More Poor Mouthing, A Case for Heartburn, A Calculator at Belmont, and Other Matters. The Two-Minute Sprint

(FORTUNE Magazine) – At precisely 1:10 on a recent Monday afternoon, your correspondent accessed an elevator at Belmont Park, pushed the ''B'' button, and soon enough found himself hiking through an eerily lit basement corridor, totally devoid of humanity, that stretched ahead for several furlongs. Figuring that the Mutuel Department was already suspicious enough of his enterprise, he manfully repressed the urge to scream and ultimately did reach the department's office. Whether the documents he there accepted will enrich our understanding of the efficient market hypothesis is what this item is all about--that and whether Bill Ziemba is right. William T. Ziemba is a professor of management science at the University of British Columbia and the author of such scholarly works as Stochastic Optimization Models in Finance, Generalized Concavity in Optimization and Economics, and (with Donald B. Hausch) Beat the Racetrack. Some of the research for this last-named exercise was made possible by a grant from the Canadian Natural Sciences and Engineering Research Council, which is one of several details that led us to wonder if Bill might be a bit of a promoter. Another detail is that he wants you to implement his system for equine wealth enhancement with the help of this $295 Beat the Racetrack Calculator--it's an HP-41 with a special chip for horseplayers--and while guilt by association is considered impolite in some circles, it is a fact that the last time we bought a calculator with a special chip, it said to sell stocks in July 1982. On the other hand, Ziemba's basic proposition rests on scholarly foundations. He is very much on top of a vast literature demonstrating that parimutuel betting at racetracks is an efficient market, just like the stock market. In general, the odds at post time correctly express the probabilities of the different horses winning. Ziemba notes, however, that the research effort has concentrated overwhelmingly on ''win'' bets. He believes that there are substantial market inefficiencies in place and show betting, and Beat the Racetrack is an effort to explain how you can profitably exploit these. It is intuitively plausible that the place and show betting pools might be inefficient markets. For one thing, they're markets in which it's hard to get a price quote. Prices--that is, odds--aren't quoted directly because your payoff on a show bet, say, is affected not only by how much was bet on your horse to show but by which other horses finish among the first three and how much was bet on them. The only data you have to work with before getting your own bet down are the dollar amounts already bet on each horse in the place and show pools; these amounts, updated at intervals of a minute or so, are displayed on the totalizator machines positioned on the track infield. Ziemba says that if you know how to analyze these tote boards, you will often observe betting opportunities in the place and show pools. Specifically, you will see that some horses are underbet in these pools relative to the win pools. To be sure, a horseplayer passionately committed to the efficient market hypothesis might argue that these disparities were not inefficiencies but, rather, reflections of the fact that some horses should be bet less in the place and show pools than in the win pool. Ziemba obviously disputes this & view of the case, and his system consists of formulas intended to help you spot the inefficiencies. When you do spot them, with or without the help of the man's calculator, you confront what he delicately calls ''the two-minute problem.'' Au fond, the problem is that at many tracks it will take the average overweight horseplayer about two minutes to sprint the distance between the spot where he has been observing the tote board and the betting windows; ergo, this person's analysis must reflect not-quite-final betting data. The betting data used by Keeping Up to evaluate Ziemba's system were the final figures. The data turned over by the kindly Belmont Mutuel Department showed the money bet on each horse in the win, place, and show pools during every race in the two weeks that ended May 19--a total of 108 races. The system turned up 33 situations in which place or show bets were indicated. The bad news is that anybody placing an equal amount of money on each of these bets would have been a loser. At $10 a throw, he would have bet $330 and recovered $308. Possibly the system just had an unlucky streak. Possibly a larger sample would yield different results. However, we definitely have no plans to revisit that basement.