The Effects Of Online Information On Stock Market

The impacts of information on stock market have always been a hot issue. With the development of economics and technology, the Internet has become the main channel of expressing emotion and welfare claims for the growing Chinese netizens. Therefore, it’s theoretically and practically meaningful to extract investor emotion characteristics with test mining techniques, so as to study the dynamics of the Internet and stock market.Specifically, financial press reports claim that Internet stock message boards can move stock markets. To verify this hypothesis, we extract almost6million messages of200GEM stocks posted on China’s largest stock message board in the last two years. To automatically classify these messages into three emotion categories, based on which we’ll construct an "emotion index" and an "agreement index", we adopt a text classification method called Naive Bayes algorithm.Our main results are as follows:(1) the online information variables have their own patterns. First, the message number reaches two peak in a trading day at morning and afternoon respectively. At9:00p.m., there is a third little peak.(2) The online information variables reflect fluctuations from stock market, and their correlation is higher than people’s expect. We found a big volatility or big trade volume goes along with big message number and diversified opinions. Diversified opinions also contribute to a higher daily return.(3)The online information help predict stock return, volatility and trading volume. Higher opinion index predicts higher stock return, but the predicting ability is limited to a short period, compared with a more lasting predicting ability of message number and word number.