The Wisdom Of Crowds: James Surowiecki on the predictive accuracy of the horse race betting markets
Chris F. Masse April 25th, 2007
With your permission ( 
), I am re-publishing here in one blog post the two comments from James Surowiecki, so the Midas Oracle readers who don’t subscribe to comments can see it.
[...] The crowd of bettors at the racetrack is not predicting which horse will win. It’s predicting the odds of each horse in the race winning — in other words, if a horse goes off at 2-to-1 odds, the crowd is predicting that the horse will win about a third of the time. So if you want to measure the accuracy of the crowd of bettors, what you want to measure is how well the crowd’s probabilistic forecasts map onto actual outcomes over, say, the course of an entire season.
When you do that, what you find is that racetrack bettors are uncannily good at predicting the odds of horses winning. With the exception of a small longshot bias, the crowd’s pre-race probabilities predict almost perfectly how likely horses are to win.
—
It’s simply not true that the crowd doesn’t care about subjective probabilities, since it’s those probabilities (that is, the odds) that determine how much you get paid if your horse wins. People want to bet on horses that have a better chance of winning than the odds predict: that is, to state the obvious, the only chance of making money over time. More to the point, when we’re evaluating the wisdom of crowds, it’s not the performance of the individuals in the crowd that matters — it’s what happens when you aggregate all of those individual judgments that’s interesting.
As for the idea that informed traders are solely responsible for the accuracy of betting odds, that cannot explain what happens in parimutuel markets — which is how horse-racing odds are determined in the U.S. Obviously, the more informed traders are, the better. But in a pari-mutuel market, every single bet (informed or otherwise) that’s placed affects the final odds. So if you exclude, say, the suckers, then you almost certainly change the final odds. Yet we know that those odds are remarkably accurate. It’s possible that if allowed only the crowd of informed traders to bet, the odds would improve. But you’d have to be confident in your ability to identify them beforehand — no easy feat — and in any case the room for improvement is very small. Dave Pennock [and Daniel Reeves]’s work on probability averaging is relevant in this regard.
As for the accuracy of betting odds, I don’t know about the odds “always” being close to objective probabilities. But the data in Arthur Hoerl and Herbert Fallin, “Reliability of Subjective Evaluations in a High Incentive Situation,” Journal of the Royal Statistical Society (1974) shows just how hard it is to outperform the crowd at the track, at least when the odds are set by a pari-mutuel system.
James Surowiecki is the author of The Wisdom Of Crowds (impact of the book) and a financial columnist at The New Yorker. - Section: The Talk Of The Town - (James Surowiecki is one of the columnists.)
How to subscribe to James Surowiecki’s column at The New Yorker:
Sub-Site Feed: http://www.newyorker.com/services/rss/feeds/talk.xml - [Partial feed, though. 
]
Subscribe to “Talk of Town” via Google Reader.
—
UPDATE: Niall O’Connor (of Betting Market) has a kind of rebuttal with the paper citations that goes with it. See the comment area, just below this blog post.
—
The main reason why James seems so keen on betting markets is that he believes that they aggregate the preferences of all participants rather than just the informed traders.
Whilst not doubting the information processing potential of the betting market, it is my belief that he imparts to the average bettor (noise trader) a degree of intelligence and sophistication that they simply do not have.
It is insider traders (See: Cain, Law and Peel 2000), and well-informed profit-maximising professional bettors (See: Adams, Rusco and Walls 2002) who drive track odds down to the levels implied by true win probabilities.
Craft (1985) has highlighted the fact that horses that shorten in price demonstrate a higher expected return, but most particulalry so, when insider information is likely to be prevalent.
Schnytzer and Shilony (1995) demonstrated that groups with inside information are better and more consistently able to predict the outcome of races, than a group which does not have access to such information.
Vaughan Williams and Paton (1996, 1997) highlighted the fact that an analysis of the divergence of the pricing behaviour of bookmakers relative to what would be expected were there no possibility of insider activity, implied that bookmakers perceived that between 2 and 4% of all sums staked in the betting market can be accounted for by insider trading.
In “Information Efficieny in Financial and Betting Markets” Vaughan Williams goes so far as to conclude;
” In summary, there is significant evidence for the suggestion that insiders in betting markets possess valuable information unavailable to the public, which they can trade upon, so as to earn above-average and even abnoraml returns, and to this extent the market may be considered informationally inefficient.”
Most members of the betting market have no notion of the true win probabilities, and accordingly most lose. The predictive qualities of the betting market stem from the activities of a few well informed individuals, and not from the activities of the broader crowd per se.
In the context of the UK betting market, if the broader “crowd” were in any way concerned about true win probabilities, they would all be trading on Betfair, which they are not!
The Wisdom of Crowds or of Markets?
BY MATHEW WEINSHELL
http://www.the-dissident.com/C.....kets.shtml