Olympic analytics: Data scientists crunching numbers for Canada's athletes

CALGARY — A group of credit risk assessors predicted Canadian swimmer Penny Oleksiak would win an Olympic medal in the women's 100-metre butterfly in Rio. 

They said Sweden's Sarah Sjostrom would win gold in world-record time and she did.

They were slightly off on the colour of Oleksiak's medal. She won silver instead of the projected bronze.

When Mark Merritt isn't doing credit risk assessment for Canadian Tire, he and the half-dozen people on his team are analyzing data and building projection models for Canadian athletes.

"We did forecast — and I'm going to boast here for a minute — we did forecast that Penny was going to win medals for Canada at the Olympic Games," he told The Canadian Press from Oakville, Ont.

"We predicted that Andre De Grasse was going to win medals. We didn't think he was going to win three, nor we did we think Penny was going to win four, but we predicted they were going to win medals.

"We forecasted our gold medal in high jumping. We predicted four world records that actually happened at the Games and one was right down to the one hundredth of a second."

When the women's pursuit team won bronze in track cycling, Merritt says he high-fived his wife and his child "because we predicted that."

Analytics, that buzzword of baseball and hockey, was heavily used for the first time by Canada's team in Rio.

Canada lagging behind other countries in high-performance sport data analysis was a topic of discussion at an Own The Podium board meeting a few years ago. A Canadian Tire senior executive sits on the OTP board.

The Canadian team's sponsorship agreement with the national retail chain gave athletes access to brains that predict consumer behaviour for the company's financial services arm.

"If you believe in the concept that past performance will predict future behaviour, that's exactly what we're doing," Merritt said. "We're using past performance of athletes and overlaying it to see where they are versus other world-class athletes in Canada and around the world.

"We went through many of the sports that took place at the Olympics and we forecasted who we believed were going to win the gold, the silver, the bronze, what time it took or what point value it was going to take to win that position."

In judged sports like diving, gymnastics and figure skating, analytics can calculate what combination of degree of difficulty and execution is required to win a medal.

"Many nations have in-house departments that are devoted to analytics," OTP chief executive officer Anne Merklinger said. "All the big countries do.

"If we weren't doing this, we wouldn't even be in the game."

Canada won 22 medals in Rio — four gold, three silver and 15 bronze — to match its best performance at a non-boycotted Summer Games.

In sport, there are always outliers. A sprinter catching a foot on a hurdle, for example, is something analytics can't predict, said Merritt.

"The example I can give you is (Canadian pole vaulter) Shawn Barber," he said.

"We predicted Shawn Barber was going to win a gold medal. He was on pace to do it. He was by far the class of the field, but then it ended up pouring rain during the pole vault and he just simply didn't perform. It was absolutely impossible for us to predict that."

The greater impact of data analysis on Canadian athletes — although not as sexy as medal predictions — is the ability to track which athletes are on a podium pathway very early in their careers.

"What we are trying to figure out statistically . . . are our athletes performing at the same calibre as other world-class athletes around the world and are they progressing to the point where they could potentially be on a podium in 2018, 2020, 2024?" Merritt explained.

"We have probably one of the largest Olympic sport databases anywhere, which is kind of cool."

Own The Podium directs taxpayer funding to sports federations based on the medal potential of their athletes. Data analytics is now a factor in those decisions.

"It's all about providing evidence for the recommendations we make around investments," Merklinger said. "Data shows us what the likelihood of a medal is."

Merritt's team is made up of people with advanced degrees in mathematics, statistics, econometrics and operations research. They're now building projection models for the 2018 Winter Games.

"We didn't really understand the enormity and the impact we could have until we really got into the data and saw some of the things we could deliver that some Olympic coaches could never have access too," he said. "It's sexy to be a mathematician right now. If people only knew how fun mathematics was."