Sell In May: A Strategy Worth 26% A Year Since 2001

Includes: QQQ, SPY, TLT
by: Fred Piard

Follow the Seasons

Most human activities have seasonal cycles. The stock market's seasonal cycle is not reliable every year, but it is powerful over the long term.
You probably have already heard of its four best-known effects:
1) "Sell in May and Go Away." This old saying tells that the odds are against you from May until September.
2) "Halloween Effect." The odds are favorable again in October and a rally may take place just before Halloween.
3) "Santa Claus Rally." December is usually a good month, with probably a rally before Christmas.
4) "January Effect." Big caps tend to fall in January, and sometimes in February.

In fact, if you plan to invest on indexes, doing so only in March, April, October, November and December, you double your gain and lower your risks.
Do you want numbers? The S&P500 index including dividends returned 4.43% per year on average from January 2002 to January 2012. But considering only these five months every year, the return rises to 9.56%. With money at risk less than half the time, the return is more than doubled.

The Only Thing I Remember about Macroeconomics

The only useful thing I can remember about macroeconomics is that when risky assets fall, safe assets rise. When stocks are down, bonds are up. Not always for the short term but it is a quite reliable medium-term trend, amplified by low interest rates and high stock volatility. Coming back to our seasonal strategy, what about investing in long-term T-bonds (or in the corresponding ETF) instead of keeping our capital in cash 7 months a year?

The result is that 5 months in SPY and 7 months in TLT every year average 15% a year from 2002 to 2012 .

The Nasdaq Effect

And what about replacing SPY by the Nasdaq 100 ETF QQQ in our seasonal strategy?
I summarize: January and February: TLT - March and April: QQQ - May to September: TLT - October to December: QQQ.
This means 2 orders 4 times a year. Less than 2 minutes per quarter.

My simulation gives an average return of 26% a year from March 2001 to March 2012. You can check the math, please let me know if I made a mistake. Not so bad for 8 minutes of work per year. Reinvesting the money plus earnings or minus losses for each trade would have multiplied capital by 12 in 11 years. This is better than gold. Past performance is no guarantee for the future but these are 11 years through all market conditions: crashes (2001, 2002, 2008), excellent years (2003, 2009), highly volatile years (2007, 2011) and average years. The only negative year is 2008, with a loss limited to -6.1% when the S&P500 dropped about -35%.

Here's a chart of the "QQQ-TLT" strategy return in % simulated from March 2001 to March 2012. In red is the strategy, in blue the S&P500.

About Drawdown

In 11 years the worst drawdown of the "QQQ-TLT" strategy lasted 18 months and has come down to -40%. It's not so bad considering that the S&P500 has experienced a drawdown of 5 years down to -60%. But 18 months are a long time. I am quite sure that most investors (maybe including myself) would have given up in early 2009, just before the strategy recovered spectacularly with 41.5% in 2009, 44.7% in 2010 and 48.9% in 2011 -simply its 3 best years in the decade.

The "QQQ-TLT" seasonal strategy has an outstanding track record but it is likely to know long and deep drawdowns. Even if you can stand it, please do not bet more than 12% of your account on that. Without going in details it has to do with the Kelly formula. You may believe or not in Kelly, however it's good insurance against enthusiasm.

Disclosure: I am long TLT.