Institutional investors are increasingly scrutinizing systematic strategies to ensure robust performance. Traditional due diligence primarily emphasizes performance metrics such as backtests, Sharpe ratios, drawdowns, and attribution. However, a critical gap remains unaddressed: whether the variables used in these models accurately reflect the economic forces they intend to capture. This oversight may represent a significant and undiagnosed source of risk in evaluating systematic strategies.
A concerning trend has surfaced among allocators observing systematic equity managers. For instance, when a manager adds a “quality” overlay to a value strategy, initial backtests often yield promising results—improved Sharpe ratios and reduced drawdowns. Yet, after a year, these strategies frequently underperform simpler value-only models. Allocators suspect that managers have overfitted models to historical data, though this does not fully explain the underperformance.
The culprit is often specification error, where certain factors are incorrectly included due to their correlation with returns rather than their independent significance. Research by López de Prado and Zoonekynd highlighted instances where such errors flipped the sign of factor coefficients, portraying an illusion of improved model fit that detracts from its structural integrity.
To address this, allocators should pose the question: “How did you decide which variables to include in your model, and which did you deliberately exclude?” This inquiry encourages managers to articulate the economic rationale behind their choices, rather than relying solely on statistical criteria.
As pension funds increasingly require factor transparency for capital allocation, clarity in model specifications becomes necessary. A well-specified model can distinguish between true alpha and misleading noise, which is critical as factor returns continue to decline.
Why this story matters:
- The rise of factor transparency is reshaping capital allocation decisions.
Key takeaway:
- Understanding the economic rationale behind variable selection is crucial in evaluating systematic strategies.
Opposing viewpoint:
- Some argue that statistical methods alone can provide adequate model specifications without delving into economic reasoning.
