MIAMI — In no sport is data more important than in thoroughbred racing, because billions of dollars in wagers are based on information about horses’ past performances. Yet the industry’s data-gathering is a product of the Stone Age.
While many sports (even quarter-horse racing) record the time of every finisher in thousandths of a second, thoroughbred racing doesn’t time anyone but the winner — not even the second-place finisher in the Belmont Stakes. Handicappers must estimate losers’ times with crude calculations based on the number of lengths by which they were beaten. And when the official data says a horse was 10 lengths behind in the early stages of a race, that number comes from human observation that is inevitably imprecise.
So it was a major development when a Massachusetts-based company, Trakus, developed a high-tech method of recording data in horse races. This winter, Gulfstream Park and Tampa Bay Downs became the fifth and sixth North American thoroughbred tracks to adopt the system. The industry is excited about the possibilities.
“I hope this becomes a new handicapping tool and gives people a whole new way to view a race,” said Gulfstream’s president, Tim Ritvo.
Michael Ciacciarelli, Trakus’s chief operating officer, said the methodology was originally intended for use in sports such as hockey and football. But nobody was more interested than Equibase, the company that collects thoroughbred racing’s data, which thought it saw the technology of the future.
A transmitter is inserted into every horse’s saddlecloth. It sends a wireless signal to as many as 30 antennas on towers placed around the track. Those signals allow the system to record a horse’s speed and path throughout a race.
Trakus’s most conspicuous function is to display, on the television screen, each horse’s position as a race is run. Traditionally, a human being has been responsible for posting the numbers of the horses leading a race — numbers that were often maddeningly inaccurate. Trakus eliminated the mistakes. The order of the field is displayed with each horse’s program number in a square — dubbed a “chiclet” — at the bottom of the screen. As horses change position, the chiclets move simultaneously. If a horse is not visible on the screen, a viewer still knows if he’s in 11th place because of the location of the chiclet.
The technology has been sufficiently accurate that Equibase has changed its method of charting races at tracks using the system. Historically, the details of every race have been gathered by a chart-caller who watches the action through binoculars and calls out numbers to a partner. Now Trakus produces the charts, though the data still requires oversight and tweaking by Equibase personnel.
For handicappers, Trakus is a potential source of valuable information. The ground that horses lose by running wide on the turns is a crucial factor. Trakus measures the distance that each horse travels. In a 1 1 / 8-mile race at Gulfstream Saturday, the front-running winner Teeth of the Dog covered 6004 feet, while-fourth place Went The Day Well traveled 6049 feet with a wide trip. According to Trakus, the loser ran at 37 mph, a superior performance to the winner’s 36.9 mph. Measurements such as these would be useful additions to any handicapper’s arsenal. So, too, would Trakus’s record of each horse’s speed in every segment of a race.
Yet Trakus has not made this information accessible to horseplayers. Limited data appears on the Gulfstream and Tampa Bay Web sites, but a handicapper would have to look up horses one by one and still not find the data that he might want. (It’s all on the Trakus Web site, open only to “industry professionals.”)
The company’s priorities seem misplaced — typical of the racing industry, which often doesn’t seem to understand its customers. Trakus has put a lot of effort into developing graphic simulations of races that Ciacciarelli says are aimed at a younger generation that loves video games. But the company has not focused on giving racing’s core customers the accurate information that they want.
A new technology will inevitably have glitches, and Trakus has had its share. Anyone who examines its data can spot isolated errors. For example: Tycoon did not run 73 feet farther than Mainstay, a rival who finished outside him, in the last race of the Keeneland meeting in October.
But some recurrent errors are more distressing. Trakus puts the final quarter-mile time for each horse on a TV display after a race. At Tampa Bay, Trakus’s final quarters for the 1-mile-and-40-yard distance were regularly preposterous. One day the last-quarter clockings in two races were posted as 20.32 and 20.60 seconds — faster than any thoroughbred has ever finished a route race. The next day Trakus said a final quarter was run in an equally unbelievable 32.35 seconds. These mistakes appeared on the screen for weeks.
Trakus officials acknowledge that they have encountered problems in the one-mile races at Gulfstream. They were unaware of issues at Churchill Downs, where the run-up for six-furlong races is 100 feet (as reported by Equibase and confirmed by track officials), but where Trakus records the run-up distance in most races as about 220 feet. Apparently, no one spotted the disparity in all of 2011. While Trakus has the ability to re-analyze races and correct errors, too many errors are going unnoticed.
As a handicapper and speed-figure maker, I want racing to have precise data and I want Trakus to succeed. So I offer a modest proposal. The company should formulate a report that allows users to log in to, say, the Gulfstream Web site and see the key Trakus measurements in the previous starts of every horse on the Gulfstream card. It should invite users’ feedback — letting thousands of eyeballs look for errors that might otherwise go undetected. The company needs to be less focused on developing whizbang graphics and more concerned about giving the best, most accurate data to handicappers.
For Andrew Beyer’s previous columns, go to washingtonpost.com/beyer.