| Essentially,security investment is a behavior choice made by people influenced by sentiment.Therefore,sentiment of investors deserves high attention and consideration in financial asset pricing and return analysis.Traditionally,security issues and assets pricing theories are analyzed in scopes of macro economy,financial analysis,etc.However,lots of financial anomalies in stock market cannot be explained with traditional models.As a result,behavioral finance combined with sociology,psychology and cognitive behavioral science showed up under the mutual study of domestic and foreign scholars.Compared with traditional models,behavioral finance gives out a better explanation for economic problems such as herding effect,by analyzing the mechanism and channels of the effect of investors’ sentiment and behavior on the future trend in stock market.With the development of the Internet,Internet media,such as blog,microblog and Internet forum,has become the main way to express and transmit information.Individuals release and accept all kinds of information at the same time.Internet media greatly enhances the possibility of individual sentiments and behaviors to develop into group sentiments and behaviors.In financial market,sentiments and behaviors of tens of thousands of investors are related and mutually influenced over the Internet,leading to fluctuation of assets prices.In addition,the text information on the Internet media is a direct expression of investors’ emotion and truly reflects investors’ viewpoints and sentiments.The gradual maturation of related theories and technologies about text mining makes it possible to mine sentiment information of investors on the Internet media in this paper.Under the theoretical framework of behavioral finance,this article mainly studies the effect of sentiments of investors on stock price.In traditional assets pricing models,the three-factor model developed by Fama and French,Nobel laureates in economics in 2013,has a great influence.Later on,plenty of scholars and securities practitioners studied,applied and perfected this model.Depending on previous studies,this paper adds sentiments of investors into this model as the fourth factor and keeps other factors as control variables to analyze how sentiments of investors influence stock prices.Considering the great influence of Internet media,this paper adopts the method of text mining,applies the text mining packages Urllb2 in Python2.7.11 software,and collects related text information from stock reviews in Sina Finance.Appling the text mining software ROST,I abstract information of emotion trend among investors and then build an sentiment index to quantify investors’ sentiments.This paper has following innovations: Firstly,this article directly mines information of sentiments of investors from Internet,lowering errors due to indirect data and improving temporal effectiveness and accuracy of investment analysis and quantification transactions.Secondly,this paper applies sentiment index made by text mining to price stock portfolios,breaking through the method of purely studying the relationship between investors’ sentiment and market index.Additionally,this article expands traditional Fama-French three-factor model from the viewpoint of behavioral finance,making it more accurate to select and construct portfolios.This paper comes to a conclusion: Compared with traditional three-factor model,the four-factor model,including investors’ sentiments built by text mining and based on behavioral finance,better accounts for the returns fluctuation of stock portfolios,especially for small-cap stocks and growth stocks.However,this model cannot accurately explain the return fluctuation of large-cap stock portfolios.In the prediction of data beyond samples,four-factor model including investors’ sentiments makes a better prediction than traditional model in the direction of stock returns.It turns out that sentiment factor is an indispensable pricing factor and deserves high attention and study. |
PDF Full Text Request