Phil Housley and his Buffalo Sabres players were at a loss for words after one of their more bizarre losses during their free fall. Analytically, they were the far superior team in a 3-1 loss to the lowly New Jersey Devils in Prudential Center on Monday night.
The Sabres had 20 more shot attempts during 5-on-5 play, and was better in nearly every other statistical category, including high-danger scoring chances. Their on-ice performance passed the eye test, yet their only goal came on Sam Reinhart's dump-in from center ice.
The game powerfully illustrated the sport's ongoing challenge of relying on numbers to evaluate performance. After all, the Sabres committed two egregious defensive errors. Still, analytics have become a pillar of every NHL organization, helping general managers construct organizations and coaches prepare players for each game.
Hockey's information age has led owners to hire based on who is well-versed in analytics and accepting of how data can create a competitive advantage. Still, those numbers have not been accepted by those who are measured.
Buffalo Sabres winger Jason Pominville, a 36-year-old with 16 years of NHL experience, gave a blank stare when broached with the topic of analytics. Neither he nor most of his teammates pay attention to shot-attempt differential or any of the other available statistics. Most players don't even know where to find the data.
They leave that up to the coaching staff and management, who use a vast amount of information – some of which is not available to the public – to determine an opponents' weaknesses or construct a lineup.
"I'm not sure what they really mean," Pominville said. "To be honest, I don't really look at them. I'm sure most guys are the same way. To me, analytics show shots but it doesn’t show battle, compete, those types of things. It doesn’t really mean much."
That stance is in stark contrast to the one held by many players around Major League Baseball, which has used analytics in some form since 1971. The innovation of Statcast, a state-of-the-art tracking technology that collects a remarkable amount of data that can be viewed in real time, has allowed Major League pitchers and positions players to discover how to enhance their play or better prepare for an opponent.
Universal acceptance did not happen overnight in baseball. There was resistance when data analysts were first involved in scouting meetings and their findings led to significant changes, such as infield shifting. Now, those employees travel with the team and have the ear of players and managers.
On the other hand, the NHL is still learning how to best use analytics. Players are not given data-driven scouting reports, and they typically have no interest in knowing their own production or usage. Each team's analytics department uses the information to help coaches game plan and identify trends to help put players in the best position to succeed. Pertinent information for an upcoming opponent is then relayed to players.
Teams can measure specific tendencies such as 5-on-5 shot attempts from the slot, or offensive-zone possession time per player. Unlike baseball, the NHL does not have tracking technology to alleviate the need for in-depth film study, and the sport's physicality adds a different element when measuring player performance.
Like traditional hockey stats, some analytics can paint an incomplete picture, as Sabres goalie Carter Hutton explained.
"I think the traditional save percentage is kind of going extinct a little bit, too," Hutton said. "You have to see games. Quality of chances are so different for each team depending on defensive structure. I think from a goalie’s standpoint you can probably dive in a little more to find out performance-wise, team-wise, (the quality of chances) given up.
"It’s the same league but teams play such a different game. I think you need to watch video of games to really have an assessment of guys."
The most well-known analytics only touch the surface of player or team performance. Much like plus-minus, Corsi, a measurement of 5-on-5 shot attempt differential, can be deceptive since the performance of one player can affect teammates on the ice. The same goes for Fenwick, which is similar to Corsi, except blocked shots are not part of the equation.
Both stats have been enhanced in recent years with adjustments that provide context. For example, a player's Corsi can be measured relative to his teammates, which can reveal who has the most positive outcome on a team's performance. There is also a way to measure statistics per 60 minutes, which accounts for ice time.
While Corsi and Fenwick are starting to become part of discourse among fans, neither takes into account shot-attempt quality, which is important when measuring a team's success during 5-on-5 play. "Expected goals" is a metric that measures unblocked shot attempts by the quality of chances – location, second chance, etc. – though it isn't necessarily indicative of future performance.
While management finds the information beneficial, most players prefer to only pay attention to data that can help prepare for an opponent.
"You try not to look at it too much," center Evan Rodrigues said. "I think it kind of just gets in your head a little bit. Media, stats, that’s more stuff to think about than going out and playing your game. For me, personally, that stuff gets in my head more than I’d like it to. I try to ignore the outside noise, all the numbers, all the things like that, and go play my game."
Whether presented in numerical form, bar graphs or heat maps, data has become increasingly available in recent years, despite NHL teams plucking many independent analysts for front office jobs.
For example, with save percentage a flawed statistic, goalies can now be measured by "goals saved above expectations," which takes the league’s average save percentage and applies it to the amount of shots a particular goalie has faced.
Though many NHL players admit that a statistic such as plus-minus can be bothersome, most don't pay mind to analytics they perceive as flawed data that can create a false narrative among fans or media who can't always pay attention to the nuances of each game.
"Personally, I think it’s looked at a little bit too much," Sabres winger Conor Sheary said. "At the end of the day, it’s about results. It’s not about Corsi or – I don’t even know the deeper analytics that people use. I think they’re good indicators, but I don’t think they impact the outcome as often as some people might think."
In addition to revealing tendencies, basic analytics can reveal player usage, such as defensive-zone starts. That information can be resourceful for in-game strategy, particularly when having the second line change at home.
Unlike their players, the Sabres' coaching staff, management and analytics department pays close attention to Corsi since 5-on-5 goal differential is indicative of overall success. Entering Monday's games, not one of the 16 teams holding a playoff spot had a negative differential, and the Sabres had success in that regard when they were sitting atop the NHL on Nov. 28.
There is not an overreliance on data. Sabres coach Phil Housley and the rest of the team's staff marry the numbers with the eye test and film study. Together, those elements help construct forward lines and defensive pairings, while helping with in-game strategy.
"There are so many different ways to go about analytics," Housley said. "There is so much information out there. We try to piece it together where we look at an individual’s Corsi rating, a team’s Corsi rating, chances for from the inner slot or on matchups, whether it’s D pairings or line matchups during games. The biggest thing for me, Corsi does tell a story. It shows what is going on at the other net compared to what is coming at our net. When you’re on the plus side, a lot of good things are going on. Sometimes during the game it doesn’t tell the whole story, but it certainly gives you a good indication where the play is at and where our team is at."
When offering a brief scouting report to reporters on an upcoming opponent, Housley will briefly cite data, such as scoring chances from the slot or 5-on-5 shot and goal differential. The latter two points often illustrate his "shot mentality" coaching philosophy. He wants his players to put pucks on the opposing net, while creating traffic in front of the other goalie.
"You marry it with what you see with your own eyes, you marry it with the film," Housley said. "Certainly you look at those analytics after the game with matchups, zone starts and all of those things that are indicators. If we’re on the plus side that means we’re doing some good things well. ... When it’s coming back and you’re below, there are a lot of things coming at you. We take analytics very seriously."
The conversation about data grew louder when NHL Commissioner Gary Bettman announced in January the league planned to deploy puck and player tracking technology next season.
The technology will include 14-16 antennae installed in the arena rafters; four cameras to support the tracking functionality; one sensor placed on the shoulder pads of every player on each team; and 40 pucks manufactured with a sensor inside for each game. All information will be provided to fans in real time.
The league wants to provide viewers additional insight into the sport, however, like the data that's being used to construct and manage organizations, the process will be met with some skepticism by the sport's traditionalists.
"I think it’s good for the game," Pominville said. "It’s something for the fans to enjoy. But again, it won’t necessarily show the entire picture. It can show that Connor McDavid is this fast, but it can’t show how well a guy forechecks or that sort of thing."