Deep Dive: Low-volatility investing — what the latest research reveals
- Robin Powell
- Sep 30
- 11 min read
Updated: Oct 2
Welcome to Deep Dive, a new TEBI series where we take a closer look at the latest research shaping investing, cutting through complexity to show what really matters for investors. The first in the series looks at low-volatility investing. For 50 years, academics and investors argued about whether low-volatility stocks deliver superior returns. New research solves the paradox: both sides were right about different halves of the strategy. Here's what that means for your portfolio.
$392 billion sits in low-volatility investment strategies worldwide. Major institutions from MSCI to Axioma include low volatility as a distinct risk factor. Yet if you open any major asset pricing model, the low-volatility factor is conspicuously absent.
Industry embraced low-volatility investing after 2008 saved portfolios from catastrophic losses. Academic papers from Eugene Fama, Kenneth French, and Robert Novy-Marx claimed the premium was "explained away" by profitability and investment factors. Investors found themselves caught between conflicting advice.
New research published in June 2025 finally explains this standoff. Both sides were right — about different halves of the same strategy. What researchers discovered changes how defensive equity investing should work.
The pattern every investor has noticed
If you've owned low-volatility investments over the past 25 years, you've experienced a confusing pattern. Sometimes it works brilliantly. Other times it fails spectacularly.
During the 2007-09 financial crisis, the FTSE 100 collapsed as banks imploded. Royal Bank of Scotland fell 95% from its peak. Lloyds required government rescue. Low-volatility strategies, underweighting cyclical and leveraged sectors, offered substantial protection. Portfolios tilted toward pharmaceuticals, consumer staples, and utilities preserved capital when markets panicked.
The dot-com crash told the same story. The NASDAQ fell 78% peak-to-trough. Low-volatility strategies, with minimal exposure to speculative technology stocks, delivered far smaller drawdowns.
The initial COVID-19 crash in March 2020 followed the familiar script. Defensive sectors cushioned portfolios during the rapid decline.
But the recoveries disappointed. After the Brexit referendum in June 2016, the pound's collapse boosted UK multinationals. Low-volatility portfolios, overweight domestically-focused stocks, missed the rally. Between 2017 and 2019, markets rotated into cyclicals and technology. Low-volatility lagged badly. The COVID recovery proved even more frustrating as markets rallied behind high-beta names.
The cumulative damage appears in the data. Between May 2015 and October 2024:
Standard S&P 500: 14.3% annual return, 0.82 Sharpe ratio
Low-volatility S&P 500: 9.5% annual return, 0.63 Sharpe ratio
Lower volatility, yes — but also lower returns and lower risk-adjusted returns. The worst of both worlds.
What is low-volatility investing?
Research dating to Fischer Black in 1972 documented that stocks with low historical volatility have systematically delivered higher returns than high-volatility stocks. This contradicts traditional finance theory.
The Capital Asset Pricing Model predicts: take more risk, earn higher returns. The low-volatility anomaly says the opposite: take less risk, earn higher returns.
Consider a concrete example. Stock A fluctuates ±5% monthly. Stock B fluctuates ±15%. Over ten years, Stock A experiences smaller drawdowns and delivers superior cumulative returns. Under standard theory, this shouldn't happen.
Yet empirical evidence shows this pattern persistently across markets and time periods. The anomaly appears in equities worldwide, in bonds, commodities, and currencies. It holds after controlling for size and value, in large-caps and small-caps, across industries and countries.
The standard construction ranks stocks by historical volatility over one to three years. Buy the calmest stocks. Either short the most volatile or avoid them. Rebalance periodically.
The promise is compelling for retirees or anyone who lost sleep in 2008: equity-like returns with bond-like volatility. Less risk and more return.
This pervasiveness made the academic dismissal puzzling. If hundreds of researchers document the same pattern across dozens of markets, how can others claim it doesn't exist?
The academic dismissal
In 2015, Fama and French published their five-factor model, adding profitability and investment factors. When they tested whether low-volatility survived after accounting for these factors, it did not.
Robert Novy-Marx reached the same conclusion in 2016. High-volatility stocks were predominantly unprofitable companies with aggressive investment spending. Low-volatility stocks were predominantly profitable companies with disciplined capital allocation. Once you controlled for profitability and investment, the apparent premium disappeared.
The logic was compelling. High-volatility stocks aren't expensive because of volatility — they're expensive because they're poorly run. Low-volatility stocks aren't cheap because of low volatility — they're cheap because they're well-run. Volatility is a symptom, not the cause.
Therefore, investors should simply buy profitable, conservatively-investing companies directly. Building portfolios around volatility was redundant.
This consensus had real consequences. The factor was excluded from standard models despite $392 billion in assets. Advisers couldn't justify low-volatility recommendations when clients asked why they weren't buying quality funds instead.
Industry practitioners pointed to decades of data. Academics pointed to regression tables. Neither could explain why the other was wrong.
The standoff persisted until June 2025.
The breakthrough: what they measured wrong about low-volatility investing
Amar Soebhag, Guido Baltussen, and Pim van Vliet asked what seems obvious in retrospect: What if academics and industry are both right — about different parts of the strategy?
The standard method constructs a long-short portfolio: buy low-volatility, short high-volatility. Measure the combined return. Test whether other factors explain it. If they do, conclude low-volatility is redundant.
Using this method, academics consistently found other factors explained the returns.
Soebhag and colleagues did something different. They split the strategy in half, measuring each leg separately and accounting for real-world costs.
For the long leg, they held low-volatility stocks and hedged market exposure. For the short leg, they shorted high-volatility stocks and went long the market to neutralize exposure. Each leg could be evaluated independently.
The results were striking:
Long leg: 3.18% gross, 2.70% net (after costs)
Short leg: 1.83% gross, 0.52% net (after costs)
The long leg survives implementation costs and delivers genuine alpha. The short leg gets destroyed — shorting fees average 1% annually, spreads are wider, high turnover generates substantial costs.
This asymmetry explained everything. As the table below shows, once you factor in real-world costs, every major model gives meaningful weight to the low-volatility long leg but none to the short leg. The academics were right to dismiss the short side; the industry was right to value the long side.
The short leg is redundant, exactly as academics claimed. High-volatility stocks correlate strongly with low profitability and high investment. Once you include those factors, the short leg adds nothing. The academics were correct about this half.
But the long leg is different. Low-volatility stocks contain distinct information not captured by quality or profitability alone. The long leg deserves its own place in models. The industry was correct about this half.
For 50 years, researchers averaged together a brilliant long leg and an expensive short leg, then concluded the average was mediocre. The new study shows why that conclusion was misleading
When tested properly, the low-volatility long leg should constitute 26-29% of an efficient portfolio. Adding it to standard factor models increased Sharpe ratios by 13-17%. Every major model showed statistically significant improvement.
The researchers validated findings across multiple robustness tests: extended samples back to 1930, seven different risk measures, 4,096 different construction methods. The conclusion was unambiguous.
The measurement error stemmed from assuming both legs contribute equally. In frictionless academic models, they do. In reality, they don't.
For 50 years, researchers averaged together a brilliant long leg and an expensive short leg — then concluded the result was mediocre. The new study shows why that conclusion was misleading.
Why single-factor low-volatility fails
Understanding asymmetry explains the academic confusion. It doesn't explain why your low-volatility ETF has underperformed for a decade.
That failure stems from valuation.
Most low-volatility products hold long positions sensibly, avoiding the expensive short leg. They construct diversified portfolios and rebalance systematically. What they don't do is screen for valuation, momentum, or quality beyond volatility itself.
Research by Pim van Vliet in 2012 documented that low-volatility strategies spend approximately 62% of time in a "value regime" where these stocks trade at discounted valuations. During value regimes, strategies outperformed by roughly 2% annually. The remaining 38% of time, they trade in a "growth regime" at premium valuations, underperforming by 1.4% annually.
The contrast is clear. As the chart below shows, low-volatility strategies deliver meaningful outperformance only when they trade at discounted valuations — and actually underperform when they trade at a premium.
The evidence is clear: low-volatility only works when it’s cheap. When investors pay full price for safety, the strategy disappoints.
A 2014 study found alpha disappears entirely except when stocks are extremely cheap. The premium exists, but it's conditional on valuation.
Where do valuations stand today? As of June 2025, the Invesco S&P 500 Low Volatility ETF trades at a P/E ratio of 22.3. The S&P 500 itself trades at 22.3. Identical valuations.
Low-volatility stocks aren't cheap. They're priced identically to the market, meaning investors pay full price for defensive characteristics without any valuation discount.
This explains disappointing 2015-2024 performance. The strategy's post-2008 popularity attracted massive inflows. Assets grew from under $50 billion in 2010 to nearly $400 billion by 2024. This influx pushed valuations higher, eliminating the discount that historically drove outperformance.
As Howard Marks noted: "No asset is so good that it can't become a bad investment if bought at too high a price."
Single-factor strategies ignore this insight. They buy regardless of price, momentum, or quality. When you pay market valuations for below-market growth rates, mathematics dictates disappointing returns.
The multi-factor solution
In October 2024, Lodewijk van der Linden, Amar Soebhag, and Pim van Vliet tested whether adding value and momentum screens could enhance low-volatility strategies.
Start with the 500 lowest-volatility stocks. Select the top 100 by combined net payout yield (cheapest) and 12-month momentum (strongest).
This prevents buying expensive low-volatility stocks likely to mean-revert downward and avoids stocks in downtrends.
Results across 1990-2023:
Standard low-volatility: 10.8% return, 0.51 Sharpe, -50.4% max drawdown
Enhanced (adding value/momentum): 13.6% return, 0.75 Sharpe, -37.2% max drawdown
The enhanced strategy delivered 2.8 percentage points higher annual return, 47% better Sharpe ratio, and 13 percentage points lower maximum drawdown.
The enhancement worked by refining when and how exposure was taken. Buy defensive stocks when they're cheap and displaying positive momentum. Avoid them when expensive or falling, even if volatility characteristics remain low.
One additional consideration matters: term premium exposure. Low-volatility stocks behave somewhat like long-duration bonds. They're large, stable, dividend-paying with mediocre growth — characteristics making them sensitive to interest rates.
The 2022 bond collapse illustrated this. Long treasuries fell 29.5%. Low-volatility equities fell 4.8% — far less than bonds but substantially more than equity beta alone suggested.
This doesn't make low-volatility investing inappropriate. It means investors must account for implicit leverage and duration exposure. If adding low-volatility equity, consider reducing bond holdings slightly to maintain intended overall risk.
Fischer Black suggested this approach in 1993, arguing investors could use low-beta equities as implicit leverage, achieving equity returns at lower volatility by combining defensive stocks with reduced fixed income.
What, if anything, should you do?
Before implementing any defensive strategy, determine whether it suits your situation.
Low-volatility makes no sense for young accumulators with decades until retirement. Volatility creates opportunities to purchase at lower prices. If you're systematically contributing and won't need capital for 20-plus years, hold market-cap-weighted indices.
Low-volatility investing becomes relevant if you're near retirement, drawing income, and large drawdowns threaten financial security — and if volatility causes actual behavioural errors like panic selling. If volatility is purely intellectual but you don't actually change behaviour, defensive strategies add cost without benefit.
If low-volatility suits you, two paths exist.
The patient approach suits most investors. Accept that single-factor strategies are expensive now. Wait for valuation regime shift. When low-volatility ETFs trade at meaningfully lower P/E ratios than the market, the value regime has returned.
Until then, simply reduce equity allocation and increase bonds if you need lower volatility. This achieves the same reduction without paying for a premium unavailable at current valuations.
This isn't market timing. It's price discipline. You're not predicting when the regime shifts — that's impossible. You're observing current valuations and responding rationally.
The committed factor investor approach requires more sophistication. This makes sense only if you're philosophically committed to multi-factor investing, accept that factors rotate unpredictably, and believe maintaining exposure through cycles generates superior long-term returns.
Don't attempt timing when low-volatility will be cheap or expensive. No evidence suggests anyone can predict regime shifts with useful consistency. Use multi-factor strategies including low-volatility alongside value, momentum, and quality.
The implementation challenge is finding appropriate products. Few UK retail vehicles combine low-volatility with value and momentum screens as research suggests works best.
A pragmatic approach combines a core global multi-factor fund with a smaller low-volatility satellite when valuations permit. Regular rebalancing maintains exposure without attempting timing.
The uncomfortable truth: right now, low-volatility is expensive. The premium has been arbitraged away by popularity. Patient investors should wait.
Monitor relative valuation of low-volatility products against the market. When you observe persistent discounts — low-volatility trading at materially lower earnings multiples — conditions supporting outperformance have re-emerged.
Until then, multi-factor strategies including modest low-volatility as one component remain most defensible.
Why this took 50 years to discover
The measurement error stemmed from three institutional blind spots.
The academic blind spot derived from standard assumptions: frictionless markets with no transaction costs, costless shorting, unlimited leverage at the risk-free rate. Under these assumptions, both legs contribute equally. In reality, shorting costs average 1% annually plus wider spreads and higher turnover, destroying the short leg while leaving the long leg intact.
When academics dismissed low-volatility as "explained away," they examined gross returns in a frictionless world. Investors lost access to what should have been 26% of an efficient portfolio.
The industry blind spot was different. Providers sold single-factor products because they were easy to explain, demand was enormous post-2008, and fees could be justified. They didn't emphasize that much return came from value, quality, and term exposure — available more cheaply elsewhere. They also didn't explain the strategy only worked when valuations supported it.
Behavioural patterns sustained both blind spots. Recency bias led investors to remember 2008 salvation while forgetting 2017-19 underperformance. Hindsight timing encouraged believing "I'll buy when markets get scary," ignoring that by then, you're buying expensive. Complexity aversion worked in industry's favour: single-factor is simple, multi-factor requires explaining interactions and rotation.
The 2025 research questioned a fundamental assumption: what if we split legs apart and measure each separately, accounting for real costs?
The answer revealed the factor works, but not how anyone measured it. The long leg contains genuine information. The short leg is redundant and expensive. For 50 years, we averaged them together and concluded the result was mediocre.
What low-volatility investing means for your portfolio
Three truths emerge:
First: The factor works when measured correctly — long leg only, after costs. Low-volatility contains distinct information not captured by quality or profitability alone.
Second: Single-factor implementation fails because it ignores valuation and momentum. Buying regardless of price or trend has disappointed as valuations expanded to market parity.
Third: Right now, low-volatility is expensive. Patient investors should wait for the value regime to return.
What changes with this research?
Before, confusion reigned. After, we have clarity that low-volatility investing works under specific identifiable conditions.
Before, investors chose between academic dismissal and industry hype. After, we understand both sides were half-right.
Before, investors bought single-factor ETFs and hoped. After, we know multi-factor approaches with valuation awareness are essential.
Low-volatility investing isn't broken. The way most investors access it is.
When valuations shift back — and eventually they will — you'll be ready to implement correctly. You'll seek multi-factor strategies combining low-volatility with value and momentum. You'll monitor relative valuations rather than trying to time cycles. You'll account for term exposure and adjust overall allocation accordingly.
Until valuations shift, patience beats expensive mistakes. The premium exists, but like all premia, it's not available at all prices. Right now, the price is too high.
The research has explained the 50-year paradox. The measurement error has been identified and corrected. The knowledge exists to implement defensive equity properly.
What remains is waiting for the price to be right. That's not timing. It's refusing to overpay for assets whose valuation has expanded beyond fundamental characteristics.
When the opportunity re-emerges, you'll recognize it. And you'll avoid the mistakes that cost investors the past decade of underperformance.
Resources
Black, F. (1993). Beta and return. The Journal of Portfolio Management, 20(1), 8-18.
Fama, E. F., & French, K. R. (2015). A five-factor asset pricing model. Journal of Financial Economics, 116(1), 1-22.
Soebhag, A., Baltussen, G., & van Vliet, P. (2025). Factoring in the low-volatility factor.
Van der Linden, L., Soebhag, A., & van Vliet, P. (2024). Leveraging the low-volatility effect. Robeco working paper.
Van Vliet, P. (2012). Enhancing a low-volatility strategy is particularly helpful when generic low volatility is expensive. Journal of Portfolio Management, 39(4), 47-57.
RECENTLY ON TEBI
TEBI ON YOUTUBE
Have you visited the TEBI YouTube channel lately? There’s already a wide selection of high-quality videos on there. Why not subscribe and be one of the first to see our latest content? You'll also find our videos on Instagram and TikTok.
