New insights challenge traditional factor investing by emphasising fundamental drivers over valuation effects

According to the original report by Arnold et al. (2025), commonly reported factor premiums overstate the portion of returns attributable to structural, fundamentals-driven risk premia because they conflate those premia with gains (or losses) from changing valuation multiples. The analysis decomposes long-run factor returns into a structural component , driven by changes in fundamentals such as book value growth and dividends , and a revaluation component that arises from shifts in price-to-book and similar multiples. For the 14 factors examined between July 1973 and December 2022, the authors report an average published factor premium of 4.0% per annum, with a revaluation drag of roughly 0.4% per annum, implying a structural premium nearer 4.4% per annum.

Research Affiliates’ related research supports and extends this perspective. Their work shows that characteristics-based measures of factors (for example, value measured by fundamentals, or profitability measured by underlying earnings) provide more informative signals about expected returns than simple factor betas, and that thoughtfully constructed multi-factor portfolios can produce meaningful, statistically and economically significant returns. According to Research Affiliates, focusing on fundamental factor characteristics rather than on raw exposures helps distinguish whether observed returns are likely to persist as structural premia or are instead transitory valuation effects. Industry studies also highlight that sensible multi-factor design , balancing exposures across value, profitability, investment, size, low beta and momentum , lowers volatility of outcomes and can improve risk-adjusted returns over time.

The practical implications for investors are threefold. First, measured factor returns contain two distinct drivers: fundamentals-led structural returns and valuation-driven revaluation returns. The latter can be sizable over shorter horizons and may even reverse; they are not the same as compensation for bearing a persistent risk premium. Second, the longer the investment horizon, the less dominant the revaluation component becomes and the more reliable the structural component is as a guide for portfolio construction. Arnold et al. show the revaluation contribution falls materially as holding periods lengthen, suggesting long-term investors should place greater weight on structural estimates when sizing factor exposures. Third, while investors cannot directly “hedge away” revaluation effects in realised returns, they can and should use characteristics-based analytics and balanced multi-factor design to tilt toward the structural drivers that are more likely to persist.

The report and Research Affiliates’ prior work also sound several cautionary notes. Factor correlations and return dynamics are time-varying: in stress episodes previously low or negative correlations can converge, reducing the diversification benefits of multi-factor portfolios. Moreover, portfolios that merely target factor betas without incorporating the underlying characteristics tend to earn returns that are indistinguishable from zero after accounting for valuation effects and noise, according to Research Affiliates. Finally, estimation of the structural component requires choices about accounting measures, universes and sample periods; different implementations will produce different decompositions and optimal weights.

Taken together, the evidence suggests a modest but important refinement to factor investing practice: treat reported factor premiums as a mixture of fundamental and valuation-driven returns, and rely on characteristics-based, multi-factor portfolio design , with an emphasis on long horizons and rebalancing , to capture the more durable structural premia. According to the original report, doing so produces more robust factor weights and improves multi-factor portfolio outcomes versus naïve or purely beta‑driven approaches.