Application Of Machine Learning Based On Investor Sentiment In Stock Market Prediction

The stock market is often called the barometer of economy.The financial market undoubtedly has an important influence on our daily life.Therefore,the stock market is also a research field that can not be ignored in the academic circle.From the earliest Dow Theory to the later K-line chart analysis and technical index analysis,now with the support of machine learning,the research of quantitative investment has been widely accepted by the market.In this paper,NLP sentiment analysis is used to establish public opinion indicators in the periodic reports of listed companies,together with the fundamental indicators and technical indicators which are used to predict the market's stock price trend in the next few days through modeling.The main work of this paper includes the following two aspects:First,this paper selects all Shanghai Stock Exchange 50 constituent stocks from January 2011 to November 2019,and extracts emotional scores from the research reports through NLP emotional analysis technology.The research reports are crawed from Sina Finance and economics Website.Different from the microblog comments or stock bar messages which are now widely researched in the market,the advantage of using the Research Report of securities companies is that their views represent the institutional sentiment with stronger capital strength.Secondly,after obtaining the emotional score,together with the fundamental indicators and technical indicators such as ROE,ROA,MACD,KDJ,we use SMV and neural network algorithm to build a model forecasting future share price.