Split Decisions
Improving the classic BSP strategy

By Ethan Haskel (TMF Cormend)

BALTIMORE, MD (September 1, 1999) -- Two weeks ago we made the observation that, although the Beating the S&P (BSP) strategy had outperformed the Standard & Poor's 500 index since 1987, a small subset of BSP stocks had fared quite poorly. Should we change the BSP strategy?

Let's quickly review our previous article, Runaway Dogs. Recall that, out of 65 stocks in the annual five-stock BSP portfolios since 1987, we found 11 stocks that had split their shares within the previous year. Contrary to our usual out-of-favor "dogs" that we like to buy when no one else will touch them, these recently split stocks were flying high; their prices had been lowered by the artificial process of the stock split, rather than due to a pessimistic outlook for the company. We hypothesized that eliminating these companies from our portfolio might enhance the BSP portfolio returns.

The findings were dramatic. The average return of our 11 recently split stocks was 8.5%. Furthermore, nine of these 11 stocks were dead last among their BSP peers for the year. Excluding just 11 stocks improved the compound average return since 1987 from 19.9% to 25.2% (the S&P 500 returned 17.3% annually for the same period).

A proposal was made that we should revise the BSP strategy to incorporate these observations. As I see it, here are the pros and cons for making such a change.

The Pros:

1. The major advantage of a revised BSP would be the potential for improved returns. A five percent enhancement in annualized returns would have a humongous effect on a portfolio's total value when spread out over many years. For instance, $10,000 growing tax-free at 19.9% per year would grow to about $337,000 after 20 years. Assuming a revised BSP return of 25.2%, the same $10,000 would become $715,000. The difference isn't exactly chump change.

2. The performance of the stocks that had split was quite consistent, with 10 of the 11 underperforming the S&P 500, often by a large margin. The improved returns weren't merely due to one or two outliers, but seemed to be part of a regular pattern.

3. The elimination of a stock that recently split makes conceptual sense, based on our framework of why high-yield investing works It doesn't represent an arbitrary rule used merely to find a spurious anomaly.

4. There are numerous precedents for improving on well-established mechanical investing strategies. The Foolish Four is a prime example. Originally inspired by O'Higgins' 1991 book Beating the Dow, the strategy underwent multiple refinements to evolve into its current format. The Foolish Workshop is chock-full of other mechanical investing strategies that have evolved from earlier models. (Note: Changing a strategy just because you find a factor that improves the returns -- without a good understanding of why it improves the returns and why it should continue to improve the returns -- would be dangerous and silly.)

The Cons:

1. The improved returns for the refined BSP model could be construed as "data mining," a topic near and very dear to our hearts. Data mining involves examining a large number of variables to find correlations that exist solely on the basis of chance, rather than any meaningful association. This topic has been discussed extensively, so I won't belabor the point.

2. There isn't enough data to make definitive conclusions.

3. By using the new BSP strategy, one wouldn't necessarily be investing in the same number of stocks each year.

4. The new strategy is more complicated.

Let's consider the "con" arguments. I reject the idea that data mining is involved here. I started out with a single hypothesis, performed a single test, and found dramatic results. That's not data mining.

True, there are only about 13 years of data available, but a few basic statistical tests on the data do show a significant effect. (I don't pretend to be an expert here, so if there's anyone out there who would like to run some statistics, let's hear from you.)

The revised BSP strategy would be slightly more complicated, perhaps adding an additional five minutes each year to check if any of the BSP stocks had split in the past year. (Most standard stock tables in newspapers include an "s" designation next to any stock that split within the year.) Yes, some years there would be fewer than five stocks in the portfolio, again adding a mild degree of complexity.

Everything considered, I think the preponderance of evidence supports dropping any BSP stock that has split within the past year. We'll still keep tabs on the "BSP Classic" portfolio, but henceforth the BSP portfolio rules will change to incorporate stock split information.

How does this change affect current and future BSP investors? If you already own a BSP portfolio, I'd sit tight. You've got yourself a group of stocks that likely will do just fine. If you're renewing your portfolio, or considering investing in a new one, just eliminate any of the current stocks that have split during the past year.

For the record, the BSP stocks currently yielding over 1% that have split within the past year include the following: Bristol-Myers-Squibb (NYSE: BMY), split 3/1/99; Xerox (NYSE: XRX), 2/24/99; and Sprint (NYSE: FON), 6/7/99. Of these, only Xerox is a current Classic BSP stock, so I'd ignore it and buy the other four.

[Editor's Note: Following are the returns of a hypothetical Beating the S&P portfolio. The stocks were selected December 31, 1998, and "purchased" in equal dollar amounts to be "held" for one year. To see a list of stocks for portfolios starting now, see Today's Stock Lists. Note that the stock split change has not yet been incorporated into this format.]

Beating the S&P year-to-date returns (as of 08-31-99):

Schlumberger (NYSE: SLB)      +44.6%
Kimberly-Clark (NYSE: KMB)     +5.4%
Campbell Soup (NYSE: CPB)     -18.4%
Ford Motor Co.  (NYSE: F)      -8.8%
Bank of America  (NYSE: BAC)   +2.1%
Beating the S&P                +5.0%
Standard & Poor's 500 Index    +8.0%

Compound Annual Growth Rate from 1-2-87:
Beating the S&P               +25.2%
S&P 500                       +17.5%

$10,000 invested on 1-2-87 now equals:
Beating the S&P             $172,300   
S&P 500                      $76,900

Today's Stock Lists | 1999 Dow Returns

Read More Foolish Four Reports

Foolish Four Portfolio

9/1/99 Closing Numbers
Ticker Company Dly Pr Chg Price
IPINTL PAPER9/16$48.63
JPMMORGAN (JP)7/16$129.63

  Day Week Month Year
To Date
Foolish Four 1.49% -4.18% 1.49% 24.02% 25.86% 39.57%
S&P 500(DA) .80% -1.27% .80% 8.86% 8.92% 13.19%
NASDAQ .42% -.29% .42% 25.45% 26.62% 40.78%
DJIA (DA) .99% -1.36% .99% 19.96% 20.20% 30.56%

Trade Date # Shares Ticker Cost/Share Price LT % Val Chg

Trade Date # Shares Ticker Cost Value LT $ Val Ch
  Cash: $80.43  
  Total: $5,034.44  

• S&P 500 (DA) = dividend adjusted. Dividends have been added to the total return of the index.
• DJIA (DA) = dividend adjusted. Dividends have been added to the total return of the DJIA.

The Foolish Four Portfolio was launched on December 24, 1998, with $4,000. Additional cash is never added, all transactions are discussed and explained publicly before being made, and returns are compared daily to the S&P 500 and the Dow. (Dividends are included in the yearly, historic and annualized returns.) Stocks are chosen once per year using a formula based on dividend yield and price. See The Foolish Four Explained for details.