On selling in May

Little remark from Steve Reitmeister of Zacks:

Every year I get more emails on this one subject than any other: “Hey Reity, should we sell in May?”

I try and douse that farcical flame with facts such as: Stocks have been up 59% of the time in the May through October period going back to 1950. Yet, it doesn’t stem the tide. So for all those whose heads were not up their rectums, congratulations for owning stocks this May as they put in another positive performance. For everyone else, it’s time to wise up.

Best to everyone,




I think it may be more instructive to consider actual data rather than passing comment by Zach’s.

Consider for example this more scientific evaluation of the issue of selling in may:


The average difference between November-April and May-October returns is 6.25% over the past 50 years. A Sell-in-May trading strategy beats the market more than 80% of the time over 5 year horizons.

To my knowledge, these investigators didn’t have their heads up their rectums.

1 Like

A Sell-in-May trading strategy beats the market more than 80% of the time over 5 year horizons.

That paper looks interesting and I look forward to reading it. But my own personal results would be much poorer if I had taken the “sell in may and go away” approach.

I’ve only kept track of my monthly portfolio returns since the beginning of 2011, but here is what I have for those years:

2011 May - October: +4.7%
2012 May - October: +0.3%
2013 May - October: +26.4%

So 2011 helped, 2012 was neutral, and I’d certainly be much poorer today had I sold in May 2013.

It’s just so much simpler to ride out the dips and keep your focus on the long term. Does it really matter if your portfolio drops a little for a few months over the summer? Is trying to avoid that really worth risking the occasional 25%+ return?



Stocks have been up 59% of the time in the May through October period going back to 1950.


A Sell-in-May trading strategy beats the market more than 80% of the time over 5 year horizons.

The clear and obvious explanation for the contradiction in those two statements is that stocks have been up 59% of the years, but when you look at 5 year segments, it expands the effects of outliers. Thus, for example, the huge one day drop in Oct 1987 affects 5 different groupings 1983-87, 84-88, 85-89, 86-90 and 87-91. Similarly, another such outlier month that the authors of your paper refer to in 1998 influences 5 more years the way those authors calculate it. Now if you remember back to the 1987 one-day crash (largest percentage one day fall of the Dow ever, something like 30or 40% as I remember it. Can that be right?), it had nothing to do with a May effect or anything else, it was a one day insanity, that some ascribe to Iran shelling two US ships, others ascribe to program trading (the 1987 equivalent of a flash crash), but no one really knows why it crashed. The market gradually increased back from there so May to October got credit for the fall and November to April got credit for the gradual rise.

Would you base your 2014 trading on that?

And consider that if you decided to look at May to Sept instead of May to October, you’d get an entirely different picture (moving the 1987 drop out of the summer and into fall and winter).

I mean really Duma, think about it….



Hey Saul.

Thanks for your kind words :wink:

I think my point is that you post an excerpt that merely represents “confirmation bias”. That is, your author says that the “market goes up 58% of the time” in the summer so if you sell you have your head up your rectum.

I am merely pointing out that this statement is nonsense and not supported by any factual representations.

For example, what if the 58% of the time, the market is up, it is only up an average of 2%, but the 42% downtimes, it is down 15%. His statistical commentary is, as a standalone, worthless information. Worse yet, it comes with a negative connotations of rectums, etc…classic confirmation bias.

So I post an actual study that proves that there is a summer affect across ALL country’s stock markets and…you can see their methodology and critique it (unlike your confirmation guy). Makes for a better conversation than rectums doesn’t it?

But keep in mind that this isn’t the only study. The more famous one is perhaps Bouman and Jacobsen (2002) that showed the same findings.

Or how about other independent research as Forbes reports:

Jeff Hirsch at the Stock Trader’s Almanac calculates that the Dow Jones industrial average has an average return of just 0.3% during the worst six-month period (May through October) since 1950. Conversely, during the best six months (November through April), the Dow has an average gain of 7.5%.

Sam Stovall at S&P Capital IQ says the S&P 500 has risen by a mere 1.2% during the average worst six-month period, while rising 6.9% during the average best six-month period. (Sam’s numbers go back to 1945.)

Now going back again to the study I referenced, they proved that:

The average 6-month winter returns (November through April) are 6.93%, compared to the summer returns (May through October) of 2.41%. The overall Halloween effect that measures the difference between winter and summer returns is 4.52%, with a t-value of 9.69.

As regards your supposition that this 5 year interval creates some anomaly in the data, that is NOT supported by the data and for reasons that should be obvious. If one uses a one year buy and hold vs. sell in May…that one year could be a statistical anomaly. But everything over a one year buy and hold proved a May Affect as the unusual year was diluted out.

And before you suggest it, I acknowledge that the May affect may be nothing more than an illusion related to the winter months being so good in comparison.

So you can take it or leave it…doesn’t matter to me. But let’s try to be as objective as we can about the truth and not minimize thoughts and data by bringing up rectums. Maybe you could reference the actual data that your guy is pulling his numbers from so we could examine it?

Because otherwise, I have provided you with 4 studies that are consistent with the contention that a May affect exists across all stock markets…the proviso of course being that past performance does not guarantee future results :wink:

The beauty of the market is that we don’t all have to agree. In fact, it is paramount that we don’t, otherwise everyone sells in May and the market crashes every single time.

1 Like

Hi Duma, it wasn’t me who put in the rectum remark, and it seemed inappropriate to me.

On the other hand, it’s hard to argue with the statement that starting fall and winter in October rather than November gives you an entire different picture because of October 1987 (Winter doing a lot worse, summer doing a lot better).

You also have to take into account the effect of 9/11

“On the first day of NYSE trading after 9/11, the market fell 684 points, a 7.1% decline, setting a record for the biggest loss in exchange history for one trading day. At the close of trading that Friday, ending a week that saw the biggest losses in NYSE history, the Dow Jones was down almost 1,370 points”

This has no effect on how future Septembers will do.

It’s like saying in my sample of 60 left handed people 6.9% more preferred ketchup to mustard on their hamburgers, but in this other sample of 60 right handed people only 2.4% more preferred ketchup, so left handed people are more likely to prefer ketchup. Where’s the cause and effect??? It’s just a correlation, like 99% of left-handed people drink water, so drinking water makes you left handed. You need causation.

And doesn’t world-wide summer effect make you wonder? It’s winter in New Zealand and Australia in those months (May to October). How about the fact that the New Zealand market was down 60%!!! in October 1987, in response to the US crash. Markets all over were hugely down. Isn’t that a better explanation than a summer and winter effect? How is that going to affect future summer markets???


1 Like

Thanks Saul.

You bring up very appropriate questions and some of them have been addressed in prior studies trying to understand why such a “summer” affect should take place.

It seems the prevailing thoughts are that this is vacation time for a large % of the world that would otherwise be moving a lot of money in the world markets. There is therefore less attentiveness to the market and more diversions until the kids back to school, etc.

Whether that is true or not, no way of knowing of course.

I liken this phenomenon to probabilities but not certainties.

FWIW, I do not sell everything in May but I do often increase cash to buy later when the doldrums more commonly (statistically) are improving.

I notice you re-allocating often enough that you may not notice the seasonal impact regardless. Ultimately though, the seasonality issue is not a primary investment strategy but rather, it is an awareness that the perhaps one’s more speculative stocks can grind downward for no other apparent reason.

How one responds to that probability/awareness is a matter of personal investment preference.


Duma, Fascinating discussion even if we don’t totally agree.

A point: You mention that the one year anomalies are diluted by taking 5-year samples. BUT, what that study was saying was that x number of 5-year samples of 60 were down. Now if they were just using single years, the 1987 “flash crash” and 9/11 would each be just one of 60 single years, but by using them in running 5-year groups each pulls down 5 samples (to a diluted amount of course) and allows them to say a bigger “x number of samples were down”. It’s manipulating statistics.

Another point: Big black swan events like 9/11 have a big affect on statistical averages but have no predictive value. The next terrorist attack (God forbid) could just as well be in December or March.

Third point: You wrote trying to understand why such a “summer” affect should take place. In the 1700’s and 1800’s people probably wrote about selling in May and going away, because that’s what they did. Literally. Rich people and aristocrats in England went to the country for the summer, or the beach, or Southern France. If you were in your country house in England it probably could have been two days each way to get a message back and forth to London, and across the channel proportionally more. A company could lose three of their big sailing ships and go broke before you’d find out on your vacation. Of course you’d lighten up during the summer.

Even in the 1950’s I remember that a long distance call was a project. And expensive. In the late 50’s I spent a summer in France (I was in my early 20’s). You didn’t call back to the US. You sent post cards. To make a call you’d have to go to the post office and schedule a call. In the late 1970’s, I had a plane canceled while I was in Frankfurt, and had to call back to the States. I had to have the operator place the call and it cost me $30 (probably $75 in current dollars). This was in a big modern city. Sure people who were going on vacation would lighten their positions.

People still take vacations and go to the beach and don’t want to follow the market as closely, but if they want to they can even call their broker on their cell phone from the beach. It’s a different world and they are not so cut off, and I doubt the effect is anything like it was 50 years ago or 100 years ago.

The statisticians you refer to almost certainly started to investigate a SUMMER effect. I doubt they thought there was something mystical about May that was going to make the market do worse. They were looking at the effect of summer vacations.

But I was a professor for a number of years and know about writing academic papers. I must have written 30-40 of them. You don’t fake data but you look at the data to see how you can best get a paper out of it. Your guys, who started out writing about the effect of summer and summer vacations, thought to themselves “Our data is showing just a tiny effect if we look at the summer. I know everyone is back at work in September so it doesn’t fit our model, but if we include it, we can pick up the start of the 1929 market crash AND 9/11, so even if it has nothing to do with summer, and certainly nothing to do with May, it sure helps our data.” And then they figured that if instead of a summer effect, they used the excuse of looking at 6-months, they could include October as well, and capture the 1987 “flash crash” (and the second month of the 1929 crash if they were going back that far). While you could say a few people were still on vacation in September, there’s no way to justify October, a regular work month that has much more in common with winter than summer, in a study of a summer effect, except to capture the 1987 drop and make your figures look better.

As you can see, I’m not a believer.



People still take vacations and go to the beach and don’t want to follow the market as closely, but if they want to they can even call their broker on their cell phone from the beach. It’s a different world and they are not so cut off, and I doubt the effect is anything like it was 50 years ago or 100 years ago

When I am on vacation, I can follow my stocks more closely than when I am in office. I don’t have those meetings that i would have in office. I can sit in the beach and watch the market with my ipad.

I was going through my first-half results yesterday, and I was struck by how much of my year-to-date gains have come just since May 1st. Overall, my total portfolio is up around 10% from Jan 1 through the end of June (I’m about 80% invested and 20% in cash), but up over 8.6% just since May 1st. So had I decided to sell in May and go away, I’d be up less than 1.5% and would be badly lagging the market instead of beating it.

It’s early yet, and it’ll be interesting to see where we are in October. But the lesson is that timing the market is tough: it’s best to just stick to your strategy and take advantage of opportunities whenever they happen to come your way.



I was going through my first-half results yesterday, and I was struck by how much of my year-to-date gains have come just since May 1st.

I agree Neil. As I wrote in my article about my “End of the Month Positions”

Seven weeks ago, at the bottom of that horrible growth-stock sell-off (which now is just a distant memory), when someone on the board was panicking, saying that he was down so much that he was afraid he’d never get it back, I answered that I was down 13%, but it had never even occurred to me that I’d end up down for the year. It was true. I wasn’t exaggerating. It actually had never even occurred to me. These were good stocks with great prospects, and I knew I wasn’t going to be down for the year. I’m now up 5.7% (which is up 21.5% from where I was when I said that, less than two months ago). The take-away is don’t panic at irrational sell-offs. If it’s one stock, then you have to make sure there’s not something wrong with that particular company. When it’s all your stocks and the pundits are all saying the big crash is coming, relax. It’s just something that happens now and then.

And that bottom was in early May. If I had “sold in May and gone away”, I would have missed the biggest rise of the year for my stocks. Don’t pay any attention to that stuff!