Is Factor Investing Dead? What the Evidence Says for DIY Investors

An international small-cap value fund like AVDV has had a strong run, and a chart like that is a tempting thing to point at. It is also not evidence that factor investing works. One good year proves about as much as the 2010s, a weak decade for factors, proved that factor investing was dead. Factor premiums are cyclical, and a single chart at either extreme tells you almost nothing about the next twenty years.

So “is factor investing dead?” is usually the wrong question. The useful questions are narrower: does the factor have an economic reason to exist, does it survive out-of-sample and after trading costs, can it be implemented cheaply, and can you hold it through long stretches of ugly tracking error? This guide works through the evidence on both sides and turns it into a decision a DIY investor can act on.

Disclosure: I own Avantis-style factor funds (AVUV and DFSV). I hold them because the logic is sound and the implementation is diversified and cheap, and I treat the premium as uncertain possible upside, not a planning assumption. This is educational, not individualized advice.

Factor evidence is not the same as a factor product

Factor investing means owning groups of stocks that share a characteristic linked to higher average returns: value (cheap), size (small), profitability or quality, conservative investment, and sometimes momentum or low volatility. The academic evidence describes broad return patterns in the data. What you buy is an ETF with a fee, turnover, taxes, imperfect exposure to the factor, and your own behavior under pressure. Most of the gap between “factors work in the data” and “factors worked in my account” lives in that second list.

The case for factors

The serious evidence starts with Fama and French. Their 1992 and 1993 work showed that size and book-to-market captured cross-sectional returns, and their 2015 five-factor model added profitability and investment.¹² The patterns appear outside the United States too, with regional differences. Fama and French’s international tests find value, profitability, and investment effects in North America, Europe, and Asia Pacific, while Japan shows weak profitability and investment relations.³

The modern case is more specific than “small beats large” or “cheap always wins.” Novy-Marx showed that gross profitability predicts returns about as well as book-to-market, so cheap and profitable is a stronger combination than cheap alone.⁴ Asness and co-authors found that the size premium, weak and unstable on its own, becomes much more reliable once you control for quality, or junk.⁵ And value and momentum show up in many countries and asset classes beyond US stocks, and tend to be negatively correlated with each other, which is why a diversified multi-factor tilt is more defensible than a single-factor bet.⁶ The honest summary of the bull case: cheap, profitable, non-junk companies have had higher expected returns than expensive, unprofitable, speculative ones.

The case against factors

The skeptical evidence is real and specific. Researchers have published hundreds of return predictors, and testing that many candidates guarantees false positives. Harvey, Liu, and Zhu argued that, after accounting for all that data mining, a newly proposed factor should clear a t-statistic of about 3.0 rather than the usual 2.0 to be believed.⁷ When Hou, Xue, and Zhang replicated the “factor zoo” with stricter standards (NYSE breakpoints and value-weighted returns, which reduce the distortion from tiny illiquid stocks), about 65% of 452 anomalies failed to clear even the t ≥ 1.96 hurdle, and roughly 82% failed under the stiffer multiple-testing hurdle.⁸

Even predictors that are real tend to weaken once they are published. McLean and Pontiff found that documented anomalies earned about 26% less out-of-sample and about 58% less after publication, as data mining washes out and investors arbitrage away the easier edges.⁹ Many “factors,” in other words, were never alive, or were too expensive to capture.

The strongest skeptical case: Andrew Chen and anomaly decay

The most useful skeptic here is Andrew Chen, an economist at the Federal Reserve Board. His position is more interesting than “nothing replicates.” In his open-source work with Tom Zimmermann, almost all published predictors do reproduce, and a companion paper argues that most claimed findings in this literature are likely true.¹⁰ Replication is not the problem.

The problems are decay and cost. Chen and Zimmermann find that predictor returns fall by roughly 50% out of sample (about 65% for risk-based predictors), and that publication bias explains only a small part of it, so the decline is mostly real.¹¹ Chen and Velikov then show that after realistic trading costs, the average anomaly nets only a few basis points per month, and in the period since the mid-2000s, when trading technology and anomaly publishing both accelerated, the individual strategies earn close to nothing net of costs.¹² Uncomfortably, factors with elegant economic stories did not clearly survive better than data-mined ones: Chen and co-authors find peer-reviewed, risk-based predictors decayed more, not less.¹³

Read the scope carefully before you panic. Chen studies academic equal-weighted long-short strategies that trade illiquid micro-cap stocks on both the long and short side, which is where most of the cost drag lives. That is not a long-only, low-turnover, large-cap-tilted fund like AVUV or a Dimensional strategy. His work is the strongest reason to haircut historical factor premiums, distrust backtests, and demand cheap, diversified, low-turnover implementation. It is not evidence that a long-only value or profitability tilt has zero expected premium for a retail investor. Those are different claims, and conflating them is the most common mistake in this debate.

Why factors can disappoint for a decade

Even defenders of factor investing agree the premiums are lumpy. David Blitz of Robeco documented that the standard Fama-French factors had negative average returns over 2010 to 2019, then pointed out that this lost decade looked a lot like the 1990s, another weak stretch that was followed by strong factor returns.¹⁴ Dimensional’s own research makes the same point from the other direction: negative ten-year premium periods are common enough that investors should expect them, and in the historical record there has been no ten-year window where three or four of the major premiums were all negative at once.¹⁵ Both are factor proponents, and both are telling you the same thing: a factor premium is uncertain, lumpy, and sometimes absent for ten years or more.

How big does the premium have to be?

Factor investing is not dead, but the expected premium is probably smaller than the old backtests, and a large part of it can be eaten by fees, taxes, and turnover. That reframes the decision. The question shifts from “do factors work?” to “how large does the premium need to be, net of costs and behavior, to be worth the tracking error?” The calculator below lets you test that directly: set an assumed gross premium, subtract your fee and tax drag, and see the net edge, the break-even premium, and the wealth it would add.

What I recommend for DIY investors

For most DIY investors, the default should still be a global market-cap index portfolio plus the right amount of bonds and cash for your risk capacity. Factor investing is optional. Vanguard itself describes factor funds as active tilts toward characteristics like momentum, quality, or lower valuations, and warns they carry significantly more risk than broad-market investing, with sharp and lengthy stretches of underperformance.¹⁶

If you do want a tilt, favor broad, low-cost, rules-based funds from firms with a clear academic process. AVUV runs a US small-cap value strategy at 0.25%, and AVDV an international small-cap value strategy at 0.36%.¹⁷¹⁸ Those are not free next to a total-market fund, but they are reasonable if you are deliberately buying a small, value, and profitability tilt. Keep the tilt modest, perhaps 20% to 40% of equity, skip tactical factor timing, and do not save less or retire earlier on the assumption that the premium will arrive. Treat it as possible upside the plan does not depend on.

Good fit: a high savings rate, a long horizon, real risk tolerance, an already globally diversified portfolio, the temperament to tolerate ten or more years of underperformance, an understanding of tracking error, low-cost ETFs, and mechanical rebalancing.

Poor fit: a short horizon, a habit of performance chasing, anyone who benchmarks to the S&P 500 month to month, a tax-sensitive investor in high-turnover funds, anyone likely to quit after three bad years, or anyone who needs the premium for the plan to work.

Measure the tilt you actually own

Many investors who say they own “small value” or “quality” do not know how much exposure they hold. A factor regression separates the label on the fund from the reality in your account, and it is a reminder that factor funds are still mostly equity risk: a value ETF usually moves with the global market and will not cushion an equity drawdown. The question worth answering is whether the tilt you hold still leaves your plan intact if the premium turns out to be zero for a decade.

Frequently Asked Questions

Is factor investing dead?

No, but naive, backtest-driven factor investing is on weak ground. The durable premiums look smaller than old backtests imply, and much of the edge from academic strategies disappears after trading costs. A modest, low-cost, diversified, long-only tilt held for the long run remains defensible; a portfolio built on a single backtest is not.

Do factor premiums still exist after costs?

For the academic long-short strategies Andrew Chen studies, individual anomalies net close to nothing after realistic trading costs, especially since the mid-2000s. But those strategies trade illiquid micro-caps on both sides. Long-only, low-turnover funds like AVUV or DFSV face far lower costs, so they are a different and more investable case. Assume a haircut, not a zero.

Should DIY investors tilt to small value?

Only if you understand why you own it and can hold it through a decade of underperformance. A small-value tilt has academic support and a sound economic story, but it adds tracking error and can lag the broad market for years. For many investors a total-market index fund is the better fit.

How long can factors underperform?

A decade or more. The Fama-French factors had negative average returns across the 2010s, much as they did in the 1990s. Negative ten-year premium periods are common enough that you should plan for them rather than treat them as a broken strategy.

Is a total-market index fund enough?

Yes. A globally diversified, low-cost market-cap portfolio is a complete equity strategy on its own. A factor tilt is an optional, evidence-based upgrade for investors who can hold it through pain, not a requirement.

Key Takeaways

• Factors are not dead, but naive factor investing is. One hot chart or one bad decade settles nothing.
• Haircut the premium. Out-of-sample decay and trading costs make the durable edge smaller than backtests suggest.
• Chen’s scope matters. His near-zero net returns are for long-short micro-cap strategies, not long-only funds like AVUV or DFSV.
• Index first, tilt second. A modest, cheap, diversified, tax-aware tilt is reasonable; a backtest-driven all-in is not.
• Don’t build the plan on the premium. Treat it as possible upside, and make sure the plan works if it is zero for a decade.

Related Guides

• Position Sizing: if you do pick stocks, how to size the bet so it can survive and matter.
• AVUV vs BSVO vs DFSV: how to implement a small-value tilt if you decide to hold one.
• Concentration Risk: why a tilt should improve the whole portfolio, not chase a factor.
• An Index Fund Is Not a Financial Plan: where market-cap indexing ends and planning begins.

Sources

1. Eugene F. Fama and Kenneth R. French, “The Cross-Section of Expected Stock Returns,” Journal of Finance 47(2), 1992. wiley.com.
2. Eugene F. Fama and Kenneth R. French, “A Five-Factor Asset Pricing Model,” Journal of Financial Economics 116(1), 2015. sciencedirect.com.
3. Eugene F. Fama and Kenneth R. French, “International Tests of a Five-Factor Asset Pricing Model,” Journal of Financial Economics 123(3), 2017 (value/profitability/investment present outside the US, weak in Japan). sciencedirect.com.
4. Robert Novy-Marx, “The Other Side of Value: The Gross Profitability Premium,” Journal of Financial Economics 108(1), 2013. sciencedirect.com.
5. Cliff Asness, Andrea Frazzini, Ronen Israel, Tobias Moskowitz, and Lasse Pedersen, “Size Matters, If You Control Your Junk,” Journal of Financial Economics 129(3), 2018. ssrn.com.
6. Clifford Asness, Tobias Moskowitz, and Lasse Pedersen, “Value and Momentum Everywhere,” Journal of Finance 68(3), 2013. ssrn.com.
7. Campbell Harvey, Yan Liu, and Heqing Zhu, “... and the Cross-Section of Expected Returns,” Review of Financial Studies 29(1), 2016 (a new factor should clear roughly t ≥ 3.0). ssrn.com.
8. Kewei Hou, Chen Xue, and Lu Zhang, “Replicating Anomalies,” Review of Financial Studies 33(5), 2020 (about 65% of 452 anomalies fail t ≥ 1.96; ~82% fail the multiple-testing hurdle). academic.oup.com.
9. R. David McLean and Jeffrey Pontiff, “Does Academic Research Destroy Stock Return Predictability?” Journal of Finance 71(1), 2016 (~26% lower out-of-sample, ~58% lower post-publication). wiley.com.
10. Andrew Y. Chen and Tom Zimmermann, “Open Source Cross-Sectional Asset Pricing,” Critical Finance Review, 2022 (most predictors reproduce); and Chen, “Most Claimed Statistical Findings in Cross-Sectional Return Predictability Are Likely True.” ssrn.com.
11. Andrew Y. Chen and Tom Zimmermann, “Publication Bias and the Cross-Section of Stock Returns,” Review of Asset Pricing Studies 10(2), 2020 (~50% out-of-sample decay; publication bias explains ~12%). ssrn.com.
12. Andrew Y. Chen and Mihail Velikov, “Zeroing in on the Expected Returns of Anomalies,” Journal of Financial and Quantitative Analysis 58(3), 2023 (net of costs, the average anomaly earns a few basis points per month; near zero post-2000s). See also Rational Reminder ep. 316. ssrn.com.
13. Andrew Y. Chen, Alejandro Lopez-Lira, and Tom Zimmermann, “Does Peer-Reviewed Research Help Predict Stock Returns?” 2022 (theory-backed predictors did not survive better than data-mined ones). ssrn.com.
14. David Blitz, “Factor Performance 2010-2019: A Lost Decade?” Journal of Portfolio Management, 2021 (negative FF factor returns in the 2010s, similar to the 1990s). ssrn.com.
15. Dimensional Fund Advisors, “Perspective on Premiums” (negative ten-year premium periods are common; no historical case of three or four premiums simultaneously negative). dimensional.com.
16. Vanguard, “What are factor-based funds?” (active tilts with significantly more risk and cyclical underperformance). vanguard.com.
17. Avantis U.S. Small Cap Value ETF (AVUV), net expense ratio 0.25% as of 01/01/2026. avantisinvestors.com.
18. Avantis International Small Cap Value ETF (AVDV), net expense ratio 0.36% as of 01/01/2026. avantisinvestors.com.
19. Factor return data: Kenneth R. French Data Library, Tuck School of Business at Dartmouth. tuck.dartmouth.edu. Figures are point-in-time and were verified against journal, SSRN, and sponsor sources; confirm current fund data before investing.

Disclosure: the author owns AVUV and DFSV and does not own AVDV. This article is educational and is not financial or tax advice. Factor premiums are uncertain and can be negative for long periods; fund data changes, so confirm expense ratios and holdings with each sponsor before investing.