Research On IPO Earnings Based On Data Mining

As an important component of financial market,stock market has played a huge role in the process of social and economic development.Along with the rapid development of the IPO,the number of new shares issued by the GEM is increasing fast.The probability of investors "getting new shares" is gradually increasing,at the same time the IPO continuous limit days decrease and the time of opening the daily limit is in advance.The situations of a dozen or even twodozen daily limits after the IPO in the past are gradually reducing.Some new shares will seal after opening the daily limit,while some of them will be marginalized and the revenue will also fall.Therefore,although the probability of playing the new ballot has been improved,the income has been reduced.Therefore,the analysis of the factors affecting the expected return of IPOwill help to improve the yield of IPO,especially by mining the emotional data of investorsand clarifying the relationship between IPO yield and investor sentiment.The traditional study on stock returns forecast usually starts from the technical index of the stock market,but the noise factors like investors emotion are also indirectly reflected by the indexes.So the stock prediction can usually not get a particularly good effect.Micro-blog,as a social platform based on sharing,disseminating and acquiring information,broadcasts short and real-time information by the attention mechanism.There is a lot of information about the stock market in micro-blog.Taking this kind of micro-blog information as the research example,it is of practical significance and application value to analyze the emotional tendency.This paper will focus on the emotional analysis.Emotional tendency classification,also known as opinion mining or sentiment classification,can be understood as a user to express its views aboutone object like support,oppose or neutral.That is often said positive emotion,negative emotion and neutral emotion.This paper first captures the related contentsof well-known stock bloggers and its followers,and constructs the financial field emotion dictionary to mining sentiment which can reflect sentiment.Then,we buildemotional index MSI and investor sentiment coverage MC to combine these two indicators with the listed company related technical indicators such as market characteristics,market control variables.Finally,we analysisthe first day underpricing of IPO of the selected 540 listed companies in the gem of multiple regression analysis.Then,we put forward SI-SVM model and MS-SVM model to forecast IPO underpricing rate.The study found that emotional index and emotional coverage are of positive relevance with IPO underpricing rate.When the positive emotional tendency is more serious,the IPO underpricing rate is high,the IPO returns become lower;when the negative sentiment tendency is more serious,he IPO underpricing rate is low,the IPO returns become higher.