Ventura goes to Penguins
In the spring, Carnegie Mellon Ph.D. and junior faculty member Sam Ventura took a consulting job with the Pittsburgh Penguins. His path there is exemplary as a Carnegie Mellon success story. He started as a hockey fan and eventually, through working with former Carnegie Mellon statistics professor Andrew Thomas, turned that interest into being employed at the highest level of the sport.
Ventura’s rise to the Penguins began with hockey statistics site War-on-ice, a site he created with Thomas. “Andrew and I first started talking about hockey when he taught [a] sports statistics class,” said Ventura, referring to a mini course offered by Thomas in Ventura’s senior year.
Sports statistics have a long history, but until recently, as Ventura pointed out, many methods are not sufficient to tell the story of what occurs on the ice. “We were talking about how existing methods to rate players offensively and especially defensively were lacking.”
From those conversations emerged War-on-ice, which became the most well-known and comprehensive database for hockey statistics on the web.
Ventura and Thomas started by downloading the NHL’s play-by-play files and logging not just goals and assists but also things like takeaways, blocked shots, and the location of the puck during each of the events.
Through all of the information, the two created “a set of metrics that would provide a statistical rating for players offensively and defensively,” according to Ventura. That statistical rating became what many believe is hockey’s most powerful tool for evaluating players.
The metric measures the change in a team’s goal scoring rate when a player is on the ice for offensive rating. For defensive rating, it measures the rate of goals conceded by a player’s team when that player is on the ice. It controls for factors like “score state… and also which players they’re on the ice with…and which players they are playing against.” The paper describing the method was published in The Annals of Applied Statistics.
The site’s expansion began when Thomas organized a hockey analytics panel at the Joint Statistical Meetings in 2014. Michael Schuckers, Brian McDonald, Kevin Mongeon, and Mike Lopez — some of the foremost minds in hockey analytics — joined Ventura, and Thomas discussed hockey analytics on the panel. “There was a lot of question and answer from the audience and the question that eventually came up was, ‘If what you’re doing is so great… then why does nobody use it?’ … and we realized we weren’t doing a very good job of marketing what we had done,” Ventura said.
At the same time — maybe even the same day, according to Ventura’s memory — the person who ran the leading hockey analytics website, Extra Skater, was hired for a consulting job. “Basically in a weekend,” War-on-ice was born. Since most of the work was done in R, it could be easily put onto the web.
The site blew up fast. “We got a thousand Twitter followers without having sent a tweet out yet,” Ventura said.
The site was also interactive with the community. They responded quickly to people’s requests, kept the site up to date, and even posted metrics that followers may have thought up. “That sort of got us a lot of trust within the online community, which I think played a big role in popularizing the website,” he said.
Ventura and Thomas built the website to survive after they had moved on. War-on-ice is still being run despite both statisticians now working as consultants for teams, making the website a conflict of interest.
Ventura also acts as the faculty advisor for Tartan Sports Analytics, a club on campus for students with an interest in sports analytics. “They all have great ways of thinking about the different sports and they’re all pretty talented from the statistics perspective as well, which is hard to combine,” Ventura said about the articles the club publishes.
Ventura is currently an assistant coach for the club ice hockey team, and he is trying to recreate War-on-ice for their league. However, getting a consulting job for a professional hockey team is a strong affirmation of how important Ventura’s thoughts have been to the world of hockey analytics.