Portfolio optimization using Genetic algorithm incorporating Value-at-Risk
Sun, Fei (2009) Portfolio optimization using Genetic algorithm incorporating Value-at-Risk. [Dissertation (University of Nottingham only)] (Unpublished)
In the traditional mean-variance portfolio optimization model, variance is as a risk measure based on the assumption of normal distribution on asset returns. However, most of empirical returns on assets are not normally distributed. The fat tails and skewness appear in the distribution of asset returns that makes the portfolio optimization model with variance as a risk measure inaccurate. With the Value-at-Risk (VaR) widely employed by financial institutions as a measure of risk, this paper presents a mean-VaR portfolio optimization model with VaR as a risk measure. The portfolio optimization problem is a two-objective optimization problem. Since genetic algorithm is a stochastic search algorithm based on the mechanism of natural selection, it is good at solve multi-objective optimization problem and has been applied into many financial areas. This paper will design a multi-objective genetic algorithm to optimize a hypothetical portfolio problem based on the mean-VaR model. Also, mean-variance model will be applied to the proposed optimization problem to compare the performance of two different portfolio optimization models.
Actions (Archive Staff Only)