Portfolio Optimization -The Mean-Variance and CVaR approach

Abstract: The recent economic turmoil has increased volatility on the Swedish stock market and made investors more exposed to risk in an uncertain environment. This research will investigate if the quantitative portfolio optimization models Mean-Variance and CVaR can produce risk-adjusted returns for investors acting in the Swedish stock market. From the classic Mean-Variance model different investment strategies with restrictions on short-selling are applied and the CVaR approach is applied on confidence levels of 95% and 99% respectively. The optimized portfolios are constructed using 3 different input periods of 1, 2 and 3 years that are rebalanced on a monthly basis. To be able to grasp the results a benchmark index and an equal weight strategy are included. We show, by using the Sharpe-ratio as an evaluation method, that the equal weight strategy produces the most efficient risk/reward during the time-period 2005-2009. The optimization models Mean-Variance and CVaR, with their applied strategies, turned out to underperform both the benchmark index and the equal weight strategy in the risk/reward universe. Finally, when analyzing the 3 different lengths of input periods it is found that no length is superior to be used in the investment strategies.