Internet Financial Products Multi-dimensional Investor Sub-Sentiments And Market Performance Based On WeChat Text Mining

Internet finance combines the penetrating and rapid characteristics of the Internet on the basis of the nature of finance,breaking the traditional financial model(Xie et al.,2012;Luo and Lin,2013;Zavolokina et al.,2016;Arthur,2017).It is reflected in the booming new financial format of Internet financial products.Internet financial products make full use of Internet big data,artificial intelligence,machine learning and other technologies combined with traditional financial products,relying on low thresholds,wide variety,good liquidity,short redemption period(Barasinska and Schafer,2014;Li Kemu,2016;Chen et al.,2016;Guo et al.,2016;Yermack,2017;Tang,2019;Teply et al.,2019),is a financial product that better meets the needs of investors and constantly challenges traditional financial products.Taking China's Internet financial products market as an example,in 2013,Ant Financial Group launched the financial product “Yuebao”,which was marked as the first year of Internet financial products.The “Internet Finance Index Report” released by the National Finance and Development Laboratory showed that the Internet The financial index increased from 100 points in 2013 to 563 points in 2018,and increased nearly six times in five years.However,with the development of China's Internet financial products market,the inherent risks of the Internet financial product platform itself are gradually exposed.According to the statistics of online loan homes,in recent years,China's Internet financial platform has experienced frequent risk events,especially in the short two months of July and August 2018,as many as 218 Internet financial platforms have run and invested.There are many risk events such as concentration and cash withdrawal(Collier and Hampshire,2010;Huang et al.,2015;Liu et al.,2018;Chen et al.,2019).Moreover,the above literature also pointed out that it is difficult to fully explain the risk and return characteristics of Internet finance by using empirical conclusions based on traditional financial markets.A large number of documents also confirm that individual investors have different professional level differences and relatively limited information in financial markets,and their investment decisions are easily influenced by sentimental state and background factors.Similarly,in the China's Internet financial products market,the proportion of individual investors is large and the spread is rapid.The investors of Internet financial products are more susceptible to sentiments,and the sentiments of Internet financial investors are more likely to spread,resulting in product return abnormal fluctuations.Therefore,when analyzing the financial risk events generated by the Chinese Internet financial products market,it is necessary to combine the theory of investor sentiment behavior.In the current study,we describe the characteristics of the Internet to indirect financial investor sentiment index method,but the drawbacks of indirect indicator approach is to index data for the monthly or even annual data can not accurately describe the instant fluctuations in investor sentiment.The time-varying characteristics of investor sentiment do not match.Secondly,there is another part of the research,using text mining method,trying to use the latest news text data,microblog data,etc.to build Internet financial investor sentiment.However,the existing literatures all describe the investor sentiment based on Internet finance from a single dimension,and the sentimental state information of other dimensions of investors may be difficult to reflect.In addition,whether using indirect index method or text mining method,the Internet financial investor sentiment has neglected the multi-level sentimental state of sentiments,and the investor sentiment of different sentimental states such as happy,anger,sad,good,etc.Therefore,how to accurately describe and deeply portray the investor sentiment of Internet financial products,that is,analyze the impact of investor sentiment on the Internet financial product market and Internet financial products from a deep and multi-dimensional perspective,which needs to be resolved.This paper considers the two unique advantages of the We Chat public number users and the better information quality.The application of text mining in the We Chat database is selected from May 1,2017 to January 18,2019,including "Internet wealth financial products,online loans,online lending,P2 P,P2P online loans,network wealth financial products,online financial products." We Chat articles with keywords,a total of 35,419 articles on the original data of We Chat articles on Internet financial products,processed to obtain daily Internet financial products investor of Happy,Good,Angry,Sad,Fear,Bad,Calm and Amazing data total 6290 article.At the same time,combined with the sentimental dictionary in the field of Internet financial products constructed in this paper,based on the three deep information of positive class,negative class and comprehensive class,we build Multi-dimensional perspective of Internet financial products investor sentiment.And select the first online loan.As a characteristic variable of the market return of Internet financial products,the comprehensive return index of Internet financial products systematically explores the interactive relationship between the sentiment of Internet financial products and the market returns of Internet financial products.The empirical results show that: different from the above-mentioned conclusions of the single dimension of the comprehensive Internet financial products investor sentiment,1)Investors' Sub-Sentiments of Good,Happy,Fear,Calm and Bad of Internet financial products are the reasons for the changes in the comprehensive return index of Internet financial products.2)Investors' Sub-Sentiments of Happy,Angry,Sad,Fear,Calm,Bad,Amazing will increase market risk,leading to a decrease in expected returns.The rest of the Good sub-sentiments are in line with the principle of “the greater the risk,the greater the returns.” This phenomenon is consistent with psychological cognition,that is,when people evaluate something they like,they think it is “high-returns and low-risk,” but things that they do not like are considered low-returns and high-risk 3)Only the Good sub-sentiment has leverage effects on returns of Internet financial products and Fears,Angry,Amazing sub-sentiment have volatility spillover effects.,which reflects the characteristics of Internet financial products,and indicates that it is not accurate to barely consider the comprehensive sentiment to describe the relationship comprehensive return index of Internet financial products,Therefore,this study can deeply and multi-dimensionally analyze the co-movement between the Internet financial products investor sentiments and the Internet financial products market by accurately describing and deeply portraying the high-quality Internet financial products investor sentiments.