Sell in May Doesn't Work. Except Maybe This Year.
The data killed the saying years ago. The midterm election cycle might bring it back.
Four million people will Google “sell in May” this month. Most of them will read a headline, nod, and do nothing. A few will actually sell. Almost none will look at the data.
I did. Here’s what it says.
From 2000 to 2025, the November-to-April period outperformed the May-to-October period. That’s the headline. Both periods were positive. If you just held the S&P 500 through all of it, you ended up doing better than either seasonal slice on its own. Edge? Technically yes. Tradeable edge? Barely.
Break it into eras and it gets clearer. From 2000 to 2010, the pattern was real. The dot-com crash and the financial crisis both hammered the May-to-October window. That’s probably where the saying got its teeth. From 2011 to 2020, the gap shrank. Over the last four or five years, it’s been roughly even. Classic arbitrage decay. If enough people know about an edge, the market eats it.
But this year has a wrinkle. Midterm elections.
When you isolate midterm years from the dataset, the seasonality effect comes back hard. The reason is structural, not mystical. Large institutions invest for quarters and years. If they don’t know which party is writing the rules, they wait. They don’t place sector bets into uncertainty. That creates a natural deceleration in the May-to-October window of midterm years that doesn’t show up in other years.
I’m not selling everything. I’m not making a calendar-based trade. But if we start to see weakness in the next few months, this is context that keeps me from panicking. It might also mean leaning harder on the shorter-term systems. Pullback setups, mean reversion, day trading. Less on the investment side until the election picture clears up.
The systems I run at Stats Edge have been through every midterm cycle since 2000. They’ve had drawdowns during those windows too. I publish them in The Drawdown Memo when they happen. That’s the difference between seasonal speculation and actual positioning.
The framework for how I test claims like this, including walk-forward methodology and the regimes that broke most strategies, is in the free 25-Year Backtest PDF at letters.statsedgetrading.com. This is exactly the kind of question pages 2-3 were designed to answer.
For real-time alerts with entries, stops, and sizing across all three systems, that’s Stats Edge Pro at $149/month with a 30-day money-back guarantee.
— Michael Nauss, CMT, CAIA, CDMS