Kevin Lowe has seen it all with the Edmonton Oilers.
The club’s first-ever draft pick dressed as a player for more than 1,000 regular-season games with the team, and in the years since his retirement he’s served as a coach, a manager and now an executive with an amorphous role in the affairs of the club. A first-hand witness to five Stanley Cup championships, he’s also been there for the lows, watching as a Steve Tambellini–led rebuild started, floundered and ultimately failed.
Now Lowe is seeing something genuinely new: the adoption of hockey analytics as a tool for making better decisions.
“We addressed the new-wave stats coverage of the game, the analytics side,” Lowe told Edmonton radio host Bob Stauffer in August, summing up the off-season. “You can’t have your head buried in sand.”
More NHL on Sportsnet:
Subscribe: Rogers GameCentre Live
Rogers Hometown Hockey | Broadcast Schedule
Sportsnet Fantasy Hockey
It’s taken a few years of flirting with the numbers for the Oilers to get their heads free. The club took tentative steps in the fall of 2011, setting up an advisory group to consult on analytics and appealing directly to its fan base with a contest that opened up its statistical database in the hopes of getting valuable analysis. But those efforts were often at odds with the statements and decisions of the team.
Tambellini rarely acknowledged statistics publicly, and when he did, he showed no grasp of what was actually important. Referring to the newly acquired Jerred Smithson at the 2013 trade deadline, Tambellini noted the centre’s faceoff win percentage over a five-game span the previous year as though it were somehow relevant. The comment stood out, but only because the rare nod to the numbers was so completely misguided.
When Craig MacTavish was brought in as Tambellini’s replacement two weeks later, it became apparent the club’s attitude had shifted. Like Tambellini, the new GM mentioned faceoffs, but MacTavish’s casual reference was to the effect of a defensive-zone loss on goals against in a 30-second window—a relevant comment that showed the organization’s statistical resources didn’t begin and end with a visit to NHL.com.
Quietly that fall, Edmonton brought in stats blogger Tyler Dellow to present at the team’s coaching clinic. It was a bold move: Dellow had frequently taken the team to task for decisions made far out of sync with the best available statistical data.
But, as coach Dallas Eakins later noted to the Globe and Mail, “I’m like, ‘How can he not be highly critical of our team? We’re in 28th place.’”
Dellow’s presentation impressed Eakins enough for the coach to stay in touch with him during the year and push for his eventual hiring this August.
Eakins’s interest wasn’t an aberration, either, as the Oilers’ summer plan drew heavily on analytics. Benoit Pouliot’s performance by shot metrics such as Corsi has long surpassed his reputation as a player; he got a long-term contract.
So did Mark Fayne, the former New Jersey Devils defender who combines strong possession numbers with an extremely high Quality of Competition rating. Both additions—and that of Teddy Purcell—showed the signs of significant analytical research. Lowe admitted as much on the radio.
“Tyler Dellow wasn’t involved in our decision-making as far as free agency was concerned, but no question analytics played a big part,” he said. “We went after puck-possession guys with size.”
Player-personnel decisions—who to play, who to trade, who to sign—are an obvious place to use data. Three summers ago, right before the team started dabbling with numbers, Edmonton spent significant money on free agents Éric Bélanger and Cam Barker.
Performance metrics indicated that Barker was a terrible gamble for the team, while Dellow was critical of Bélanger’s three-year contract because, despite decent numbers, NHL aging curves suggested an imminent drop-off for the 33-year-old. For the modest price of an analytics guy in the front office, Edmonton could have avoided blowing $7.5 million and two roster spots on the ultimately ineffective duo.
The impact of analytics goes far beyond personnel decisions, however. Dellow’s work was unique in that it combined analytics and video to identify the effectiveness of given strategies.
It’s widely known that the data suggests the best tactics are those with an emphasis on puck possession (e.g., carrying the puck into the offensive zone rather than simply dumping it in), but what Dellow brings to Edmonton is the ability to drill beyond the general and look at specific situations where tactical problems are costing the team.
An example would be the 30 seconds following a defensive-zone draw. In 2013–14, the Oilers sported one of the league’s worst shot differentials in that time frame, even as other objectively bad teams found ways to excel in that aspect of the game.
After using data to identify those successful teams, the next step is to turn to game tape to identify why those teams were effective, with the aim of mimicking their success. Findings can be implemented almost immediately, creating improvement without personnel changes.
Certainly the Oilers hope their new focus on the data can improve the on-ice product in Edmonton after five straight seasons of sub-.500 hockey, but for an outsider it may be difficult to spot the influence. As Lowe acknowledged, analytics is just one of several tools at the team’s disposal and won’t always drive choices.
Case in point, the team signed Keith Aulie this summer, a player who fares poorly by the numbers but is a known quantity to professional scout Duane Sutter from their time together in Calgary.
Even ideas that harmonize with the data may come from other sources. The Florida Panthers are NHL bottom-feeders who have done good work right after defensive zone draws. New Oilers assistant coach Craig Ramsay happened to work in the Sunshine State last season and doubtless will bring his own ideas from the Panthers (and previous stops) to the mix.
None of this minimizes the importance of analytics as part of the process, though. Edmonton has been prone to avoidable mistakes during its post-2006 irrelevance—mistakes that in many cases a commitment to working more closely with data would have prevented. A data-driven culture doesn’t just serve as a restraint on other approaches; it actively fosters improvement.
NHL teams don’t turn on a dime, but if the Oilers are as serious about this as they seem to be, they should see some immediate results.
“We’re hoping [the analytics] lead us to be a much more successful team,” Lowe said in August.
When? “This year.”