Portfolio Selection Redux, or, How the Paradox of Portfolio Efficiency can be used to Improve Portfolio Performance

Portfolio performance is determined by both stock selection and portfolio construction. Mean–variance portfolio construction as formulated by Markowitz showed many years ago that efficient portfolios most accurately reflect expected returns in portfolio performance, whereas inefficient portfolios always incur unnecessary risk and leave potential outperformance on the table, thereby wasting some of the portfolio manager's skill. Standard portfolio optimization, however, is widely regarded as problematic, so many managers adopt heuristic portfolio construction methods instead. This article argues that the underlying problem is not with the idea of portfolio efficiency, but with its implementation. A simple analysis of the risk structure of a portfolio can identify which holdings are most inefficient and which are "efficient enough" given the limitations of any manager's forecasting ability. This information can then be used in rebalancing to ensure the greatest increase in overall portfolio efficiency for the least amount of turnover while considering the inherent uncertainty in all forecasts. For active managers with stock selection skill, more efficient portfolios mean better performance.