Quality Minus Junk (QMJ): What the Quality Factor Adds to Small-Cap Value

Everyone agrees a profitable, growing, financially sound company deserves a higher price than a shaky one. The question Cliff Asness, Andrea Frazzini and Lasse Heje Pedersen asked in “Quality Minus Junk” was whether the market pays enough of a premium for those traits. Across 24 countries and six decades, their answer was no: high-quality stocks earned more than their prices implied.¹

That result now shapes how most systematic small-cap value funds pick stocks, and it gets cited as a reason to buy “quality” ETFs. Two newer pieces of evidence change how much weight it deserves. The factor has been close to flat since the paper’s sample ended in 2016. And when we regressed fund returns on it, the profitability-screened small-cap value funds carried as much quality exposure as the ETFs sold as quality.

What QMJ measures

The paper defines a quality stock as one investors should be willing to pay more for, all else equal, and scores every stock on three groups of characteristics:¹

• Profitability: gross profits over assets, return on equity, return on assets, cash flow over assets, gross margin, and low accruals.
• Growth: the five-year change in five of those measures (all but accruals), computed on residual profits per share.
• Safety: low market beta, low leverage, low bankruptcy risk (Ohlson’s O-score and Altman’s Z-score), and low volatility of return on equity.

An earlier working-paper version also included payout: low net equity and debt issuance and a high share of profits returned to shareholders. The published version drops payout from the headline score and tests it only as an alternative.¹

Each component is converted to a rank, the ranks are standardized and averaged, and the result is a single quality score. The factor itself is built the way Fama and French build theirs: stocks are split into small and large, then into high-, middle- and low-quality groups, and QMJ is the average return of the two high-quality portfolios minus the average of the two junk portfolios. It is long-short and self-financing, so its return is the spread between two groups of stocks. No single fund delivers that spread.

“Quality” means different things to different index providers, and the differences matter more than the label:

QUAL’s three measures overlap with QMJ’s profitability and safety legs, but it has no growth leg, no short side, and holds large and mid caps weighted to match the market’s sectors. Evidence for QMJ is not evidence for any particular fund with “quality” in its name.

Why quality should cost more, and why it hasn’t cost enough

A stock’s price-to-book ratio should rise with how much profit the company earns on its capital, how fast that profit grows, how much of it reaches shareholders, and how safe the stream is. Fama and French used the same valuation logic to predict that, holding price-to-book and investment fixed, more profitable firms should have higher expected returns.⁷ A high price for quality is expected. The puzzle is whether the price is high enough.

Asness, Frazzini and Pedersen found that quality stocks do trade at higher prices, but quality explained only about 10% of the variation in price-to-book across stocks on average, and at most about 49% in the US once other controls were added. A portfolio buying quality and shorting junk earned a cumulative four-factor alpha over the five years after formation of about 21% in the US and 22% globally, which is what you would see if the market underreacts to quality.¹

The evidence in the paper

The published sample covers 54,616 stocks in 24 developed markets: the US from June 1957 and the broad global sample from June 1989, both through December 2016.¹

Source: Asness, Frazzini and Pedersen (2019), Tables 4 and 5.¹

The alpha rises as factors are added because QMJ loads negatively on the market, on size and on value, and in the paper’s data on momentum too: quality stocks tend to be larger, lower-beta and more expensive than junk. Controlling for those exposures makes the result look stronger. QMJ returns were positive in 23 of the 24 countries (New Zealand was the exception), and the four-factor alpha was statistically significant in 18.¹

The last row is the one to keep in mind. When Fama and French’s profitability (RMW) and investment (CMA) factors are added, the US alpha falls from 0.60% to 0.33% a month and QMJ’s loading on RMW is 0.55. The residual is still statistically significant, so QMJ is more than RMW, but profitability accounts for a large share of it.¹

Our rerun on AQR’s current QMJ file and Ken French’s factors comes close: a four-factor alpha of 0.53% a month for 1957 to 2016, and a five-factor-plus-momentum alpha of 0.36% a month for 1963 to 2016 with an RMW loading of 0.56. The four-factor gap comes mostly from momentum: the paper found a negative momentum loading, while the current file loads positively on Ken French’s momentum factor. AQR rebuilds the full history on each update, and its current file also shows a higher raw return for 1957 to 2016 than the paper did (0.36% a month against 0.29%).⁸

Risk or mispricing

A factor with a return premium is usually explained as pay for bearing a risk that hurts in bad times. QMJ is awkward for that story. It had negative market beta, so it tended to gain when the market fell, and its long side holds the firms with the lowest leverage and bankruptcy risk. The paper tested whether quality does especially well in the worst markets, beyond what its beta implies. The evidence was weak: the effect was marginally significant for the profitability component in the US sample (t = 2.4) and insignificant for QMJ as a whole.¹

The paper offers three pieces of evidence that point toward mispricing. Sell-side analysts set higher target prices for quality stocks, but their targets implied lower returns for quality than for junk, the reverse of what followed. Short sellers concentrated on junk, and junk was more expensive to borrow, which limits how much arbitrageurs can push junk prices down.¹ The safety leg overlaps with the low-beta anomaly, which Frazzini and Pedersen attribute to investors who cannot or will not use leverage and so bid up high-beta stocks to get more expected return.⁹ None of this proves the premium is mispricing. An unmeasured risk could still be at work, and the paper does not claim a single settled explanation.

Which parts of quality earn a premium

Jason Hsu, Vitali Kalesnik and Engin Kose sorted the many definitions of quality used by index providers and researchers into groups and tested each. Profitability, accounting quality, payout and dilution, and investment all showed return premia. Capital structure (leverage), earnings stability and growth in profitability showed little evidence of one, and profitability and investment captured most of the quality premium.¹⁰

That finding cuts into two of QMJ’s three legs. The growth leg is growth in profitability, and much of the safety leg is leverage and earnings volatility. It also cuts into QUAL, which ranks stocks on two of the three weak categories: earnings variability and leverage.

What QMJ has done since the paper

The published sample ends in December 2016. AQR still updates the series, and we used its file (data through July 2026) to measure what came after. These are raw long-short returns, which is what an investor in the factor would have earned before costs, not alphas.¹¹⁸

Summitward calculation from AQR’s Quality Minus Junk: Factors, Monthly file, USA column, downloaded September 29, 2026. AQR reconstructs the history on each update, so later downloads can differ.⁸

The average fell by more than half and the volatility rose by half, so the Sharpe ratio dropped from 0.58 to 0.15. The factor lost money in five of the nine full calendar years from 2017 through 2025, with the worst year in 2020 and the best in 2022. And July 2026 alone, a +14.3% month that was the largest in the 69-year US series, turned a near-zero post-2016 average into a small positive one. Ken French’s independently built RMW factor gained 11.1% the same month and his momentum factor lost 12.2%, so the swing shows up outside AQR’s file too.

Global QMJ held up better: 0.47% a month from July 1989 to December 2016 against 0.23% from January 2017 to July 2026, a Sharpe ratio of 0.33 in the later period.⁸ Regressing the post-2016 US series on the five Fama-French factors plus momentum gives an alpha of 0.20% a month with a t-statistic of 1.35 and an RMW loading of 0.86, higher than before 2017. In that regression, the RMW exposure and the (insignificant) alpha each account for about 0.20% a month, and the negative market beta subtracts about 0.30% in a period when stocks did well.

Nine and a half years is not long enough to reject the factor. If QMJ’s true Sharpe ratio were 0.47 (the paper’s US figure) or 0.58 (AQR’s current file for the same years), a 9.6-year stretch at 0.15 or worse would happen by chance about 9% to 16% of the time. The broader research on published anomalies gives two reasons to expect less than the backtest, though. David McLean and Jeffrey Pontiff found that across 97 return predictors, returns fell by 26% out of sample and by 58% after publication.¹² Kewei Hou, Chen Xue and Lu Zhang replicated 452 anomalies and found 65% insignificant at the usual 5% threshold when portfolios use NYSE breakpoints and value weights, which limit the influence of microcaps, and 82% below a stricter multiple-testing cutoff.¹³ On the other side, Theis Jensen, Bryan Kelly and Pedersen found that most factors replicate in US and global data covering 93 countries, grouped into 13 themes that include quality and profitability.¹⁴

Quality and small-cap value

QMJ’s negative size and value loadings make it look like the opposite of small-cap value. The research points the other way. In “Size Matters, If You Control Your Junk,” Asness and coauthors showed that small stocks are disproportionately junk, and that the junk was hiding the size premium. (That paper used the earlier QMJ definition, which included payout.) From 1957 to 2012, the size factor’s alpha against the market (current and lagged), value and momentum was 0.14% a month with a t-statistic of 1.23. Adding QMJ raised it to 0.49% a month with a t-statistic of 4.89. Using Fama and French’s RMW and CMA in place of QMJ, from 1963, raised it from 0.16% to 0.33% a month, so profitability and investment do much of the same job.¹⁵ Robert Novy-Marx found the same pattern for value: controlling for gross profitability sharply improved value strategies.¹⁶ Fama and French’s five-factor model struggles most with small stocks that invest heavily despite low profitability.¹⁷

The practical version of that research is already in the funds. Avantis and Dimensional both rank small-value stocks on profitability as well as price. The small-cap value guide, the VBR vs. AVUV comparison and the RMW guide cover that argument in detail. The question here is narrower: how much quality exposure do those funds end up with, compared with the funds sold as quality?

How much quality your funds already hold

We regressed monthly returns of three screened small-cap value funds, three index small-cap value funds, three quality ETFs and the total market on two models. The first is the Fama-French five factors plus momentum, where the RMW loading measures profitability exposure. The second swaps RMW and CMA for QMJ, where the QMJ loading measures quality exposure as AQR defines it. A loading of 0.13 on QMJ means the fund moved as if it held 13 cents of the long-short QMJ portfolio for every dollar invested.⁸

Summitward calculation, November 2019 to July 2026 (81 months). *DFSV from April 2022 (52 months) and DFAT from August 2021 (60 months), their full histories. SMB, HML and RMW from the five-factor plus momentum model; QMJ from a model with market, SMB, HML, momentum and QMJ. Newey-West t-statistics in parentheses. Fund returns from Yahoo Finance adjusted closes; factors from Ken French and AQR.⁸

The screened small-value funds carried QMJ loadings of 0.13 to 0.17, the same range as QUAL and JQUA, and higher RMW loadings than any of the quality ETFs. They got there while also holding size loadings near 0.9 and value loadings of 0.4 to 0.55. The unscreened Russell 2000 Value fund had no measurable quality or profitability exposure in this window.

Some rows need care. These windows are short, and the t-statistics on loadings around 0.1 are modest. IJS shows almost no quality exposure from 2019, but over its full history since 2000 its QMJ loading is 0.13 (t = 3.3). The S&P 600’s positive-earnings requirement, which applies only when a stock is added to the index, may explain a tilt that shows up in some periods and not others; our regressions do not test that.¹⁸ SPHQ’s loading is 0.04 since 2019 and 0.13 over its full history from 2006, the latter with a t-statistic of only 1.7. And when QMJ and RMW are both in the model, QMJ adds nothing significant for any of the small-value funds, which means their quality exposure is profitability exposure. JQUA is the one fund whose QMJ loading survives with RMW in the model (0.19, t = 3.4).⁸

Because QMJ loads negatively on value, a quality fund might be expected to cancel part of a small-value portfolio’s value tilt. In these numbers it would not: QUAL, SPHQ and JQUA had value loadings between −0.02 and 0.01. Adding one dilutes a small-value tilt the way adding any market-like fund would, by taking up portfolio weight, and adds a small amount of profitability exposure the screened funds already provide.

No retail fund holds the paper’s portfolio. Capturing QMJ as measured requires shorting junk, which is where borrow costs are highest. The nearest AQR mutual fund, AQR Equity Market Neutral, blends several signals and reported total annual expenses of about 6% in its 2026 prospectus once dividends paid on short positions were included.¹⁹

When quality justifies an allocation

The research supports tilting toward profitable, conservatively investing firms, especially within small and cheap stocks. It does not support treating every quality ETF as a source of the paper’s returns, and QMJ’s post-2016 record is a reason to size any tilt modestly. How that plays out depends on what you already own:

• You own a screened small-cap value fund (AVUV, DFSV, DFAT, or their international versions). You already carry about as much quality exposure as QUAL provides, alongside the size and value tilts you chose. A separate quality fund adds cost and complexity and dilutes those tilts. We would not add one.
• You own an unscreened small-value index (IWN, and VBR to a lesser degree). You hold the part of the small-cap universe the junk research warns about. The direct fix is replacing the fund with a profitability-screened one, subject to taxes on any gains in a taxable account. A large-cap quality ETF next to IWN would not remove the junk inside IWN.
• You own only a total-market fund. A quality ETF is a defensible way to add a profitability and low-leverage tilt without taking on small-cap volatility. Expect a small effect: QUAL’s QMJ loading of 0.14 is a mild tilt, and it comes with tracking error against the market.
• You invest in long-short or alternative strategies. Full QMJ is a diversifier with negative market beta and a long in-sample record. It is available only inside multi-signal market-neutral funds with high all-in costs, and its recent decade has been flat.

In every case, pick the allocation and hold it. A factor that can go nine years with a Sharpe ratio of 0.15 will tempt you to drop it at the wrong time.

How Summitward helps

The Portfolio page runs a Fama-French factor regression on your own holdings and reports the RMW loading, the profitability exposure behind much of QMJ. If your small-value sleeve shows an RMW loading near zero, it is closer to IWN than to AVUV.

Key Takeaways

• QMJ had a strong in-sample record. From 1957 to 2016 it beat a four-factor model by 0.60% a month in the US (t = 9.95), with positive returns in 23 of 24 countries.
• Much of it is profitability. Adding RMW and CMA cuts the US alpha from 0.60% to 0.33% a month, and the quality categories with the best independent evidence are profitability, accounting quality, payout and investment.
• It has been flat since 2016. AQR’s US QMJ averaged 0.14% a month from January 2017 to July 2026 (t = 0.45), and nearly all of that came from July 2026.
• Screened small-value funds already carry it. From November 2019 to July 2026, AVUV’s QMJ loading of 0.13 matched QUAL’s 0.14, while IWN’s was about zero.
• Fix junk at the fund level. If your small-value fund does not screen for profitability, switching funds addresses the problem more directly than adding a quality ETF.

Frequently Asked Questions

What is the quality minus junk factor?

A long-short portfolio from AQR researchers Asness, Frazzini and Pedersen that buys stocks scoring high on profitability, profit growth and safety and shorts the lowest scorers, within both small and large stocks. AQR publishes its monthly returns for the US, 23 other countries and several regions.

Is QMJ the same as the Fama-French profitability factor?

No, but they overlap heavily. RMW sorts on one measure, operating profitability. QMJ combines six profitability measures with growth and safety. QMJ’s RMW loading was 0.55 in the paper and 0.86 in our post-2016 regression, and its alpha after controlling for RMW is about half its four-factor alpha.

Is QUAL a quality minus junk fund?

No. QUAL holds US large and mid caps ranked on return on equity, earnings variability and leverage, with sector weights matched to the market and no short positions. From November 2019 to July 2026 its QMJ loading was 0.14, a mild tilt toward the factor.

Does AVUV screen out junk stocks?

It ranks small-value stocks on profitability, defined in its prospectus mainly as adjusted cash from operations to book value, alongside valuation. In our regression from November 2019 to July 2026 its RMW loading was 0.23 and its QMJ loading was 0.13, against about zero for the Russell 2000 Value index fund IWN.

Has the quality factor stopped working?

Its US version has been close to flat since the paper’s sample ended, averaging 0.14% a month from 2017 to July 2026 with a t-statistic of 0.45. Global QMJ did better at 0.23% a month. A decade is too short to reject a factor with a 60-year record, but the post-2016 period is consistent with the research showing that published anomalies earn less after publication.

Should I buy a quality ETF if I already own small-cap value?

If your small-value fund screens for profitability, probably not; it already carries similar quality exposure. If it does not, replacing it with a screened fund addresses the junk problem more directly than adding a large-cap quality ETF.

Related Guides

• Beyond Fama-French: q-Factors and AQR’s Factors for how QMJ overlaps with RMW and the q-factor ROE factor, and how fund alphas change across seven factor models.
• Cheap for a Reason: Value Traps for the value-trap evidence: distress, the F-score, and a 63-year test of cheap, unprofitable stocks.
• RMW Explained: The Profitability Factor for the factor behind most of QMJ, and how Avantis and Dimensional measure it.
• Small-Cap Value for the long-run evidence on the size and value premiums.
• VBR vs. AVUV for the cost of profitability screening and when the index fund is still the right choice.
• Minimum Volatility Investing for Betting Against Beta, which overlaps with QMJ’s safety leg.
• Cash-Based Operating Profitability for the profitability measure Avantis uses.
• The ABCs of Systematic and Factor Investing for where quality sits among the other documented factors.

Sources

1. Clifford S. Asness, Andrea Frazzini and Lasse Heje Pedersen, “Quality Minus Junk,” Review of Accounting Studies 24(1), 2019, 34–112. Tables 2, 4, 5, 6, 7 and 19; Section 6.3 for payout. doi.org
2. iShares MSCI USA Quality Factor ETF, summary prospectus (SEC EDGAR accession 0001193125-25-302174). sec.gov
3. Invesco S&P 500 Quality ETF, summary prospectus dated August 28, 2026. sec.gov
4. JPMorgan U.S. Quality Factor ETF, fact sheet as of August 31, 2026. am.jpmorgan.com
5. Avantis U.S. Small Cap Value ETF, summary prospectus dated January 1, 2026. sec.gov
6. Dimensional U.S. Small Cap Value ETF, summary prospectus dated February 28, 2026, and supplement dated September 25, 2026 moving the DFSV ticker to the ETF share class of the U.S. Small Cap Value Portfolio with an expense cap of 0.27% from November 1, 2026. sec.gov, supplement
7. Eugene F. Fama and Kenneth R. French, “Profitability, Investment and Average Returns,” Journal of Financial Economics 82(3), 2006, 491–518. doi.org
8. Summitward calculations: QMJ and BAB returns from AQR, factor and T-bill returns from the Ken French data library (202608 CRSP database), fund returns from Yahoo Finance dividend-adjusted closes. Script, committed French data and printed output (AQR data is downloaded by the script, not redistributed): summitward-research
9. Andrea Frazzini and Lasse Heje Pedersen, “Betting Against Beta,” Journal of Financial Economics 111(1), 2014, 1–25. doi.org
10. Jason Hsu, Vitali Kalesnik and Engin Kose, “What Is Quality?” Financial Analysts Journal 75(2), 2019, 44–61. doi.org
11. AQR Capital Management, “Quality Minus Junk: Factors, Monthly,” updated through July 31, 2026, and “Betting Against Beta: Equity Factors, Monthly.” aqr.com
12. R. David McLean and Jeffrey Pontiff, “Does Academic Research Destroy Stock Return Predictability?” Journal of Finance 71(1), 2016, 5–32. doi.org
13. Kewei Hou, Chen Xue and Lu Zhang, “Replicating Anomalies,” Review of Financial Studies 33(5), 2020, 2019–2133. doi.org
14. Theis Ingerslev Jensen, Bryan Kelly and Lasse Heje Pedersen, “Is There a Replication Crisis in Finance?” Journal of Finance 78(5), 2023, 2465–2518. doi.org
15. Clifford Asness, Andrea Frazzini, Ronen Israel, Tobias J. Moskowitz and Lasse H. Pedersen, “Size Matters, If You Control Your Junk,” Journal of Financial Economics 129(3), 2018, 479–509. doi.org
16. Robert Novy-Marx, “The Other Side of Value: The Gross Profitability Premium,” Journal of Financial Economics 108(1), 2013, 1–28. doi.org
17. Eugene F. Fama and Kenneth R. French, “A Five-Factor Asset Pricing Model,” Journal of Financial Economics 116(1), 2015, 1–22. doi.org
18. S&P Dow Jones Indices, “S&P U.S. Indices Methodology,” July 2026: positive GAAP net income for the most recent quarter and the trailing four quarters, applied at addition, not for continued membership. spglobal.com
19. AQR Equity Market Neutral Fund, summary prospectus, 2026. sec.gov

Author disclosure

Summitward has no business relationship with AQR, Avantis, Dimensional, Vanguard, BlackRock, Invesco, JPMorgan or any firm mentioned here and receives no compensation from them. Figures labeled as Summitward calculations are historical measurements over the stated window, not forecasts. Nothing here is investment advice.