From 1993 to 2026, SPY (the S&P 500 exchange-traded fund, or ETF) returned more from November to April than from May to October. The mean was 7.28% against 4.28%, with dividends counted. A mean is the sum of the returns divided by how many there are. Each figure averages 33 half-years.
The saying "sell in May and go away" describes that gap. The gap is 3.0 percentage points. It is real in the data, but small enough to be luck. Selling every May also earned less than staying invested.
Does "sell in May and go away" work? The saying holds that stocks earn less from May to October than from November to April. In SPY since 1993 that was true on average, but luck can explain the gap. Staying invested earned more than selling each May.
November to April returned a mean of 7.28% against 4.28% for May to October. It was higher in 19 of 33 years.
We handed each year's two returns to the two halves at random, 200,000 times. A gap of 3.0 percentage points or more in either direction appeared in 17% of tries. That assumes the 33 years are independent.
Selling every April and buying back every October earned 6.86% a year with the cash earning nothing. It earned 8.19% with the cash earning the interest rate on 3-month US Treasury bills. Staying invested earned 10.80%.
What does the saying tell investors to do?
The saying claims that stocks earn less from May to October than from November to April. Followed literally, it means selling shares at the end of April and buying them back at the end of October.
How did we test it?
We used SPY's daily prices from 29 January 1993 to 26 October 2026, with dividends counted.
We fixed the dates before running the numbers. The May to October half runs from the closing price on the last trading day of April to the closing price on the last trading day of October. The November to April half runs from the last trading day of October to the last trading day of the next April.
Only finished halves count. The first starts on 30 April 1993 and the last ends on 30 April 2026. That gives 33 halves of each kind.
The May to October half of 2026 is unfinished, so we left it out of every figure in this post. One year means a May to October half and the November to April half that follows it.
The median is the middle return once the 33 are in order. A percentage point is the gap between two percentages.
Did November to April beat May to October?
On average, yes. The mean was 7.28% for November to April and 4.28% for May to October. The medians are 8.18% and 4.64%, a gap of 3.54 percentage points. Chart 1 shows both.
Source: YX Insights price data
The average favours November to April, but the pattern is patchy by year. November to April came out higher in 19 of the 33 years, so May to October was higher in the other 14. November to April rose in 26 of the 33 years. May to October also rose in 26.
The worst May to October fell 29.2% in 2008. The best May to October rose 23.7% in 2025.
Which years favoured November to April?
Chart 2 gives each year's gap. A bar to the right means the November to April half returned more than the May to October half before it.
Source: YX Insights price data
Across the 33 years, the gap ran from 22.4 percentage points in favour of November to April in 1998 to 20.5 against it in 2021.
Did the gap hold in every period?
The table splits the 33 years in other ways.
Years | Number of years | Mean, May to October | Mean, November to April | November to April higher in |
|---|---|---|---|---|
All years, 1993 to 2025 | 33 | 4.28% | 7.28% | 19 of 33 |
1993 to 2009 | 17 | 2.77% | 6.36% | 11 of 17 |
2010 to 2025 | 16 | 5.88% | 8.27% | 8 of 16 |
All years except 2008 | 32 | 5.32% | 7.77% | 18 of 32 |
SPY with dividends counted. Each year pairs a May to October half with the November to April half that follows it. Source: YX Insights price data
The gap shrank after 2009. From 2010 to 2025, November to April had the higher mean, but it came out higher in only 8 of the 16 years.
Dropping each of the 33 years in turn leaves a gap between 2.4 and 3.7 percentage points, always in favour of November to April.
The count of years depends on which halves we pair. Our main pairing puts each May to October next to the November to April that follows it. That gives 19 of 33.
Pairing each May to October with the November to April before it gives 22 of 32.
Could chance explain the gap?
Yes. We ran a shuffle test. In each of the 33 years, we randomly decided which of that year's two returns counted as May to October and which as November to April. We recomputed the mean gap. We repeated that 200,000 times.
A gap of 3.0 percentage points or more in either direction came up in 17% of tries. Chance alone produced a gap this large in about one try in six.
The test assumes each year is independent of the others. It also assumes a year's gap is equally likely to favour either half if the saying is false. Real markets may break both assumptions.
Is May to October the weakest six months to hold?
The saying picks one six-month window out of twelve. We ran every window that starts at the end of a calendar month. Each has 33 observations.
On the mean, the window that starts at the end of April, which is the May to October half, ranks twelfth of twelve at 4.28%. The window that starts at the end of October ranks first at 7.28%. Chart 3 shows all twelve.
Source: YX Insights price data
The median gives a different order. The weakest window by median starts at the end of November, at 4.07%. The end of April is second weakest, at 4.64%.
The twelve windows overlap, because each shares five months with the next. They are not twelve separate tests. The ranking shows that May to October had the lowest mean of the twelve in this sample. It does not show that the shortfall is beyond what chance produces.
What would selling in May have cost?
We followed $10,000 from 30 April 1993 to 30 April 2026, which is 33.0 years. The annual rate is the single yearly return that turns the starting sum into the ending sum.
A Treasury bill is a loan to the US government that lasts under a year.
Chart 4 shows three paths. One stays invested. One sells with the cash earning nothing. One sells with the cash earning the 3-month Treasury bill rate. Each month out of the market, that cash earns one-twelfth of the month's 3-month bill rate.
Source: YX Insights price data; Federal Reserve Board
Staying invested grew the $10,000 to $295,400, an annual rate of 10.80%. Selling at the end of each April and buying back at the end of each October left $89,198, or 6.86% a year, with the cash earning nothing. Adding the bill rate lifts the selling path to $134,379, or 8.19% a year. Staying invested still finished ahead.
Selling lost ground because the May to October half was positive on average. It returned a mean of 4.28%, while cash earned a mean of 1.25% over the same months.
What does the test leave out?
The test covers one fund over one 33-year window. Trading costs and taxes are left out. Both would lower the selling paths further.
Our guide to how to read a backtest covers the general problem of judging a pattern found in past prices. For how a few days can drive a long-run return, see time in the market.
In SPY since 1993, November to April returned more than May to October on average, but the gap fits chance. Staying invested earned more than selling each May.
Get the next explainer
The Academy is free to read, here on the web. The Long Trend, our free weekly letter, comes by email.
Learn more with YX Insights
This explainer is part of the YX Insights Academy. Each one takes a single idea and checks it against real data.
The same approach runs through everything else we publish:
Systematic Portfolio: ready-made portfolios for Macro & Megacaps and for Commodities. We publish the holdings and every change.
Multi-model Signals: a daily long-or-flat call on every name we cover. Each call shows how strongly our four models agree.
Research: company deep dives, macro commentary and essays on how we test.
Good places to start on the website:
Academy: more explainers like this one.
Start here: get our free weekly letter by email.
Macro & Megacaps Systematic Portfolio and Commodities: the two portfolios in detail.
Track record: live results since 3 August 2026, plus the ten-year backtest.
Upgrade: get the Systematic Portfolio, Multi-model Signals and research in full.
Common questions about selling in May
Has SPY been weaker from May to October?
Yes, on average since 1993, but the gap fits chance. SPY is the S&P 500 exchange-traded fund. The May to October half returned a mean of 4.28% with dividends counted, against 7.28% for November to April. May to October still rose in 26 of 33 years.
Does selling in May beat staying invested?
No, not in our test. From 30 April 1993 to 30 April 2026, selling every April and buying back every October turned $10,000 into $89,198 with the cash earning nothing. With cash earning the Treasury bill rate it became $134,379. Staying invested gave $295,400.
Is the sell in May effect statistically significant?
Not in our test. We randomly reassigned each year's two half-year returns to the two halves, 200,000 times. We recomputed the mean gap each time. A gap of 3.0 percentage points or more in either direction came up in 17% of tries. Chance alone produced a gap this large in about one try in six.
Does November to April beat May to October every year?
No. November to April returned more than May to October in 19 of the 33 years, so May to October came out ahead in the other 14. From 2010 to 2025, November to April was higher in only 8 of 16 years, which is half.
DISCLAIMER: This article is intended for educational purposes only. Any information or analysis in this note does not constitute an offer to sell or a solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice, nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.
