For decades, investors and academics have been locked in a hunt for alpha—the holy grail of risk-adjusted returns that can’t be explained by known market factors. The quest has grown so frenzied it spawned what John Cochrane famously called the factor zoo: a sprawling menagerie of hundreds of documented anomalies, each claiming to produce excess return. Yet, despite all this effort, alpha has been shrinking. Mutual fund alpha has declined, as has hedge fund alpha. The playbook that worked in 1993 looks considerably less potent today.
So, is alpha dead? Andrew Berkin and Christine Wang of Bridgeway Capital Management, authors of the study “The Incredible Structural Alpha,” published in the Spring 2026 issue of The Journal of Beta Investment Strategies, believe the answer is an emphatic no. However, finding it requires a fundamentally different mindset.
Alpha: What did the authors examine?
Berkin and Wang set out to show that meaningful, persistent alpha is available not by discovering exotic new factors or mining ever-larger datasets but simply by being more thoughtful about constructing portfolios.
Their laboratory was the classic 5×5 grid of portfolios sorted by size and value—the same analytical framework that Fama and French used in their landmark 1993 work. Using 60 years of US stock data (July 1963 through June 2023), they walked through four straightforward and easy-to-understand portfolio construction scenarios, measuring how each incremental design improvement affected both raw returns and risk-adjusted alpha.
The four levers they pulled were:
1. Deeper factor exposure: Concentrating portfolios more tightly in the extremes of size and value (true small cap, truly cheap stocks), rather than broader splits.
2. Timely data and more frequent rebalancing: Using the most current market cap available rather than stale year-old data and switching to quarterly accounting data when it becomes available.
3. Removing stocks with unwanted factor exposure: Screening out the worst-momentum stocks from value portfolios to avoid the drag of stocks that are cheap but falling for good reason and junk from small caps.
4. Using multiple value metrics: Combining four measures of value (book/price, sales/price, earnings/price, and cash flow/price) rather than relying on book/market alone.
What did they find?
The results are striking, and they build on each other.
1. Deeper exposure is worth it: Starting with the baseline Fama-French approach (Scenario 1), the smallest and deepest value stocks returned over 16.0% annually across the 60-year period—while small-growth stocks returned just 3.65%. Even after adjusting for their known factor exposures, the smallest, deepest value corner of the market generated a statistically significant alpha of nearly 2% per year. The smallest growth corner, meanwhile, produced a statistically significant negative alpha of 6%. The implication: The nonlinearities at the extremes of the size and value spectrum are real, persistent, and economically large.
2. Fresh data changes everything (especially through the lens of momentum): When the authors switched to quarterly data and current market cap (Scenario 2), something interesting happened. Returns for the deepest value portfolios actually fell slightly—but their measured alpha rose sharply, from 1.97% to 4.53% for the smallest deep-value portfolio. Why? Because more timely classification of stocks as “value” means you’re often catching stocks that have recently dropped in price—stocks with negative momentum. Those stocks carry a momentum headwind that suppresses returns but isn’t fully accounted for in simple alpha calculations. The alpha is real; the momentum drag is a separate, identifiable cost.
3. Removing the “cheap for a reason” boosts returns directly: Screening out the worst quintile of momentum stocks (Scenario 3) produced higher returns across nearly all 25 portfolio squares—with the gains largest among smaller stocks, where momentum is the strongest. The smallest deep-value portfolio’s return rose to 17.85%. Notably, alpha itself barely moved (4.53% to 4.57%), but its statistical significance increased substantially, with the t-statistic rising from 4.54 to 5.96. The message is clear: Removing poor-momentum stocks raises returns through better factor exposure, not through some mysterious new source of alpha.
4. Multiple value metrics unlock the large-cap value premium: This is perhaps the most practically important finding for many investors. It is well known that using only book/market as the value measure works poorly for large-cap stocks—the traditional high-minus-low-value premium mostly comes from smaller companies. But when the authors combined four value metrics (Scenario 4), the largest deep-value portfolio’s return jumped from 9.73% to 12.65% annually. The value premium was restored across all size segments.
The mechanism is transparent: Metrics like sales/price, earnings/price, and cash flow/price also capture profitability. Stocks that look cheap on these measures tend to have solid underlying businesses. Because the standard alpha model uses only book/market as its value factor, the extra returns from these other measures show up as alpha rather than factor compensation—which is exactly what the authors observe.
The structural size and value factors
Having demonstrated the benefits in portfolio terms, Berkin and Wang took the analysis one step further and constructed improved versions of the famous small minus big and HML factors themselves.
The results are sobering for fans of the conventional Fama-French factors—and encouraging for those willing to invest the effort in better construction:
- The standard SMB returned just 0.17% per month over the full 60-year period and was not statistically significant. The structural version returned 0.37% per month, rising to 0.44% per month when “cheap for a reason” low-value stocks were excluded from the small-cap side.
- The standard HML returned 0.29% per month. The structural version returned 0.33% per month.
The gap widens dramatically in the second half of the sample (1993–2023), when the conventional factors have struggled badly. The standard SMB averaged just 0.06% per month. The structural version earned 0.26%–0.33% per month. Similarly, standard HML earned just 0.13% per month in this period, while the structural version maintained 0.28% per month.
This is the paper’s most important practical finding: When the conventional factors were weakening, thoughtful portfolio construction preserved most of their return premium.
Finally, their findings were basically unchanged in a robustness check using the Fama-French six-factor model.
Key takeaways for investors
1. Alpha from construction, not from new factors: The factor zoo is partly an illusion. Many apparent new anomalies may simply be capturing the structural alpha that already exists in well-known factors when they’re properly constructed. Investors should be skeptical of claims that any new metric is truly distinct from size and value.
2. Go deeper, not broader: Concentrating in the true extremes of size and value—not the broadly defined “smaller half” or “high-30%”—produces meaningfully higher returns and alpha. Most mainstream indexes and many exchange-traded funds don’t do this. Index funds cannot capture this structural alpha.
3. Pay attention to momentum interactions: A stock that looks cheap may be cheap for a reason. Screening out the worst-momentum stocks from value portfolios is a practical, disciplined way to improve returns without abandoning the value thesis. The momentum and value factors interact—ignoring that interaction leaves money on the table.
4. Use multiple value metrics, especially for large caps: Book/market alone is a poor value signal for large-cap companies. Combining it with earnings yield, sales yield, and cash flow yield produces a more robust value signal across the full market-cap spectrum, particularly in the large-cap space where book value has become a less reliable anchor.
5. Implementation is not a footnote: The authors are explicit that disciplined execution matters as much as smart factor design. Patient trading, broad stock-selection pools, position bounds, securities lending, and—critically—closing a fund before it grows too large to execute its strategy are all levers that can add or destroy the structural alpha that portfolio design creates. A beautifully designed strategy, poorly executed, will still underperform.
6. Fees matter, but less than you might think; design matters more: The authors note that structural alpha can amount to several percent annually—an order of magnitude larger than typical fee differences between funds. Investors fixated on minimizing basis points may be optimizing the wrong variable if they’re ignoring the quality of portfolio construction.
7. The conventional factors aren’t broken; they’re just poorly measured: The well-documented weakness of size and value in recent decades is largely a story of how those factors are conventionally defined and measured. Structural improvements to construction preserve the premium substantially. The factors work. The definitions need updating.
8. AUM matters: As my September column “The Hidden Costs of Passive Investing” explains, large systematic managers attempt to reduce slippage by remaining flexible and limiting their participation in daily trading volume, typically to 1% to 3% of average daily volume. For smaller managers, or highly liquid large-cap stocks, this constraint is immaterial. However, for megafirms trading less liquid small-value securities, it creates a problem called latency—positions may take quarters, or even years, to build or unwind. The result is lower trading costs but also lower exposure to the factors with premiums.
The alpha bottom line
The search for alpha doesn’t have to mean chasing increasingly exotic signals in increasingly crowded corners of financial markets. Berkin and Wang make a compelling case that the building blocks—size and value—still work, and work considerably better when handled with care.
The “incredible structural alpha” isn’t magic. It comes from the disciplined application of things most sophisticated investors already know they should do: Concentrate on the true extremes of your target factors, keep your data current, avoid stocks with compounding headwinds, diversify your signal, and execute with precision. The paper’s contribution is to quantify exactly how much each of those choices is worth—and the answer is: quite a lot.
For investors navigating a world where conventional factor premiums have compressed, that’s a meaningful source of hope.
Larry Swedroe is a freelance writer and author. The views expressed here are the author’s. For informational and educational purposes only and should not be construed as specific investment, accounting, legal, or tax advice. The author does not own shares in any of the securities mentioned in this article.