Intersectional vs Sequential Multi-Factor Models

And the winner is?

May 2025. Reading Time: 10 Minutes. Author: Nicolas Rabener.
SUMMARY

  • The intersectional and sequential multi-factor models generate comparable results
  • The sequential ordering of factors matters less than expected
  • The intersectional model features lower turnover

INTRODUCTION

Pursuing factor investing is a bit like crafting the perfect dish: it requires a solid base, a careful balance of salty and sweet, and just the right hint of umami. In the world of stock selection, this translates to identifying stocks that outperform the market, aren’t overly expensive, exhibit high-quality traits, and maintain below-average volatility. As in cooking, achieving this blend is no easy task.

Investors often turn to smart beta ETFs for factor exposures. However, these products can carry conflicting factor tilts, which may cancel each other out at the portfolio level (read Factor Exposure Analysis 114: Factor Offsetting). A more effective approach is to build customized portfolios tailored to specific factor goals (read Smart Beta ETF vs Customized Factor Portfolios).

There are three main methods for constructing multi-factor portfolios using stocks: the combination model, the intersectional model, and the sequential model (read Multi-Factor Models 101). Since the combination model often results in the same factor offsetting issue, we will set it aside.

In this research article, we focus on comparing the intersectional and sequential models to understand their respective advantages and trade-offs.

PERFORMANCE COMPARISON

Our analysis focuses on selecting U.S. stocks based on three key factors: quality (Q), value (V), and momentum (M). We construct portfolios using the top 100 stocks meeting these criteria, each with a minimum market capitalization of $1 billion, and rebalance them monthly with 10 bps of transaction costs.

We compare two multi-factor approaches: