data sources and processing methods,and the concrete steps of constructing indirect composite investor sentiment by principal component analysis.After getting two kinds of investor sentiment indicators,this paper analyses the advantages and disadvantages of investor sentiment indicators from several angles.Starting with the traditional qualitative analysis method(correlation analysis with return rate,Granger causality test,etc.),this paper makes quantitative analysis based on AH stock premium index test.AH stock premium is a phenomenon of price difference between twin stocks.This paper introduces the phenomenon of price difference between twin stocks and their prices,and uses AH stock premium index to test the effectiveness of investor sentiment indicators.The empirical results show that although the investor sentiment index based on principal component analysis performs better in Granger causality test,the investor sentiment index based on text analysis is better in the test of the relationship between the more important economic significance and Shanghai Stock Index and the test based on AH stock premium index.Therefore,it is concluded that the direct measure of investor sentiment index based on text analysis is better than the indirect measure of investor sentiment index based on principal component analysis.
Keywords/Search Tags:Investor sentiment, principal component analysis, text mining, ah premium index
PDF Full Text Request
Related items