Sell in May and go away

Brooke Thackray put out an interesting report (click here)  on the health of the stock market. It’s worth reading. His basic premises is that this year there is an even greater potential for a weak summer (i.e. the “worst six months”).


Today, I’ve posted a chart, courtesy of, for the DJIA’s monthly seasonal patterns since 2000 (tech bubble). As you will note, May can be one of the weaker months, with only about half of all May’s being positive (56% of  May’s since 2000). June is the more concerning month, given its performance since 2000. Only 29% of all June’s have been positive since that year – which puts it ahead of September for the “Most likely month to fail” award. It’s probably worth noting that, despite the greater number of June failures, those poor performing Junes are on average not as bad as the poor performing Septembers. So it’s an arithmetic (i.e. the average number of poor month’s) problem that June faces, not a geometric (i.e. the extent of the pullbacks, if one occurs) problem. September tends to contain some of the bigger market meltdowns on a percentage basis –  despite having less of them.


Seasonality chart of the DJIA Industrials from 2000 to 2017 - in consideratio of the Sell in May and Go Away phenomenon


Thackray’s 2017 Guide tells us that since 1950, the average return for June has been 0.0% to 2016.  Meanwhile the average September’s return was -0.5% since 1950. Mathematically inclined readers will see that September’s bigger pullbacks offset the fact that there are less of them than in June. Whatever the case, this study suggests a greater probability of weakness in June – even if its typically not an aggressive pullback.

We’re up to 28% cash in the ValueTrend Equity Platform at this time. I’d expect to be selling more equities in the coming days and weeks.


  • Keith could you comment on Jason Goepfert’s comments about seasonality in the Sentiment trader? Perhaps it is what we wish to look and pay attention to when we read?

    • I talked to Brooke Thackray about his comments. Bottom line– if you look at the numbers, it has worked about 70% of the time–that is, “worst 6 months” do actually underperform 70% of the time. However, they are not always losers. They can merely underperform with smaller gains. The big problem, and the reason the strategy works, is that IF the market does in fact sell off in the worst six months, it sells off much bigger than it has during “best 6 month” corrections. So again, its a geometric problem –not as much an arithmetic problem.
      So it may not be worth it to sell before the worst 6 month period unless you use other indicators that signal greater risk. This is why I created the “Bear-o-meter”. It contains seasonality–but has a full menu of other indicators to indicate risk. If we get a poor signal, it only enhances the chance of a “worst 6 month” correction.

    • If you are not interested in the joy ride of volatility, sell in May looks like a good strategy.

  • Hi Keith,
    In the past you have spoken about inverse ETF’s to hedge and reduce volatility in the portfolio. At this time of year, in addition to raising cash, I have selectively used the HIX or HIU. I have never felt totally comfortable doing this extensively however because of the decay that takes place as these instruments are reset daily. My target this summer is to get to 15 % of my equity components in my portfolio to cash. Ordinarily, what percentage of your cash would you allocate to inverse ETF’s such as HIX or HIU?
    Thanks for your response.

    • Single inverse ETF’s don’t decay as aggressively as leveraged ones. For this reason, I have only done the single inverse ETF’s as you describe.
      I use them strictly to offset existing positions–not as an outright bet. I offset stocks that I know have at least market beta (1.0) or greater. Some readers may know that I own Google—it has a high beta. I might want to keep GOOGLE, but i think it will fall as much or more than the market in a correction (assuming I feel strongly a correction is imminent). The inverse ETF–perhaps a NASDAQ one would be best–would help offset a decline in Google. Once the danger period ends, I get rid of the inverse ETF–no specific time in mind- although I don’t want to hold these instruments for more than 2 months really.
      The bottom line–I use inverse ETF’s to create a “market neutral” situation surrounding a portion of my stocks that I want to hold, but I feel are most vulnerable–by owning inverse ETF’s I can (not perfectly) at least partially offset the losses, if any- in a correction. If markets go up, my stock goes up, the inverse ETF goes down, and I have offset the gains on the stocks. So its a relatively neutral position–you dont gain or lose much –except in the case where you miscalculate the extent the stock would move. So if Google fell 10% in a 5% market decline, my equal weighted inverse ETF only offset half of its loss–same with a gain (the reverse happens).
      Hope that helps

  • Hi Keith,

    Now we are about to enter the weak season, I have a question regarding selling an ETF after a long-term break-out:

    I’m looking at an ETF that I bought last year, which has since rallied and broke-out. I hesitate to share the ticker, because I don’t want to appear like I’m “Selling It”, but since an ETF rather than a stock, I highly doubt sharing it would “move the price”, so here it is: ZEQ. There is nothing wrong with the trend. It’s beautiful. The RSI is right on 70.00, but trends often stay over-bought, so I don’t think we want to sell based on that. I have once seen a trader draw a line at the long-term break-out level, then draw a line at the lows below that line, measure the distance, and set that as the price target from the break-out. Unfortunately, I cannot find the post or tweet that explains this. Does that make sense to you? Basically, instead of selling based on a percentage move away from the long-term moving average (which you have talked about before), you sell when the prices moves as much as the price move from the recent lows to the break-out.

    I’d much appreciate a few words on this, if that is something you have seen or tried,
    Thanks Keith,

    • Matt- an excellent question. I think what you are referring to is a measured move based on taking the prior move and extrapolating it forward. Its logical–the past volatility shows us how much it could move in the future–after all, that past volatility is based on what investors were willing to move the stock (ETF) by. So if you measure from the prior low to the top of a former trading range (before a breakout) that shows us how much the market was willing to swing this stock (or ETF) in the past. Now, its broken out. But market participants are still likely to move it to the same degree as they have in the past before running out of enthusiasm. Note that this techniques is better used with stocks–ETF’s are less concentrated in their market participants, being made up of many stocks–and in your case, many country stocks.
      I like to use that sort of thing as a target potential–but if there is no overhead resistance and you are in an uptrend, I prefer to ride it as long as I can and sell when it rounds over.
      Hope that helps

      • Super clear reply Keith. Thank you.
        I love your willingness to transfer your knowledge to the less experienced.

        • Matt–be sure to tell others to follow the blog–the greater the community, the better.


Leave a Reply

Your email address will not be published. Required fields are marked *

Never miss another blog post!

Get the SmartBounce blog posts delivered directly to your inbox.



Recent Posts


UK & Canada: the next value plays?

Hiu to gold

Value plays

Ask us anything


Long bond setup

NAZ futures

Opportunity in the fall, gold, and why risk-on matters


Just asking

Keith's On Demand Technical Analysis course is now available online

Scroll to Top