Resumen de Portfolio optimization and out-of-sample performance

Francisco J. Nogales

• In this work, we study the practical performance of one of the most known optimization problems in Finance: the Portfolio Selection Problem. This is an optimization problem with uncertainty, coming from unknown future asset returns. To alleviate the impact of estimation error on this problem, several approaches have been recently proposed.

  Among the best approaches are: those based on ignoring the expected asset returns (imposing a factor structure on the associated covariance matrix; shrinking the sample covariance matrix; adding short-sale constraints; constraining the portfolio norm), those based on optimizing the worst-case performance (robust portfolio optimization) and the naive 1/N rule.

  We compare empirically (in terms of portfolio variance, Sharpe ratio, value-at-risk, and turnover) the out-of-sample performance of these strategies across several real datasets.

  From these results, we give some insights on how to improve practical performance of proposed policies.