A1632
Title: Maximum drawdown-optimized portfolios
Authors: Philipp Staehli - University of Basel (Switzerland) [presenting]
Dietmar Maringer - University of Basel (Switzerland)
Abstract: The maximum drawdown (MDD) is the largest cumulative loss from a peak to the following trough within a given period of time. It is one of the most widely used path-dependent risk measures in the fund management industry, and it is often used as an additional criterion to assess a portfolio or strategy. However, there is little empirical research on portfolios explicitly optimized for MDD. MDD is used as an objective for portfolio optimization. Based on S\&P 500 Health Care stocks' data for 2012-2020, an empirical study is performed with 2000 random combinations of assets for different time windows. For each of these situations, gradient-based sequential least squares were used to minimize the in-sample MDD and the in-sample variance, respectively. These optimized portfolios were then analysed for their out-of-sample performance. As expected, the out-of-sample return of MDD-optimized portfolios was higher, and the out-of-sample MDD was lower than their minimum-variance (MVP) counterparts. At the same time, however, MDD-optimized portfolios typically outperformed their MVP counterparts by having lower Value-at-Risk and lower Expected Shortfalls for out-of-sample windows up until the end of 2019. These advantages do not prevail for out-of-sample windows in 2020, i.e., after the covid-crisis had begun while portfolios were optimized predominantly on pre-covid data.
