North American racing suffers from a dearth of public data. Computer-readable versions of basic race results — equivalent to the box scores that other sports share freely — are available only for a fee from the Equibase service run by the Jockey Club.
Equibase does publish free PDFs of race results, but, when developer Robin Howlett published free software to convert PDF results to a computer-readable format, he received a cease and desist order from Equibase.
Data restrictions are especially tight when it comes to equine fatalities. Since July 2008, the Jockey Club has collected data on equine fatalities through the Equine Injury Database, but the data are confidential and only the barest summaries are made public.
Once a year, the Jockey Club puts out a short press release giving the year’s number of fatalities and summarizing the fatality rate for dirt, turf and synthetic surfaces. It also discloses the annual injury rate for individual race tracks, if they agree, but only 35 out of 113 North American tracks are willing to disclose even that much.
If all this secrecy is meant to protect racing’s reputation, it’s not working very well.
Earlier this year, a national scandal erupted at Santa Anita, where, according to media coverage, the fatality rate had spiked to unprecedented levels.
In fact, though, over the past 12 months, Santa Anita’s fatality rate was a little below average, and the two-month spike in death rates that occurred in January and February of 2019 was similar to streaks that had occurred in 2014 and 2016 without much notice being taken. But for months nobody knew that, because nobody had the data.
In June, Hall of Fame trainer Jerry Hollendorfer was banned from Santa Anita and ten other tracks, allegedly for his outsized fatality rate. In fact, though, over the past five years, Hollendorfer’s fatality rate has been close to the industry average. But again nobody knew that, either, because nobody had the data.
The Jockey Club doesn’t publish fatality rates for individual trainers and is unable to share the data that would let anyone else calculate them.
An alternate data source
What can we do when the industry won’t disclose data that’s relevant to a scandal threatening its very existence? We can look for other data sources. And the most valuable data on North American horse racing fatalities appears on a website called Horse Racing Wrongs.
Horse Racing Wrongs is a non-profit founded by animal rights activist Patrick Buttuelo in 2013. It submits freedom of information requests to the state agencies that oversee racing and posts the results on its website. For every year since 2014, the site has published a ‘kill list’ giving the names of horses who died of injuries, along with the date of each injury, the track where it occurred, and an indication of whether it occurred during training or racing.
I cleaned data from Horse Racing Wrongs and used them to write two stories on the history of fatal injuries at Santa Anita. Then, with an assistant, I looked up the trainer of every fatally injured horse, and I used that to write a story showing that Hollendorfer’s fatality rate was no higher than average.
The data on racing deaths are thorough and accurate
I did wonder about the quality of the data on Horse Racing Wrongs. The site’s ‘kill lists’ are inconsistently formatted and required a lot of cleaning. And there are reasons to wonder if the lists are too long or too short.
Horse Racing Wrongs campaigns openly for the ‘eradication of horse racing’, so it certainly has no incentive to understate the fatality rate. But it relies on public information requests to state agencies, and if those agencies are less than forthcoming, the data could be incomplete.
Despite these concerns, though, Horse Racing Wrongs’ data seem to be accurate and nearly complete, at least when it comes to fatal injuries sustained during racing.
Over the five years from 2014 to 2018, Horse Racing Wrongs listed 2,538 horses who were fatally injured in races. Over the same period, the Equine Injury Database reported 2,536 fatal racing injuries — just two fewer. In each year, the number of racing deaths on Horse Racing Wrongs was within eight percent of the number in the Equine Injury Database.
Fatal injuries sustained in Thoroughbred races, 2014-2018
Note. Before calculating these counts for Thoroughbreds, I removed quarter horses and Standardbreds from the kill lists on Horse Racing Wrongs.
I wouldn’t expect these counts to agree exactly. The Equine Injury Database uses a very specific definition of a fatal racing injury. According to the database, an injury is only fatal if death occurs within 72 hours. Under this definition, the 2006 Kentucky Derby champion Barbaro, who suffered a catastrophic injury in the June 2006 Preakness Stakes and died of related complications in January 2007, would not not be classified as having died of a racing injury.
I don’t know if the state agencies queried by Horse Racing Wrong subscribe to that same definition. If they don’t, there might be discrepancies. But apparently the discrepancies are rare.
I also looked up about 50 individual horses listed by Horse Racing Wrongs as having died of racing injuries. In every case, the Equibase results chart confirmed the story. Nearly all the horses “did not finish (DNF)” their final race, and trip notes often indicated that they “broke down”, were “in distress”, or were “vanned off” the track.
The data on training deaths are not as good
While Horse Racing Wrongs’ data on racing deaths are excellent, its data on training deaths are not as good. The site complains that “some states withheld, or claimed they had no records of, training deaths”, and in general training deaths are harder to track.
Some deaths occur at private training facilities rather than race tracks. There are no trip notes to confirm that a horse was injured in training, and it can be hard to know whether a later death was related.
Overall, Horse Racing Wrong lists only a quarter as many training deaths as racing deaths. This is almost surely an undercount. For example, the California Horse Racing Board, which is one of the state agencies that does report training deaths, typically finds that the number of horses who die in training is comparable to the number who die in races.
It’s not great for the sport’s image when investigators seeking data on equine welfare have to turn to a group that campaigns for the abolition of horse racing. But the industry doesn’t leave us a lot of choice. It would be better if the Jockey Club was allowed to disclose more of the injury data it collects.
About the author
Paul von Hippel is associate professor of public policy, sociology, statistics, and data science at the University of Texas, Austin. He tweets about horse racing at @equinometrics.