Sell in May Anomaly: What the Evidence Shows

The saying “sell in May and go away” suggests that stock returns are weaker from May through October than from November through April. This sell in May anomaly is one of the most famous seasonal patterns in finance. The evidence for it is real and consistent across many markets, yet it persists despite decades of academic study and widespread awareness—a puzzle that challenges the theory of efficient-market pricing.

The Pattern: Returns by Season

Academic researchers have confirmed that if you divide the year into two halves—“winter” (November–April) and “summer” (May–October)—the winter half has historically delivered higher average returns in US equities and many developed markets.

Using long-run data, the historical pattern is stark. In the US stock market from 1950 to 2020, the sp-500-index gained an average of roughly 8% per annum in the November–April period and about 3% per annum in the May–October period. That is not a minor difference; it is a 5-percentage-point annual gap. Compounded over decades, the impact is enormous.

The effect is strongest in May, June, and September. July and August are surprisingly positive, bucking the summer slump slightly. This nonuniformity—the pattern is not a clean split—is a clue that the explanation is more subtle than a single cause.

Historical Persistence

The pattern has been documented back to the 19th century in both the US and UK markets. Academic studies have uncovered seasonal returns in data spanning 1820–2020. This is not a recent, statistical artifact; it is a deep historical phenomenon.

However, the strength of the pattern has varied. In some decades (notably the 1950s–1960s), the May–October gap was pronounced. In the 1980s–1990s, it was less consistent. Since 2000, the effect has weakened further, though it has not vanished.

This gradual attenuation is important: it suggests that awareness and trading by sophisticated investors have eroded the anomaly, reducing—but not eliminating—the opportunity. If the pattern were a true market inefficiency, you would expect it to disappear entirely once it is widely known. That it persists in diluted form is evidence that multiple factors drive it, not a single simple mispricing.

Geographic Evidence

The sell-in-May pattern is not unique to the United States. Researchers have found versions of it in:

• United Kingdom: Similar November–April outperformance, documented since the 1700s.
• Australia: Clear seasonal effect, though slightly different timing due to Southern Hemisphere seasons.
• Canada: Robust May–October underperformance.
• Japan: Present, though weaker than in Western markets.
• Europe: Mixed evidence; some countries show it, others do not.

Emerging markets show weaker or inconsistent seasonal effects. This geographic variation is another clue: the anomaly appears correlated with developed, institutional markets where many investors have the same vacation calendars and tax years.

Competing Explanations

Why does this pattern exist? Multiple theories have been proposed, and none fully explains it:

Vacation and Rebalancing

Institutional investors often take August off. Summer vacations reduce trading activity and market-maker participation. Investors also conduct portfolio rebalancing in October (fiscal year-end for many institutions), triggering buying ahead of the winter months.

Bonuses and Tax Timing

Many financial professionals receive bonuses in January or early spring. They deploy this cash in equities around February–March, boosting winter demand. Conversely, in October and December, many investors realize losses for tax-loss-harvesting, selling positions ahead of year-end. Summer sees less of this tax-driven turnover.

Window Dressing

Fund managers want their year-end (or quarter-end) holdings to look good to clients. In late October and December, they buy “quality” stocks and sell “junk.” This demand spike for blue-chip names in Q4 boosts returns. Conversely, they trim weaker positions in spring, creating selling pressure.

Economic Seasonality

Real economic activity is indeed seasonal. Consumer spending peaks in November–December (holiday shopping). Corporate earnings announcements cluster in specific months. If earnings announcements tend to be more positive in Q1 and Q4 than in spring and summer, momentum-investing strategies would naturally ride those cycles.

Expiration Effects

Options and futures contracts expire monthly and quarterly, with concentrations in March, June, September, and December. These expirations can trigger unusual volatility and order flow, pushing prices temporarily. The pattern of expirations is not uniform across the year, potentially explaining some seasonal spikes.

None of these explanations is bulletproof. Together, they paint a picture of a market shaped by institutional calendars, tax codes, and human behavior in ways that create predictable—but mild and fragile—seasonal tilts.

Statistical Strength and Robustness

When researchers run statistical tests, the May–October underperformance is real at conventional significance thresholds. A raw average difference of 5% per annum compounded over 70 years is not noise.

However, the effect is not consistent month-to-month or year-to-year. Some Mays are fine; some Novembers are dire. The anomaly is a tendency, not a law. In any given year, you can encounter exceptions. This is why it persists: it is too weak and erratic to be exploited profitably once you account for transaction costs, taxes, and execution-risk.

Why It Persists Despite Being Known

The anomaly is famous. Articles appear every April warning investors to “sell in May.” Ironically, this awareness has likely eroded the effect. If everyone tries to sell in May, the May selling is already priced in. The fact that the pattern survives at all is puzzling.

One interpretation: the anomaly reflects true seasonal patterns in economic activity and investor behavior that are so deep (tax codes, vacation timing, fiscal years) that they cannot be traded away. Another: the anomaly is simply not profitable after costs, so traders do not arbitrage it hard enough to erase it.

A third possibility is that the anomaly is partly self-fulfilling. Retail investors and some institutions still believe “sell in May”; their compliance with the rule keeps the pattern alive. But the effect is small enough that it does not compensate for the costs of moving in and out of positions, so most rational money leaves it alone.

Modern Evidence and Decline

Post-2000 studies show the anomaly is weaker than historical data suggests. Some researchers argue it has mostly disappeared in recent decades, especially after 2010. Others contend it still shows up in factor returns (e.g., value stocks underperform in summer more than growth stocks) and in bond markets.

The decline in the anomaly is consistent with increasing market efficiency. As more investors have access to data and compute power, obvious seasonal anomalies become arenas for competition. The half-life of a known anomaly is typically 10–15 years; once published, it attracts capital and degrades.

For a practical investor today, the takeaway is not to trade mechanically on the May calendar. But it is also not to dismiss seasonality entirely. Real institutional and real-economic rhythms still exist. They may not be tradeable as simple long–short seasonal bets, but they appear in specific market segments and interact with other factors like momentum-investing and value-investing.

See also