The dawn of “advanced statistics” in hockey is thought to have occurred more than a decade ago, when progressive NHL front-office types and avid fans took baseball’s Moneyball approach and applied it to pucks.
Jim Corsi’s approach has gained the most traction. Now the goaltending coach of the St. Louis Blues, Corsi developed a formula that measures every player by the number of shots (on goal, off-target and blocked) vs. the number of shots against while that player is on the ice. The result: the Corsi number.
So-called hockey analytics have not always been welcomed by the old guard. Many still bristle when a number other than goals, assists and points is used as the measuring stick.
But the new metrics are not going away.
“You’re always looking for more information,” Blue Jackets general manager Jarmo Kekalainen said. “And you’re always trying to keep up with what other clubs are doing, make sure that you’re the team with the advantage.
“It’s something we’ve looked at more and more. It’s something we talk about now whenever we’re looking at players, not just within our organization but players who are free agents or trades that we may have discussed.”
Three years ago, only four NHL clubs had a staff member dedicated to analytics. Now, 13 clubs list a front-office member who deals with advanced statistics. The Blue Jackets are one of those 13. Josh Flynn was hired by previous general manager Scott Howson but has remained on staff to handle salary-cap issues and analytics.
“I’m spending more time paying attention to analytics than I used to,” Flynn said.
On Tuesday, the Toronto Maple Leafs hired 28-year-old Kyle Dubas as assistant general manager. He is considered a progressive follower of Corsi’s. As for the other 17 clubs, some pay outside firms for statistical analysis, some refuse to discuss their level of interest publicly, and two — Colorado and Detroit — have shown little interest.
Several clubs have made it a priority this summer. When the Pittsburgh Penguins fired Ray Shero as general manager this spring, team ownership cited a desire to be more engaged in hockey analytics as one of the reasons. The cash-strapped Arizona Coyotes have added money to their budget to address the issue.
Even the slow-to-change New Jersey Devils, under general manager Lou Lamoriello, have taken the plunge.
“I’ve found it easier to find teams who are interested in talking to me,” said Eric Tulsty, a San Francisco-area chemist and math whiz who has worked for several NHL clubs, including the Blue Jackets. “Teams that wouldn’t even listen a few years ago are now actively seeking help.”
The basic Corsi premise is that possession of the puck is tantamount to winning hockey, and that shots on goal are representative of possession.
“Often, those two points are true,” Flynn said. “But those aren’t universal truths in hockey. Some teams dump and chase by design.”
Corsi does not figure favorably for a fourth-line player who dumps the puck in deep and forechecks. It does not give any benefit to defensemen who play heavy minutes against the other team’s best players, when simply keeping them off the score sheet is deemed a success.
For instance, Blue Jackets defenseman Jack Johnson is a frequent target of the hockey analytics crowd. His 47.9 Corsi rating last season — meaning the Jackets took 47.9 percent of the shots when he was on the ice — was 13th on the Jackets and 317th in the league.
“Give me a break with that,” Kekalainen said.
Another issue is that the Corsi ratings are produced from the play-by-play sheets provided by the NHL, as recorded by off-ice officials.
“There’s a pretty clear definition of a shot, a missed shot and a blocked shot,” Tulsty said. “But it’s not an exact science, and the people working at certain rinks have different definitions for shots and other statistics.”
Or, as Flynn put it: “It can be pretty crude.”
Almost every NHL club is working with Corsi numbers and trying to find ways to manipulate them into a more accurate formula. The hope is that further advancement will render irrefutable statistic figures.
The NHL might get involved, too. League officials have met with at least two companies that would install high-definition cameras in every rink, similar to what the NBA has done the past three seasons. The cameras would monitor every foot of the ice, taking 25 pictures per second.
“It would measure where players are positioned when their team scores, where the other team scores, how fast a player is moving, how much better a player would have gone if a player was one foot to the left,” Tulsty said. “This is where it’s headed.”
One of those companies, PowerScout, has worked with the Blue Jackets on a limited basis the last few years.
“We’re never going to do away with scouts,” Blue Jackets president of hockey operations John Davidson said, “because you measure the player in so many different ways, and that’s never going to change. But you have to keep your eyes open. This league, this game … it’s always changing.”
The Corsi method
A system developed by Jim Corsi, a former goaltender who now coaches for St. Louis, measures how many shots on goal, missed shots and blocked shots are generated vs. how many shots on goal, missed shots and blocked shots are allowed when a player is on the ice.
Thus, a rating of better than 50 percent means the player’s club has the puck and shoots more often when he’s on the ice. A rating below 50 percent indicates that the opposition has the puck and shoots more often when a certain player is on the ice.
According to ExtraSkater.com, Boston’s Patrice Bergeron was tops in the NHL last season with a 61.2 percent Corsi rating, and Edmonton’s Luke Gazdic was last at 36.8 percent.
Among Blue Jackets players, defenseman James Wisniewski (54.1 percent) had the best rating and right wing R.J. Umberger — since traded to Philadelphia — was last at 46.4.