The Evolution of “Sell in May”
One last bit of follow up to my previous post questioning the “sell in May” rule.
In the table to the right (click to zoom) I show how the stock market’s best 6 months of the year has evolved over the last 80+ years. Each row represents 20 years. Red cells denote the “best” 6 months.
Recall that the “sell in May” rule is based on November through April being the best half of the year.
Given the results in my previous post, I was surprised by how consistently Nov – Apr (or at least Dec – May) was the best performing period.
Happy Trading,
ms
OOPS: in my rush to get this post out I botched the table I initially posted. I was showing the best 6 months from 1930 to whatever year was listed on the table (rather than that particular 20 year period). Sincerest apologies!
Geek note: the “best” 6 month period was chosen based on volatility-adjusted (not absolute) dividend-adjusted S&P 500 returns.
Filed under: Time-based | 12 Comments
That’s beautifully visualized.
Whatever happened to the “Summer Rally?”
Hello Russ – I was under the impression that the summer rally is a myth. A quick Google search turns up…
http://www.washingtonpost.com/business/economy/dont-buy-into-wall-streets-summer-rally-promos/2011/05/31/AGsxjgFH_story.html
And if the mainstream guys don’t even buy it, I have to think it’s bunk =)
michael
Wow, I never would have thought it was that consistent. I’m sure it would be a lot noisier but it would be interesting to see a yearly chart. Did something change in the late 1930′s to 1950′s that could account for this? More vacations? Changes to the due date for taxes? Something that could have effected pension fund buying?
Very interesting. Any chance you can re-run this with shorter periodicity? Perhaps five year and ten year stretches? Output probably won’t be as clear-cut as with the 20 year periods, but perhaps some time-varying trends will emerge.
Very interesting post!
It would be interesting to see if this cycle is present in the southern hemisphere markets (eg. Brazil, Argentina), and if there’s a phase shift from the north hemisphere to south hemisphere markets. If so, it would be possible to make a trading strategy rotating between both markets?
Or maybe, this phenomena is not present through all the US market, maybe there are some anti-cyclical sectors?
How have you “volatility adjusted” the returns? I’ve recreated the table based on absolute returns, and – while the seasonality is evident – it is not quite a clear cut as portrayed here.
Hello DT – I hadn’t looked at abs returns before, but you’re right, slightly less clear but same basic pattern.
To calculate vol adjusted returns I divided annualized returns over each period by the annualized standard deviation of returns.
I think this is always a better assessment of the “goodness” of a set of returns (than abs returns) b/c it helps to balance differences in vol between different instruments, market regimes, etc.
michael
Great post as always!
Years ago I made this graphic:
http://www.market-trends.net/wp-content/uploads/2011/01/DRV.png
It shows an average of each days relational value to the corresponding years high & low: 0% is equal to the lowest low and 100% is equal to the highest high for that year.
The graph shows some interesting seasonal effects – for example the very significant November-December rally, but also March to May rally before heading into the quiet summer months.
If you have time and interest I would love to see the same graph, but with all the data you have going back to the 30′ies :)
I have tested the “Sell in May and go away” rule for Equity Mutual fund flows for the last 10 years of weekly data and my results show exactly the same results…Developed market flows follow this trend more closely thantheir EM counterparts…