Construction Of Investor Sentiment Index Based On Text Mining And Its Application In Quantitative Investment Strategy

With the rapid development of the Internet and the rise of social media,stock bar,as an important information exchange platform,can not only provide investors with a lot of valuable information,but also become a digital platform for investors to express their views and emotions.When sudden events occur in the capital market,there will be a large number of investor comments in the comment section of the stock bar.Most of these comments have strong subjective emotions,and the aggregation of these emotions will directly affect the investors’ strategy and the stock price.Since 2018,the international political and economic situation has been facing increased uncertainties,and sudden events such as trade frictions,COVID-19 and the Russia-Ukraine war have occurred frequently.Stock bar comments have become the hot spot of market public opinion.Therefore,it is very important to pay attention to the investor sentiment of stock bar for the construction of investment strategy and the supervision of public opinion.This paper takes the event-driven market as the research object,constructs the investor sentiment index by improving the sentiment dictionary,and compares and analyzes the investment strategy before and after the improvement of the sentiment dictionary.Based on the rare earth export event in the trade war between China and the United States in 2019,first of all,this paper selects 10 stocks related to this event as the research object,and uses Python to climb the comment post of each stock in the Oriental Wealth Stock Bar as the sentiment text.On the basis of the sentiment dictionary of Taiwan University,it improves the sentiment dictionary and constructs a new sentiment index.Secondly,by using the moving average strategy of "sentiment factor + bullish indicator",we compare the returns of the investment strategy before and after the improvement of the dictionary.The results show that:(1)the emotional factors constructed by the improved dictionary have higher benefits in the application of the strategy;(2)The comparison strategy with or without emotion index showed that Sharpe ratio and return rate were significantly improved by adding emotion factor;(3)In order to verify the robustness of the algorithm,we replace the sample events for testing,and the results show that: the emotion factor constructed by the improved dictionary can achieve better strategy effect.The innovations of this paper are as follows: First,in terms of dictionary innovation,the improved sentiment dictionary has a better backtest effect.This paper improved the NTU Dictionary by referring to the words such as comment post,research newspaper and news,and then calculated the emotional score of the emotional text by using the form of cumulative score.The research method has been further expanded than the previous research.Second,in the aspect of strategy research,this paper introduces the sentiment indicator as the object of strategy observation,and constructs the moving average strategy of "sentiment factor + bullish indicator".Through the comparative analysis of different dictionaries and strategies before and after the presence or absence of emotional indicators,the research Angle is more comprehensive and the research content is more abundant.