Growth Investing - Be Careful About Extremes

by: Marc Gerstein

On March 15, 2012, a web site posted and article wondering aloud if silver-miner Pan American Silver (NASDAQ:PAAS) had become "the Perfect Stock." That's a pretty high-blown label and probably one I'd never consider for any stock, and the article did sort-of chicken out suggesting that although "some stocks are a lot closer to perfection than others," there are no sure things. But the author did post a fundamentals-based scorecard, so to speak, and used it to show that PAAS passed eight out of his nine tests. Two of those considerations related directly to revenue growth and others related to bottom-line strength (e.g., margins, return on equity, P/E).

The market, however, has not only failed to appreciate PAAS as a perfect stock, it hasn't even been seen it as a perfect silver-mining investment. Figure 1, shows it's been lagging the iShares Silver Trust ETF (NYSEARCA:SLV).

(Click charts to enlarge)

Figure 1 - From Yahoo! Finance

Has Mr. Market been failing to look at the numbers, or is something else going on? For one thing, there is a planned acquisition, but I'm not sure that's it (and it may actually be a good thing). As to the numbers, PAAS has, indeed, been doing well; quite well. There have been some unusual items on the income statement, but not enough to detract from the company's basic fundamental strength, helped obviously, by general strong silver prices.

Interestingly, though, the strong growth performance may hold the key, not to an opportunity but to the problem. The same may be true of such companies as Agrium (AGU), a provider of agricultural products and services that has been benefiting from an acquisition and very low crop inventories that have been boosting demand from farmers for seed and fertilizer; CF Industries (NYSE:CF), another acquisitive company that has been profiting handsomely from strong fertilizer demand; Core-Mark Holding Co. (NASDAQ:CORE), which serves convenience stores with merchandising programs, information systems etc. and has seen growth strengthen due to an acquisition and a major new contract; Darling International (NYSE:DAR), which has benefited from an acquisition and from strong demand for the tallow, fats, bone meal etc. it produces from recycled beef, poultry and pork waste streams; Holly Frontier corp. (NYSE:HFC), a petroleum refiner that benefited from a merger and strong refining margins; and Rex American Resources (NYSE:REX), which has interests in ethanol production facilities and has benefited from increased investment in various plants.

Figure 2 summarizes key growth statistics for all of these companies.

Figure 2

Those are pretty impressive numbers, right? Maybe they are. Maybe they aren't. Take a hint from the fact that I used red font to note the most dramatic growth rates, the kinds that attract attention on stock screens and which tend to be spotlighted in computer-generated research "reports," (such as one I created when I worked at Reuters) and in articles that follow the numbers very closely such as often appear on web sites today. Take another hint from the fact that I made reference to merger-and-acquisition activity in connection with each stock.

Do you really think that 1,615.5% or 2,499% are likely to be reasonably representative of the sales growth rates we can expect from Pan American Silver in the future? Mr. Market, which did not seem to think PASS was the perfect stock, or even that it was close, is probably skeptical. I like Core-Mark, but I don't think I can make any analytic use of the 487.8% year-to-year EPS growth rate achieved in the latest quarter. etc., etc., etc.

All of these stocks have something else interesting in common. All appeared on a growth-oriented stock screen I adapted from one I created for StockScreen123. The original screen, which requires percentile ranks of 50 or better relative to the market for each of the above growth rates after which I select the top 15 based on a Quality-Value-Growth ranking system I created, was not intended to be used as an investing strategy; it was actually a demonstration (the adaptation I made for use here was to restrict consideration to Russell 3000 stocks). But even so, the screen's five-year performance record is decent, as can be seen in Figures 3.

Figure 3 - Five-Year Backtest ($100 invested in model on 3/27/07 grew to $206.0 on 3/26/12)

The one-year backtest, however, was lackluster.

Figure 4 - One-Year Backtest ($100 invested in model on 3/27/11 shrank to $93.6 on 3/26/12)

We all know that growth is good for stocks. I think we and just about all stock-market participants and observers should be able to agree that companies that grow briskly ought to be stock-market winners. That theory could be trumped by episodes of absurd valuations, bad balance sheets, etc., but the ranking system I used here to select the top 15 pretty much guards against that.

For growth investors, the challenge lies in the legal mantra: Past performance does not assure future outcomes. If a growth strategy disappoints, it will likely be because future growth failed to measure up to expectations established by historical data.

This will always present a challenge with every stock. We, as humans who can't simply peer into the future, must always be on our guard. But there are some things we can do to tilt the odds in our favor. One such solution is to eliminate from consideration historical growth rates that are so extreme as to make it very unlikely, if not impossible, that the future can measure up to the past. Sometimes, the extreme will come from a non-recurring gain. Sometimes, it will come from corporate structural moves, such as M&A. And other times, it can even come from unusually strong business, as we've seen in the markets served by the companies referred to above. Whatever the case, the key is the notion that extremes tend to be unsustainable.

I created a second version of the above-mentioned stock screen wherein I eliminated all extreme growth rates. There are many ways to define extreme. For this exercise, I used Z-scores and filtered out all companies that were above 3 or below -3 with respect to any of the six growth statistics considered. (In one of the most annoying instances of popular nomenclature, Z-score is used in two unrelated ways. One is the Altman bankruptcy test. That's not the one I'm using. The one I use here is the quant technique that normalizes a dataset by re-stating a number in terms of how many standard deviations above or below the mean it is. I eliminate stocks with any growth rate that's more than three standard deviations, either way, from the mean. I filter out upper extremes because, as noted, they are least likely to be sustained. I also eliminate downward extremes because they are likely to be associated with unusual corporate developments that may eventually bear on future results.)

Figures 5 and 6 show better backtest results for the new approach, which bars extremes (the stocks listed above in Figure 2 are those that made the first screen but fail to make the revised version).

Figure 5 - Five-Year Backtest ($100 invested in model on 3/27/07 grew to $251.7 on 3/26/12)

Figure 6 - One -Year Backtest ($100 invested in model on 3/27/11 grew to $116.8 on 3/26/12)

According to this backtest, eliminating extremes added about 4% per year to overall performance. That's not stupendous, but it is worthwhile and hard to ignore, especially since the change is so consistent with common sense. The one-year benefit was much more pronounced, as the strategy shifts from a 6.4% loss to a 16.8% gain. This is noteworthy not only because of the magnitude of improvement, but also because of the timing. The latest year was one during which many quantitative strategies struggled - being sandbagged by unsustainable extreme data-points may have been especially troublesome for many during this period.

Figure 7 lists the stocks that make the new less-tolerant-of-extremes version of the screen.

Figure 7

It's still not perfect. There are still numbers about which I'm uneasy (in terms of being willing to assume sustainability). But it's definitely an improvement over the original screen, as suggested by eyeballing and the backtests.

Going forward, I'm satisfied with most of the sales growth rates. But EPS growth is a ratio that is subject to breathtaking extremes, and while plus-or-minus 3 might make for adequate z-score boundaries around sales growth rates, I probably need to do more work with EPS. I looked at year-to-year EPS growth rates from the latest quarter and here's what I found for the Russell 3000 companies:

  • Average growth rate: +7.3% (That's reasonable.)
  • Maximum Growth Rate: +97,233.3% (That's ridiculous, and remember, we're looking at Russell 300 companies, not penny stocks.)
  • Minimum Growth Rate: -18,696.7% (That's also absurd.)
  • Standard Deviation: 2,114.05% (OUCH! Restricting z-scores to the -3 to +3 range barely scratches the surface.)

From this, I'll assume the benefit of improved screen performance seen above came from the use of z-scores to limit the range of permissible sales growth rates, and that I need to go back to the drawing board regarding EPS.

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.