Academic research in equity factor models have long focused on linear models, where stocks are sorted and ranked by a single characteristic. This makes sense when the objective of the exercise is to explain the cross-section of returns in as parsimonious a manner as possible, but may not be the idea approach to developing portfolios.
Non-linear sorts – for example, those based upon decision trees – may introduce greater nuance and may be able to better pick up on meaningful interaction effects between characteristics.
In this week's research note, we explore the application of machine learning techniques to create defensive equity portfolios based upon a set of per-engineered security and firm characteristics. With characteristics ranging from value to quality and momentum to risk, we hope to identify a more nuanced definition that can help increase the certainty that the portfolio will actually provide protection during equity drawdowns while reducing the implicit cost of that protection. (PDF).
✉️ Join Newsletter | ✎ Send us your thoughts | ✄ Share URL
What We're Reading
→ FINANCING RISK: "A risk factor linked to aggregate equity issuance conditions explains the empirical performance of investment factors based on the asset growth anomaly of Cooper, Gulen, and Schill (2008). This new risk factor, dubbed equity financing risk (EFR) factor, subsumes investment factors in leading linear factor models." Equity Financing Risk