Recap : A Hypothesis for ‘Sell in May’


“Sell in May and go away, come back on St. Ledger’s day” is an old adage on Wall Street. There have been a lot of statistical studies on it, (and I’ve done some myself) and to the best of my knowledge, it is statistically robust, generally speaking. In other words, there is a low chance that the effect is due solely to randomness. Much less studied is WHY it happens. While a complete study of market price action will always be somewhat incomplete, (i.e. when there is a transaction, is it driven by the buyer or the seller? What if one side is an HFT? Etc.) I have a few thoughts I’d like to share on why the seasonality occurs, when it may be less effective, and why it is likely to continue to happen in the future.

One popular hypothesis of why it works is simple mean reversion. A strong move through April certainly increases the chance of a pullback in May. The trouble with this hypothesis is that mean reversion effects tend to only last for a month or two. So it doesn’t explain why asset prices are flat for over 4 months. In addition, it doesn’t explain why asset prices tend to do better the other months of the year.

My take on it is that this effect is tied to seasonality in the Federal budget, which is itself tied to deadlines for taxes. Before we get into the details, recall that a worsening in the federal budget deficit is a net INCREASE in the system for credit, while an improving deficit is a decrease. Using Bloomberg data going back to 1968, we see a seasonal tendency for a worsening deficit in 1Q, but a sharp improvement in 2Q, which is probably due to the tax deadline, a weak tendency in 3Q, and another worsening in 4Q. Incidentally, the 1Q and 4Q periods have also been the best periods for risk assets, historically. Now, the magnitude of the effect is certainly not very large, but it is subject to a couple caveats. First, it is not necessarily the change that matters, but rather the rate of change. Second, the market cap of stocks as a percentage of GDP has historically been much smaller. (i.e. amounts measured in pct of GDP may have had a larger impact)


So if we make a leap and assume that this hypothesis is roughly accurate, what does that mean for asset prices going forward? Specifically, when is this pattern likely to be stronger or weaker? One thought of course is to adjust based on how quickly the deficit changes. However, that data is only available with a substantial lag. (the most recent print was from Dec) What may be more useful, and timely, is the level of cash balances held by asset managers going into 2Q. The rationale is that lower cash balances (as a percentage of assets) mean that market participants are more vulnerable to negative surprises, whether they come from momentum reversal or decreases in system credit. And the picture there seems to be better than usual, given survey data. Additional offsets may come from private credit growth. The amount is somewhat questionable this year, given the weak revenue trends the big banks just reported. But perhaps 2Q will be better. In aggregate, it seems reasonable to expect a more muted ‘sell in May’ effect this year.

Separately, Scott Skyrm posted a good piece on the potential impact of the 5% Dodd-Frank Leverage ratio on the repo market:

In order to calculate the impact of the 5% Dodd-Frank Leverage Ratio, it means that cost of $100 million net Repo assets becomes $5 million, and at 10%, the annual cost is $500,000. Then, $500,000 is the equivalent of 16.4 basis points on $300 million gross assets. Therefore, on Tuesday, April 8th, the cost of Repo increased by 6.4 basis points for banks doing business in the U.S. The new cost, under the 5% Leverage Ratio is 16.4 basis points.


  • BoE Minutes: While inflation was likely to pick up somewhat in the coming months, all members agreed that the probability that it would be above 2.5% in 18–24 months time remained less than 50%.
  • PMI:
  1. China HSBC Mfg PMI improved to 48.3 as exp vs 48 prev
  2. EU Mfg PMI improved to 53.3 vs 53 exp and prev. Germany was strong, while France was weak.
  3. EU Svcs PMI improved to 53.1 vs 52.5 exp and 52.2 prev. Again, Germany strong, France weak.
  4. US Markit Mfg PMI was stable at 55.4 vs 56 exp and 55.5 prev

Australia CPI increased to 2.9% vs 3.2% exp and 2.7% prev. The Trimmed Mean measure was stable at 2.6% vs 2.9% exp

New Home Sales dropped sharply by 14.5% MoM. Vs 2.3% exp. An improvement in the Northeast was swamped by weak prints elsewhere.

Canada Retail Sales growth slowed to 0.5% MoM as exp

Upcoming Data:

  • Thu: Germany IFO, Draghi Speaks, US Durable Goods Orders, Jobless Claims, Japan CPI
  • Fri: UK Retail Sales, US Markit PMI
  • Mon: US Pending Home Sales

2 thoughts on “Recap : A Hypothesis for ‘Sell in May’

Comments are closed.