| In the era of big data and artificial intelligence, we make use of data mining to explore the correlations between investor emotions and volatility about stock markets and better predict trends of stock markets. The study analysis will help to solve macro-control and even national security problems.Firstly, we extract some hot stocks on the basis of number of discussions from professional investment forums in March 2016 and find that in short-term hot stocks do not change significantly and the discussions about suspension of stocks do not decrease significantly. The trend of stock prices and market index are basically same and later t-test parameters of OLS regression is significant so that it finds no obvious anomalous fluctuations. Secondly, the paper calculates the Pearson correlation coefficient between the market and emotion signals is 0.7.The correlation coefficient between China Merchants Bank and emotion signals is 0.53 and the correlation coefficient is 0.52 to BYD. They both have a clear correlation but the market has a stronger correlation. Emotion signals have some features that it is sparse and signals mostly occur in the time that trend will be reversal or break and sometimes is ahead of the market reaction about one or two days.This paper selects transactions data of Shanghai index from November 2, 2015 to March 4, 2016 and we use emotion signals, the highest price, the open price, the lowest price, the close price, trading volume and turnover seven dimensions together as input variables of SVM. The lag period is five days and genetic algorithm can help to optimize parameters. The trend prediction on the Shanghai index obtain MSE = 0.00381 and then we don't add emotion signals so that get a result that MSE is 0.004371.It indicates that emotional signals can help investors in the market predict future trend. Emotional signals can be served as warning signs of investors and also can be served as an analysis, monitoring and management tools of public opinions. |
PDF Full Text Request