Stock Market Trend Forecast Based On Internet Reviews

As the network gets a rapidly popularity in China, and the amount ofinformation and propagation velocity are other sources can not match, the internetbecome one of the most important source of information. The network is also animportant distribution center for information of the financial sector, especially withWEB2.0technology development, forums, blogs, chat rooms and other interactivetechnologies are emerging, so that investors can participate in the process ofnetwork information creation, dissemination and access. Stock forum is one of themost popular online community, a large number of investors in the stock forumexchange information and share their experiences to assist investment decisions.Accessing to information in the online forum is an important way of understandinginvestor psychology and behavior.Compared to the centuries-old mature financial markets abroad, China’sfinancial markets still in development stage.The regulatory system is imperfect, andthe majority in the market are speculators whose behavior is susceptible to thepsychological impact. Get psychological and behavioral information of investors inthe stock forum, and study their mutual influence on the stock market has practicalsignificance.Behavioral finance theory believe that the stock market will be subject topsychological and behavioral impact of irrational investors. Our study on China’sstock market will base on this theory. This paper presents a method of automaticallyremoving the domain-independent reviews, which successfully removes84%of thestock market irrelevant reviews, and retains information for more than90percent ofthe stock market. This paper compares the semantic analysis methods, machinelearning methods and N-Gram methods, and support vector machine combined withthe information gain method can obtain better experimental results than the othertwo.We analyze factors affecting stock prices, and develop a regression model topredict the stock price. We have analyzed single stock, the communications sector and the Shanghai Composite Index based on stock reviews. The results showed thatthe lag stock’s closing price, the sentiment index, institutional rating and the numberof lag day news can effectively explain the closing stock price changes. Withone-way ANOVA analysis on communications sector, the results show that thesentiment index is an important factor to affect the rate of return and volatility. Wehave done correlation analysis between sentiment index and the leading and thelagged of the Shanghai Composite Index, also done on the individual stocks. Theresult show that investors sentiment will be affected by the Shanghai CompositeIndex, and have effect on single stock price.