Factor Investing After Fifteen Years of Underperformance

Somewhere between 2018 and 2020, “factor investing is dead” became a genuine headline rather than a provocation. Value funds had trailed the S&P 500 for over a decade. Smart-beta ETF sponsors were quietly renaming products. Financial advisors who had built client portfolios around size and value tilts in the 2000s were fielding uncomfortable questions at annual reviews. Then 2022 happened, value crushed growth by one of the widest annual margins on record, and the obituary got shelved. By 2026, with AI-driven concentration back in force and value lagging again, the question has resurfaced — this time from investors who lived through both the drought and the whiplash and want a straight answer about what actually happened and whether the strategy still deserves a place in a portfolio.

Why This Debate Keeps Coming Back

Factor investing traces back to a specific academic finding: in 1992 and 1993, Eugene Fama and Kenneth French showed that a market-beta-only model failed to explain a large chunk of the return difference between stocks. Small companies beat large ones over long stretches. Cheap companies, measured by book value relative to price, beat expensive ones. Mark Carhart added momentum in 1997 — stocks that had recently risen kept rising, for a while. By 2015, Fama and French folded in profitability and investment patterns, producing the five-factor model that underpins most of today’s “smart beta” and factor-tilted index products.

Asset managers translated the academic long-short portfolios into products retail and institutional investors could actually buy: long-only index funds tilted toward cheap stocks, small stocks, high-momentum stocks, or high-quality, low-volatility stocks. Assets in these funds grew from a rounding error in the 1990s to several hundred billion dollars globally by the late 2010s. Then the premium that had justified the whole category — value, specifically — spent thirteen years going nowhere or worse, and the category had to explain itself in a way it never had before.

Investors who wanted no part of that argument mostly defaulted to two choices: buy the total market and stop thinking about factors altogether, or lean on something simpler and more familiar, like a plain, rules-based dividend index fund, which captures some value-like characteristics without the explicit factor-fund branding or the tracking-error stomach test. Both are reasonable responses. Neither answers the underlying question of whether the academic premiums are still there.

How a “Factor” Is Actually Built

The word “factor” gets used loosely enough that it is worth pinning down the mechanics before arguing about whether the strategy works. A factor is a characteristic of a stock — cheapness, size, recent price trend, profitability, volatility — that has historically correlated with returns beyond what plain market exposure explains. Building an investable factor involves four decisions, and each one changes the fund you actually end up holding:

• The metric. “Value” can mean book-to-price, earnings-to-price, sales-to-price, free cash flow yield, or some blend of all four. MSCI, S&P Dow Jones, Russell, Dimensional, and AQR each define it differently, and the same stock can show up as cheap in one methodology and expensive in another.
• The universe and ranking. Academic long-short portfolios rank the entire investable universe and take the top and bottom deciles. Retail long-only funds typically overweight the cheapest third or quintile of a benchmark and underweight or exclude the rest, which mutes both the premium and the volatility relative to the pure academic version.
• Rebalancing frequency. Most factor indexes rebalance semi-annually or quarterly. Faster rebalancing captures a fresher signal but multiplies trading costs; slower rebalancing lets winners drift away from the factor definition between reset dates.
• Implementation as long-only vs. long-short. The academic factor premium is measured as a long-short spread — long the cheap decile, short the expensive decile. A long-only retail fund only captures roughly half of that spread, because it gives up the short leg’s contribution entirely.

That last point explains a lot of investor confusion. When a research paper reports a value premium of 4% to 5% a year, it is describing a hypothetical long-short portfolio that most people cannot replicate cheaply, with leverage, shorting costs, and borrow fees stripped out of the math. The value ETF sitting in a brokerage account is a diluted, long-only approximation of that idea, and it was never going to deliver the full academic number even in a good decade.

The Value Factor’s Fifteen-Year Drought, In Numbers

Using the standard Fama-French data series, the value factor (high book-to-market minus low book-to-market, or HML) averaged an annualized premium of roughly 4% to 5% from the early 1960s through the mid-2000s. From 2007 through 2020, that premium turned negative on a cumulative basis, and value underperformed growth in most of those calendar years. Several papers from AQR and Research Affiliates — the two research shops that have argued most loudly both for and against the “value is broken” thesis — independently measured this stretch as the worst drawdown in the value factor’s history by depth, and the longest by duration, worse than its decline during the early 1930s.

Three explanations dominate the debate over why this happened, and they are not mutually exclusive:

1. Structurally cheaper money favored growth. A long stretch of falling and then near-zero interest rates mechanically raises the present value of cash flows expected further in the future — exactly the profile of high-growth companies. Cheap stocks, by definition, tend to be mature, capital-intensive businesses whose cash flows arrive sooner and grow more slowly, so they benefit less from a falling discount rate.
2. Book value stopped measuring what it used to measure. Traditional value screens rely on book-to-price, but accounting rules expense most intangible investment — research, software development, brand building — immediately rather than capitalizing it. A software or platform company that plows profits into R&D looks like it has almost no “book value” even while building a durable competitive moat, which mechanically excludes it from cheap-stock screens and mechanically includes capital-heavy, often-declining businesses instead.
3. Crowding compressed the premium after the fact. A widely cited 2016 study by McLean and Pontiff found that roughly half of the return associated with a published academic anomaly disappears after publication, as more capital chases the same signal and prices adjust. Value was one of the first and most heavily marketed factors; by the mid-2000s it had a decade of academic papers, mutual funds, and eventually ETFs built around it, all trading the same handful of cheap-stock signals.

None of those three explanations is airtight on its own, which is exactly why the argument has run for years rather than being settled in a single paper. The rate-based story struggles to explain why value continued to lag through parts of 2022 and 2023 even as rates rose sharply. The intangibles story is compelling but doesn’t fully explain why size and low-volatility factors also went through their own weak stretches on different timelines. The crowding story fits the data reasonably well but is hard to prove definitively, since crowding is difficult to measure directly.

Crowding, Turnover, and the Costs a Backtest Never Shows You

A backtest of a factor strategy typically assumes trades execute at the closing price with no spread, no market impact, and no tax drag. None of those assumptions survive contact with an actual portfolio. Factor funds turn over their holdings far more often than a plain cap-weighted index fund — a total-market index fund might turn over less than 5% of its holdings in a typical year, driven mostly by index additions and removals, while a value or momentum factor fund frequently turns over 20% to 60% of its portfolio annually as stocks migrate in and out of the top ranking bucket at each rebalance.

Each rebalance trade incurs a bid-ask spread and, for less liquid small-cap or micro-cap value and momentum names, real market-impact costs as the fund’s own buying and selling moves the price against it. Momentum strategies are particularly exposed to this because the entire signal is built on recent price movement, meaning momentum funds are structurally buying stocks that have just gotten more expensive to trade and selling stocks that have just gotten cheaper to trade, right as everyone else with the same signal is doing the same thing on the same rebalance date.

Capacity is the other cost nobody puts in a marketing brochure. A factor premium that shows up cleanly in a backtest with a modest amount of capital chasing it can shrink or vanish once tens of billions of dollars try to exploit the identical signal on the identical rebalance schedule. This is arguably the single best argument for the crowding explanation of value’s drought: the value factor became one of the most heavily productized ideas in all of asset management right before its worst multi-year stretch on record.

Worked Example: The Real Cost of a 30% Value Tilt, 2007–2021

Numbers make this concrete faster than another paragraph of theory. Assume an investor starts with $50,000 in January 2007 and picks one of two paths, using round, illustrative figures consistent with the historical pattern described above rather than any single fund’s exact realized return:

• Path A — 100% total market index: compounds at roughly 10.3% annualized over the 15 years from 2007 through 2021, a period that includes the 2008 crash and the long bull market that followed.
• Path B — 70% total market, 30% value-factor tilt: the value sleeve returns roughly 7.5% annualized over the same period — a meaningful but historically realistic shortfall versus the market during value’s weakest stretch — for a blended portfolio return of roughly 9.5% annualized.

Compounded over 15 years, Path A turns $50,000 into approximately $217,700. Path B turns the same starting balance into approximately $194,000. The 30% tilt toward value, in other words, costs this investor roughly $23,700 in opportunity cost over a decade and a half — not because the tilt was reckless, but because it happened to span almost the entirety of value’s worst historical drawdown.

Now extend the example three more years. In 2022, value’s relative return swung sharply positive as interest rates rose and growth stock valuations compressed; Russell index data for that year showed large-cap value losing roughly 7.5% while large-cap growth lost roughly 29%, a spread of more than 20 percentage points in a single calendar year. That one year alone would have clawed back a meaningful chunk of Path B’s cumulative shortfall. Then in 2023 and 2024, growth reasserted itself as a small number of AI-linked mega-cap stocks drove most of the market’s return, and the gap widened again. The lesson from stretching the example to 2025 is not that either path is “correct” — it’s that the value premium behaves like a series of long, uneven waves rather than a steady tailwind, and an investor’s real experience depends enormously on which years they happened to hold through.

Annualized Value Premium by Period

Factor by Factor: A Side-by-Side Scorecard

Value gets most of the press because its drawdown was the longest and most dramatic, but it was not the only factor that struggled. Here is how the five most commonly productized factors compare on the metrics that actually matter to someone deciding whether to hold one.

Read across that table and a pattern emerges: the factors with the steadiest long-run record — quality and low volatility — are also the ones with the least dramatic story attached to them, which is precisely why they generate fewer headlines and less product proliferation. Momentum has a strong long-run number but the highest turnover of the group by a wide margin, which quietly eats into the premium every single year regardless of whether the strategy is “working.” Size has arguably the weakest current case of all five, having drifted for close to two decades without a comeback on the scale of value’s 2022 rebound.

Common Mistakes Investors Make With Factor Tilts

1. Treating a backtest as a forecast. A 60-year average premium tells you almost nothing about the next five years, and the gap between the long-run number and any given decade can be enormous, as value’s own history shows.
2. Chasing whichever factor just outperformed. Buying a momentum fund because it had a great trailing year, or rotating into value right after 2022’s rebound, tends to buy in near a local peak rather than at the start of a new multi-year run.
3. Assuming “value” means the same thing across providers. Two funds both labeled value can hold almost entirely different stocks depending on whether the index uses book-to-price, a multi-metric composite, or a sector-relative screen.
4. Stacking overlapping factor funds without checking real diversification. Owning a value ETF, a dividend ETF, and a “quality value” ETF can amount to one concentrated bet dressed up as three separate positions, because the underlying holdings overlap heavily.
5. Underestimating the emotional cost of tracking error. A 13-year underperformance stretch is easy to accept in a spreadsheet and very hard to sit through in a real account, especially when a client, spouse, or your own anxiety keeps asking why the “smart” strategy is losing to the plain index fund.
6. Ignoring turnover-driven tax drag in taxable accounts. A factor fund with 40% annual turnover realizes far more short- and long-term capital gains than a cap-weighted index fund, a cost that compounds silently in a non-retirement account.

A Practical Checklist Before You Tilt

• Confirm your time horizon is genuinely decades, not years. If you cannot commit to holding a tilt through a stretch of underperformance lasting five, ten, or even fifteen years, a factor allocation is the wrong tool regardless of the academic case behind it.
• Read the index methodology, not just the fund name. Find out exactly which metric defines the factor, how often it rebalances, and how concentrated the resulting portfolio is relative to the broad market.
• Check the fund’s actual turnover and expense ratio. Compare it against a plain index fund in the same asset class; the difference tells you how much of any future premium will be consumed by costs before it reaches you.
• Size the position modestly. A 10% to 30% tilt relative to a core market-cap-weighted holding captures most of the diversification benefit without turning the whole portfolio into a single-factor bet.
• Diversify across factors rather than concentrating in one. A blended value-quality-momentum approach smooths out the individual factor cycles better than a single-factor fund, since the factors are not all weak or strong in the same years.
• Rebalance on a schedule, not on sentiment. Decide in advance when you will trim or add to the tilt, and stick to that calendar regardless of whether the factor just had a good or bad quarter.
• Place high-turnover factor funds in tax-advantaged accounts when possible. This keeps the turnover-driven tax drag from compounding against you in a taxable brokerage account.
• Review performance on a multi-year, not quarterly, cadence. Judging a factor tilt every three months all but guarantees an emotional decision at exactly the wrong moment.

Key Takeaways

• The value factor’s 2007–2020 drawdown was likely the worst in its recorded history, exceeding its decline during the Great Depression by most measures used in the academic literature.
• Value rebounded sharply in 2022 as rates rose, then lagged again from 2023 through 2025 as AI-driven mega-cap growth stocks dominated index returns.
• Real-world factor funds capture only part of the academic long-short premium, because they are long-only and diluted relative to the theoretical portfolio in a research paper.
• Turnover and trading costs — often 20% to 90% annually depending on the factor — erode a meaningful share of any premium before it reaches an investor.
• Quality and low-volatility factors have shown the steadiest long-run records, while size currently has the weakest case among the five major factors.
• A modest, diversified, multi-factor tilt held for a full market cycle or longer is a more defensible approach than a concentrated bet on any single factor’s recent performance.

Frequently Asked Questions

Is factor investing dead?

No. The underlying academic premiums have shown up in return data going back to the early 1960s and, for some factors, even earlier. What changed is that several factors, value most severely, went through a multi-year stretch weak enough to make the strategy look broken. The premiums have historically reappeared after long droughts, most visibly in 2022, but there is no guarantee of when or how strongly that happens for any given factor going forward.

Why did value investing underperform for so long?

Three overlapping explanations dominate the research: persistently low interest rates mechanically favored growth stocks’ distant cash flows, accounting rules for intangible assets like software and brand-building understate the true value of modern growth companies, and heavy institutional crowding into a well-publicized signal likely compressed the premium after the fact.

Did factor investing recover in 2022?

Value staged one of its strongest single-year comebacks on record in 2022, as rising interest rates hit high-multiple growth stocks especially hard. That rebound partially offset years of prior underperformance but did not fully erase the cumulative gap built up between 2007 and 2020, and value lagged growth again in the years that followed.

What is factor crowding?

Factor crowding describes what happens when a large amount of capital chases the same investment signal at the same time, particularly around the same rebalance dates. It can compress the premium a factor delivers going forward, since a signal’s edge partly depends on other investors not already trading on the identical information.

How much of a portfolio should be tilted toward factors?

Many practitioners suggest keeping factor tilts to roughly 10% to 30% of an equity allocation, layered on top of a core market-cap-weighted holding. That range captures a meaningful diversification benefit from the factor’s different return pattern without turning the whole portfolio into a concentrated bet on one signal’s timing.

What’s the difference between “smart beta” and factor investing?

The terms overlap heavily in practice. “Factor investing” usually refers to the academic framework and the specific characteristics being targeted, such as value or momentum. “Smart beta” is the marketing label most fund providers use for long-only index products built to capture one or more of those factors, and it can include additional rules like sector caps or volatility constraints that aren’t part of the underlying academic definition.

References

1. Fama, E. and French, K. (1992, 1993). Foundational papers establishing the size and value premiums beyond CAPM beta, and the original three-factor model.
2. Carhart, M. (1997). “On Persistence in Mutual Fund Performance,” introducing momentum as a fourth factor.
3. Fama, E. and French, K. (2015). “A Five-Factor Asset Pricing Model,” adding profitability and investment factors.
4. Asness, C. and AQR Capital Management research library. Papers addressing the 2007–2020 value drawdown, including analysis of its length and depth relative to prior history.
5. Arnott, R., Research Affiliates research library. Analysis of the value factor’s historical drawdowns and the intangible-asset accounting critique of book-to-price screens.
6. McLean, R. D. and Pontiff, J. (2016). “Does Academic Research Destroy Stock Return Predictability?” Journal of Finance.
7. MSCI and FTSE Russell factor index methodology documentation, describing construction rules, rebalancing schedules, and turnover characteristics of published factor indexes.
8. Dimensional Fund Advisors research library, on long-only implementation of academic factor premiums and the gap between long-short and long-only capture.