First, we'll discuss the first quarter (January through March), and then we'll look at January-specific trends.

The first quarter has averaged about a 3.5% gain since 1994.  Our list of winning indicators is topped by cheap stocks, beating out the next best indicators by a very hefty margin.  Depending on exactly how you played things, you could've averaged better than 20% per first quarter since 1994 with this strategy.

Going back to 1985, the average gains in the period are around 5%.

The trend toward gains via cheap stocks also has the virtue of being quite consistent.  As a side experiment, we calculated the % gain you would have gotten by simply going with the cheapest 4% of stocks in every first quarter since 1994, sitting out the other 3 quarters, and reinvesting your earnings in successive first quarters.  The answer:  1050% in 10 quarters of investing from 1994 through 2003!

Further down the list of winning indicators, you see volatile stocks, small-caps, and stocks with nice gains on the last day of the year.  Prior to 2005, you'd see the traditional January indicator of yearlong losers on the list...but that group performed horridly in 2005, wiping it off the list.  In fact, one strategy that has worked nicely for shorts has been to short stocks with nice gains in the fourth quarter, but poor overall performance in the previous year.

For shorts, look for expensive stocks.  Non-volatile stocks have been weak.  Food processing industries have been weak.  Of course, the period in general has been quite positive, making shorting a difficult endeavor.

Looking below to January-specific trends shows you that a very large chunk of gains in cheap stocks occurs in the first few weeks of the quarter, meaning that you might actually be able to improve on the aforementioned 20%!  In fact, the trend toward gains via cheap stocks or beaten-up stocks has a way of reversing in February.  To be perfectly clear:  unless you're inclined to minimize the turnover in your portfolio (something we wouldn't argue with...there's nothing wrong with trading some potential gains for sanity), "playing the quarter" is probably a sub-optimal strategy compared to "taking your January effect gains and moving on".

Those most interested in avoiding portfolio churn should note that stocks of low marketcap perform well, a trend that is also evident in the second quarter.

As for industries, semiconductors have fared well, while utilities and transportation-related stocks perform poorly.  Again though, you see these same groups in our January-specific tables, so they may have spent most or all of their gains by the end of January.

In tests where we looked at the performance of stocks in the previous 3 first quarter periods, we found no correlation between strong past performance and future first quarter performance.  For intrepid followers of such a version of seasonality, however, we offer the following table of stocks that have performed particularly well in the quarter:

In terms of "industry momentum", stocks whose industries groups finished strongly on the last day of the year have a tendency to fare well in the quarter.  This is somewhat at odds with the fact that stocks that lose in a large way in the closing days of December also tend to prosper...but so be it.

Another test we performed looked at a selection of large caps, going all the way back to 1985.  Here, there was a tendency for stocks that performed well in this time slot to repeat their performances.  Volatile stocks, those with poor performances the week before the first quarter, and stocks with nice, if not spectacular, yearlong gains (!) have fared well.  Underperformers are dominated by expensive stocks and non-volatile stocks.

As for the month of January...

We'll cut right to the chase:  yes, you can make a lot of money by buying stocks that were beat up in December or over the course of the year, but there are a couple strategies that work quite a bit better for gains in January.

The first is to simply buy cheap stocks.  We've filtered out all stocks that made extreme single-day gains in the month on the off chance that we might have missed a reverse split or two, but we still get the same eye-popping results:  in a month that has averaged less than a 2% gain since 1994, choosing the 4% cheapest stocks would've given you an average 14%+ gain.  The statistical significance here is off the charts.  Virtually all the gains occur in the first 10 days of the month, making the results all the more eye-popping.

The top 4% of stocks in our "prcvd_risk" grouping have also been quite strong.  Here, we divide volatility by price, finding stocks that are simultaneously cheap and volatile.

As we hinted in our December summary, don't try to get in on the January effect prematurely.  Buying cheap stocks in late December is generally a losing strategy, though our data shows you could probably pull it off on the last day of December and maybe add a couple percentage points to your January gains.  Even buying these cheap stocks on the second to last day of the year could be a mistake.

The first trading day of the year begins the January effect in style, however.  Despite a slight tendency toward losses in the general market, stocks with recent losses or cheap prices post gains with very high statistical significance.  Interestingly, the second trading day of the year has been the most positive day of the year going back to 1950, averaging about .5% gains.  The next best days of the year average only about .3% gains.

Another supposed January effect angle, that of focusing on stocks of low capitalization, doesn't find itself in our tables of top strategies.  This accords with a number of academic studies that haven't found much correlation between capitalization and January performance, at least in recent years.  Bear in mind that we don't use actual capitalization data in our tables, just an oblique estimate (recent average volume X recent price).

Further down the significance ladder you see the classic January effect indicators...big yearlong losses, a low 100 day moving average, late December losses, etc.  Volatile stocks also perform well, but they don't make our "risk-adjusted" tables.  Again, the aforementioned gains occur almost entirely in the first half of the month.  It should be noted that seeking gains via stocks with high losses over the course of the previous year isn't a strategy that's immune to reversal...in 2002, for example, yearlong losers continued to underperform the general market.

On the short side, our list is topped by oil stocks, which have actually averaged a 1% loss per January.  The losses are strongest in the second half of the month. Utilities also perform poorly.  Historically, shorts could do well simply by reversing the above long indicators:  sell expensive stocks, and stocks with big yearlong or December gains.  None of these short indicators are nearly as significant as the long indicators, though...i.e. they're more likely to vary in any given year.

The following table lists stocks that have performed well or poorly in January on a historic basis: 

Dividing the month into halves, the first half accounts for most of the gains, averaging a tad better than 1% in a two week span.  Long strategies for the second half are basically the same as those for the full month...buy cheap or beaten-up stocks, but you've got to reduce your expectations of massive profits at this point.

If you'd like to combine two indicators for second half gains, we've got a schizophrenic array of them in our data tables.  The best in the list is gotten by combining computer stocks and a 20 day moving average well below the 100 day moving average.  The strategy has averaged better than 10% gains in the second half of the month over the last 10 years and is fairly significant from a statistical perspective.  A bit risky, though.

On the short side, combining yearlong losses with 3 month gains seems to work decently.  Again, it's risky.  

Retail stocks, particularly apparel, tend to suffer in the first half of the month, continuing their December woes.  Regional banks and insurance stocks also fare poorly.  Broadcasting and communications stocks tend to fare poorly in the second half, reversing December's gains.  Biotechs outperform semiconductors in the second half.

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

Most of our data tables go back no earlier than 1993.  However, we do have access to data that goes back quite a bit earlier.  We looked at a sampling of the S&P 500 going back to 1985, and found the same sorts of trends as mentioned above...rather surprising, since we were looking only at large-cap stocks.  Stocks that had split in the year prior to the January measurement were tossed.

As usual, cheap stocks were strong on a risk-adjusted and non-risk-adjusted basis...the cheapest 4% averaged a 10% gain over the month.  Volatile stocks were strong, even after risk-adjustment...not surprisingly, computer stocks led our industry groups.  Late December losers were strong.  Yearlong gainers outperformed the general market, but weren't prominent in our tables.

On the negative side, retails were weak, actually losing a tad in a month that averaged about 2% gains over the course of the experiment.

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

The problem with the "January effect" is that it has numerous definitions.  For some, it's a general tendency for the broad market to gain in January.  For others, it's a tendency (one we'd dispute, actually) for small-caps to gain in particular.  For still others, it refers to a tendency of yearlong losers to rebound in January.  A final "January effect" is the tendency for the market to continue to gain in years with positive Januaries...again, we dispute the significance of this "fact" (see below).

Regarding the actual causes of the rebound in losing stocks, most academicians cite tax-loss selling and the tendency of money managers to perform "window-dressing" (a.k.a. "gamemanship") at the end of quarters and years:  the argument seems to be over which cause is stronger. The internet is rife with arguments both ways.  Some have suggested that low holiday trading volume, combined with the above two causes, may further amplify gains...an intriguing notion.

Below is a table that details the performance of yearlong losers and cheap stocks in every January since 1986.

Yearlong gainers and stock prices were divided into deciles...the second and third columns show results for the stocks with the lowest 10% of yearlong gains and stock prices.  Results would certainly be even more dramatic if, say, we looked at the lowest 4% of yearlong gainers and stock prices.

Given the tax-loss thesis, one would expect that January effects would be particularly strong when coming off years where stocks in general suffered.  Obviously, our strongest "January effect" occurred in January of 2001.  This was coming off a 2000 in which the majority (56%) of stocks actually gained.  Note also that although yearlong losers outnumbered gainers by a 2:1 ratio in 2002, January 2003 didn't show a strong January effect. 

The window-dressing hypothesis predicts that the riskiest stocks would be dumped at the ends of accounting periods (e.g. the end of the year) and repurchased at the beginnings.  One would then expect that purchasing the most volatile stocks at the beginning of January would be a particularly strong strategy.  It is a good strategy...but again, not the best.  Note, however, the cheap stocks have a reputation of being "risky", whether or not the actual standard deviations indicate riskiness.

Whatever happens in January, it doesn't happen quite as mechanically as some might suggest.  The fact that cheap stocks as a group outperform yearlong losers suggests some other factors besides December tax-loss selling and window dressing.  Perhaps the difficulty of shorting stocks under $5.00 has some role in the effect...it's more difficult to defer tax-losses via "shorting against the box" with cheap stocks, meaning that these stocks could be relatively immune to some of the selling pressures that more expensive stocks would experience.  

Perhaps there are some simple post-Christmas psychological factors in play...people go for bargains following their Christmas splurges.  Perhaps people are simply feeling happy and are a bit more likely to indulge in risky behavior.  In fact, there's some evidence that analysts are more likely to issue positive reports on companies around the new year.

We should point out that in January of 1994, and 1996-1998, a good strategy for January gains was to buy stocks that lost in the closing days of December.  These Januaries followed years in which the general market performed strongly.

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

Looking at the table above, 2005 stands out as a dramatic exception to the usual patterns.  In fact, the patterns reversed...stocks with large yearlong gains were quite strong compared to the market, while semiconductors, yearlong losers, cheap stocks, and volatile stocks topped our list of losing groups.

Why the 2005 reversal?  The simple answer would be that the "January effect" is bunk to begin with, so one shouldn't be surprised if an upcoming January reverses some of the supposed trends of the month.  We find this explanation difficult to accept, given the statistical significance of some of these effects in our own tests.  Excluding that, one's got to wonder why the usual "tax loss selling" would actually reverse.

One explanation might be that a large number of yearlong losers and cheap stocks were semiconductors and tech stocks, which had longer term (over 1 year) gains.  In such a case, it wouldn't make sense to sell these stocks in December for the sake of tax writeoffs...better to sell them in the new year.

Those interested in further analysis of this strange January might want to note the large difference in the "% of stocks that lost money in the previous year" column in 2003 versus 2004.  There, you go from 67%, the highest figure in the table, to 9%, the low in the table.

Further comments on January 2005 can be found here. 

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

There's yet another "January Effect" of which we've heard mention.  It's the "reverse January effect".  The idea here is that the previous year's winners have a tendency to get burned in early January, but will bounce back thereafter.  We haven't investigated this phenomenon, but even those who promote this strategy don't argue for massive gains...something along the lines of 7% average gains in the six months following mid-January.

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

Regarding the "January effect" that forecasts continued yearlong gains following a positive January, we're less than impressed.

A rather eye-catching statistic we've seen tossed out by the pundits is the fact that (as of Jan 2004), the market has risen 92% of the time in the year of a positive January.  Problem is, the year that is being counted includes January results.  To take this sort of logic to an extreme, it's easy to see the fallacy of saying something like "in 99% of the years in which the January-November period is positive, the year itself is positive".

Now, what happens in the 11 months following a positive January?  We count 36 such Januaries since 1950, and in 29 of those 36 years, the market has gone up (80%).  But the market goes up 69% (37 of 54) of the time anyway.  With a little concerted data snooping, we could probably find far better predictors of a positive upcoming year...and they'd be just as worthless as this silly version of the "January effect".  Note:  actually, we have found a few such indicators...check it out.

Here's another chunk of so-called wisdom:  Since 1937 the January indicator [of January gains/losses predicting yearlong gains/losses] has a perfect record predicting market direction in odd-numbered years.  That observation was made at the end of January 2002 in a prominent newsletter.  The rest is history...from now on the above quote will have to include an "except", given 2003's losing January and strong yearlong performance.

Copyright © 2008 MarketSynopsis.com. All rights reserved.