Key Takeaways
- • Larry Swedroe, writing for Alpha Architect, examines a July 2026 paper by Anthony Renshaw, Yurong Gu, Matt Candar, and Andrew Ang that applies diversification within factors, across factors, and across time in a multifactor equity portfolio.
- • The combined multifactor portfolio posted an information ratio of 1.00 over the full 2004 to May 2026 sample, ahead of enhanced value at 0.94, quality at 0.73, momentum at 0.65, and cyclical value at -0.20.
- • The portfolio benchmarks to the STOXX US Universal Index, targets 1% ex ante tracking error, rebalances quarterly, and caps quarterly one-way turnover at 5%.
- • Timing overlay added little over the full sample and improved results only over the most recent three years, so the authors position it as a small enhancement rather than a core driver.
- • The authors find low correlation among factor sleeves is the main source of the combined portfolio's risk-adjusted outperformance.
Summary
A July 2026 paper, "Diversification all the Way Down: Multifactor Investing Within Factors, Across Factors and Across Time," by Anthony Renshaw, Yurong Gu, Matt Candar, and Andrew Ang, argues that many factor portfolios remain concentrated in a single signal, style, or market regime, and that diversification should be layered at several levels. As summarized by Larry Swedroe in an Alpha Architect article published October 16, 2026, the authors diversify signals within each factor, diversify across factors, and diversify timing decisions.
The portfolio splits value into two sleeves: cyclical value, based on book-to-price and earnings yield, and enhanced value, which uses economic profit, intangible capital, and free cash flow. Momentum is broadened beyond price trend to include idiosyncratic momentum, short interest, hedge fund positioning, analyst revisions, and economic-profit momentum. Quality is measured with the Axioma US5.1 Profit Quality factor. Size and low volatility were excluded because they would have introduced excessive beta relative to the tracking-error budget.
According to the paper as summarized, the multifactor portfolio outperformed each individual sleeve on a risk-adjusted basis. Enhanced value behaved differently from cyclical value and was more resilient during extended periods when traditional value underperformed. The factor sleeves were weakly correlated, which the authors identify as the engine of the combined portfolio's results. A timing overlay averages bounded tilts from three models, each able to double or halve a factor's nominal weight but not eliminate it. The overlay improved performance over the most recent three years but added little over the full sample.
The backtest covers 2004 through May 2026, benchmarked to the STOXX US Universal Index, with a 1% ex ante tracking-error target and concentration and industry constraints. Swedroe notes that these are hypothetical, unaudited results prepared by Alpha Architect, and past performance is not indicative of future results.
The portfolio splits value into two sleeves: cyclical value, based on book-to-price and earnings yield, and enhanced value, which uses economic profit, intangible capital, and free cash flow. Momentum is broadened beyond price trend to include idiosyncratic momentum, short interest, hedge fund positioning, analyst revisions, and economic-profit momentum. Quality is measured with the Axioma US5.1 Profit Quality factor. Size and low volatility were excluded because they would have introduced excessive beta relative to the tracking-error budget.
According to the paper as summarized, the multifactor portfolio outperformed each individual sleeve on a risk-adjusted basis. Enhanced value behaved differently from cyclical value and was more resilient during extended periods when traditional value underperformed. The factor sleeves were weakly correlated, which the authors identify as the engine of the combined portfolio's results. A timing overlay averages bounded tilts from three models, each able to double or halve a factor's nominal weight but not eliminate it. The overlay improved performance over the most recent three years but added little over the full sample.
The backtest covers 2004 through May 2026, benchmarked to the STOXX US Universal Index, with a 1% ex ante tracking-error target and concentration and industry constraints. Swedroe notes that these are hypothetical, unaudited results prepared by Alpha Architect, and past performance is not indicative of future results.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.