Apr 16, 2005

We've begun generating "stock summaries" that show the tendencies of single stocks (as opposed to large groups of them) over history.  You can read more about this new feature here.

When we generate these tables, there's a bit of a dilemma regarding stock splits.  Should we adjust our historical data for splits, or simply toss out data around the time of a stock split?  We've been going with the latter approach (we're forced to toss this data because otherwise, for example, a 2:1 split appears to be a 50% loss, and thus one might wrongly conclude that the stock in question has a tendency to decline in value when the price is high).  There are some other workarounds to this problem...we know.

Why not adjust for splits?  Well, one of our underlying "beliefs" on this site is that the price of a stock matters.  We're not entirely sure why price matters...the answer probably has more to do with human psychology than logic...but the data on this site confirms this belief again and again.  Some stocks may tend to gain or lose when they're particularly cheap or expensive.  Obviously, though, you'll eradicate any chances of detecting these tendencies if you naively adjust for splits.  Microsoft appears to be a penny stock if you look at split-adjusted data going back to the 1980's...but it has never traded anywhere near penny-stock levels.

This "non-adjustment" procedure is only used when we generate our "stock summaries".

March 3, 2004

One recurring theme on this site is the advantage of simply buying cheap stocks culled from the Russell 3000.  The effect occurs on a time scale of anywhere from one day to a full year.  It's often so strong that we wonder if maybe we've built in some kind of error or oversight in our algorithms.

We've written our programs so that stocks that trade below $.50 during the period of time we examine get removed from our data.  Stocks that begin above $.50 and go below this figure get removed, but then so do stocks that begin below this figure and go up.  We're not so naive to think that the two groups simply negate each other, but it's hard to believe that our cut-off price makes a tremendous difference.  Note also that it's often the case that the second and even third cheapest groups often find themselves in our winning groups tables.

We usually do adjust for splits.  Whereas a stock's "real" price may have been $100 last year, if it split 2:1 between now and then, we record the stock's price last year as $50.  In other words, whenever a stock splits, the entire data file associated with that stock gets adjusted, line by line.  Thus, a particularly successful stock that has undergone a series of splits over the last 10 or so years might have a very cheap stock price 10 years ago, at least as far as our data files are concerned.

In the future, we can do some experiments to determine what role, if any,  the above procedure has in biasing the data.  Our guess...not much.  Note that we're not predicting 10 year gains in our tables.  The maximal period we look at is 1 year.  One of the most significant trends we've identified is that for cheap stocks to prosper in the first couple weeks of January.  On several occasions your webmaster used real money to take advantage of the effect, with very nice success...no illusion there! 

Cheap stocks performed extremely well in 2003...I opened a data file generated at the beginning of 2003 (not in retrospect, where split-adjustments might be applied) and looked at the cheapest 100 stocks (out of more than 2000) individually to see if the 200% gains connected to this group were somehow "artificial" (e.g. we missed a few reverse splits).  But the gains were absolutely real.

Nevertheless, it would be worthwhile to examine stocks that we list as "cheap" as a result of splits versus stocks that are "naturally" cheap.  In fact, there are a number of studies that show that stocks that have split do indeed outperform the market in general.  One showed that stocks that have split have historically outperformed by an average of 5% per year...quite significant when most indices have gained an average of about 11%/year over their lifetimes.  The effect can be raised to 11% if the split is accompanied by an upward earnings revision.

Another study showed average gains of 3% in the 5 days following a split.  The same study showed gains above the market averages of 8% in the first year, 9% in the second, and 12% in the third.

Another effect of particular relevance to option traders is the fact that volatility goes up considerably (around 35%) following a split.  In theory, it shouldn't.  In fact, the increase in volatility is maintained over the year following the split, with little subsidence.

Naturally, there's a lot of debate as to why the "post-split" effect is so strong.  The best explanation, as far as I'm concerned, is simply that good, solid, growing companies are in a position to split.

Note that if the positive effects of splits were largely responsible for the gains in cheaply priced stocks, the effect would diminish from, say, 10 years ago to the present.  Because of its many splits, Microsoft would be quite a cheap stock 10 years ago (by virtue of data adjustments) and so would be considered a "buy" at that time.  But no company splits from $100 to under $5, so stocks that traded cheaply in more recent times (say, the beginning of 2003) won't fall into our cheapest decile or 4% by virtue of stock splits.  Yet cheap stocks have shown a stronger-than-ever propensity to gain.  Again, witness 2003's tripling of value in the group...a gain that we examined with an archeologist's brush.

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

Still more on cheap stocks...check out the following link:  http://moneycentral.msn.com/articles/invest/extra/6266.asp .  The idea, in case the link goes belly-up at some point in the future, is that the authors examined 491 stocks trading under $5 and compared them to the broader market.  The cheap stocks outperformed by an average of 10%.  More radically, though, the authors proceeded to group the cheap stocks according to ValueLine's rating system (a rank of "1" is best, "5" worst).  ValueLine didn't give a "1" rating to a single sub-$5 stock in this period, but the 30 stocks ranked "2" averaged 56% yearly gains! Those ranked "3" averaged 28%.  Those ranked "4" averaged 13%, while those ranked "5" shot back up to 32%.

Note that all our own cheap stocks are drawn from the Russell 3000, which doesn't list micro-cap, "junky" stocks.  Chances are, if you referenced our cheap stocks against ValueLine, you'd find a disproportionate number of them with a #2 or #3 ranking.

Now, if you make 56% per year for 11 years, you walk away with a 133-fold gain.  Not bad.  Our own data shows that you get further advantages by going considerably below $5.00, so one can only imagine how much more radical the gains would be if the selection process were further restricted.

By the way, shame on ValueLine for excluding any stock under $5 from its #1 ranking.  These stocks are precisely the ones that make the most impressive gains.

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

Mid January, 2004

By the way, your webmaster is putting his money where his mouth is by picking up three cheap stocks at the end of December...kndl, acpw, and oplk.  They're all up about 15% in this 2 week span.  

It's a bit odd that these "cheap stock" strategies haven't gotten much media ink.  In this day and age, bid/ask spreads and commissions are quite minimal regardless of the price of the stock, so all but the tiniest smidgeon of that 15% is profit.  For many a bond holder, that's 2 years of waiting.  There's also the general impression that cheap stocks are, without exception, garbage, so they're better ignored.  So be it.  It's better to stay quiet about some of these trends!

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

Jan 3, 2004...more on the subject of stock pricing...

Below is a table of industry groupings ranked according to percentage gains in 2003.   The industries were divied up into 50 groups.  Inside the parentheses, we've listed the ranking of the industry in question according to price (the industry with the cheapest average price receives a "1", the most expensive a "50"). 

Average Gain: 53.93 Group Stdev: 24.55 Revised h/l prob: 0.40, 0.12 Market Focus: 1.2+0.6 (0.00000, 3.72) Market Focus5: 0.4+0.3

Top Gaining Industries 2003 116.57 industry 72.24 to 72.60 (3) 105.45 industry 77.18 to 79.00 (12) 99.85 industry 66.51 to 67.61 (10) 99.14 industry 67.61 to 68.68 (9) 97.79 industry 72.60 to 73.20 (2) 96.84 industry 70.67 to 72.24 (4) 81.53 industry 75.94 to 77.18 (8) 81.04 industry 61.58 to 61.93 (6) 78.56 industry 1.12 to 8.06 (17) 78.24 industry 73.20 to 74.38 (1)

Losing Industries 2003 18.17 industry 51.14 to 53.86 (31) 22.11 industry 8.06 to 10.15 (42) 23.84 industry 13.20 to 17.48 (29) 26.15 industry 17.48 to 18.47 (40) 26.63 industry 91.51 to 91.90 (46) 28.00 industry 38.17 to 43.32 (38) 28.70 industry 89.35 to 91.17 (50) 29.90 industry 94.34 to 94.80 (28) 30.68 industry 43.32 to 44.64 (36) 31.52 industry 91.17 to 91.51 (45)

Amazingly (we think), none of the top-performing industries ventured into "expensive" territory (a price rating greater than 25), AND none of the bottom-performing industries could be accorded a "cheap" designation.  

The most "out of sync" industry would appear to be REIT's, which were our 8th worst industry despite being only the 22nd most expensive group (or 28th cheapest, as the table shows).  Utilities and some retail groups were also somewhat "out of sync" with what you'd expect given their prices.

Again, you can see our 2003 year-end result tables here.  One final experiment, the results of which aren't shown here, was to divide all our indicators into 50 groups (as opposed to our usual 10 or 25).  Of the top 25 predictors of 2003 gains, NONE involved industry groups!  The best predictor, of course, was the top 2% cheapest stocks in our database, averaging a 230% gain for the year.

Now, what are currently the cheapest industries when we divide industries into 50 groups?  The cheapest are software stocks, followed by computer hardware, semiconductors, scientific equipment, and business services. If you're looking for cheap stocks that aren't particularly high-tech, try oils.  Retail in general and REIT's are also somewhat cheap.  On the expensive side, the list is topped by computer service stocks, construction services, investment and consumer services, insurance, banks, aviation, and food processing.

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

Below (Jan 1), we discussed cheapness as an indicator of future gains.

On one hand, a stock's price doesn't seem like a particularly logical predictor of future gains.  At a company's IPO, there isn't any particular rule for setting the price of a stock.  What's more, stocks split all the time, so price has little to do with the past success of a company.  In itself, the price doesn't tell you much about the company.

On the other hand, price is certainly the most basic criterion in stock selection.  There's a very different "feel" to a $130 stock than a $3 stock.  Regardless of fundamentals, the former stock doesn't feel like a bargain, while the latter does.  You might look at the $130 stock and think, "hell, is there any way it could double at that price?".  You might look at the $3 stock and think "great deal".  Then again, you might think "junk".  Price is a potent psychological criteria in stock selection.

Of course, there are some practical considerations related to pricing.  Small guys can't or don't buy round lots of expensive stocks, though one would presume these guys don't have much influence in the market.  Bid/ask spreads are often larger for cheap stocks.  Stocks under $5 are more difficult to short and are less likely to have options trading under them...cheap stocks are less likely to be used as hedging instruments.  Particularly cheap stocks are in danger of being delisted.

Jan 1, 2004...we've posted results for the year.

As has so often been the case, cheap stocks dominated the list of top gaining stocks.  It's not so much "market-capitalization", all you gurus out there.  It's CHEAPNESS!  

While virtually every major market service's year-end "wrap up" focuses almost entirely on industries and possibly small caps vs large caps, these groupings aren't heavily represented in our data tables for the year.  The real gains are found in cheap stocks, volatile stocks, and stocks with heavy losses in 2002.

Below, we show the results of an experiment where (read carefully) we used volatility, marketcap, and industry data to predict stock price (just the price of the stocks, not the % gains over a future period). You can see that the 4% of stocks with the highest capitalization were the most expensive, averaging $39.48 at the end of 2002.  The 4% most volatile were the cheapest, averaging a mere $3.83.  Only at the bottom of the lists do you find any industry representation. 

The case we're hinting at here is volatility and capitalization is primary to industry groupings in correlating with a stock's price.  Further, in 2003 at least, cheapness was by far the most significant predictor of future gains, not any particular industry grouping.

Now, assuming that cheapness will again be a predictor of gains in 2004, what groupings are currently the cheapest and most expensive?  

You can see that semiconductors are found in the cheapest industry grouping we have, averaging $17.27 per share.  Still, they're a good deal more expensive than last year ($10.34).  Note also that the most volatile stocks increased in price from less than $4 to nearly $10.  Finally, it's interesting to see that the basic arrangement of the tables from 2002 and 2003 are quite similar.

Of course, the absolute cheapest group of stocks is...the cheapest group of stocks.  What we're doing here is looking at "secondary" predictors of cheapness.

So, if you want to make a 2004 prediction, you might venture a guess that semiconductors will prosper.  If you turn out to be right, you'll win special kudos for picking this group at this time, since a number of prominent gurus have been predicting a lousy 2004 for the group.  Be sure to make the prediction with a good deal of aplomb (try "Contrarian Pick of the Decade" or "Why the Experts are Wrong...Semiconductors will Sparkle in 2004") and don't mention that you're simply picking the cheapest industry...better to dazzle with your knowledge of book-to-bill ratios.

If you turn out to be right, you can bang investors over the head with this prediction for at least the next ten years (hey, some pundits are still bragging that they predicted crashes in the 1980's).  If you turn out to be wrong, don't worry about it, since nobody is likely to remember your prediction anyway.

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

Worried that somehow our analysis program has some sort of bug that overemphasizes gains in cheap stocks, we went back to a list of stocks we created in January of last year (2003).  We sorted the stocks by price, and looked at the individual performance of the cheapest stocks, in order...oplk, tmr, cpst, scon, hand, dgit, mrvc, ocpi, stor, tmta, dyn, wstl, sww, vign, epic, lu, vlnc, sfe, onnn, mslv, egov, imny.  Check out the gains for yourself...they're phenomenal!

Note that the above list actually contains some fairly large-cap stocks (e.g. Lucent, Dynegy).

We stop our list at imny for a reason...it's the first stock (#22) on our list that actually showed a loss for the year (-35%)!  Two of the stocks ceased trading during the year (hand and stor).  But both of these stocks ceased trading at prices above their January prices (each with gains around 50% before they went off the radar screen).

It should be noted that these stocks were the cheapest stocks in the Russell 3000, not the cheapest stocks in the American market.  The fact that these stocks were actually strong enough to be listed in this index, which won't accept micro-cap stocks, might mean that the larger body of cheap stocks didn't/wouldn't fare as well...one should not draw the lesson that buying the absolute cheapest stocks in the entire American market will produce the most phenomenal gains.  It's safer to choose your cheap stocks from an established index like the Russell 3000.

Copyright © 2010 MarketSynopsis.com. All rights reserved.