Feb 13, 2004...We've seen some rather silly statistics regarding the probabilities of positive and negative years.  For example, if a calendar year is odd and its January is positive, the market has made a net yearlong gain in every year since 1937, with a 2003 exception.

Presumably, most of these "discoveries" are made by folks who actually believe they've stumbled onto something of importance.  In the spirit of fun, we thought we'd use our software to dig up some of our own "rules".  Just to be clear, we place no little or no faith in the predictive power of these rules.  The main point is to illustrate how a bit of "data snooping" can result in rules that seem important.  All the rules below involve Dow data from 1930 through 2003.  Bear in mind that the Dow has gone up 50 out of 74  (68%) of these years anyway.

• In every year where dividing the year by 8 leaves a remainder of 7, the market has gone up (1999, 1991, 1983, etc).

• In every year where dividing by 9 leaves a remainder of 8, the market has gone up (1997, 1988, 1979, etc).

• In every year where dividing by 3 leaves a remainder of 2 AND dividing by 7 leaves a remainder of 2,3, or 4, the market has gone up (1997, 1991, 1985, 1976, 1970, 1964, 1955, 1949, 1943, 1934)

• Here's a REALLY silly rule:  the Dow has finished up in every year where it finished between 2950 and 7550.

• In 31 of 39 years in which May has been a positive month, the market has gained (this actually beats the esteemed January indicator which has predicted yearlong gains in 30 out of 41 years since 1930). 

• 29 of 38 years in which a Democrat has been President were positive years.

• The first two years of a Republican presidency were negative in 9 out of 16 such years.

If we were to actually give credence to any particular rule, it would be this:  years prior to election years (i.e. dividing the year by 4 leaves a remainder of 3) have gone up 17 out of 19 times.  We won't claim to understand why (the ready-made explanation is that the president is working extra hard to improve the economy, with next year's election looming), but at least there's some potential psychological/political rationale that goes beyond mere numerology.

Further election cycle observations (made by others) are the facts that the pre-election year gains tend to occur in the first half of the year, and election year gains tend to occur in the second half of the year.

*****************

April 10, 2004:  Regarding our May indicator, it qualifies as quite "silly" because, of course, a good portion of the year has already passed by by the time May rolls around.  

If you look at the 11 months that follow May, as opposed to the calendar year it is within, May serves as a very average predictor of future gains.  Looking at the market this way, in fact, January is certainly the best month-wise predictor of gains in the next 11 months.  

But we're still far from impressed with this version of the "January effect".  The end of the calendar year serves as a natural point at which market reversals can occur.  Thus we have the well known phenomenon whereby yearlong losers tend to fare well in the first few weeks of the new year (which also goes by the name "January effect"...don't be confused).  But, of course, the 11 months after January don't contain such a reversal point.  And, of course, the 11 months following all the other months do contain this reversal point.

We haven't tested it, but we'd guess that if you were to look at the 12 months after a positive January, or the 2 or 3 years after a positive January, this version of the "January effect" would be rendered sterile.

Copyright © 2008 MarketSynopsis.com. All rights reserved.