Research On Investment Value Based On Textual Information Of Firms’ Announcement Titles

With the rapid development of computer science and technology,the frontier computer technology promotes innovation of fintech,which significantly expands the data used for empirical financial research.Financial text data has become the data that financial scholars can analyze.Among the numerous financial text data,the announcement of listed companies is an essential approach for listed companies to broadcast their operating information,covering a more comprehensive range of investors.Therefore,it is of great significance to study the changes in stock returns via the announcement text information.Based on the text data of listed companies from 2000 to 2020,this paper uses the text analysis method to classify the announcement’s title and constructs the sentiment dictionary.Based on the information contained in the announcement,this paper proposes the quantitative investment strategy for empirical analysis,evaluates the effectiveness of the trading strategy,explores the information value of the announcement of listed companies,and furtherly studies the investors’ response to the announcement information and the changes in the market.Based on disclosure items in company announcements,the announcements are divided into 61 categories,like general meeting of shareholders,board announcement,intermediary announcement and so on.We segment the announcement title and construct the emotional dictionary.After improving the jieba word segmentation using the financial dictionary,the words are combined with stock returns to determine its words sentiment.A rolling annual update emotional dictionary of announcements is constructed.Finally,this paper proposes three investment strategies based on the above two different processing of announcement information: the strategy based on announcement classification,the strategy based on announcement text sentiment,and the strategy based on the combination of announcement category and sentiment.The specific conclusions are as follow:(1)Information contained in company announcements comprise economic value,and the three strategies based on the announcement information obtain statistically significant excess returns;(2)There are significant differences in the earnings of different types of announcements.Bullish announcements have excellent performance in stock returns,and the returns are positive,while bearish announcements have terrible performance,and the returns are negative.(3)Announcement sentiment dictionary can clearly distinguish the direction of the announcement,the annualized stock returns of positive words are 39.23 %(=5.17),and the annualized stock returns of negative emotions are-2.17 %( =-0.29);(4)Among different types of announcements,sentiment dictionary has different distinguishing ability,especially in uncertain announcements,the positive and negative words of the sentiment dictionary will distinguish such announcements and obtain apparent benefits,and the positive and negative words cannot affect the essential common judgment of investors’ bullish and bearish announcements.