Old Investment Saw
I have often heard the investment saying "Sell in May and go away." However, being a committed long-term investor, it was an anathema for me to think about pulling funds out of the market for extended stretches of time. After all, doesn't the market go up 2/3rds of the time? (Which is yet another common investment saying.)
On the other side of the coin, I remember all too well a number of horrific periods during the summer months where I was wondering what I was doing in the markets. At times, it seemed the market was capable of only one behavior and that was to go down.
So with that in mind, I thought I would look at some recent history to understand whether there is any truth to "Sell in May." Or whether the approach of not making any seasonal adjustments is a more rational choice. I also wondered whether sector rotation, also something I have not practiced on a seasonal basis, made any sense.
For this study, I wanted to focus on a strategy that if found, would be easy for investors to implement. I selected the ETF SPY to represent the S&P 500. For the sectors, I used the popular State Street Global Advisors ETFs, the ones whose symbol begins with "XL." For example, XLE for Energy and XLI for Industrials. Given that sector ETFs are a relatively recent phenomenon, this means looking at the last 14 years of data, a sample size, though not ideal, is large enough to at least give an indication as to whether there is a workable investment strategy that can improve returns and mitigate risks.
The ETFs used in the study are listed in the following table.
The Data - November through April
Let's begin by jumping right into the actual data. Below are the 14 year returns, year by year for SPY and each of the sector ETFs. The first table is the period November 1st to April 30th. The returns are inclusive of dividends.
The Data - May through October
The next table is the period May 1st to October 31st, again inclusive of dividends.
The Data - Comparing May to October against November to April
Finally, let's look at the average return in the May-October period vs. the November-April period in summary from across all 14 years. We will examine average, median, minimum and maximum returns respectively for the S&P 500 and for each sector.
The last set of rows in the table shows the relative performance of each sector vs. the S&P 500 in each of the two seasonal periods.
What does the Data Say?
That's a lot of data. Let's try to distill it into some investable take-aways.
First off, there is some truth to the "Sell in May" notion. There is no getting around the fact that the returns in the November through April time frame trounce the returns in the May through October timeframe. Using SPY as a proxy for the S&P 500, the average performance was 3.4% better in November-April vs. May-October. That is a significant difference.
More importantly, the worst November-April performance was -5.4%, whereas the worst May-October performance was -35.1%. That is a staggering difference in decline. If you pick the wrong year to be fully invested in November-April, you get nicked, in May-October, you get obliterated.
How about when things go well? The best November-April was 15.0% and the best May-October was 20.1%. So at least the best year for May-October held slightly more upside than November-April. However the volatility is much greater. The average price change (up or down) in the May-October period is 10.7% vs. only 5.9% for November-April.
Takeaway #1: November-April is a much better time to have more exposure to the market than May-October.
Does May-October Make Sense?
Aside from the obvious performance disadvantage May-October has relative to November-April, does it even make sense to be in the market in May-October? Again let's look at the data.
Of the 14 periods, eight of them (57%) generated positive returns and six periods (43%) generated negative returns. The overall average return in the period was positive by 0.5%. Not exactly stellar returns considering the risk of drawdown. For SPY, the three worst periods over the last 14 years were -35.1%, -11.6% and -8.8% (average 18.5% down). For comparison, the three best periods were 20.1%, 11.2% and 10.6% (average of 14.0% up). In short, the upside rewards in good years are good, but they are eclipsed by the downside risk in bad years.
Takeaway #2: May-October returns overall are slightly positive, but given the risk of investing in that period, it may be prudent not to maximize exposure to the market during this time frame.
Can Sector Rotation Help?
We have been focusing on the S&P 500 via SPY, but how about some wise sector rotations to help mitigate risk and improve return. Let's take a look.
The first big worry is the really bad periods which result in significant capital destruction, and from which it can take a long time to recover. So with that in mind, let's find our safest sectors.
Well, perhaps unsurprisingly, they would be Staples (worst case -18.8%) , followed by Health (-21.6%) and Utilities (-26.9%). None great, but all far better than the worst case -35.1% performance of the S&P 500. These sectors seem like they might help mitigate some risk against profound capital loss.
Exploring further, let's look at the worst three periods for these three sectors relative to the S&P 500:
Here, unfortunately it gets murkier. While Staples continues to look good, and Healthcare is OK, Utilities had two pretty bad years, and the third one wasn't so great either.
Takeaway #3: Except for Staples, sector rotation may be of somewhat limited value in mitigating the risk of loss relative to the S&P 500.
S&P 500 Sector Weightings
Let's now consider the attributes and limitations of each sector when evaluating a sector rotation strategy for May-October. We need to think in terms of overweighting, equal weighting or underweighting specific sectors relative to the S&P 500 to improve return and reduce capital loss risk.
To know whether to underweight or overweight a sector, it is helpful to know the weightings of the benchmark itself. The following table shows the weightings of each sector within the S&P 500. This can act as a reference from which to plan overweighting or underweighting of individual sectors. Note that Technology includes Telecoms. The weightings are as of May 3rd, 2013.
Sector Rundown - Overweight
A couple of interesting facts on the Staples sector. First, during the S&P 500 disastrous decline of -35.1% (2008), Staples only declined -14.6%. Also Staples have been up 10 out of 14 periods in May-November (71%) vs. only eight out of 14 for the S&P 500 (57%). Finally the average gain in the period is 2.9% vs. 0.5% for the S&P 500. More gain and less risk. What's not to like?
Utilities is a mixed bag. Its max decline is significantly smaller than the S&P 500, but the second and third biggest declines are significantly larger. Not exactly what we are looking for to mitigate risk. However if you can stand the risk, Utilities were up in 10 out of 14 periods and booked an average gain of 2.1% or 1.6% better than the S&P 500.
Tech averaged the best average performance in the May-November period, tying with Staples at 2.9%. And its best period (41.3% in 1999) was far better than any other best period of any other sector. Of course, that was the internet bubble era. However, it shows when bubbles form, the upside is huge. At least until things pop. Tech has been up in 9 of the 14 periods (64%), which is the third best sector score.
Sector Rundown - Equal Weight
Discretionary runs like the S&P 500, up the same 8 of the 14 periods. And average gain of 0.4% vs. 0.5% for the S&P. The worst period is also similar at -36.1% vs. -35.1%. Clearly an equal weight sector.
Energy matches the S&P 500, up the same eight of the 14 periods. And it has an average gain of 0.9% vs. 0.5% for the S&P. However, the worst period (2008) was down -41.2% vs. -35.1%. That is the third worst sector decline, exceeded only by Financials and Materials. On the other hand, the best period of 27.1% in 2004, is the second best sector gain in the period behind Technology. So a higher risk/reward sector that again probably deserves equal weight at best.
Industrials also performs somewhat similar to the S&P 500, up in nine of the 14 periods with the same average gain of 0.5% as the S&P. The worst period was slightly worse at -39.9% vs. -35.1%, but the best period slightly better at 23.8% vs. 20.1% for the S&P. Another equal weight sector.
Materials has offered an average return of 1.5% and outperformance vs. the S&P 500 of 1.0%. Like the S&P 500, it was up in eight of the 14 periods. However, the worst decline of -47.4% in 2008 was brutal. It is difficult to recommend overweighting such a sector with a year like that, even if the other years have been far better. Equal weight, but tread carefully.
Sector Rundown - Underweight
Financials have been a real underperformer in the May-November period. While there is still a 50/50 chance of a gain, the losses at -47.9% (worst performance of all sectors) more than trump the best gain at 20.9%, which basically matches the S&P. The average gain is negative at -1.9%, again worst of all sectors. More risk with less gain is not exactly a winning formula. Best to reduce/avoid this sector in May-October.
How about Health? Not nearly as good as Staples. In fact, the average return in the period is actually negative 0.7% and there is only a 50/50 chance of gains. The only thing Health brings to the table is a decline that is not quite as bad as in the worst year. So it turns out not to be a sector to overweight, but rather to actually underweight.
The last thing I want to look at is if there is any clear pattern between the performance of the preceding November-April period and the performance of the subsequent May-October period. Maybe we can get an edge by finding a pattern predicated on the preceding six months of performance. The following table illustrates performance momentum data for the S&P 500 via the SPY ETF.
Does anything stand out? Well, first off the S&P 500 only went down three times in the last 14 years between November and April. Interestingly, for those three down periods, in the following May-October period, the S&P 500 also went down in all of them. So a clear warning here. If November-April is not good, the negative momentum will very likely continue through the following May-October.
In the 10 periods the S&P 500 went up from November-April, it went up in the following May-October period 7 out of 10 times. The 11th will be decided six months into the future. However, the fact that the preceding 6 months have been positive is also a positive sign for the next six months.
Yet on the pessimist side of the ledger, there is another fact that cannot be ignored. Historically, the two biggest gains during November-April period resulted in negative returns in the following May-November period. And the period we are in now, is actually the new second biggest gainer. So that is a pretty big counter trend for the normally bullish scenario of May-October being positive when the preceding November-April is positive. Clearly a big caution sign flashing here.
Summary - How to Invest the Seasons
So how do we sum this all up? Based on the last 14 years of data, the moves an investor should consider for the May to October timeframes are:
Reduce market exposure relative to the exposure maintained in November through April
Overweight Staples, and to a lesser degree, overweight Utilities
Overweight Technology, but jump ship first with this sector if things head south
Underweight Financials, and to a lesser degree, underweight Health
Based on historical behavior, these adjustments should improve the likelihood of better return during the market's fallow period, while also reducing risk.