Price on Value
November, 1999

Dogs of the Dow: More Bark Than Bite?

John Price,  Ph.D.

Wouldn’t you love to have an investment strategy that, with just a few minutes work each year, completely eclipses the Dow Jones Industrial Average. This is possible according to the claims made by the supporters of the various Dogs of the Dow strategies.

The original Dogs of the Dow or Dow Dividend strategy sorts the 30 stocks in the Dow Jones Industrial Average DJIA by their dividend yields at the beginning of each year. A portfolio is formed consisting of equal weights of the 10 stocks with the highest yields. This portfolio is held for a year at which time it is liquidated and a new portfolio formed. This strategy was popularized in a 1992 book called Beating the Dow by M. O’Higgins and J. Downes.

After the original success came the innovations so that now there are more variants than there are Dow stocks. The first of these is the Dow Five. It starts with the ten Dogs of the Dow stocks and restricts those to the five stocks with the lowest prices.

The Dow Four refines this further by dropping the lowest priced stock. The Foolish Four tweaks this by adjusting the weights of the stocks in the Dow Four. It doubles the weight on the lowest-priced stock in the Dow Four.

The first row of the next table gives the average annual returns of these four Dow strategies from 1973 to 1996 along with the that of the DJIA. The second row gives the standard deviations of these returns.

In their book The Motley Fool Investment Guide David and Tom Gardner report that for two decades, from 1973 to 1993, the Foolish For strategy returned an average annual return of 25 percent and that "it should grant its fans the same 25 percent annualized returns going forward as it has in the past."

Did alarm bells start going off in your head when you read this?

If only it was this easy. No matter what patterns, correlations or successful trading rules are observed in past stock data, why should they carry into the future?

Data mining

Finding patterns in stock data is called data mining. Patterns can be found by eye-balling stock charts but more likely by using computers to search through large data bases.

A recent paper called Mining Fool’s Gold by Grant McQueen and Steven Thorley (Journal of Investment Management and Research, 1999) describes a number of warning signs for detecting data mining. The first is the use of a high number of variables. The more variables you use the more chance of finding suggestive patterns.

The second is when strategies are built on top of one another with each additional strategy tweaking the steps of the previous one. This is particularly dangerous when there is no motive for the adjustment apart from the fact that it would have provided better profits.

A trading strategy with any claim to reliability and longevity needs a plausible theory. The argument for the Dogs of the Dow starts with the observation that having a high dividend yield is an indication that a company is going through a difficult time. But the Dow companies are large and powerful with the expertise to pull out of this. Hence the Dogs of the Dow represent companies that are on the verge of turning themselves around. Or so the argument goes.

Even if you find this argument convincing, what possible reason could there be to reduce these to the five stocks with the lowest prices? How could raw stock prices contain any information regarding the future success or failure of an investment?

One way to test for the validity of investing strategies is to run them on out-of-sample data. This is usually done in one of two ways. First, run the strategy over a different period of time than was used to find the strategy. Second, use different times of the year to start and rebalance the portfolios.

McQueen and Steven Thorley did both these tests for the Foolish Four. They examined its success over the years 1949-72. Then they examined it over the original sample period, 1973-96, while forming the new portfolios on the first business day of each July instead of the first business day in January.

They obtained the following results.

The first thing to notice is that the performance of the Foolish Four is much reduced. It had almost identical results for as the DJIA for the years 1949-72. For the years 1973-96 with the mid-year starting times, the Foolish Four outperformed the DJIA by around 3 percentage points. This is a handsome victory, but a far cry from the 12 percentage points when January was the starting time.

Starting your own doggie portfolio

Whatever you might think of the success and usefulness of the Dow strategies, human nature being what it is, it is somehow satisfying to search for patterns and successful trading strategies in past data. Maybe, just maybe, one of those patterns will be the one to carry into the future. And make you rich. Or at least famous.

Here are a few tongue-in-cheek ideas to get you started. The first is called the Fractured Four and is described in the paper by McQueen and Thorley. It consists of holding the Foolish Four stocks in equal weights in even years and buying only the second-to-lowest stock in the remaining years. It outstripped the other Dow strategies with a return of almost 35 per cent per year for 24 years.

An even more nutty strategy was mentioned by Jason Zweig. It consisted of holding the bottom 25 per cent of stocks by market value without repeat letters in their names. This one beat the market by six points per year for two decades.

What is the lesson here? Just this. If you are willing to spend enough time and dig deep enough through historical data, you are certain to come up with some outstanding trading strategies. The only problem is that the success of the strategies is hypothetical. Reliable success in the future needs a more substantial basis than past patterns and correlations.
______________

Next month’s article "Surprise! Surprise!" will look at the gains to be made when the stock you are holding gets a positive earnings surprise. I will also describe a recent study that claims you can improve your ability to predict which are the companies that will have these surprises.