A Novel Quantitative Behavioral Framework for Financial Markets Prediction

Abstract

Effective prediction of financial asset prices has become a challenge in the present day volatile world. The use of mathematics have become very extensive in the financial world, most of the mathematical models concentrates on the market data rather than the behavior of the market from which the data has been generated. An attempt has been made for the first time to model the prediction of asset prices based on both the market data and the behavior of the market participants. The participants in the financial markets behave differently from each other, these behavioral differences can be attributed to the participants understating or/and his perception about the market. Each investor has his own perception about the market and he feel it is close to reality, but truly speaking it is not so. Each participant has his own impact on the market and the reality is the aggregation of each participant’s perception. The impact of the investor’s behavior has been modeled in the present quantitative behavioral approach by dividing the participants into broad categories based on their trading behavior. To model the participant’s impact first one should predict the proportion of participants in each category. Most of the times, finding the exact number of participants in each category is not easily available from the market data, so an evolutionary based swarm intelligence model has been adopted in the present framework to find the proportion of the participants in each category. Finally the whole methodology has been applied to gold asset class (because gold is an international asset with increasing volatility these days) to validate the present method. The model is tested rigorously using different time varying samples to validate the present methodology; some interesting results have been obtained from the present study. The back testing results prove that the model presented in this paper is very effective in predicting the prices close to reality. The present frame work is very generic and can be applied to any financial asset class to estimate the returns close to reality.