Drawdown measure and market timing skills : an international empirical study

Abstract

Despite the fact that classical timing measures have unique and effective features for the evaluation of a portfolio manager's timing skill, they suffer some limitations. These limitations versus the potential features of other risk measures, e.g., dynamic risk measures, provide the possibility to propose alternative timing measures. A set of potential reasons such as the unique features of technical timing measures, the superiority of average drawdown measure compared to variance (or standard deviation), the lack of sufficient positive timing evidence resulting from the existing timing measures, the ignoring of the effects of worst market losses occurring over the holding period of an investment, the low forecasting power of the existing timing measures stimulate us to propose the average drawdown market timing measures. Subsequently, this study adds the average drawdown market timing measures to the Carhart (1997) standard model for estimating selection and market timing skills of a portfolio manager. This study collects data of 3087 stocks from 23 developed countries and constructs the 23 country-level portfolios. The purpose is to compare performance of the average drawdown timing approach and the traditional timing approach. The timing measures are constructed using two most-common timing measures of Treynor and Mazuy (1966) and Henriksson and Merton (1981). Another purpose is to compare the daily performance and the monthly performance of average drawdown timing models. The results show that portfolio managers have significant timing and selection skills using the average drawdown timing measures. The predictive power of these timing measures is better than the power of traditional timing measures based on higher statistical and economic significance. The daily performance of the average drawdown timing measures is better than the monthly performance. These results are robust to different checks, e.g., different lag lengths, simulation tests, and alternative monthly returns.

Date of Award	2018
Original language	English