The so-called "Dogs of the Dow" strategy calls for buying the ten highest yielding stocks of the Dow Industrials in order to try to outperform the whole index. This strategy has not worked well since the early 1990s, but has come close to matching the whole Dow so far in 2009. There may be a reason, based on year-end 2008 valuations. An example from chess will illustrate the point.
Suppose I am an objectively inferior chess player to a particular opponent, Ms. D(ow). Still, I have a chance to win if she gives me a handicap. If that handicap is a knight (the equivalent of three pawns) I will win the majority of the time. If it is only two pawns, I will win sometimes, but less than a majority. A one pawn handicap makes me a huge underdog (only an occasional winner), a knight and a pawn (four pawns' equivalent) makes me a decided favorite, and a rook (five pawns) handicap makes me almost a lock. "Even odds" may be about 2 1/2 pawns (we alternate handicaps between two pawns and a knight).
Suppose each percentage point dividend yield premium of the ten dogs stocks compared to the whole Dow is the equivalent of a pawn. With a 1-2 percentage point premium, the dogs will usually or almost always lose; which was the case for most years of the past decade. With a three point or more premium, the dogs will usually win, as was the case for most of the 1970s and 1980s. At the end of 2008, the dogs' yield premium over the Dow was 2.37%; not enough to prevail, but enough to make the "game" close (so far this year). If the dogs had been priced lower, to yield 3.00% over the Dow, they would probably be ahead by now.
Now, the difference in caliber of the giver and receiver of "knight odds" is quite large. If I am an average club player, Ms. D is probably a "master." And if the World Champion were giving knight odds, someone who could take them and win would probably be at least an expert (just below master).
But the Dow stocks usually share common characteristics, insofar as they are large, established, and at least somewhat mature companies. There are large and important differences in the robustness of one group of companies to another. (And these differences could be particularly pronounced for two particular companies.) Still a group of the ten least robust ... not be that much less robust than the Dow as a whole; at least not if you give the first group "knight odds," or a three percentage point yield differential
There is another wrinkle. Suppose this was a gambling proposition (these exist informally at tables all around New York City), but Ms. D, like Ben Graham's Mr. Market, would announce the odds in advance, and allow you to decide which rounds (years) to play. The "odds" (yield premium on dogs stocks) is known at the beginning of each "round" (December 31st of the previous year). You can choose not to play (by buying the whole Dow), which counts as a draw, or you can choose to play (by using the "dogs" strategy in years when the yield is at least three percentage points higher than the Dow as a whole. Thus, you have a way of getting the best of both worlds, and probably outperforming.