Key takeaways
- Five iShares factor ETFs trailed the index each one tracks by 0.13 to 0.21 percentage points a year since launch, measured to 30 June 2026, against a 0.15% fee. The funds did their job.
- All four MSCI US factor indices in this piece show a smaller edge over the plain MSCI USA index since they launched than across their full published record: quality goes from +0.43 to -0.52 points a year.
- MSCI's momentum index beat its parent by 2.65 points a year over a record starting in 1994, and by 0.44 points across the twelve full calendar years since its 2013 launch.
- The Enhanced Value index is ahead by 1.26 points a year over its full record and behind by 4.27 points across the eleven calendar years since its December 2014 launch.
- Between 52% and 59% of each index's published record predates its own launch date. MSCI labels that stretch back-tested on every factsheet.
The funds tracked their index. The index didn't track the backtest.
You are looking at a momentum fund or a value fund, the long-run chart is persuasive, and you want to know whether the product delivered what the research promised. The answer splits in two, and only one half is about the fund.
The fund half is clean. Five iShares factor ETFs trailed the index each one tracks by between 0.13 and 0.21 percentage points a year since inception, measured to 30 June 2026. Each charges 0.15% a year. Whatever went missing, it wasn't lost in the plumbing.
The index half is where the story is. Take four MSCI US factor indices from the same family: momentum, enhanced value, sector neutral quality and minimum volatility. Every one of them looks better over its full published record than over the calendar years it has actually existed. MSCI's momentum index goes from 2.65 percentage points a year ahead of the market to 0.44. Its value index goes from 1.26 points ahead to 4.27 points behind. That is the gap between a backtest and a live record, and it's the gap that decides what you get.
Factor ETF returns landed within 0.21 points a year of the index each fund tracks
Start with the part that works. Every iShares factsheet prints the fund's annualised return since inception next to the index it is benchmarked to, on the same page and the same date. Here is that comparison for the five US single-factor funds plus the plain S&P 500 tracker, all to 30 June 2026.
| Fund | Launched | NAV since inception | Benchmark | Gap | Fee |
|---|---|---|---|---|---|
| MTUM, momentum | Apr 2013 | 16.79% | 17.00% | 0.21pp | 0.15% |
| VLUE, value | Apr 2013 | 13.79% | 13.99% | 0.20pp | 0.15% |
| QUAL, quality | Jul 2013 | 13.81% | 14.00% | 0.19pp | 0.15% |
| SIZE, low size | Apr 2013 | 11.89% | 12.03% | 0.14pp | 0.15% |
| USMV, min vol | Oct 2011 | 11.60% | 11.73% | 0.13pp | 0.15% |
| IVV, S&P 500 | May 2000 | 8.41% | 8.46% | 0.05pp | 0.03% |
The five factor funds averaged 0.17 percentage points a year of shortfall against a 0.15% fee. Two of them, SIZE and USMV, came in under their own fee. BlackRock doesn't break out securities lending revenue on the factsheet, so the reason isn't visible there, and it would be guesswork to name one. What is visible is the size of the number.
Factor ETF returns are not where the promise leaked. A fund whose annual gap to its benchmark runs about 0.02 points wider than its fee is doing what a tracker is built to do. If you want the fuller version of that measurement, the difference between a fund's realised gap and the noise around it is the subject of tracking difference vs tracking error.
Every one of the four MSCI factor indices did better before it existed
Now the index. MSCI publishes two things on the same factsheet that are rarely read together: the index's launch date, and a performance record that starts decades earlier. The gap between them is back-tested. MSCI says so in the small print: "There are frequently material differences between back-tested performance and actual results."
Between 52% and 59% of each of these four records predates the index's own launch. So the record can be split. The chart above puts the two halves side by side. Every figure is a net return, and in each row the index and the MSCI USA parent are measured on the same basis.
| Index | Launched | Full record vs MSCI USA | Live calendar years vs MSCI USA |
|---|---|---|---|
| MSCI USA Momentum | Feb 2013 | +2.65pp a year, from 1994 | +0.44pp a year, 2014 to 2025 |
| MSCI USA Enhanced Value | Dec 2014 | +1.26pp a year, from 1997 | -4.27pp a year, 2015 to 2025 |
| MSCI USA Sector Neutral Quality | Dec 2014 | +0.43pp a year, from 1998 | -0.52pp a year, 2015 to 2025 |
| MSCI USA Minimum Volatility | Jun 2008 | -1.01pp a year, from 1988 | -3.37pp a year, 2012 to 2025 |
Four indices, four deteriorations. In raw terms, momentum compounded at 13.34% a year from 2014 to 2025 against the parent's 12.90%. Enhanced Value compounded at 8.64% from 2015 to 2025 against 12.91%. Quality managed 12.39% against the same 12.91%, and minimum volatility 10.96% from 2012 to 2025 against 14.32%.
None of that says the indices are broken. It says the part of the record that was computed after the rules were fixed reads differently from the part that was computed before, and the direction is the same in all four cases.
The premium by decade, straight from the Fama-French library
Before deciding that something decayed, it helps to see how much these premia move about on their own. The Fama-French factors are the standard academic measurement. SMB is small stocks minus big ones, HML is value minus growth, and Mom is recent winners minus recent losers. Kenneth French publishes monthly factor returns from July 1926 and annual ones from 1927, and the table below compounds his annual rows into a return per year for each decade.
| Period | SMB, size | HML, value | Mom, momentum |
|---|---|---|---|
| 1930s | 9.23% | -1.87% | 7.42% |
| 1940s | 5.53% | 10.18% | 6.85% |
| 1950s | -0.64% | 4.85% | 11.19% |
| 1960s | 5.36% | 3.71% | 11.93% |
| 1970s | 3.38% | 7.32% | 8.78% |
| 1980s | -0.57% | 6.25% | 9.26% |
| 1990s | -2.54% | -0.41% | 15.27% |
| 2000s | 4.51% | 7.20% | -12.08% |
| 2010s | -0.76% | -2.84% | 2.92% |
| 2020 to 2025 | -4.16% | -5.39% | 1.24% |
| 1927 to 2025 | 1.68% | 3.06% | 6.70% |
Read down the value column. Four of the ten periods are negative, including the two most recent. The full-sample figure of 3.06% a year sits on top of decades that range from 10.18% to -5.39%. The momentum column is wilder still: 15.27% a year in the 1990s, then -12.08% a year across the 2000s. One year did most of that damage. French's momentum factor returned -83.48% in 2009. These are historic measurements of what happened, not forecasts of what comes next, and the spread between decades is the reason that distinction matters. Our separate readings of the value premium, the momentum premium and the size premium go through each series on its own terms.
So a weak live decade for a factor index is not automatically evidence that a factor index is a bad product. Some of it is the factor itself having a weak decade, which the table says it does regularly.
A long-only index is not the long-short portfolio the premium was measured on
Here's the mechanism that connects the two halves, and it's the thing most factor marketing skates over. French defines HML as an arithmetic difference: "the average return on the two value portfolios minus the average return on the two growth portfolios." That construction comes from Fama and French, 1993, "Common Risk Factors in the Returns on Stocks and Bonds", and the data library still points at it. The short leg is not optional. It is half the definition. Momentum is built the same way: "the average return on the two high prior return portfolios minus the average return on the two low prior return portfolios", ranked on what French calls "prior (2-12) returns", meaning the twelve months ending two months back.
A factor investing ETF has no short leg. The MSCI USA Momentum SR Variant Index, which MTUM now tracks, held 126 stocks out of the parent's 525 at 31 August 2026. It is the parent index with the weights pushed around, so it picks up part of the factor and keeps most of the market underneath it. That is what makes the arithmetic comparable but not equal.
Put numbers on it. Over the twelve calendar years from 2014 to 2025, the Fama-French long-short momentum factor returned 1.21% a year. MSCI's long-only momentum index beat its parent by 0.44 points a year over the same twelve years, roughly 36% of the long-short figure. A long-only tilt can only load partly on a long-short factor, which follows from the definition above, and it is the honest reason a smart beta fund cannot hand you the number in the academic table.
The value case is less comfortable. The Fama-French HML factor lost 5.15% a year from 2015 to 2025. MSCI's Enhanced Value index trailed its parent by 4.27 points a year over the same eleven years, which is most of the long-short loss rather than a fraction of it. That's what concentration does. It works in both directions, and the same concentration shows up on the momentum sheet, where information technology was 47.63% of the index at 31 August 2026 and Micron Technology alone held 6.84% against a 1.63% weight in the parent.
The strongest objections to reading it this way
Three of them, and the first is the most serious.
Minimum volatility was never built to beat the market. Judging it on return alone is judging a seatbelt on lap times. Over the record from 1988, MSCI's min vol index carries a beta of 0.72, a Sharpe ratio of 0.59 against the parent's 0.56, and a maximum drawdown of 47.18% against 55.36%. Its ten-year annualised standard deviation is 12.23% against 15.59%. USMV's own three-year equity beta was 0.49 at 30 June 2026, against 1.22 for MTUM. On the measure it was designed for, it did the job, and the -3.37 point return gap is partly the price of carrying a beta of 0.72 through a strong decade for the market. The low volatility anomaly was always a risk-adjusted claim.
Eleven years is not a verdict. Two of these four indices launched in December 2014. A premium measured across 99 years cannot be refuted by a sample that short, and the decade table above shows why. Each of these three factors has at least one losing decade in the table, and each has followed one with a positive decade.
Value has already turned. Over the five years to 31 August 2026, MSCI's Enhanced Value index returned 16.29% a year against the parent's 11.85%. VLUE's one-year NAV return to 30 June 2026 was 80.03%, against 22.29% for IVV. Anyone who read the 2015 to 2025 row as a death notice missed that. A live record that flips this hard in twelve months is a live record that is still too short to settle anything.
What this evidence cannot tell you
These are the limitations worth carrying away, and the first one is a caveat about the arithmetic itself.
The live-era figures in the index table are ours, not MSCI's. We compounded the annual net returns MSCI prints on each factsheet, starting from the first full calendar year after each index launched. MSCI does not publish a "since launch" column, which is precisely the point, but it does mean the figures carry our compounding rather than its own.
One row needs a footnote of its own. MTUM's benchmark today is the MSCI USA Momentum SR Variant Index, which MSCI launched on 19 May 2020 and which "selects top 125 securities from the MSCI USA Index". The momentum row in the table above measures the MSCI USA Momentum Index, launched on 15 February 2013, because that is the one with a long enough live record to measure. They are two different indices with two different launch dates.
Minimum volatility is the weakest row. Its factsheet's annual table starts at 2012, so the 2012 to 2025 measurement drops the first three and a half years of a record that began in June 2008, and those years contain the recovery from the financial crisis.
The value comparison mixes a total-return factor with a dividend-paying index. Enhanced Value yielded 1.91% against the parent's 1.10% at 31 August 2026, so the two are close on a net-return basis but the underlying factor is not measured the same way. Nothing here is risk-adjusted either. Comparing a 1.22-beta momentum index to its parent is not a like-for-like test, and the same applies in reverse to min vol.
Everything above is US large and mid cap, in dollars, over a period dominated by one market's technology sector. It's a single sample of a single market. Historic premia are measurements of the past and carry no forecast of the future, which is a rule MSCI states on its own factsheets and the Fama-French decade table demonstrates without needing anyone's help.
What would change the conclusion
If the next decade's factor returns look like the 1970s rather than the 2010s, the live records repair themselves and this reading dates badly. HML paid 7.32% a year in the 1970s. It has paid -5.39% a year since 2020. Nothing in the table rules out a reversal, and the 16.29% five-year run in Enhanced Value may already be one.
If the fund-level gap widened past the fee, the conclusion would move from the index to the product. It hasn't: the widest of these five gaps is 0.21 points against a 0.15% charge. A gap of 0.40 or 0.50 points would be a different conversation about turnover and capacity.
If an index kept its back-tested edge through a full live cycle, the pattern here would look like a coincidence of one weak decade rather than anything structural. That test is furthest along for minimum volatility, live since June 2008, and so far it has gone the other way.
The number to watch isn't the ten-year chart on the fund page. It's the launch date printed two lines below it, and how much of the chart sits to the left of that line. Every MSCI factsheet gives you both, and a fund's factsheet gives you the gap between the index and what you actually received. LedgerTouch tracks the second number continuously. The first one is printed on the factsheet and never changes.