Tuesday, March 24, 1998

The Daily Workshop Report
by Robert Sheard (TMF Sheard)

LEXINGTON, KY. (March 24, 1998) -- In our ongoing Workshop research, I'd like to present the results of yet another back-tested model based on relative strength. But before anyone gets excited, this isn't a model I'd necessarily recommend. While the numbers are fairly impressive, the basic 26-week relative strength model we already track in the Workshop has tested out better than the model I'll present today. Part of research, however, is discovering what doesn't work perfectly as well as what works very nicely.

The screen I tested starts again with the 100 stocks ranked highest for timeliness by Value Line. (I also tested using both 1s and 2s on the timeliness scale and the years I tested both versions, the group of 1s alone performed better.)

Then I accepted from those 100 stocks only those that had a #1 ranking on Value Line's technical rank. This typically pares the group down to 30 - 50 stocks, which I then sorted by 26-week returns to stress the relative strength. The holding period was one year, from 12/31 to 12/31.

The first full year this technical ranking was available was 1988, so we have only ten full years for the test. The following table includes the returns for the top five stocks, all ten stocks, and the Standard & Poor's 500 Index:

Year  Top 5  Top 10 S&P 500 
 1988 (10.21) (3.14)   16.81  
 1989  44.64  36.29    31.49  
 1990   8.17  22.70    (3.17) 
 1991 120.33 105.67    30.55  
 1992  11.54  22.24     7.67  
 1993  57.66  45.05     9.99  
 1994 (11.57)  7.08     1.31  
 1995  41.67  19.63    37.43  
 1996  47.51  40.67    23.07  
 1997  60.57  47.74    33.21  
 1998  41.65  34.25    12.14 (through 3/18/98) 
 

The annualized returns from January 1, 1988 through March 18, 1998 are:

35.85%  Top Five 
 34.79%  All Ten 
 18.95%  S&P 500 
 

Some of the models Jim Lynn discussed recently began in 1989, so if you're trying to make a comparison, here are the annualized returns starting one year later, on January 1, 1989:

42.11%  Top Five 
 39.72%  All Ten 
 19.19%  S&P 500 
 

While these numbers are impressive, and I'm particularly impressed with the returns for the ten-stock group in the weak market years of 1990 and 1994, the overall returns significantly lag those put forth by the straight 26-week relative strength model (skipping stock #1 and buying stocks #2 - #6).

I'll continue to caution readers that such short periods for our back-tests on these models leave all historical conclusions open to exaggeration by the past decade's terrific bull market. Even though this approach I've featured today gained 22% while the market lost 3% in the last recession (1990), it's still a remarkably volatile strategy. And who knows how it'll fare in an extended bear market?

In any given year, you might have one stock gaining 160% while the one next to it in the rankings loses 60%. The overall gain of 50% is sweet if you are able to ignore the individual components and look only at the aggregate return, but my experience reading the message folder here for the last several years is that very few investors are disciplined enough to do that. When one stock is losing money hand over foot, human nature's first impulse is to try to tweak the system to avoid such a stock, though there's likely nothing useful that could be done other than to envision the entire portfolio as one unit rather than a series of individual components.

We have ample evidence that relative strength is a powerful tool, and nothing we do here is going to add significantly to that fact (given Jim O'Shaughnessy's much longer studies using the Compustat database). But how we consider implementing such an approach is still very much worth looking into.

My personal belief is that if you're going to use such an approach, you're better off not getting greedy and going for the ultimate in returns. Instead, buy a few more stocks (like the 10-stock version here instead of the five-stock version) to smooth out the risk and volatility of the entire portfolio. And then marry that to a more stable approach like Keystone or the Dow models. If sacrificing one or two percentage points a year allows you to stay with the approach in bad years, you're better off than using the more concentrated approach and bailing out when it drops 15% or 20% in a single quarter or year.

It's good to keep in mind Aesop's fable of the Dog and His Shadow. While crossing a footbridge with a nice piece of meat in his mouth, the dog noticed his reflection in the water. Thinking it was another dog with an even bigger piece of meat in its mouth, the dog so wanted the bigger piece he began to bark, dropping his meat into the stream and losing it altogether. Fool on!

Go Kentucky -- Final Four-Bound Again!

Check out the latest file updates for the Workshop:
New Rankings | 1998 Returns | New Database

[Robert Sheard is the author of the forthcoming book, The Unemotional Investor, due out from Simon & Schuster on May 12. To pre-order your copy, please visit Amazon.com, where it's available at a discounted price.]