City Research Online

Abstract

Markowitz advanced the theory of portfolio construction, dividing the work into two stages of security selection and capital allocation. He did not address security selection, but established a new method to allocate capital. Since then, much has been written regarding each stage while the investment industry grew and its performance was examined through a variety of lenses. This thesis contributes to the literature by examining the interaction between these two stages of portfolio construction. Each stage impacts performance, but an examination of the interaction is missing from the literature. A deeper understanding of the interaction provides opportunity for practitioners to improve portfolios and gives a base for future research to refine the understanding.

The research begins by holding 1,000 US large capitalization stocks constant while applying a panel of capital allocation methods. Despite starting a fixed set of securities, differences in performance and risk characteristics are documented. Markowitz identified the objective superiority of risk adjusted returns over just returns, so my work applies the Sharpe ratio as the measurement unit. Differences in Sharpe ratios across the panel of allocation methods are tested for robustness using a bootstrap test.

With a hierarchy of Sharpe ratios established, the next step varies the security selection and observes the change in Sharpe ratios. Security selection is implemented with one year perfect foresight which is a limit condition for the potential of stock picking. When applying good and bad security selection it is observed that the hierarchy of Sharpe ratios is unstable in the presence of security selection. The bootstrap robustness test shows the hierarchy can invert with statistical significance. This is the first step in understanding the dependence of the optimal capital allocation method in the presence of security selection skill.

Toward the goal of optimal Sharpe ratio portfolios, I examine portfolios constructed with perfect foresight into return, volatility, and correlations. Optimized portfolios are dominated by low correlation securities, not by high returning securities. Conversely, by first selecting the highest returning securities possible, you crowd out the optimal Sharpe solution. Reward curves are built for selecting securities based on returns, volatility, and correlation. The shape of the correlation reward curve is like the low volatility anomaly.

Because low correlation securities dominate optimized Sharpe ratio portfolios, perfect correlation selection is applied. Stock level and portfolio level characteristics are documented in the cross section of years. Then the method is applied to the panel of allocation methods to create return time series and Sharpe ratios. The bootstrap robustness test applied to the Sharpe ratios of the allocation methods shows that even in the absence of perfect correlation selection, nearly any level of success achieved in correlation selection creates robust improvements in the Sharpe ratio.

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HG Finance
Departments:	Bayes Business School > Bayes Business School Doctoral Theses

Export

Downloads