Grey Wolf Optimization Model for the Best Mean-Variance Based Stock Portfolio Selection

Portfolio diversification has long been known and prevalent as a risk minimization method. The goal of this work is to develop a model to use Grey Wolf Optimization (GWO) algorithm to create an investment portfolio with optimal risk-return ratio. The initial set of equities we propose to be selected using Self-Organizing Maps (SOM). We have highlighted the set of factors to use as input variables for SOM. When the equities in portfolio is defined, we need to decide the proportion of our capital distributed among the portfolio equities. For this task we employ the GWO algorithm, which let us find the optimal weights assignment along the portfolio shares based on Mean-Variance portfolio minimization condition. As we know the Variance and Mean respectively express the Risk and Return rating of portfolio. Our research investigates the sensitivity of the GWO algorithm to number of iterations, wolf herd size and shares weight limits. Just knowing these parameters best values, we can expect for optimal portfolio diversification. The model verification was performed on stocks from S&P500 index. The GWO was applied for determining the weights for portfolio equities to get optimal risk-return ratio. The comparison of GWO balanced portfolio performance with the direct investment to S&P500 index, let us conclude the advantage of our proposed model.