Study of stock market movement and crashes using agent based modelling

Abstract

The random walk of the stock market is simulated using historic information rather than using a simple random function. The historic stock market index information is analyzed to determine behaviors/ rules of buying, selling and holding of stocks by individual investors. Agent based model (ABM) is developed to incorporate the rules identified and executed resulting in a random walk of the stock market. The input parameters to the ABM are varied randomly and different tests are conducted. Logistic regression is used to predict the probability of a stock market crash.