Sunday, November 28, 2010

Is horse racing merely unpredictable or really non-predictable?

One of the things I like about Thoroughbred horse racing is handicapping, or trying to predict the winner in advance. This is a tough business. Just ask anyone who has tried.

As a rank novice in this game, I have a pretty low bar and a lot to learn. I go to the track as often as I can. I talk to anyone who will engage with me about horses and racing. I read the classic literature. I watch races on TV. I study The Daily Racing Form. I download past performances and result charts both for races I attend and others I have an interest in. I even (sometimes) risk my own hard earned money at the mutuel windows.

As an antidote to my natural urges toward self-deception, and as a more advanced learning tool, I also participate in two on-line handicapping contests. Both are pretty straightforward, asking the participants to predict the outcomes of between one and four races a week. The one I like best is run on Dan Illman’s handicapping blog on

While my results over the past year or so have been dismal, I have learned a lot from the other participants. These folks include owners, trainers, pedigree analysts, lifelong aficionados, degenerate gamblers, and novices like myself.  The blog and its comments cover a number of racing topics each week, but the focus is usually the contest – take a mythical $100 and wager it to the biggest payoff on a race chosen by the winner of last week’s contest. Illman weighs in with his analysis and so do dozens of commenters. Very few of them are the kind of get-a-lifers you might expect. Everyone is free to enter and encouraged to offer some thoughts on the race. The prize is some free DRF downloads, bragging rights, and the chance to choose the next race in the contest.

This week’s race and blog contest results made me really stop and think. The race was the Grade 1 Hill ‘N’ Dale Cigar Mile from Aqueduct with a purse of $250,000. A top ranked race at a top racecourse featuring some of the most widely known Thoroughbreds in America. These included some rising three year olds as well as veteran four, five, and seven year old runners. Three of the horses had won over $1 million in purses. The result? The highest odds horse, Jersey Town, won paying $71.50 for a $2 win wager. Another way to say this is that the horse chosen by the fewest number of bettors (by far) won the race.

So we have a group of enthusiastic, experienced analysts looking at a field of nine well-known horses. The standard past performance data contains thousands of data points for the field. Statistical summaries and “expert analysis” are available for free all over the place. How did we do?

Not one of the eighty DRF contest participants predicted the winner. No-one would have cashed a ticket of any kind if they were playing with their own money. If we were choosing the winner at random, then eight or nine players should have chosen correctly, but none did.

Jersey Town was not some dark horse. He has good breeding and is trained by a well-respected veteran. He is a four year old who had raced eleven times previously, finishing in the money all but one of those times. He raced in New York less than a month ago, finishing second. The first and third place horses in that race were also racing in the Cigar.

The racing press sees it as an upset, but no more. These things happen.

It makes me think hard when a race that should have been formful winds up like this. Is this why racing is dying? Is it that gamblers really prefer the predictability of losing at the slot machines and roulette wheels to the non-predictability of the pari-mutuel windows?