Market Anomalies: Patterns That Challenge Market Efficiency

The Efficient Market Hypothesis makes a testable claim: sort stocks by anything publicly known, and the resulting portfolios should show no predictable return spread — anything predictable should already have been traded away. Yet researchers running exactly these sorts on decades of US data have documented a set of anomalies: groupings where the return spread persists, with a stable sign.

Set expectations first: an anomaly existing does not mean you can profit from it. Transaction costs, risk, and competition routinely eat the paper profits. What every anomaly does do is force a choice among three explanations — compensation for a risk we can’t measure, the footprint of investors repeating the same mistake, or a statistical fluke that was bound to turn up once researchers tried enough variables. That three-way test is the framework this whole article runs on.

One variable, three horizons

The classic sorting variable is past return itself, and it points in opposite directions depending on the window.

At 1 to 12 months you get momentum: recent winners tend to keep rising, recent losers keep falling. Jegadeesh and Titman (1993) supplied the standard evidence — buying the past 3–12 months’ winners and selling the losers earned roughly 1% per month over 3–12 month holding periods, before transaction costs. The behavioral story is underreaction: prices digest news gradually rather than instantly, and herding amplifies the drift.

Shrink the window to a week or a month and the sign flips. That’s short-term reversal (mean reversion): stocks that just got hammered tend to bounce. Stretch it to 3–5 years and it flips again — very long-term losers outperform very long-term winners. Both ends share one explanation: overreaction to news creates price overshoots, and overshoots eventually correct.

So “past returns predict future returns” carries no information until you name the horizon: reversal inside a month, continuation inside a year, reversal again at three to five years. Any strategy built on price history owes you a statement of which horizon it lives on.

Cheap stocks beat expensive ones

The second family of sorting variables is price relative to fundamentals. Stocks that are cheap relative to earnings, book value, or dividends — low P/E, low P/B, high dividend yield, in other words value stocks — have historically outperformed growth stocks, the ones priced expensively relative to fundamentals. Fama and French (1992) put the historical spread at roughly 5–7% per year in US equities.

Why should that spread exist? The behavioral version: overconfident investors overpay for “exciting” growth stories while excessive skepticism turns struggling companies into bargains. The risk version: cheap stocks may be cheap for a reason, the extra return compensating a hard-to-measure risk. That argument has run for three decades without a verdict — the three-way test in the wild.

The size effect: more useful as a cautionary tale

Small-cap stocks historically outperformed large caps by roughly 2–3% per year. The explanations hold together: small companies are less liquid and riskier, so investors demand a premium, and institutional position-size limits keep big money out of small caps, depressing prices further.

The second half of the story is the part worth remembering: since the 1980s — that is, since the effect was documented and became common knowledge — the small-cap premium has been much smaller, and negative in some sample periods. How much of a pattern survives publication and the capital that chases it is itself a test of what the pattern was made of. The size effect did not score well.

The January effect

The calendar can serve as a sorting variable too. The January effect is the tendency of small-cap and value stocks to do particularly well in January, especially the first few trading days. The usual mechanical story: tax-loss selling in December pushes prices down, buying returns in early January, and early buyers capture the rebound. On magnitude, studies have put the first trading day of January at something like 5–10x the average daily return historically — but such figures are very sensitive to the sample period, so don’t treat them as constants.

It’s also the standard demonstration of the gap between paper profits and realized ones: the effect concentrates in the least liquid small caps, exactly where spreads and market impact are most expensive. After transaction costs and taxes, most investors keep little of it.

The anomaly zoo

The anomalies above are just the most heavily studied. Grouped by where the sorting variable comes from, the literature splits into three families.

Price-based (technical):

Fundamental-based:

Corporate events:

The longer the list gets, the more suspicious you should be. Researchers have tried hundreds of candidate variables, and even if returns were pure noise, some sorts would look “significant” in historical data.

Why arbitrageurs haven’t flattened them

Which brings back the unavoidable question: if these anomalies are in the textbooks, shouldn’t arbitrageurs trade against them until the profit opportunity is gone?

Part of it has been traded away — the size effect is the example, and competition does grind down published patterns. What remains survives for different reasons. Some anomalies may not be free lunches at all but compensation for risk we can’t measure. Some are unprofitable after costs: between the paper anomaly and an actual order sit spreads, market impact, and capacity limits, and nobody has an incentive to flatten a pattern that isn’t worth trading. And a good share were never real — statistical flukes from data mining that fall apart out of sample, kept looking alive by publication bias, since journals prefer to publish an anomaly’s discovery rather than its disappearance.

So the right takeaway is not “pick an anomaly and trade it.” It’s to run every pattern through the three questions from the top — risk premium, behavioral bias, or statistical artifact? — and then, whatever the answer, ask how much survives costs and implementation. Haugen’s The New Finance assembles this evidence into a frontal assault on the old paradigm; Shiller’s Irrational Exuberance is the fullest statement of the behavioral side. Both are worth reading before you settle your own view on how efficient markets really are.

References

• Fama, French, “The Cross-Section of Expected Stock Returns.” Journal of Finance, 1992.
• Jegadeesh, Titman, “Returns to Buying Winners and Selling Losers.” Journal of Finance, 1993.
• Haugen, The New Finance. Prentice Hall, 1999. ISBN: 978-0-13-081117-4.
• Shiller, Irrational Exuberance. Princeton, 2015. ISBN: 978-0-691-16417-6.