Skip to main content

Quality Factor

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,094 words

The quality factor is the empirical tendency for stocks of profitable, financially sound, well-managed companies to deliver higher risk-adjusted returns than stocks of unprofitable, fragile, "junk" companies. Its core tension is definitional: unlike value (book-to-market) or size (market cap), there is no consensus formula for "quality" — each major provider builds it from a different bundle of accounting characteristics. Quality is therefore better understood as a family of related signals (profitability, earnings stability, low leverage, conservative investment, growth, payout) than as a single clean factor, and much of the academic debate centers on whether it is a distinct premium at all.

How it's calculated / formed

There is no canonical construction. The three most-cited academic definitions are:

  • Novy-Marx (2013) — gross profitability. Gross profits (revenue minus cost of goods sold) divided by total assets. Novy-Marx argued this is "cleaner" than bottom-line earnings because items like R&D and SG&A — expensed under accounting rules — are actually investments in future profits. He found gross profitability had roughly the same power as book-to-market for predicting the cross-section of average returns (Novy-Marx 2013).
  • Fama–French (2015) — RMW (Robust Minus Weak). Operating profitability: revenue minus COGS, interest expense, and SG&A, scaled by book equity. RMW is the average return of high-profitability portfolios minus low-profitability portfolios, one of the two factors added to extend the three-factor model into the five-factor model (Fama-French 2015).
  • AQR QMJ (Asness, Frazzini & Pedersen 2019) — Quality Minus Junk. The broadest definition, motivated by the Gordon growth model: a quality stock is one an investor should rationally pay more for, scored across four pillars — profitability (gross profits, ROE, cash flows, etc.), growth (trailing improvement in those metrics), safety (low beta, low leverage, low bankruptcy risk, low earnings volatility), and payout (shareholder-friendly management). Stocks are ranked into a composite z-score; QMJ goes long the top ~30% and short the bottom ~30% within large- and small-cap universes (Asness et al.).

Commercial indexes (e.g. MSCI Quality) typically use a narrower trio: high ROE, low debt-to-equity, and low earnings-variability.

How it's used in practice

Quality is used three ways. As a standalone tilt, ETFs and index funds (MSCI Quality, etc.) overweight high-scoring names for a defensive equity sleeve. As a complement to value — its most respected use — because quality and value are negatively correlated: value buys cheap stocks, quality screens out the "value traps" that are cheap because they are deteriorating. Combining "cheap and good" (sometimes called quality-at-a-reasonable-price) has historically improved the value strategy's drawdowns and Sharpe ratio (Alpha Architect). As a multi-factor ingredient, quality/profitability sits alongside value, momentum, size, and low-vol in quant blends, where its low correlation to the others is the main draw.

Quality is also a defensive signal: QMJ tends to perform best in market downturns and "flights to quality," when investors reprice fragile balance sheets, giving it a payoff profile that partly hedges value's pro-cyclicality.

Adoption, debate & evidence

The evidence for the profitability component is among the stronger in the factor zoo. Fama–French report RMW as a robust, persistent premium; Novy-Marx showed gross profitability survives controls; AQR found QMJ delivered positive returns in 23 of 24 countries studied and significant risk-adjusted returns over long, broad samples (AQR). Adoption is wide: quality is a standard sleeve at most large factor shops.

But two honest caveats dominate the debate:

1. Is "quality" even one thing? Hsu, Kalesnik & Kose ("What Is Quality?", Financial Analysts Journal 2019) found that the major providers (MSCI, FTSE Russell, S&P, Research Affiliates, etc.) use substantially different characteristics, so their quality indexes do not proxy a single hidden factor. Decomposing the candidate traits, they reported that profitability, accounting quality, payout/dilution, and investment were associated with a premium, while capital structure, earnings stability, and growth showed little evidence of one — and concluded that index design "seems driven more by marketing optics than theory or data." So some components are well-supported while "quality" as a marketed umbrella is heterogeneous. 2. No risk story. AQR's own authors concede they "cannot tie the returns of quality to risk" and that prices vary "puzzlingly" little with quality — i.e. the premium looks more like a mispricing/anomaly than compensation for risk. The leading explanation (Novy-Marx) is behavioral: markets underappreciate how persistent high profitability is, because competitive moats keep it from mean-reverting as fast as models assume.

So the folklore "buy great companies and you'll beat the market" is only half right — cheap great companies have beaten the market; great companies bought at any price have not reliably done so.

Strengths & limitations

Strengths: strong and globally consistent profitability evidence; genuinely low/negative correlation with value (the diversification is real); defensive payoff in crises; intuitive and economically sensible.

Limitations: definition risk — your result depends heavily on which metric you pick. The #1 misuse is conflating "quality" with "good companies I admire" and paying any price for them; quality screens say nothing about valuation, and high-quality names can stay expensive and underperform for years (notably the period following the late-2010s, when crowded quality/growth names lagged). Quality is also exposed to accounting manipulation (reported profitability can be managed) and definitional crowding once a metric becomes popular.

Sources

Disputes flagged: (1) whether "quality" is a single coherent factor — contested (Hsu et al. say no; profitability specifically is well-supported). (2) Risk-based vs. behavioral explanation — unresolved; AQR concede no risk story. The 4.7%/year RMW figure cited by some secondary sources is a commonly quoted long-sample average, not a guaranteed forward return.