Predictive analytics and big data have been used in sports for well over a decade. Using advanced statistical analysis allows professional sports franchises to better identify talent and predict future player performance.
In 2002, the trend entered the mainstream when Billy Beane, the Oakland A’s general manager, used it to create a cost-effective baseball roster. He used the data to predict that a player’s on-base percentage was more important for team success than traditional stats such as their batting averages or home runs.
Beanes strategy was successful and led to sustained success by the As, a small-market team with a payroll exponentially smaller than many of their competitors such as the New York Yankees. Since Beanes Moneyball era, predictive analytics and big data have taken over the sports world.
Beyond baseball, another sport where such tools are especially prominent is horse racing. Data mining, statistics, modeling, machine learning, and a variety of other tools are used by owners, trainers, race course management, bookies, and bettors alike.
Equibase and the Origins of Big Data in Horse Racing
In 1990, the Jockey Club and the Thoroughbred racing associations collaborated to form Equibase as means to collect and share thoroughbred data. In the two and a half decades since its inception, Equibase has collected roughly 1.7 million races and over 14 million starts. In total, these races represent well over one billion individual data elements that have been collected, quality controlled, and made available to industry partners and racing fans.
In 2010, STATS LLC partnered with Equibase in an effort to better unlock the insights available in this vast trove of data. During a 2015 global symposium on big data in horse racing, Jim Corelis, Senior VP of Product Development for STATS LLC, emphasized the current emphasis on data science in horse racing:
We have a data science team. As we’ve gotten more big data across all the sports, we’ve had to employ more folks that aren’t just doing research and just digging into the data, but data science teams that are creating algorithms and using machine learning to really dig in and pull the insights out, create predictive analytics and other insights. We’re setting those guys loose on the horse racing data to create more win probabilities and things to help the end users.
Breaking Down Bio-metrics for a Safer Sport
Among the wave of wearable technology products to debut over the last several years, equine fitness monitoring devices such as the E-Trakka system have captured tens of thousands of biometric readings for horse racing.
These advanced readings, obtained via a custom saddle blanket equipped with built-in GPS and a heart-rate monitor, offer insights into ability, fitness, lameness and suitable race distances. As the volume of data on equine performance during races grows, the ability to use predictive analytics to better understand that factors that influence race horse performance will improve as well.
Additionally, more data on horse biometric readings, combined with modeling and analysis of track conditions, diet, weather, and other factors will enable trainers to more accurately identify and scratch horses at a high risk of injury prior to starting a race.
The Financial Impact of Predictive Analytics
Analytics-backed betting is commonplace among serious horse racing bettors. Since the success of predictive analytics relies on looking at the right kinds of data, its not always a perfect system. For those who know what to look at however, theres a wealth of data that can be used in horse racing to better predict outcomes.
With so much data publicly available from sources like Equibasess Stats Central, savvy bettors break down factors such as:
- The success of the jockey riding the horse.
- The diet of the horse.
- The mood of the horse.
- The weather during races.
- The length of the track and/or race and if that helps or hurts a horses chance of winning.
- The racing pattern of the horse in relation to winning percentage for specific lengths of races.
During the 2015 symposium on big data in horse racing referenced earlier, the panel moderator Dick Powell observed the challenge individual bettors face when trying to break down all this data without the aid of a computer and an algorithm.
The beauty of some of these programs is the programs can have enough data that they’re able to analyze and they’re able to do all these things, but with myself, there’s a real fear of over-handicapping and overanalyzing, having too much data 2015 Global Symposium on Racing & Gaming Page 28 that can cloud your judgment, cuz as Dan pointed out with the art and science, there is an art. The computer can just rely on science, but as a handicapper, I have to rely on some art, and if I get down to A and B, who do I pick?
Early adopters using predictive analytics tools in horse racing betting had a significant advantage over regular players in the pari-mutuel system. However, as industry organizations such as Equibase work to make that information more accessible to the general public, the scales begin to balance back out.
While the future of horse racing is uncertain, it is a sure thing that big data and predictive analytics will be a big factor in the development of the sport in the coming years.
