Horse

Racehorses appear in the book not as individual animal portraits but as the subjects of a demanding prediction test: can the collective judgments of bettors anticipate outcomes more reliably than any single observer? In an analysis of 1,825 races at Aqueduct and Belmont Park in 1970, market odds closely matched both horses’ finishing order and their actual frequencies of victory. The example makes horse racing a vivid case of collective intelligence in a setting where forecasts can be compared with definite results.

A Test of Betting-Market Predictions

Horse racing offers a ready-made way to compare market expectations with outcomes. The odds assign horses relative chances of winning before a race; after the race, those probabilities can be checked against how often horses actually won. The analysis discussed in the book examined every race run at Aqueduct and Belmont Park in 1970, asking whether odds predicted both the frequency of victory and the ranking of finishers.

The study covered 1,825 races, with fields ranging from five to twelve horses. This breadth matters: the pattern was not confined to races of one standard size. Instead, the market’s assessments could be tested across a varied set of fields, against the same concrete measure—the horses’ actual results.

Odds and Finishing Order

With only a few exceptions, the market’s ranking predicted the order in which the horses finished. The favorite finished first most often, the second-favored horse finished second most often, and the same descending pattern continued through the rankings. When the researchers compared horses’ average finishing positions, those positions declined monotonically as the horses’ market ranks fell, with no mistake in that measure.

That result concerns more than whether the favorite won a particular race. It shows that the odds contained information about the field’s relative ordering: across the races, horses placed higher by bettors tended to finish correspondingly higher. The market’s collective ranking was therefore informative even when considered across horses that did not win.

Probabilities Against Results

The odds also closely tracked how often horses won. In 312 seven-horse races, bettors’ forecasts assigned the favorite a 33 percent chance of victory; the favorite won 34 percent of the time. For the second-favored horse, the predicted and observed rates were 22 and 21 percent. For the third, both were 16 percent; for the fourth, both were 12 percent. The fifth was forecast at 9 percent and won 8 percent of the time, the sixth was forecast at 6 percent and won 8 percent, and the seventh was forecast at 3 percent and won 2 percent.

Across those seven-horse races, the small gaps between predicted and observed rates make the comparison unusually clear: subjective estimates formed through betting were close to the objective frequencies recorded afterward. Results in races with other field sizes were similar, though less striking. The example thus supports the book’s broader argument that, when many separate judgments are aggregated, their combined forecast can hold substantial information about an uncertain outcome.

Predictive Strength and Its Limits

The book places horse-racing odds within a larger discussion of betting markets, whose predictions can be evaluated because races end with an observable result. It characterizes the final odds as reliable guides to both the order of finish and the probability of winning. The 1970 study provides the specific evidence behind that claim, rather than relying only on the general observation that favorites tend to win more often.

This success does not mean that betting markets are free of bias. The book also points to the long-shot bias, documented in research on parimutuel betting markets. That qualification keeps the horse-racing example from becoming a claim of perfect judgment: the odds can closely calibrate probabilities and rankings while still displaying a systematic weakness. Their significance lies in how much collective information they capture, not in an assumption that every market estimate is flawless.