'As January Goes, So Goes the Year' - True or False?

The above quote is one of the most cited seasonal trends in the market, but does it work? Going back to 1962 (we realize we could have easily gone farther back, but 45 years is an adequate sample), we compared the S&P 500's performance in January with its performance during the rest of the year and found that 71% of the time, the market follows the same path in the February through December period as it does in January.

When we broke out up years versus down years however, we found that the indicator is much more reliable predicting market gains following a positive January (86%) than it is in predicting losses following a down January (47%).

All this means is that if the market has a positive January, bulls will cite the reliability of this indicator, and if January is down, the bulls will cite its unreliability in predicting declines.

Comments

[Eric Boughton_:]

So much confusion of correlation with causation in the investment business! People need to learn statistics. The January "directional indicator" is an obvious statistical consequence of two facts: 1) January is up more often than other months (65% of the time vs. 58% for a random month); 2) The market is up in most 11 month periods. The correlation of these two variables, even if both were random, would result in the appearance of predictive value.

Also, I updated your stats back to 1928 and found that January offers a correct prediction of the following eleven months only 67% of the time (down from 71%) for the larger sample size. Over this larger sample, an up January is followed by an up eleven months 78% of the time (down from 86%).

Further arcanities: an up May (which only happens 56% of the time) is followed by an up June through April 77% of the time, almost as good as January's "power".

(All statistics use S&P 500 Index data without dividends.)

2007 Jan 05 07:36 PM

[Vladimir Se...:]

Eric, Thanks for pointing out facts 1 and 2 :)

Statistics is just wonderful! With as much data as history of the market provides all you need is to select time frames and filter the data to your liking and you can make any angle or point of view look really good and logical.

Every investor on the planet is supposed to know that past performance is no guarantee of future results, yet statistics is being applied over and over in trying to predict the future.
If anything was predictable in such a way, it would obviously be already priced into the market rendering any such prediction useless. People who believe otherwise must really believe that they are the only ones who can do the math in the entire market. Market is smarter than that i'm afraid.
I'm not saying market is 100% efficient at all times, "events" make it inefficient. By "event" i mean something that can not be effectively predicted. It is a very broad definition, any individual order is an event.
So place your bets on events that best fit your understanding of what the future is going to look like and leave statistics to "texas sharp shooters" and MPT followers of this world :)

2007 Jan 06 01:10 PM

[Malkiel:]

Norman Fosback took on this myth in his 1976 classic "Stock Market Logic"--see pages 151-154 in the last 1996 edition. His conclusion after an analysis similar to Eric's was that one could "flip a coin on New Year's Eve and you will have just as good a chance of being right about the February to December market trend as if you used the January barometer...and you will have your answer 31 days sooner".

2007 Jan 06 04:42 PM

[kennebunk2002:]

A favorite course in my college was 'How to Lie in Statisitcs. Nothing seems to have changed in 40 years. One might try another age old truism- check the first ten trading of of the new year. I seem to remember that this has some merit.

2007 Jan 06 05:07 PM