Strategies · Long-term investing
Low-volatility investing: the rules, and USMV against SPY
How minimum-volatility funds are built, and 15 years of USMV against SPY: smaller swings and falls, level with the market until 2019 and far behind it since.
On this page
Low-volatility investing means owning the stocks whose prices move the least. Finance theory says that should earn less: risk and return go together, so calm stocks should pay calm returns. For decades the data disagreed. Portfolios of low-volatility stocks earned about as much as the market, sometimes more, with much smaller swings. Researchers call it the low-volatility anomaly, and funds built on it gathered tens of billions of dollars in the 2010s.
This lesson states the rules those funds follow and tests the largest of them, the iShares MSCI USA Min Vol Factor ETF (USMV), against SPY from November 2011 to October 2026. The fund did what its name says: it swung less and fell less in every decline. Whether it also delivered the anomaly depends on the half of the test: yes through 2019, and then a long way behind.
The idea
Volatility measures how much a price moves around. The textbook relationship in Risk and return: volatility, drawdowns and the equity premium is that investors must be paid to hold what fluctuates more. Across asset classes that holds: stocks have out-earned bonds, and bonds cash. Within the stock market it has not held. Baker, Bradley and Wurgler sorted US stocks by volatility each month from 1968 to 2008 and found that the calmest fifth had far outgrown the most volatile fifth, which lost money after inflation.
Two explanations are offered. One is behavioral: investors overpay for exciting, volatile stocks in the hope of a large win, as they do for lottery tickets. The other is structural: professional managers are judged against an index, so a stock that returns the same as the market with less risk does nothing for them, and many cannot borrow to scale it up. Frazzini and Pedersen showed that investors who cannot use leverage bid up risky stocks instead, which lowers their future returns.
Note what the anomaly claims. It does not say low-volatility stocks earn more than the market. It says they earn about the same with less risk, so the return per unit of risk is higher. In a strongly rising market a portfolio with a beta of 0.7 is expected to trail. That expectation matters for reading the test.
The rules
There are two common ways to build the portfolio, and they are not the same strategy.
Minimum volatility (USMV's index, MSCI USA Minimum Volatility)
- Start from large and mid-sized US companies.
- Estimate how volatile each stock is and how each moves with every other.
- Have an optimizer find the combination with the lowest expected volatility for the whole portfolio, subject to limits: each sector within a few points of its market weight, a cap on any one stock, and a cap on turnover.
- Rebuild twice a year.
- Reinvest dividends.
Low volatility by ranking (SPLV's index, S&P 500 Low Volatility)
- Take the S&P 500.
- Rank the stocks by how much their daily prices moved over the past year.
- Hold the calmest 100, weighted so that the calmest get the most.
- Rebuild every three months.
- Reinvest dividends.
A person holding either fund follows one rule: buy it and keep it.
The difference is the sector limit. The ranking method has none, so it can hold 20% or more in utilities and almost nothing in technology. The optimizer is held near the market's sector mix and looks for stocks that offset each other. This lesson tests USMV and shows SPLV as the neighbour.
What the backtest shows
The test, stated in full so it can be checked:
- Data. Month-end closes of USMV and SPY from November 2011, USMV's first full month, to October 2026: 180 months.
- Trading. Each fund is bought once and held.
- Costs. One purchase each. The funds' costs are inside their prices: USMV charged 0.15% a year, SPLV 0.25% and SPY 0.0945% in their 2025 prospectuses. Turnover at each rebuild is paid inside the fund.
- Left out. Dividends on both sides, and taxes. Low-volatility funds have yielded a little more than the market (USMV's trailing 12-month yield was 1.37% at the end of October 2026, according to iShares, somewhat above SPY's), so the omission counts against USMV by a few tenths of a point a year. The gap in the table is more than three points.
| Measure, price only | USMV | SPY |
|---|---|---|
| Total return | 295.9% | 504.6% |
| Return per year (CAGR) | 9.7% | 12.8% |
| Deepest fall from a peak | −19.6% | −24.8% |
| Longest time below a peak | 25 months | 23 months |
| Volatility, annualized | 11.3% | 13.9% |
| Beta to SPY | 0.69 | 1.00 |
| Return per unit of volatility | 0.85 | 0.92 |
| Value of $1 at the end | $3.96 | $6.05 |
The risk numbers are what the fund promised. Volatility was about a fifth lower, and the fund moved 0.69% for each 1% move in the market. In the 58 months when SPY's price fell, it fell 3.4% on average and USMV 2.2%. In the 121 months when SPY rose, it rose 3.2% on average and USMV 2.3%.
The return is another matter. One dollar became $3.96 against $6.05. And the last line that matters for the anomaly, return per unit of volatility, was lower for the calm fund than for the market: 0.85 against 0.92. Over these 15 years as a whole, the fund gave up more return than risk.
Two halves
| Period | USMV, per year | SPY, per year | USMV's deepest fall | SPY's deepest fall |
|---|---|---|---|---|
| November 2011 – December 2019 | 12.1% | 12.4% | −8.1% | −14.0% |
| December 2019 – October 2026 | 6.8% | 13.3% | −19.6% | −24.8% |
The split at the end of 2019 gives two different strategies. For eight years the anomaly appeared exactly as described: the market's return, with a worst fall of 8.1% against 14.0%. In the seven years since, the fund earned about half the market's return.
The ratio of the two prices reached 1.060 in June 2016 and fell to 0.614 in May 2026, a relative decline of 42.1%. Over every five-year window in the data, USMV's price was ahead of SPY's in 23 of 120, or 19%.
The neighbour
| Fund | Method | Fund, per year | SPY, same months | Deepest fall | Since December 2019 |
|---|---|---|---|---|---|
| USMV | Optimized, sectors near market weight | 9.7% | 12.8% | −19.6% | 6.8% a year |
| SPLV | The 100 calmest S&P 500 stocks, no sector limit | 7.7% | 12.8% | −21.8% | 4.1% a year |
The simpler ranking method did worse on every measure, including the deepest fall. Its concentration in a few calm sectors left it more exposed, not less, when those sectors fell together in 2020. SPLV's higher yield, which this test leaves out, offsets a part of the gap.
Where it fails: the losing periods
March 2020. A low-volatility fund is bought for the bad months, and this was the worst month of the period. From December 2019 to March 2020 SPY's price fell 19.9% and USMV's 17.7%. The protection was two points. Low volatility is measured from the past, and in a crisis that hit travel, property and utilities' customers all at once, yesterday's calm stocks moved with everything else.
The recovery from it. From March 2020 to March 2021 SPY rose 53.8% and USMV 28.1%. The fund took most of the fall and half of the rebound. That twelve-month stretch accounts for a large part of the whole gap.
Bull markets led by a few volatile stocks.
| Year | USMV | SPY |
|---|---|---|
| 2013 | 22.2% | 29.7% |
| 2020 | 3.5% | 16.2% |
| 2021 | 19.2% | 27.0% |
| 2023 | 8.2% | 24.3% |
| 2024 | 13.8% | 23.3% |
| 2025 | 6.0% | 16.4% |
USMV's price finished ahead of SPY's in four of the fourteen full calendar years: 2014, 2015, 2018 and 2022. Three of the four were years when the market was flat or down, which is the pattern a beta of 0.69 predicts.
Where it did work: from December 2021 to September 2022 SPY fell 24.8% and USMV 18.3%, and in 2018 USMV lost 0.7% while SPY lost 6.3%.
Other ways it fails or misleads:
- Interest rates. Calm stocks are often utilities, consumer staples and other steady dividend payers, which investors treat partly as substitutes for bonds. When bond yields rise, as in 2022 and 2023, those stocks lose that support.
- Popularity. The anomaly depends on calm stocks being neglected. After a decade of inflows into low-volatility funds they were no longer cheap, and a crowded defensive trade is not defensive.
- Trailing is the design, up to a point. A fund with a beta of 0.69 should trail a market that rises 13% a year. It should not trail by as much as it did, and nobody can say in advance which kind of decade is coming.
- A short sample. Fifteen years, against forty in the research. The fund's history contains one fast crash and one slow bear market, and no decline like 2000 to 2002 or 2007 to 2009, where the research found low volatility most valuable.
Compared with its neighbours
| Strategy | Lowers risk by | Deepest fall in its own test | What it gave up |
|---|---|---|---|
| Minimum volatility | Choosing calmer stocks | −19.6% (SPY −24.8%) | 3.1 points a year of price return |
| Dividend growth | Choosing steady dividend raisers | −42.7% (SPY −52.2%) | 0.9 points a year |
| Holding some cash | Owning less of the market | In proportion to the cash | In proportion too: no anomaly needed |
| 10-month average rule | Leaving the market in downtrends | −23.8% (SPY −52.2%) | About a point a year, and many false signals |
The third row is the honest benchmark for this strategy. A portfolio of 70% SPY and 30% cash has a beta of 0.7 by construction. A low-volatility fund earns its fee only if it beats that mix, which is what the anomaly claims and what the second half of this test did not show.
How Finabase does it
Finabase does not rank stocks by volatility and no Finabase tool says what to buy.
- Stock pages. Each stock's page shows its beta, the number closest to what these funds sort on. A beta well under 1 marks a stock that has moved less than the market.
- Chart. The chart draws Average True Range and Bollinger Bands, two ways of seeing how much a price moves (Average True Range (ATR): volatility, stops and position size, Bollinger Bands: formula, the squeeze and walking the bands). Put USMV and SPY side by side over two years to see the difference in swing.
- Screener. The screener filters on market capitalization, P/E, dividend yield and volume. It has no volatility or beta field, so it cannot build the list in rule 2.
- Backtest and Strategy. These tools replay indicator signals and if/then rules on one symbol, buying a fixed $10,000 per signal with no trading cost; Strategy covers two years. They do not build portfolios, and two years would have shown either half of this test, not both.
Frequently asked questions
Do low-volatility stocks beat the market?
Not in this test. USMV's price rose 9.7% a year from November 2011 to October 2026 against 12.8% for SPY. The research claim is narrower: about the same return with less risk. That held through 2019 (12.1% against 12.4%) and failed afterwards (6.8% against 13.3%).
Is a low-volatility fund protection in a crash?
Partly, and less than the name suggests. USMV fell 18.3% against 24.8% in the 2022 bear market, but 17.7% against 19.9% in the 2020 crash. It holds stocks, and in a fast market-wide fall stocks move together.
What is the difference between minimum volatility and low volatility?
Minimum volatility uses an optimizer to find the calmest portfolio, with sectors held near their market weights. Low volatility by ranking holds the 100 calmest stocks with no sector limit. In this test the ranked fund, SPLV, returned 7.7% a year against USMV's 9.7%.
Why did low volatility fall so far behind after 2019?
Its holdings took most of the 2020 crash and little of the rebound, then missed the technology-led gains of 2023 and 2024, while higher interest rates weighed on the steady dividend payers it favors. Some of that is what a low-beta fund does in a strong market; some is more than beta explains.
Summary
Low-volatility investing holds the calmest stocks, either through an optimizer (USMV) or a simple ranking (SPLV). Tested through USMV from November 2011 to October 2026, on prices without dividends, it delivered its risk promise: a fifth less volatility, a beta of 0.69 and a deepest fall of 19.6% against 24.8%. It did not deliver the market's return: 9.7% a year against 12.8%, and less return per unit of volatility than the market itself. Through 2019 the two were level and the anomaly looked real; since then the fund has earned half as much, after protecting little in the 2020 crash. Fifteen years of one fund cannot decide which half is the exception, and the fund's somewhat higher dividends, left out here, would recover only a few tenths of a point.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
What does the low-volatility anomaly actually claim?
- Calm stocks earn more than the market every year
- Calm stocks have earned about the market’s return with less risk, so more return per unit of risk
- Calm stocks never fall in a bear market
Show the answer
B. Calm stocks have earned about the market’s return with less risk, so more return per unit of risk The claim is about risk-adjusted return, not beating the market outright. A portfolio with a beta near 0.7 is expected to trail in a strongly rising market and to fall less in a falling one.
USMV had a beta of 0.69 and returned 9.7% a year against 12.8% for SPY, with 0.85 of return per unit of volatility against 0.92. What does that say about the 15 years as a whole?
- The fund reduced risk, and gave up proportionally more return than risk
- The fund reduced risk at no cost
- The fund was riskier than the market
Show the answer
A. The fund reduced risk, and gave up proportionally more return than risk Lower volatility and smaller falls were delivered. But the anomaly needs the calm portfolio to keep more return than its lower risk implies, and over the full period the ratio was below the market’s.
The fund matched the market through 2019 and earned about half as much afterwards. What is the sound conclusion?
- The first half is the true result and the second is noise
- The strategy stopped working in 2020
- The result depends on the period, and 15 years of one fund cannot say which half is typical
Show the answer
C. The result depends on the period, and 15 years of one fund cannot say which half is typical Either half alone would support a confident story. A test that changes its answer when split is telling you its evidence is thin, which is why the period is always split at least once.
Why is "70% SPY, 30% cash" the fair comparison for a fund with a beta of about 0.7?
- Because both have the same fee
- Because it reaches the same market exposure with no stock selection, so the fund is only worth having if it beats that mix
- Because cash never loses value
Show the answer
B. Because it reaches the same market exposure with no stock selection, so the fund is only worth having if it beats that mix Anyone can lower risk by holding less of the market. A low-volatility fund has to do better than that simple mix to justify itself, and doing better is exactly what the anomaly asserts.
Sources and further reading
- Malcolm Baker, Brendan Bradley and Jeffrey Wurgler, “Benchmarks as Limits to Arbitrage: Understanding the Low-Volatility Anomaly,” Financial Analysts Journal 67(1), 2011
- Andrew Ang, Robert J. Hodrick, Yuhang Xing and Xiaoyan Zhang, “The Cross-Section of Volatility and Expected Returns,” Journal of Finance 61(1), 2006
- Andrea Frazzini and Lasse H. Pedersen, “Betting Against Beta,” Journal of Financial Economics 111(1), 2014
- MSCI — MSCI USA Minimum Volatility (USD) Index methodology
- iShares — MSCI USA Min Vol Factor ETF (USMV) fund page: expense ratio 0.15%, 12-month trailing yield 1.37% as of October 31, 2026, inception October 18, 2011 (read November 2026)
- S&P Dow Jones Indices — S&P 500 Low Volatility Index methodology
- Finabase chart data — USMV, SPLV and SPY monthly bars to October 2026; results computed by Finabase
