Stock Market Portfolio Construction:A Four-stage Model Based on Fractal Analysis

The paper exploits fractal characteristics of time series of stock returns of Indian companies and uses them for portfolio construction. To model the dynamics of return series of the respective stocks, we use single Fractal model based on Hurst Exponent and Multi Fractal Detrended Fluctuation Analysis. For simulations, daily stock returns of fifty three Indian companies belonging to nine different industrial sectors during January, 2009 to October, 2016 have been considered. The paper considers a four stage framework for portfolio construction. The first stage is achieved by selection stocks through fractal modeling from different sectors. In the second stage, the stocks are arranged in descending order based on Hurst exponent. In third stage, pair correlation is applied for final selection of stocks in the portfolio. In the fourth stage, Evolutionary Algorithm is used for calculation of the weights of each stock in the portfolio. Our results indicate that the portfolio so chosen, outperforms the market in each year of the study.