Sell in May? The Calendar Effect That Stopped Working

"Sell in May and go away" is one of Wall Street's oldest adages. It's also one of the few that used to actually work. For decades, investors who avoided the May-October period outperformed those who held through summer. But something changed around 2014. The pattern that worked for twenty years stopped working entirely.

I tested this using SPY data from 1993 to present—33 years of daily returns broken down by month and season. The findings surprised me. Not because the seasonal effect was real (it was), but because it vanished so completely in the past decade that implementing it today would actually hurt your returns.

The adage originated in London, where aristocrats would sell their holdings before leaving the city for summer. The American version suggested buying at Halloween and selling in May. And for most of market history, this simple rule beat buy-and-hold.

That's nearly double the returns for simply being invested during the right six months. The winter period also showed a higher win rate—67.2% of months were positive versus 61.1% in summer. For a passive strategy with zero stock selection, that's a meaningful edge.

The month-by-month picture

Looking at individual months reveals where the seasonal effect actually lives. September is the problem child—the only month with negative average returns over the 33-year sample. November is the star, with 78.8% of months finishing positive.

Notice that May itself isn't particularly weak (+1.12%). The "sell in May" guidance really should be "sell before September." The three weak months are February (+0.05%), June (+0.07%), and September (-0.89%). August is borderline at +0.03%.

If you could only avoid one month, September would be the obvious choice. If you could only be invested for two months, November and April would give you the highest expected return with the least volatility.

But then something changed

Here's where the analysis gets interesting. When I split the data by decade, the seasonal pattern shows a clear trend: it's been fading steadily and has now completely reversed.

The edge was real in the 90s (+0.75 percentage points per month). It was even stronger in the 2000s (+0.88pp). But since 2014? Summer has actually outperformed winter by 0.09pp. The pattern didn't just weaken—it flipped.

What happened? Several factors likely contributed. Algorithmic trading arbitraged away the predictable pattern. Global markets now trade around the clock, reducing the "summer lull" effect. And perhaps most importantly, tech stocks—which have dominated the past decade—show the opposite seasonality pattern.

The sector divergence

Not all sectors follow the same seasonal rhythm. When I broke down the analysis by sector ETF, a stark pattern emerged: cyclicals show strong winter effects while technology actually performs better in summer.

Energy (XLE) shows the strongest winter edge at +1.14 percentage points. Materials (XLB) follows at +1.08pp. These are classic cyclical sectors that benefit from the Q4/Q1 earnings acceleration.

But look at technology: XLK has a -0.44pp winter "edge"—meaning it actually performs better in summer. QQQ shows a similar pattern at -0.16pp. Utilities show zero seasonality at all.

September: The month everything fears

September deserves special attention as the only consistently negative month. But even here, sector differences matter enormously.

Tech gets hit hardest in September (-2.36% for XLK), while gold is the only asset with positive September returns. If you must be invested in September, gold provides the only historical shelter. Everything else bleeds.

Individual stock patterns

The sector patterns extend to individual stocks. Here's where it gets actionable: some stocks show consistent winter outperformance while others thrive in summer.

Apple's summer edge (+1.60pp) likely reflects the iPhone launch cycle and back-to-school Mac sales. Google's summer strength (+1.58pp) coincides with strong Q2 advertising results. AMD's winter edge (+2.44pp) aligns with gaming console cycles and data center budget flushes in Q4.

Energy stocks (Chevron, Exxon) show classic "sell in May" behavior—winter gains, summer losses. These patterns are consistent enough across 15 years of data to be worth incorporating into tactical timing decisions.