Improvement of Portfolio Selection: A Comparison of Different Methods
Abstract
As one of the most important models in finance, Efficient Portfolio Theory pioneered by Markowitz (1952) has been developed since 1950s. Although it has been widely used in practice, Markowitz's mean-variance model has been questioned about its validity because of its bad estimation performance especially in small samples due to the parameter uncertainty problem. Many strategies have been proposed for the purpose of lower the estimation error of mean-variance model. This dissertation gives a review of the existing literature with the goal of improving the performance of the Markowitz mean-variance model. We evaluate across five empirical data sets of 11 estimation methods. Among these methods, the combination rules by Tu and Zhou (2010) are practicable in terms of Sharpe ratio, and optimal two-fund rule and shrinkage on the covariance rule are practicable in terms of CEQ return. However, in comparison with the in-sample performance, these models surely still have room to improve.