by The Editorial Team, AdvisorAnalyst
For decades, investors and portfolio managers have quietly built billion-dollar strategies around a simple, counterintuitive idea: stocks that move less tend to perform better on a risk-adjusted basis than stocks that move more. Low-volatility investing has accumulated over $392 billion in assets under management globally. Yet academic finance has largely left it out of the standard models used to explain how markets price stocks.
A new working paper, Factoring in the Low-Volatility Factor1, by Amar Soebhag, Guido Baltussen, and Pim van Vliet — researchers at Erasmus University Rotterdam and Robeco Quantitative Investments — directly confronts this gap and offers a compelling resolution.
Their conclusion: the low-volatility factor has been unfairly dismissed by academic models because those models are built on assumptions that don't hold in the real world.
The Disconnect — And Why It Existed
Standard academic factor models — the frameworks used to evaluate whether a stock return driver is truly independent and meaningful — have consistently found that low volatility adds little once you already account for profitability and investment factors. The implication seemed to be that low-volatility investing was redundant: captured by other things academics already knew about.
The authors acknowledge this. In their baseline tests, replicating the standard academic approach, they find that augmenting factor models with a low-volatility factor raises the maximum Sharpe ratio by only 0.57% on average — essentially nothing.
But the authors argue that these tests are built on two flawed assumptions. First, they treat the "long" and "short" sides of a factor as interchangeable — as if buying low-volatility stocks and shorting high-volatility stocks are equally practical. Second, they ignore real-world costs: transaction fees, shorting fees, and the many practical limits on short-selling.
"Traditional factor tests assume symmetric long-short positions and frictionless markets," the authors write. "Once we account for factor asymmetry and real-world investment frictions — such as tradeability, transaction costs, shorting costs and shorting frictions — low-volatility materially improves leading asset pricing models."
Separating the Long from the Short
The paper's central insight is about asymmetry. When the authors split each factor into its long leg (owning the attractive stocks) and its short leg (shorting the unattractive ones) and let each leg carry its own weight in an optimized portfolio, the picture changes dramatically.
The hedged long leg of the low-volatility factor — owning low-volatility stocks with market exposure removed — earns 3.18% per year on a gross basis. The short leg of the same factor, betting against high-volatility stocks, earns only 1.83%. This gap is not unique to volatility; it shows up across most factors. Long legs outperform short legs.
More importantly, when the low-volatility long leg is allowed to enter factor models separately, it earns an average portfolio weight of 26.2% in optimized portfolios — and improves model Sharpe ratios by an average of 11.9%. Across all seven factor models tested, the improvement is statistically significant at the 1% level.
"We show that this apparent subsumption is driven primarily by the short side of the low-volatility factor," the authors write, "whereas the hedged long side contains distinct information that materially improves factor model performance."
In plain terms: the reason academics thought low volatility was redundant is that the short side of low volatility looks a lot like other factors (it correlates heavily with low-profitability and high-investment stocks). But the long side — the part that actually matters for most real investors — is genuinely different.
What Happens When Costs Enter the Picture
The paper goes further by incorporating actual transaction costs and shorting fees into the analysis. This is where the low-volatility advantage becomes even clearer.
After accounting for costs, momentum (one of the most celebrated factors in academic finance) sees its annual return drop from 5.41% to 2.17%. Post-earnings announcement drift falls from 3.45% to -0.87% — it actually loses money net of costs. Low volatility, by contrast, declines from 5.01% to 3.22%, and its long leg retains strong net returns of 2.70% per year.
Short legs across almost all factors turn negative after costs. The optimal portfolio assigns zero weight to shorting in most cases. But the low-volatility long leg continues to receive allocations of roughly 26% in optimized portfolios, and factor model Sharpe improvements average 13% after costs.
"Low volatility becomes even more relevant once factor asymmetry and real-world investment frictions are incorporated," the authors conclude.
The robustness tests are extensive. The finding holds across data going back to 1930, across seven different measures of low risk (including beta, idiosyncratic volatility, and downside volatility), and across 4,096 different portfolio construction specifications. In the specification check, the low-volatility long leg improves factor models at a statistically significant level in 69% to 97% of all tested variations.
Key Takeaways for Advisors and Investors
The practical implications of this research are direct:
1. Low-volatility belongs in your factor toolkit — not as a curiosity, but as a core allocation. The research shows it improves portfolio efficiency across virtually every model tested, once you build portfolios the way real investors do.
2. The long side is where the value is. For most advisors and institutional investors who cannot or do not short stocks, this is actually good news. The performance edge in low-volatility comes from owning low-risk stocks — not from a complex long-short trade. Long-only low-volatility strategies capture most of the benefit.
3. Net-of-cost performance matters more than academic factor returns. Many factors that look compelling in academic papers dissolve under real transaction costs. Low volatility is one of the few factors that holds up. Its naturally low turnover keeps friction costs down.
4. Factor models used for portfolio construction, risk attribution, and manager evaluation should include low volatility. As the authors put it: "Low-volatility deserves consideration not only as investment factor but also as a core dimension in factor models used for portfolio construction, manager evaluation, risk attribution, and strategic asset allocation decisions."
The gap between academic theory and investment practice has long frustrated practitioners who built successful low-volatility businesses while academics told them their factor was subsumed. This paper closes that gap — not by changing the evidence, but by changing the lens through which we evaluate it.
Footnote:
1 Soebhag, Baltussen and van Vliet, "Factoring in the Low-Volatility Factor," May 2026 (SSRN Working Paper).
Copyright © AdvisorAnalyst
