Ohio State student puts a spotlight on college hockey analytics

Screenshot of the HockeyU homepage. Credit: Lantern File Photo

When Nayan Patel first came to Ohio State, he was studying engineering, had a knack for statistics and a love for college hockey.  

Although his academic path may have changed, his enjoyment for statistics and college hockey has not.

Putting those two interests together, Patel, a fourth-year in economics, dove into the sports data analytics world, and HockeyU Analytics — Patel’s website which showcases his own statistical model, national rankings, weekly predictions and player ratings — was born.

“I’m a big fan of college hockey, I grew up going to OSU hockey games,” Patel said. “So, I thought no one is really doing anything specifically for college hockey, so I thought this is the perfect place for me to start and kind of play around with stuff and see what’s out there.”

Patel launched HockeyU last year along with his statistical model, “Oliver.” Built with data from the previous five seasons and with inspiration from Ken Pomeroy’s college basketball model, Oliver is designed to be predictive.

While most human polls are a reflection of a team’s resume, Patel’s model aims to predict what a team will do going forward. When the 2019-20 hockey season was canceled due to COVID-19, the Ohio State men’s hockey team was ranked No. 10 in the United States College Hockey Online top 20 — eight spots higher than where Patel’s model ranked the Buckeyes. 

“What hurt Ohio State was just their scoring,” Patel said. “They didn’t score as much as some of the other teams. Their defense was definitely top 10 for most of the season, but they just weren’t generating the same amount of offense.” 

In order to evaluate teams using the limited statistics tracked at the collegiate level, Patel centered his model around team offensive and defensive ratings. 

“So, basically, what I did was calculate offensive and defensive ratings based off of shot attempts,” Patel said. “Basically, it takes a similar concept from basketball where shot attempts in hockey correlate to possession.”

Screenshot of the game-by-game predictions for Ohio State Men’s Ice Hockey. Credit: Lantern File Photo

The creation of Patel’s model has not been without difficulties — chief among them the access to statistics at the collegiate hockey level. Metrics such as player ice time and shot tracking, which Patel said are crucial to NHL models, are not available to him. 

The lack of publicly available data is not a problem that the NHL is immune from, however. Alison Lukan, a hockey analytics writer covering the Blue Jackets and Ohio State hockey, said that the sport as a whole is lagging behind others when it comes to publicly available data. 

“We in the sports analytics community look to sports like baseball, even now football and European football, which we, of course, call soccer, as being leaps and bounds ahead of what hockey can do and interpret,” Lukan said. “There’s just more acceptance, also, so it’s about not just what data is available but also the overall acceptance level.”

For those working with analytics outside of a team’s front office, Lukan said that the relevance of analytics comes from wanting to ask better questions and be more advanced like those other sports.

“The additional data tells us more about what’s really happening on the ice and it tells us more about the players and how they’re playing. It answers questions like, ‘Who is getting the puck into the offensive zone? Who is generating scoring chances?’” Lukan said. “Those ideas are more compelling to me than just who took this many shots.”

At the collegiate level, Lukan said that changes to data collection and availability would allow for better team-to-team comparison, which she said is “ultimately what’s the most valuable.” 

Patel’s model accurately predicted 67 percent of head-to-head matchups by the end of February, which Patel said was partly due to the bigger discrepancies between teams at the collegiate level, but it was encouraging nonetheless. 

“Nayan’s work, I would also just commend it,” Lukan said. “He is one of the people out there doing what’s really, really hard work to try to bring college hockey into the analytical space, and he should be tremendously commended for all that he’s done so far.”

Going forward, Patel said that he would like to continue the work he does with HockeyU and bring more attention to men’s and women’s college hockey rather than focus solely on the professional game. 

“I think it’s just honestly best to follow in the work they do and try to apply it and come up with new ideas for college hockey, which is a place a lot of people kind of ignore,” Patel said.