Has the Value Premium Decayed?
A retrospective on the Fama-French HML factor, the return to buying cheap stocks and shorting expensive ones, from 1926 to 2026. The full-sample premium is real and statistically significant. The last decade looks very different.
Monthly factor returns from the Ken French Data Library, rolled up in Python and rendered live with Apache ECharts. Hover any series; drag the time slider to focus a window.
file:// path, serve the folder over HTTP (the page fetches a JSON roll-up from the same origin).
A century of compounding, then a plateau
Growth of one dollar in the HML factor since 1926, on a log scale. The series compounds steadily into the mid-2000s, then flattens. Over the full sample the premium annualizes near 3.6%, with a Newey-West t-stat above 3, so the long-run effect is statistically real.
Growth of $1 in the HML factor, 1926 to 2026
Log scale. Cumulative compounded return of the long-cheap, short-expensive factor.
The rolling premium fell below zero
The trailing 10-year annualized premium. It held positive through most of history, then dropped under the zero line in the 2010s and has stayed weak. This is the decay the question asks about, and it is visible without any model.
Trailing 10-year annualized premium
Each point is the annualized HML return over the prior 120 months. Dashed line marks zero.
The premium is regime-dependent
Annualized premium by decade. Strong through the mid-century, then thinning, and negative in the 2010s. A constant-premium assumption misses how much the payoff has varied across regimes.
Annualized premium by decade
Gold bars are positive decades; red bars are negative. Decades with fewer than 24 months are excluded.
The deepest drawdown in the sample
Drawdown of the cumulative factor from its running peak. The decline that began after 2007 is the deepest and longest in the record, reaching roughly negative 58%. Investors call this stretch the value winter.
Drawdown from running peak
How far the cumulative factor sits below its prior high, at each point in time.
Methodology & data
Source: Ken French Data Library, F-F Research Data Factors (monthly), data through April 2026, retrieved 29 June 2026. HML is the high-minus-low book-to-market factor from Fama and French (1993).
Definitions: the full-sample premium is the geometric annualized HML return over 1,198 monthly observations. The rolling premium is the annualized return over each trailing 120-month window. Decade figures are the arithmetic mean monthly return annualized. Drawdown is the cumulative factor relative to its running peak. The t-statistic uses Newey-West (HAC) standard errors with 12 lags to account for autocorrelation.
Pipeline: the monthly factor file is parsed and rolled up in Python and pandas into a compact JSON of the cumulative series, rolling premium, drawdown, decade table, and summary statistics. The page renders that roll-up client-side with Apache ECharts. Everything is self-hosted under a strict content-security policy.
The companion research note works through the same question in full, with the regressions and robustness checks.
PythonpandasstatsmodelsNewey-West HAC Apache EChartsRolling-window analysisDrawdown