Who will win Canada's Olympic curling trials? A basketball stats guru might have the answer
College hoops stats pioneer has a curling model too — with some surprises
This is an excerpt from The Buzzer, which is CBC Sports' daily email newsletter. Stay up to speed on what's happening in sports by subscribing here.
In the ever-expanding world of sports analytics, Ken Pomeroy is one of the most revered pioneers. Back in late 2003/early 2004 (just months after Moneyball was published), Ken devised a statistical model for rating teams in U.S. college basketball — a sport he liked watching in his spare time from his job as a meteorologist. These ratings, published on the website kenpom.com, often differed from the conventional wisdom about some teams. Hardcore hoops fans and gamblers devoured this new info. Then the media and even some coaches caught on, seeing in Ken's numbers a way to better understand what was really happening on the court. Today, it's safe to say that anyone who follows the sport — or even puts a little effort into their March Madness bracket — has been enlightened by Ken's work in some way. You might call him the Bill James of college basketball.
And it turns out, Ken is also really into curling. He started playing a few years ago at home in Salt Lake City and wanted to figure out how to get better. But there aren't many curling experts in Utah ("It's a bunch of people who took the sport up later in life," he says) so he began watching games on TV and online to try to glean best practices from the world's top players. This led him to wonder, just like he did with college basketball all those years ago, what might be hiding behind the final scores. Which teams were really the best?
So Ken did what Ken does: create a statistical model that rates the world's top teams and forecasts who will win when they face each other. This one is simpler than his college basketball system, which accounts for myriad factors such as scoring differential, strength of schedule, pace of play and even luck. "Basically, [the curling model] just looks at wins and losses. It doesn't look at how a game was won," Ken explains. "And, naturally, it takes into account how good your opponent is, so you get more credit for beating good teams."
One of those big events — maybe the biggest in terms of prestige and sheer quality of competition — is the Canadian Olympic curling trials, which begin Saturday in Saskatoon and will decide who represents Canada in the men's and women's events in Beijing this February. I talked to Ken about what his model forecasts for the trials, and here are the most interesting takeaways mixed with some insight from him:
Tracy Fleury is the slight favourite to win the women's tournament.
This is somewhat unexpected. Yes, Fleury's Manitoba-based rink holds a massive lead in Curling Canada's official rankings, and they're on a roll this season, winning one of the two Grand Slams held so far and reaching the final of the other. But most fans probably expect Kerri Einarson and Rachel Homan to end up playing for the Olympic spot after they met in the last two Scotties finals (Einarson won them both). Ken's model is pretty high on them too — they're essentially tied for second in probability of winning the title at about 23 per cent each. But Fleury, who's never made it to a Scotties final, has the edge: her team is at 30.4 per cent to win the trials.
Ken calls Fleury's rink "the undisputed best team in the country this season." They're 30-5, including a sparkling 9-3 against opponents ranked in the top 10. This compares quite favourably with both Einarson (a stunning 0-9 against top-10 teams) and Homan (3-3). Those two have been their usual dominant selves against lesser competition, but there won't be many easy marks at the trials.
A possible red flag for Fleury is that she's started with the hammer in an unusually high 28 of her 35 games, which might mean her rating is a bit inflated. "It's pretty unsustainable to win hammer 80 per cent of the time," Ken says. "So that might bring her down to earth a little bit when we get into the trials."
The women's event looks like a three-team race — with maybe one dark horse.
After Homan and Einarson, there's a pretty big drop-off to the No. 4 skip in win probability: Jennifer Jones, who has a 9.5 per cent chance to take the title. The 2014 Olympic gold medallist and six-time Scotties champion is, at 47, on the downslope of her career. But she showed she's still dangerous by reaching the final of the Slam season opener, which she lost to Fleury. "There are still signs that she can get hot and play with really good teams," Ken says.
None of the other five skips in the field have a better than 3.4 per cent chance of winning the title. In this tier are Laura Walker, Krista McCarville, Casey Scheidegger, Kelsey Rocque and Jacqueline Harrison. If you're hunting for a long shot, Ken points to McCarville and Scheidegger because they're top-10-calibre talents who haven't played as much in the last couple of years, which has cost them in the ratings.
Brad Gushue is the clear favourite to win the men's tournament — but there are more potential spoilers than in the women's.
No big surprise here. The 2006 Olympic gold medallist and three-time Brier champ tops the official Canadian men's rankings and seems to be peaking at the right time. His St. John's-based team won the last Slam before the trials. Ken's model gives Gushue a 37.1 per cent chance of getting back to the Olympics. "He just does not have bad games," Ken says. "He hasn't lost to a team outside the top 10 on Canadian soil since November 2019."
Ken calls Gushue's rink "clearly the best team in Canada," and Brad Jacobs' the clear No. 2. The latter has a 21 per cent chance of winning the trials. "It would be pretty stunning if one of those two didn't win it. That said, you add up their probabilities and it's only a 58 per cent chance. So there's definitely room for the next tier."
That group features some big names. Kevin Koe, the 2017 trials winner and four-time Brier champion, has a 13.1 per cent shot. Brendan Bottcher, who's made it to four straight Brier finals and finally won one last season, is at 11.7. Matt Dunstone (8.0) and Mike McEwen (5.5) are also threats. The rest of the field — John Epping, Jason Gunnlaugson and Tanner Horgan — are all at 1.3 per cent or lower. While it would be a surprise if anyone from the Koe/Bottcher/Dunstone/McEwen tier wins the trials, Ken says they're "all capable of looking like the best team in Canada over any given week."
He's sticking with his model in the men's event and taking Gushue to win his first Olympic berth in 16 years. On the women's side, he's going against the system a bit and picking Homan to earn her second consecutive trip to the Games.
You can read Ken's analysis of the Canadian trials here. He'll also be appearing on the season premiere of That Curling Show to preview the trials with hosts Devin Heroux and Colleen Jones. Watch it on the CBC Sports YouTube channel on Friday at 7 p.m. ET.