Sell in May and go away? Summers really are weaker, but selling left you with less than half the money
We ran the old Wall Street saying through 55 years of S&P 500 prices and 8 stock markets. The summer slump shows up almost everywhere. Trading on it did not pay.
The belief
“Stocks do worse from May to October. Sell at the end of April and buy back at the end of October.”
The pattern is real: the S&P 500's winter beat its summer in 39 of 55 years. The trade is not: selling every May beat simply holding in only 4 of 21 full years.
What we tested
We followed the saying to the letter. On the last trading day of April we sold an S&P 500 index fund (SPY) and put the money into three-month US Treasury bills. On the last trading day of October we bought the fund back. Dividends were reinvested while we held it. Every switch cost 0.05% of the money, a fair price for a fund this large.
We started with $10,000 on 30 April 2004 and stopped on 2 October 2026. The comparison is the same fund, bought once and never sold.
What $10,000 became
- Held all year
- Sold every May, back in November
The result
Holding the fund turned $10,000 into $104,655, or 11.0% a year. Selling every May turned it into $48,125, or 7.3% a year. The seller sat in Treasury bills on 52% of trading days and missed most of the gains.
Selling came out ahead in only 4 full years: 2008, 2011, 2022 and 2023. In the other 17 the seller ended the year behind a person who did nothing.
| Year | Sold in May | Held all year |
|---|---|---|
| 2008 | −10.2% | −36.8% |
| 2011 | +9.6% | +1.9% |
| 2022 | −12.5% | −18.2% |
| 2023 | +27.8% | +26.2% |
So is the summer slump a myth? No
The weak summer is one of the most stubborn patterns in stock prices. From 1971 to 2025 the S&P 500 rose 7.0% in an average winter (November to April) and 2.2% in an average summer (May to October). Winter beat summer in 39 of those 55 years.
It is not only an American habit. In seven of the 8 big stock markets we checked, the average summer since 2005 was weaker than the average winter. The gap is widest in Europe: the German DAX gained 8.6% in an average winter and 1.7% in an average summer. The one exception is the Nasdaq 100, where summers did better.
Average winter and summer, 8 stock markets, 2005–2025
- Winter, November to April
- Summer, May to October
Why a real pattern still lost money
Weaker is not the same as negative. From 2004 to 2025 the S&P 500 rose in 17 of 22 summers, by 3.8% on average. Treasury bills paid less than that in most years, and almost nothing from 2009 to 2015, when their rate stayed below 0.2% a year. Each May the seller swapped a likely gain for a smaller sure one.
Some of the summers the seller missed were large: +22.8% in 2025, +18.7% in 2009, +13.3% in 2024 and +12.3% in 2020. The bad summers it dodged were fewer: −30.1% in 2008, −8.1% in 2011 and −6.3% in 2022.
The gap has shrunk since the idea became famous
Economists Sven Bouman and Ben Jacobsen made the pattern famous in a 2002 study, and fund managers have known it ever since. Before 2004 the average S&P 500 summer gained 1.1% against 7.8% for winter, a gap of 6.6 points. From 2004 to 2025 the summer gained 3.8% against 5.9%, a gap of 2.1 points.
Nobody can prove why the gap shrank. A trader who sells in May today is betting on the smaller, recent version of the pattern.
Every S&P 500 summer since 1971, May to October
- Summer rose
- Summer fell
- Average of the period
Who might still use it
The seller had a calmer ride. The worst fall from a peak was 36%, against 55% for the holder, and the day-to-day swings were about a quarter smaller (13.6% a year against 18.7%). That calm cost 3.8 percentage points of return every year. Counted as return per unit of swing, holding still came out ahead: 0.59 against 0.53.
If a fall of 55% would make you sell everything at the bottom, a rule that keeps you out half the year may protect you from yourself. As a way to make more money, it failed.
The bottom line
Summers really are weaker than winters, in the US and in most markets abroad. But stocks still rise in most summers, so stepping out every May has cost far more than it saved. HALF TRUE.
How we tested
- Rule
- Hold SPY from the last close of October to the last close of April; three-month US Treasury bills the rest of the year.
- Period
- 30 April 2004 to 2 October 2026, starting with $10,000.
- Costs
- 0.05% per switch, two switches a year. Taxes not counted.
- Dividends
- Reinvested in the main test. The long history and the other markets use index prices without dividends, for both halves of the year alike.
- Long history
- S&P 500 index, season years 1971 to 2025. A season year runs from the last close of October to the next last close of October.
- Other markets
- 8 indices, season years 2005 to 2025.
- Data
- Daily closing prices to 2 October 2026.
This is research on past prices. It is not advice to buy or sell anything, and past results do not promise future ones.