Curlytics: The analytics revolution sweeps into curling

Phill Drobnick is a coach for Team USA Curling, but not the hardcore numbers-crunching type. Yet in the after-hours of the 2018 PyeongChang Olympics, after the ice and arena cleared of action, he was searching for any efficiency Team Shuster could use to win.

Drobnick and team alternate Joe Polo threw rocks down the empty sheets of ice — timing them, measuring the curl — looking for the most accurate and deficient stones in the bunch. When team skip John Shuster needed to make a shot, he wanted the best rocks, and they were taking the guesswork out of it.

At a big event such as the Olympics and the World Championships, Drobnick is one of three coaches for Team USA. He compiles scouting reports and strategies, finding three to five key areas for the team to focus on in a match.

Another coach is scouting rocks. If one looks slow or seems like it turns too much, it receives special attention at night when Drobnick and Polo are testing. If a certain rock isn't hitting high accuracy percentages on the draw, it too gets a look.

The concept is simple: When Shuster is taking the final two shots of an end, with points on the line, make sure the most accurate rocks in the team's color are in his hands based on what analogies and the numbers tell them.

"There's not a lot of people that talk about that, the analytics of the stones," Drobnick said.

Curling is a sport ripe for analytics.

It's a turn-based game and strategy is dictated almost entirely by the decisions made the previous turn. For a statistician, that means there's value in understanding the odds behind every critical decision. What's the real difference between a four-point shot and a one-point shot? What happens if a team misses that four-point shot and gives a point away late in the match?

Those are all questions Gerry Geurts recognized more than a decade ago. A baseball fan growing up, he was captivated by the numbers side of the sport. He had just graduated with a computer systems technology degree from Mohawk College in Ontario when American journalist Michael Lewis published "Moneyball: The Art of Winning an Unfair Game" in 2003, considered the seminal book on baseball's first analytical revolution.

Lewis had followed Oakland Athletics General Manager Billy Beane, who after losing three of his biggest stars in free agency when on to win 103 games in 2002 with a low-budget team. He did it largely through analyzing players using on-base percentage and devaluing traditional statistics like home runs, runs batted in and stolen bases that drew larger contracts and more attention.

For Geurts, the connection to statistics and curling was evident. He was already working in the sport as a data collector and president of CurlingZone, building a live scoring system used by the World Curling Tour.

Breaking down the odds was a natural step.

"Early on, I knew there was value behind it," he said. "How do I evolve some of these concepts into curling? It's the perfect analytical sport."

On the ice, teams are now using analytics to study themselves and their opponents, looking for strengths, weaknesses and tendencies to try and gain the advantage. They study shooting percentages, aggressiveness and situational play and ask how one team is better in a certain scenarios than others?

It gets down to the nitty-gritty at times.

"If a certain player is throwing one turn better than the other, we'll force them into a shot on his weaker turn," Drobnick said. "Especially on a skip. If a skip is throwing 90 percent on an in-turn draw, but 60 on an out-turn draw, we'll really try to force him into those types of shots."

Just as importantly, he adds, is understanding how the stats relate to the basic concepts of the game.

In a tie game, the percentage to win is dead even at 50-50. A control position means a team has up to a 65 percent chance to win. A dominant control position is more than 65 percent. Scoring in the first end, and the eighth end and beyond, can dramatically shift the odds.

That makes using the stats, especially late in the game, a key piece to calling shots in a meaningful way.

A team that leads by one point after the eighth end has a 60-40 chance to win the match. So, when faced with the question of going for one point or going with four points, there's a 10 percent swing from the tied match scenario in scoring just one point. Four points bump the odds to about 90-10, but missing the four-point shot and giving away one point drives it down to a 40-60 chance to lose.

"Before, it used to be a gut feeling," Drobnick said. "Even if we give up a point, it's no big deal. Now, do we risk those 20 percentage points for a shot we might only make one out of five times?"

Adoption of statistical analytics has been slow to start. Many of the strategies used in 2018 were pushed by Geurts heading into the 2010 Olympics in Vancouver. It wasn't a surprise though. The "Moneyball" concepts of Beane took a decade to fully-assimilate into baseball.

Players and coaches implementing more analysis into their game are adapting and playing out the scenarios in real time on the ice, Drobnick said. Their level of use varies team by team.

Team Shuster is focused more on the rocks and scouting reports, but the coach points to the late round robin and medal round matches in the 2018 Olympics, when more conservative play led to lower-scoring, as evidence of the team's of approach to playing the odds.

They contrast to Team Ruohonen, skipped by Brooklyn Park's Rich Ruohonen, which represented Team USA in the 2018 World Men's Championship, and utilized the numbers in every game.

Geurts now consults with 20 teams around the world, including the 2018 women's gold-medal winning Swedish team. His early work on the data defied the sport's conventional wisdom, showing it's better to be up a point with the other team throwing the hammer, than down a point with the hammer.

In short: less control over the final shot is better.

"It's taken a while to find the right athletes to buy into it," he said. "Just like anything, it's human nature to resist change and adopt to new ideas. That was the greatest lesson I learned from Moneyball, patience. They wrote the manual on how to do it, what they were looking at and why it was important — and it took a decade."

Along with partners Kevin Palmer and Jason Gunnlaugson, Geurts is continuing a deep dive into the numbers and the sport. He isn't ruling out a full spectrum of research similar to baseball's analytical advancements, which measure everything from the spin rate on pitches to the launch angle on batted balls to the efficiency of catchers receiving pitches (thus creating more strikes).

Canadian researchers are currently putting sensors on stones to track speed and spin rates for more data to process. Geurts considers their current work the 2.0 version. Analytics 3.0 is being tested and refined, and a fourth generation of concepts are in the works.

The biggest challenge moving forward, beyond adoption, is turning the data and concepts into tangible strategies for the players once the research is complete.

"The sky's the limit, to be honest, when you consider what's going on right now," Geurts said. "The biggest thing is getting it put into play. Too much is paralysis by analysis. If you go to a skip with a big book of info, it's going to be too much, so we have to look at how we involve and adapt this stuff into the game."

How will curling analytics change the game? That's a question still to be settled.

Teams like Shuster, an offensive juggernaut on the world stage, still play their game. Drobnick said using the stats give mid-level teams a potential edge against an elite team, but it still boils down to making the shots.

Geurts said the balance of entertainment and competitiveness rests on the leagues. As baseball considers rule changes to combat specialized pitchers and defensive shifts, both the result of advanced statistics, curling has adopted the five rock rule to keep more stones and more action in play.

"It can hurt the game," Geurts said, referencing changes in baseball. "When you're an athlete, you're going out there and going out there to win. There's a difference in sport as entertainment and sports as a pursuit. The sport and management have to find ways to continue to make the game entertaining. That continues to evolve in curling."