High-Dimensionality Effects on the Efficient Frontier

• Activity Number: 676
• Type: Invited
• Date/Time: Thursday, August 4, 2016 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Rituparna Sen*
• Companies: Indian Statistical Institute
• Keywords: Efficient Frontier ; Bayesian Multiple Testing ; Portfolio Optimization

Abstract

We investigate the problem of efficient frontier calculation when the number of available assets is high. El Karouii (2010) suggested bias correction using random matrix theory for parameter estimation and risk evaluation. DeMiguel (2009) and Fan et al.(2012) discuss norm-constrained portfolios, which result in sparse solutions. Recently Yang et al (2015) have proposed a robust solution to the problem. We propose selection of high/low performing assets from a large collection using Bayesian Multiple Testing and subsequently constructing portfolios based on those assets. In the process we derive new theoretical results concerning multiple testing in the regression set-up. We compare the out-of-sample performance of the different methods in simulations and real portfolios of 100 stocks from 3 markets.