There's an old Wall Street adage....."As goes January, so goes the year." That's just a clever way of saying if the market gains ground in January, the year should be a positive one for stocks. Of course, if January ends in the red, it could spell trouble for the market. The rationale behind the idea has never really been fully explained, although there are plenty of theories revolving around investor mentality, institutional re-allocation, and the like. Frankly, I think it could be a combination of reasons - if the saying is indeed true. More than that, I don't entirely care if there's a good reason, bad reason, or no reason. My primary concern as an investor is whether or not there's any validity to the relationship. After all, my only real goal is to buy low and sell high.

That said, knowing how investors are likely to use the January indicator to make long-term decisions, today we're going to put the theory to the test. Is the January indicator for real, or is it just a hypothesis that has only become true in our heads thanks to lots of repetition?

Our research looks back on 37 years worth of January indicator data....all the way back to 1970. The methodology is simple - if the annual return and the January return were both positive or both negative (to any degree), then the January-based hypothesis was correct. If instead the returns for one time frame were in conflict with the other time frame, then the premise failed. 'Scoring' the test was as simple as it seems.

Over the last 37 years, the January indicator predicted 23 bullish years. It was right 21 times, and wrong only twice. The success rate of 91.3% is more than a little impressive. But how did it do when it came to predicting bearish years? On 14 occasions a bearish year was predicted by a losing January. Eight of those years were indeed bearish, while 6 of those years ended up being positive anyway. So, the model's bearish prediction accuracy was only 57.7%. Still though, the overall success rate of 78.3% should easily indicate yes, there is indeed something to this hypothesis.

But, you know us.....we can never leave well enough alone.

A true test of this January indicator's significance can only be complete if there is a control group, or a meaningful comparison. To establish such a control, we randomly selected four more months of the year to perform the exact same test on. That is, we're looking for correlations between one particular calendar month and the same particular calendar year. The other months we chose were March, May, June, and August. (We didn't want to choose any month after August, because the year is close to being over by that point....which would make using the tool a little pointless).

The results? The March indicator correctly forecasted only five of thirteen bearish years, but did foretell nineteen of twenty-four bullish years. May was impressive too, accurately spotting seven of its sixteen bearish expectations, and eighteen out of twenty-one bullish years. June was an outstanding barometer as well, predicting eight out of fifteen bearish years, and twenty out of twenty-two bullish years. Finally, August was wrong ten times about its sixteen bearish forecasts, but was right seventeen out of twenty-one times when it came to predicting bullish years. Respectively, the accuracy (both bearish and bullish) of these other four months was 64.8%, 67.5%, 75.6%, and 62.1%. For our four non-January months, that's an overall prediction accuracy of 67.5%. And remember, our total January-based prediction accuracy was 78.3%.

Is the difference between the January results and all the others significant enough to say the January barometer is statistically meaningful? A mathematician might say yes, but frankly, I don't think it is....not when it comes to making investment decisions. When it comes to my dollars, I want a serious edge - an edge that's at least twice as likely as a coin toss to yield the expected results. Based on the math above, I just don't see any strong benefit in buying into the theory. That's not to say it would be wrong for others to do it - I'm just saying I prefer a more meaningful tendency.

Granted, the difference between 78.3% accuracy and only 67.5% accuracy isn't chump change - I understand that. I might be a little more open-minded if it weren't for one thing gnawing at me that shrinks the size of the disparity (for me anyway). That is, the market is supposed to go up. The bullish January predictions were slightly more successful than the bullish predictions for all the other months...although only slightly better. But see, that's still not impressive, since things are supposed to be bullish given enough time. The small margin of improvement just using January as your guide could be easily dismissed as coincidental. Point being, it's not a predictive measure...that's just the norm. In other words, it's not a cause-effect relationship - that's just ongoing market expansion. You'll notice the bearish predictions of all the January's really were no better than the bearish predictions of the other four months. So, I have to question whether or not there's any real forecasting being provided by the theory.

Just some food for thought before we start getting too far into the month.