Fama-French HML Explained: What the Value Factor Means for DIY Investors

Factor premia are historical averages with wide return distributions and meaningful out-of-sample uncertainty. Past spreads are not forecasts of future ones.

HML stands for High Minus Low. In Fama and French’s three-factor framework, it is the return of stocks with high book-to-market ratios minus the return of stocks with low book-to-market ratios. In plain English: cheap stocks minus expensive stocks, where “cheap” means a high book value per dollar of market price. HML is the academic version of value investing: a transparent, rules-based long-short factor that asks whether buying cheap and selling expensive produces a return premium above and beyond what the market explains.

Value is one of the better-documented return patterns in empirical asset pricing, but the realized premium is noisy, episodic, and implementation-sensitive. HML is robust as a research finding and fragile as a personal investment strategy if the investor lacks patience, diversification, low costs, and behavioral discipline.

What HML actually is

HML is built from six value-weighted portfolios, formed at the intersection of two size groups (small and big) and three book-to-market groups (high, neutral, low). The formula from Kenneth French’s data library is exact:

HML = 1/2 (Small Value + Big Value) − 1/2 (Small Growth + Big Growth)

The factor is constructed using all NYSE, AMEX, and NASDAQ stocks with the required price and book-equity data. Breakpoints are based on NYSE stocks only, which prevents micro-caps from dominating the size split. Kenneth French data library: factor descriptions.

Two consequences follow from that construction, and both matter for any DIY investor reading about value:

• HML is a long-short factor. The factor goes long high book-to-market stocks and short low book-to-market stocks. It is a zero-investment portfolio. No retail ETF replicates HML exactly; ETFs are long-only and constrained to positive weights.
• HML is size-controlled. By averaging across small and big value, and across small and big growth, the factor isolates the value spread from the size spread. Long-only value ETFs vary in how strongly they tilt by size. That matters because the value premium has been historically stronger in small caps.

The evidence: US, international, and across asset classes

The original Fama-French paper showed that a three-factor model adding size (SMB) and value (HML) to market beta explained the cross-section of US stock returns better than CAPM alone. Fama & French (JFE, 1993). That work could plausibly have been a US-sample artifact, so the international evidence matters.

Fama and French’s 1998 paper extended the test to thirteen major non-US markets. Value beat growth in twelve of thirteen, with a global high-minus-low book-to-market spread averaging about 7.6% per year over 1975-1995. Fama & French: Value versus Growth: International Evidence. The 2012 follow-up confirmed positive value premiums in North America, Europe, Japan, and Asia Pacific over 1991-2010, with small-cap value tilts particularly pronounced outside Japan. Fama & French (JFE, 2012): Size, value, and momentum in international stock returns.

Asness, Moskowitz, and Pedersen widened the lens further. Their 2013 Journal of Finance paper, “Value and Momentum Everywhere,” found “consistent value and momentum return premia across eight diverse markets and asset classes, and a strong common factor structure among their returns.” That includes country equity indices, government bonds, currencies, and commodities. They also document that value and momentum are negatively correlated with each other, both within and across asset classes, which makes the combination more diversifying than either alone. Asness, Moskowitz & Pedersen (JoF, 2013).

Why HML is compelling

Three things separate value from many published return anomalies:

1. Economic intuition. A lower price per dollar of book equity, earnings, or cash flow generally implies a higher expected return for given fundamentals. Price matters. That is bedrock asset pricing, not a clever statistical pattern.
2. Out-of-sample evidence. Value has appeared in twelve of thirteen international markets, in country indices, in currencies, and in commodities. It is hard to dismiss as a US data artifact.
3. Plausible behavioral story. Investors may systematically overpay for exciting growth stories and underpay for boring, distressed, cyclical, or unpopular firms.

HML is one of the better-documented return patterns in empirical asset pricing, but the realized premium is noisy, episodic, and implementation-sensitive. Value is not magic. The 2010s provided a brutal counter-example, with HML running negative for most of the decade.

The five-factor model: did HML become redundant?

Fama and French’s 2015 five-factor model added two new factors: RMW (profitability, robust minus weak) and CMA (investment, conservative minus aggressive). In their US 1963-2013 tests, HML became redundant in a precise statistical sense: a four-factor model that dropped HML explained average returns about as well as the full five-factor model. HML’s average return was subsumed by its covariation with the market, size, profitability, and investment factors. HML can still characterize a portfolio’s value tilt for performance attribution; it just stopped adding incremental information about expected returns once profitability and investment were in the model. Fama & French (JFE, 2015): A five-factor asset pricing model.

This finding gets misread frequently as “value is dead.” The precise reading is that HML and the joint combination of profitability and investment carry overlapping information about expected returns. A long-only fund that targets value plus profitability plus investment captures the same expected-return information that HML alone captured, with less redundancy. That is the intellectual foundation for the modern Dimensional and Avantis approach: deliberately combining value, profitability, and investment exposures rather than chasing book-to-market in isolation. See the companion guide RMW Explained for the profitability side of the same story.

Not everyone accepts the redundancy finding. Asness and Frazzini argue in “The Devil in HML’s Details” that Fama-French HML uses stale price information by construction (book equity from December of t-1 paired with market equity from June of t, a roughly six-month lag) and that the factor excludes momentum. With a timely value signal and a momentum factor included alongside it, the value premium reasserts itself as non-redundant in their tests. The honest synthesis: HML in its original form is redundant in the US 1963-2013 sample; value as a concept is not.

Three serious caveats

1. Long stretches of underperformance

From roughly 2010 through 2020, Fama-French US HML returns were negative more often than positive, and cumulative HML returns over the decade were deeply negative. Long-only US small-value funds (DFSVX, AVUV-precursor portfolios) underperformed the S&P 500 by roughly 50-60 percentage points cumulative over the worst five-year stretches. A value tilt is a tracking-error decision, not just an expected-return decision. The investor has to be willing to look wrong for years.

2. Data mining risk in the broader factor literature

Harvey, Liu, and Zhu argued that because researchers have published hundreds of factors, statistical significance thresholds should be raised. Their paper proposes using a t-statistic hurdle of at least 3.0, rather than the conventional 2.0, to declare a new factor real. Harvey, Liu & Zhu (RFS, 2016): …and the Cross-Section of Expected Returns. HML survives that hurdle in long samples, but the warning applies to the broader value-factor zoo that mushroomed after Fama-French’s original work.

3. Anomaly decay after publication

McLean and Pontiff documented that the returns to published anomalies decline after publication, consistent with investors learning about and trading on them. McLean & Pontiff (JF, 2016). Their average finding is roughly a 32% drop in returns post-publication and a 58% drop post-publication-plus-five-years of trading. Some part of HML’s post-1993 performance may reflect this dynamic.

A fourth caveat worth flagging: traditional book equity may understate the value of intangible assets in a software-heavy modern economy. Recent research argues that adjusting book equity for off-balance-sheet intangibles (R&D, brand capital) materially changes which firms look “cheap” and can restore HML’s incremental explanatory power relative to the five-factor model. Park: Intangible Capital in Factor Models (Mgmt Science, 2024). The implication: book-to-market may be a less reliable value signal today than in 1993, and composite or intangible-adjusted value measures may work better.

Fama-French HML vs. AQR, Dimensional/Avantis, Alpha Architect

Most personal-finance writing on value treats “value investing” as a single thing. Four implementations deserve to be distinguished:

The conceptual takeaway: HML is the research factor. Long-only retail funds are investable approximations. Different sponsors weigh the tradeoffs differently. Comparing AVUV to VBR by expense ratio alone misses the deeper question of which approximation of the factor each one targets.

Long-only ETFs for systematic value exposure

For DIY investors who want to act on the framework above, the practical question is which long-only ETFs deliver enough value exposure to be worth the tracking error. A non-exhaustive map:

Expense ratios as listed on issuer pages in early 2026 (Avantis, Dimensional, Vanguard, iShares, Alpha Architect). Confirm current ratios at purchase.

For the deep ETF-implementation case, see Why I Like AVUV, AVDV, and AVGV. For the small-cap value asset-class case, see The Case for Small-Cap Value.

The Value Tilt Pain Test

The value premium is easy to understand and hard to live with. The calculator below makes the cost and the worst-case relative drawdown of a value tilt concrete in dollar terms. Move the sliders to see how much of your portfolio is being tilted, the annual fee overhead, and what a historical worst-case 5-year value-vs-growth stretch would have looked like at your chosen allocation.

What I recommend

The market portfolio remains the cleanest default. Use total global market index funds as the core. If you want a value tilt, treat it as a satellite, prefer global exposure, and prefer systematic low-cost implementation. A 5-20% tilt is easier to live with than a portfolio dominated by factor bets. Do not confuse HML with the ETFs that approximate it, and do not forecast the premium mechanically. HML is a research factor; ETFs are investable approximations.

Who this is for, and who it is not

A systematic value tilt is most useful for DIY investors with a long horizon, a diversified core, and behavioral patience to hold the tilt through multi-year underperformance. The empirical premium is positive on average, but it shows up in lumpy multi-year stretches, not in steady annual outperformance.

It is the wrong call for investors who will abandon the strategy after two or three bad years, who care primarily about tracking the S&P 500, who are highly tax-sensitive in taxable accounts and unwilling to manage turnover, or who think “cheap” is the same thing as “safe.” A bad-bargain stock in a declining business is not a defensive asset.

Frequently Asked Questions

Is HML the same as buying VBR or AVUV?

No. HML is a long-short, size-controlled academic factor. VBR is a long-only mid-and-small-cap value index fund. AVUV is a long-only small-cap value ETF that integrates value with profitability and investment screens. All three exposures relate to the same idea but differ meaningfully in construction, factor purity, and how much tracking error they carry against the broad market.

Did value “die” in the 2010s?

No, but it had a brutal decade. From roughly 2010 through 2020, HML returns were negative more often than positive in the US sample. Since 2021, the value premium has recovered somewhat, though performance has varied across regions and value definitions. The 2010s stretch is the strongest single out-of-sample stress test in HML’s public history, and it shows the premium is real but slow.

Is small-cap value the best place to seek value exposure?

Historically, yes in the US and developed international samples. The value premium has been larger in small caps, consistent with the idea that mispricings are more persistent and arbitrage is more limited among smaller, less-followed firms. The 2012 Fama-French international paper documents this pattern in North America, Europe, and Asia Pacific. Japan is the documented exception, where the small-cap-value tilt has been weaker.

Should I hold international value as well as US value?

The international evidence is part of what makes HML compelling as a real phenomenon rather than a US-only data artifact. Restricting a value tilt to US stocks ignores the strongest out-of-sample support for the strategy. A diversified global value sleeve (AVGV, or a mix of AVUV plus AVDV plus optionally AVES) is conceptually cleaner than US-only value.

Should I hold value funds in taxable accounts?

Possible but tax-aware. Value funds typically have higher turnover than broad-market index funds because the factor construction itself implies stocks moving in and out of the value band. Higher turnover generally produces more realized capital gains in taxable accounts. If you have the option, hold value-tilted funds in tax-advantaged accounts and broad-market funds in taxable. If everything sits in taxable, prefer ETF-structured value funds, which are more tax-efficient than mutual funds, and watch the distributions.

Related Guides

• Fama-French Factor Analysis decomposes a portfolio across the market, size, value, and momentum factors. Useful for measuring how much HML exposure your current portfolio actually has.
• The Case for Small-Cap Value covers the asset-class lens: why the small-value premium has been larger than the value premium alone, and the multi-year drought that paired with it.
• Why I Like AVUV, AVDV, and AVGV goes deep on the Avantis ETF implementation.
• RMW Explained covers the profitability factor that joined HML in the five-factor model.
• The Case for Global Diversification covers the broader argument for non-US equity exposure relevant to international value sleeves.
• The Risk-Free Rate Is Your Hurdle Rate covers the Treasury and TIPS baseline a value tilt has to clear to earn its keep.

Key Takeaways

• HML is the academic value factor: long high book-to-market stocks, short low, size-controlled. Long-only ETFs are investable approximations of it, not the factor itself.
• The evidence is broad: US, twelve of thirteen international markets, cross-asset. Fama-French 1993, 1998, 2012; AQR 2013. Hard to dismiss as a US data artifact.
• The five-factor “redundancy” finding is overlap, not refutation. HML carries the same information as profitability plus investment combined. Avantis and Dimensional target the joint exposure.
• Long underperformance stretches are the central risk. The 2010s decade was brutal for value, with multi-year relative drawdowns of 50 percentage points or more in some long-only US small-value funds.
• Treat any value tilt as a satellite, prefer global exposure, and pre-commit to holding through bad stretches. The Value Tilt Pain Test above makes the dollar implication concrete.