After changing the way we analyze basketball, Ken Pomeroy turns to curling

But what do you do if you’re a college basketball analytics trailblazer who suddenly is left without any college basketball to analyze? If you’re Ken Pomeroy, you turn to curling.

Yes, curling, the ice-based sport that attracts a whiff of amused attention every four years at the Winter Olympics but then retreats to its status as a niche sport played in some of the world’s colder climates. Pomeroy is attempting to statistically analyze curling the way he does college basketball, posting the results at DoubleTakeout.com, where he also blogs about the sport.

In a telephone interview, Pomeroy said his interest was piqued after he actually took up the sport in Utah, where he lives.

“I was playing for a few years and I guess I kind of figured out it was a good middle-aged-person’s sport, a nice outlet for those competitive juices,” he said. “Obviously my basketball days are behind me. I picked up hockey a few years ago and that went about as well on my body as basketball did. The place where I played hockey they had a curling league so I discovered curling through there.”

Pomeroy, a meteorologist by trade, began his college basketball ratings earlier this century as a way to more properly interpret the sport and predict results. His statistics measure how teams perform on each possession — offense and defense — rather than over an entire game, where numbers can be skewed by teams that operate at faster or slower paces. Fans, sportswriters and coaches all rely on Pomeroy’s numbers, and point spreads issued by oddsmakers often closely align with the score predictions at KenPom.com, which offers some free information but charges a subscription fee for more in-depth analysis.

Pomeroy began his curling blog last summer, after the end of the 2020 college basketball season was erased by the pandemic. He is taking a pretty basic approach in his attempt to predict outcomes and rank the world’s best teams, both men’s and women’s.

“The curling ratings are a little different [than the basketball ratings] in that they don’t look at scoring margin or anything like that,” he said. “So it just looks at who a team’s beaten and who a team’s lost to and kind of evaluates their record based on who they played. … It sounds a little like the RPI [a now-discredited metric once employed by the NCAA to evaluate teams] but it’s not computed in that way at all. It’s a very fair way to handle results and strength of schedule.”

Though professional curling has been curtailed by the pandemic, just like other sports, the Canadian women’s championship — called the Scotties Tournament of Hearts — is taking place this week. Before the tournament began, Pomeroy pegged the teams captained by Kerri Einarson and Rachel Homan as the favorites to advance to the final round, and indeed, those two teams already have qualified for the second round of round-robin play. But Pomeroy also gave a younger team captained by Mackenzie Zacharias a 13.4 percent chance of making the playoffs (the sixth-best chance out of 18 teams), and her team already has been eliminated after winning only two of its first seven games.

“Maybe I’m overrating junior teams a little bit, maybe I need to look into that,” Pomeroy said. “The goal every year is to hopefully make this a little better, a little more useful, have some features that will help people understand world curling better.

“The top two teams are pretty easy to call. After that, it got a little interesting. In the end, three teams make the playoffs and I think you’re going to get three of the top five teams on my list make the playoffs. But some of those teams a little further down are a little more challenging [to predict] with the pandemic and the lack of playing and practice for a full season that’s maybe causing a bit more uncertainty.”

Another challenge for Pomeroy is that rosters for the four-person club teams are often in a state of flux: Pregnancies and job-related absences can scramble lineups.

“That was a big challenge, trying to figure out what I consider a team on those cases, like how many absences can you have and how to rank those teams and writing the code to do that,” Pomeroy said. “There’s probably still more to do on that.

As for his basketball ratings, Pomeroy said there’s “very little human intervention needed” at this point in their evolution, though he did note that his predictions have been a little less accurate this season because of pandemic-related issues (fewer games, no fans lessening home-court advantage, rosters changing because of positive tests, etc.). And once basketball season ends, he hopes to be able to devote more time to his curling metrics, particularly with the spotlight of the 2022 Beijing Winter Olympics less than a year away.

“There is a similarity [between the two ratings systems] in that they’re kind of predictive; I’m trying to predict the future, trying to handle trends and teams,” he said. “You can make predictions in terms of, ‘Hey, this is one team’s chances of winning against another team.’ That was the goal for me because when I’m watching a sport that is definitely something I’m interested in knowing before I see two teams compete.”