Analysis of Herd Behavior in Stock Prices Using Machine Learning

In this paper, we consider the problem of herding behaviour in a Stock Exchange. Herding occurs when amateur investors follow the advice of financial gurus since they do not have the time, expertise or finances to do the research that is typically performed by these gurus. Although herding is well understood, many of the previous analyses have been through the use of statistical techniques. In this paper we have a second look using Machine Learning and demonstrate its effectiveness. We use a dataset obtained from the Singapore Stock exchange. Stocks were grouped into different portfolios based on the number of shares traded per day. Results from the algorithm show that herding is evident in each portfolio. We also find that herding is more pronounced among stocks that have higher volumes of shares traded.