Mark Hulbert asks an interesting question: What does a strong first quarter imply for the rest of the year?
On the one hand, we know that trend and persistency are definitely part of the market’s chaotic behavior; On the other hand, reversion to the mean is a powerful force whenever markets deviate too far in either direction from their historical trend.
To investigate the Q1 phenomenon, Hulbert took Dow Jones Industrial Average data back to 1897, looking for a pattern. What did he find?
"Bad news. Over the nearly 110 years since , I could detect no relationship between the market’s strength or weakness in the first three months of the year and its behavior from April through the end of the year.
Here’s one particularly revealing statistic: Consider those 50% of the years since the late 1800s in which the DJIA performed the best during the first quarter. Over the final three quarters of those years, the DJIA gained 5.2%, on average.
In contrast, the DJIA gained an average of 6.6% in the final three quarters of the other 50% of the years – the years in which the DJIA was the weakest during the first three months of the year. To be sure, this difference between 5.2% and 6.6% is not statistically significant. Nevertheless, note carefully that the DJIA has performed better following weaker first quarters than following stronger ones.
This doesn’t mean that the stock market will not do well for the rest of 2006. But if it does, it won’t be because the stock market performed so admirably during the first quarter
While that’s fairly interesting, I would quibble about merely breaking the data into two halfs. Analyzing this in terms of deciles (10 pieces) might be more revealing.
The two primary forces discussed above — trend, and reversion to the mean — might not be revealed in the two data groups. Instead, I would imagine the top and bottom 10% might be more revealing of mean reversion, while the middle group might show trending patterns.
UPDATE: April 5, 2006 :10:23am
I emailed Mark Hulbert, asking him about the decile issue versus 50% approach; Mark responded:
"I conducted a number of statistical tests in background in preparation for the column, all of which led to the same conclusion.
In a simple regression, the r-squared is less than 1% and the coefficient on the 1st quarter’s performance, while statistically insignificant, is nonetheless negative—just the opposite of those who are betting on the momentum story…"
That suggest that mean reversion is the dominant force at work.
As goes the quarter, so goes the year?
MarketWatch, 12:01 AM ET Apr 5, 2006
Can you take this one step further for us?
” The two primary forces discussed above — trend, and reversion to the mean — might not be revealed in the two data groups. Instead, I would imagine the top and bottom 10% might be more revealing of mean reversion, while the middle group might show trending patterns.”
Would love to see his data/study have a valuation overlay using the concept of price to peak earnings. I would guess that doing so would generate very negative average results, though the sample size will be very small.
Isn’t this data also indicative of the seasonal tendency of Q1 & Q4, in that order, to be the best performers for the market overall?
If you want to see mean reversion, check out the Russell 2000 over the past 15+ years or so. The bigger the first quarter, usually the worse the 2nd and 3rd quarters. And vice versa. Of course, it all comes out in the wash during the usual 4Q jam job. But the worst 2nd/3rdQ drawdowns followed the best 1Qs.
The Russell had its 2nd biggest 1Q in the past 15 years this year. No guarantees—but history is not on its side in the 2Q and 3Q.