Most investors have heard of the “January Barometer,” first five days of January rule, and “If Santa Should Fail to Call” type systems. Motivated to identify how all these systems lined up in terms of predictability, it would be prudent to turn the computer loose on all time periods of the year, ranging anywhere from one week to one quarter, in search of the time frame that had the highest correlation coefficient to the following year’s market performance. While such a study reveals interesting findings, the best predictor of future performance appears to be based on the combined two-month period from Nov. 19 to Jan. 19. “Triggering a trade,” (below) shows the one year (Jan 19 – Jan 18) S&P performance following the Nov. 19 to Jan. 19 period broken down into three criteria: Above 3%, 0-3% and negative; the three different categories of the ‘Turn of Year’ barometer. 

There was only one losing year (1987) out of 30 after a 3% plus bullish Turn of Year setup. In defense of that case, the S&P actually was up 20% from Jan. 19 through mid-August before succumbing to the double digit interest rates that fall that led to Black Monday (Oct 19). Conversely, there have been eight occasions since 1950 when the post-Jan. 19 year finished with a double-digit loss, and six of those eight occurred after one of the negative Turn of Year periods. All three of the -20% post-Jan. 19 years followed a negative Turn of Year period. In 1982 the bearish Turn of Year signals were the opposite of 1987 for the bullish Turn of Year readings as the S&P sold off in predictable fashion through the first half of 1982 before setting a major bottom in August and turning around to give the system one of its few annual boo-boos.

‘Sell in May’ system

The S&P has returned an average of 8.5% minus dividends since 1950. The “Sell in May” crowd point out that 90% of those returns have come in the six months between November and April, and what little equity return you miss out on during your equity hiatus from May through October can be more than offset via the returns of six month interest-bearing securities. 

A study of the optimal periods to be in and out of the market since 1950, assuming only two trades (in and out), revealed that the only modestly negative period for equities is, on average, from Aug. 7 to Oct. 23. So if you assume your cash goes under the couch and your objective is to maximize gains with no aversion to the risk and the volatility of doing so, you would have been served best to be invested in equities the 9½ months from Oct. 23 to Aug. 7 and in cash the remaining 2½ months.  

However, if you assume you could get anywhere from a 3% to 10% alternative return on your non-equity investment, the best alternative followed the Sell in May crowd and suggested equities from Oct. 28 through April 20 (see “Two trades,” right). 

Intermediate seasonal model 

Now the goal is to combine both of these seasonal strategies into one seasonal model. My approach was to give each day from 1950 through June 12, 2012 a rating based on both the Turn of Year barometer and the Sell in May strategy using the following criteria: 

• If a day’s preceding Nov. 19 – Jan. 19 period returned more than 3%, then that day is assigned a +1.
• If a day’s preceding Nov. 19 – Jan. 19 period returned 0.0%-3.0%, then that day is assigned a 0.
• If a day’s preceding Nov. 19 – Jan. 19 period returned less than 0.0%, then that day is assigned a -1.

In addition:

• If a day falls between April 20 and Oct. 28, then that day is assigned a 0.
• If a day falls between Oct. 28 and April 20, then that day is assigned a +1.

The ratings for each day then are added to give an overall score between -1 and +2. For example, if the active Turn of Year time frame is negative and you are trading between April 20 and Oct. 28, that day is assigned a -1. If the Turn of Year time frame is greater than 3% and you are trading between Oct. 28 and April 20, each day will receive the maximum +2.  Below are the net results for the market’s performance based on each of the four possible ratings:

Wayne’s intermediate seasonal model

So from 1950 through June 18, 2012, if you had been 100% long during the 15.50 years where both the Turn of Year and Sell in May signals were positive, and then 100% short in the 6.85 years when the net rating was -1, you would have experienced an average annual return of 25.13% during those 22.35 years you were playing the market, which is almost exactly the same net result as the Buy and Hold return over those same 62 years, obtained with one-third of the market exposure. “The power of compounding” (below) shows the results of investing $100,000 in 1950 according to this game plan. 

If Santa is mad…

There is an interesting side study that reveals itself from this study. Recall our Turn of Year statistics suggest the prognosis was not good the following year when the Turn of Year periods were negative. Our -1 ratings appear during the Sell in May periods after those negative Turn of Years. “Get out or get short” (below) shows the performance during all of the -1 rating periods. 

If you simply wished to manage, say, a retirement account, with long-only exposure involved, you could take the following approach. 

This approach also returned almost the identical result as Buy and Hold, but achieved it with an S&P Beta of 0.60%, which gives you 40% of your capital to invest elsewhere. In addition, the largest drawdown was 24.26% as compared to the 52.58% drawdown incurred with Buy and Hold from May 19, 2008 through March 3, 2009. 

Trading does not need to be overly complicated. The above seasonal model could be expanded upon and optimized but is solid and only requires three trade dates: Jan. 19, April 20 and Oct. 28. Such simplicity not only frees up time but also reduces cost from errors, slippage and fees.