Despite the rise of factor investing, most strategies fail as a result of causal blind spots. In this article, we expose the “factor mirage” and urge a shift from correlation to causation for more robust, trustworthy, and effective investment models.
Report Overview
Factor investing is a foundational paradigm in quantitative asset management. Yet despite the proliferation of factors and widespread institutional adoption, most strategies have failed to live up to their in-sample promise. Although p-hacking and backtest overfitting have received considerable blame, a more insidious source of error is rarely discussed: the uncritical application of an econometric canon that ignores causal structure. In this article, we introduce the concept of the factor mirage — a factor model that appears to be statistically valid but is causally misspecified. We show how collider bias and confounder bias, when embedded in the standard regression framework, can yield misleading inferences, poor out-of-sample performance, and misguided investment decisions. By shifting from associational to causal reasoning, practitioners can build more robust strategies, reduce false discoveries, and restore trust in factor-based approaches.
Support Research for the Global Investment Community. Your donation raises professional excellence and enables the foundation to continue its important mission to fund, publish, and distribute in-depth, high quality, independent research.