Predicting Chinese Stock Market with Internet Key Word Hotness by Statistical Time Series Regression Analysis

Recently there is a growing amount of Internet data available. And investors of the stock market have an increasing demand for using these valuable Internet data to guide their investment decision. We crawled the attention of some hot-words and plate words on different social media through python and regressed them with the stock price in the same period. It was divided into two parts: hot words and stock price, plate words, and stock price. Combined with the Granger causality test framework, it was found that the popularity of keywords had a certain impact on the relevant stock price. Still, most of its impact coincided with the impact of the market on the stock price. This also showed that the market is efficient. The influence of the attention of hot words on the market trend depends on the nature of the words. The wider the audience, the greater the impact on the market. We use the popularity of keywords to predict the stock returns by a statistical data mining model. We think that this may better predict the trend of the A-share market and has more timeliness. Simultaneously, it can more comprehensively reflect the actual operation of China’s A-share market from the perspective of investors.