Non-Linear Equity Factor Models?
Agenda
Presented by Northfield President Dan diBartolomeo.
Since the 1970s, the key property of equity factor models has been the presumed linear relationship between security returns and factor exposures. This assumption arises out of the Arbitrage Pricing Theory (Ross, 1976). However, the necessary conditions under which the APT would be assumed to strictly hold in the real world do not exist, as the APT assumes that liquidity is essentially infinite and therefore transaction costs are non-existent which allows traded markets to always achieve a clearing price. While equity factor models have been used successfully to describe the relationship between returns and sources of risk, the recent popularity of “machine learning” has spawned new interest in exploring non-linear relationships.
Inspired by our recent research into the non-normality in distribution of the cross-section of stock returns, this webinar will explore a middle ground between traditional linear models and machine learning models. We will illustrate the estimation of factor models based on simple specifications such as a linear relationship between “heavy tailed” returns and the absolute value of factor exposure, or between such returns and the square of factor exposure. While the explanatory power of such relationships is lower in traded equities than the linear model, there appears to be material information content in these alternative specifications.
The presentation will conclude with a discussion of low liquidity assets (e.g. high yield bonds) where an expectation of non-linear behavior may be more readily observed.
About Dan diBartolomeo
Dan diBartolomeo is President and founder of Northfield Information Services, Inc. He is also a former Visiting Professor at the CARISMA Research Center of Brunel University in London and serves on the Board of Directors of the Chicago Quantitative Alliance and the advisory board of the International Association for Quantitative Finance. He is a regional director of the Professional Risk Managers International Association, (PRMIA), and the Quantitative Work Alliance for Applied Finance, Education and Wisdom (QWAFAFEW). He is past president and director of the Boston Economic Club.
Dan has been admitted as an expert witness in US federal courts and state courts for litigation matters regarding investment management practices and derivatives.
In 2010, Dan received an award from Institutional Investor magazine as one of the forty most influential executives in financial technology in connection with his analytical work that helped uncover the Madoff investment fraud.
Dan is a director of the American Computer Foundation, and formerly served on the industry liaison committee of the Department of Statistics and Actuarial Sciences at New Jersey Institute of Technology. He continues his more than twenty years of service as a judge in the Moskowitz Prize competition, given by the University of California at Berkeley for excellence in academic research on socially responsible investing.
Dan has a long list of publications including books, book chapters and research papers in professional journals such as Financial Analyst Journal, Quantitative Finance and Journal of Investing. In 2017, he was named co-editor of the Journal of Asset Management.