By the late 1980s, the CAPM had a problem. Decades of accumulated U.S.

equity returns showed that the model's predictions held in spirit but failed in cross-section. Small-capitalization stocks earned more than CAPM predicted; high book-to-market (value) stocks earned more than CAPM predicted; both effects were robust, persistent, and statistically significant in long samples. Eugene Fama and Kenneth French published their resolution in the Journal of Finance in 1992 ('The Cross-Section of Expected Stock Returns') and refined it into a tradable factor model in the Journal of Financial Economics in 1993 ('Common risk factors in the returns on stocks and bonds').

Their Fama-French 3-factor model added two factors to CAPM's market beta — a size factor and a value factor — and showed that the combined model explained the cross-section of average returns substantially better than CAPM alone. The 3-factor model became the practitioner standard for performance attribution within roughly a decade, and the underlying factor return time series (along with momentum, profitability, investment, and several others) were made publicly available by Kenneth French at the Tuck School of Business at Dartmouth, where they remain the authoritative reference for academic and institutional factor work today.