There was a time when Corsi was a trailblazing statistic in the world of hockey analytics. With the NHL lacking in robust data sets found in other leagues like Major League Baseball and the NBA, Corsi could provide a rough sketch of how well a team was playing beyond its goal differential. And with goal differential highly influenced by luck over the course of a season of such a high-paced sport filled with good and bad bounces of the puck, the metric was something of a godsend.
But it’s time to realize Corsi has run its course. It’s now time to turn to more telling stats, such as scoring chances, a metric that factors in shot quality into its analysis.
One of the biggest advantages of using Corsi, as opposed to goals, was its “stickiness” factor, meaning it would largely carry over for a team from year to year as long as its core stayed intact. Roughly speaking, we can expect 57 percent of a team’s Corsi percentage difference from the average to be repeated next year. Goal percentage — a team’s scoring differential expressed as a rate of all goals scored — sees just a 42 percent carry-over, making it more fleeting and thus less predictive. The portability of scoring-chance percentage (57 percent) is on par with Corsi percentage but it correlates more closely with winning than Corsi does, making it the preferred method of team strength, both now and when considering a team’s future prospects.
Over the past five seasons, including the 2017-18 campaign, a team’s overall Corsi percentage has a weak correlation to regulation and overtime wins during the regular season. When looking to see whether two factors correlate, statheads will talk about their r-squared value, which measures on a scale of zero (not at all correlated) to one (perfectly correlated) how linked they are. Goal percentage (the percentage of all goals scored by one team) correlates at 0.9 with regulation and overtime wins, which is to be expected. By comparison, Corsi correlates at just 0.19. That’s not that strong of a link.
Scoring-chance percentage, however, is significantly better that Corsi, with a mark of 0.30. High-danger scoring-chance percentage, which tracks which team registers more of the shots most likely to become a goal, also tracks stronger than Corsi with a mark of 0.25.
Obviously, this is why NHL coaches harp on creating scoring chances, especially high-quality chances in the slot or crease, as opposed to futile shot attempts from the blue line. The latter are part and parcel of the Corsi metric, muting its effectiveness.
|2013-14 to 2017-2018||Shooting percentage on …|
|All shot attempts||5%|
|High-danger scoring chances||14%|
How does this apply to the NHL this season? The Carolina Hurricanes lead the league in scoring-chance differential (plus-389) followed by the Tampa Bay Lightning (plus-259), Boston Bruins (plus-246), Vegas Golden Knights (plus-188) and Chicago Blackhawks (plus-236). The Hurricanes and Blackhawks, as you can see, have no problem generating chances, but converting them has been an issue: Those teams are the fourth- and fifth-worst in scoring-chance shooting percentage, limiting their ability to turn shots into goals and goals into wins. That will likely correct itself next season, or perhaps earlier, but for now, it keeps them on the outside of a playoff spot. (The Hurricanes miss the second wild-card slot in the East by virtue of a tiebreaker with the New York Islanders, while the Blackhawks are 11 points out in the West.) What this means is that the Hurricanes could be a team no one wants to face in the playoffs, should they make the cut.
From the rest of that group, you see more of what you’d expect. The Lightning lead the NHL in scoring chance percentage at 9.7 percent (the league average is 8.1 percent) and also lead the East. The Bruins are right behind them. The Golden Knights lead the West and the entire league.
Not only do scoring chances give us better insight into which teams are playing better hockey than Corsi, they can also be used for more in-depth player valuation.
Vladislav Namestnikov of the Tampa Bay Lightning has taken just 184 shot attempts, but the lion’s share of those have been scoring chances (149, 81 percent), with more than half coming from the slot or the crease. As a result, his team has outscored opponents 72-34 with him on the ice. The Lightning are close to breaking even (53 percent of goals scored) with Namestnikov on the bench.
Viewed through Corsi, Namestnikov would be considered one of the worst forwards in the league at creating shots — he ranks 77th for his rate of shot attempts per 60 minutes (10.7) out of 85 forwards skating at least 1,000 minutes — but he moves up to 44th among this group for his rate of scoring chances per 60 minutes. Namestnikov’s rate of chances created in the slot or crease ranks 17th (4.8 per 60).
Compare that to Washington Capitals center Nicklas Backstrom, who takes a similar amount of shot attempts (188) but registers fewer scoring chances (111, 59 percent). As a result, the Capitals aren’t as good as outscoring opponents with Backstrom on the ice (69-45) and see more goals go in their favor with him on the bench. To be fair, Namestnikov does see a higher share of his zone starts in the offensive zone than Backstrom, but that equates to just one extra offensive-zone faceoff per game.
|2017-18||Shot attempts||Scoring chances||High-danger scoring chances||Scoring chance %||High-danger %||Scoring chances that are from the slot or crease|
Entering Monday’s games, there were 13 defensemen who had created 100 or more scoring chances. The league leaders include Brent Burns of the San Jose Sharks (138), Washington’s John Carlson (124) and Victor Hedman of the Lightning (122). Also on the list is Calgary Flames rear-guard Dougie Hamilton (114), who creates a higher percentage of high-danger chances (19 percent) than anyone else of that group.
Defensemen should be judged not only on how well they suppress shot attempts, but also how well they discourage opponents from creating scoring chances. And if they can create scoring chances themselves, all the better.
Washington’s Brooks Orpik, on the other hand, has created just 20 even-strength scoring chances all season, the lowest among defensemen playing at least 20 minutes per game, two more than Jonathan Ericsson of the Detroit Red Wings. Not only does Orpik not create scoring chances for the Capitals, he isn’t good at preventing them, either: Among that same group, only Michael Del Zotto of Vancouver and Nick Leddy of the New York Islanders are on the ice for more scoring chances against than Orpik (41.8 per 60). That results in a league-low 42 percent of scoring chances in Washington’s favor, and just 37 percent of the high-danger chances, when Orpik skates. While the team has often lauded Orpik’s value as a leader and presence in the locker room, the numbers paint a bleak picture about his ability to perform on the ice.
Goaltenders don’t create scoring chances, but how well they defend them goes a long way to determining the success of a team. For example, the Chicago Blackhawks are in a down year, in danger of missing the Stanley Cup playoffs for the first time since 2009. But don’t place much blame on goaltender Corey Crawford. The 33-year-old netminder has a league-high .854 save percentage against high-danger scoring chances this season and even a higher save percentage (.929) than expected after factoring in the quality of all shots against (.908 expected save percentage). Should he return to the lineup (he has been out of action since Dec. 23 with an upper body injury), perhaps the Blackhawks can make a late-season push.
Nashville Predators netminder Pekka Rinne has the second-highest high-danger save percentage (.843), followed by Ben Bishop of the Dallas Stars (.840), giving each team confidence in their upcoming postseason bids. The Islanders, on the other hand, have a hard time overcoming the issues Jaroslav Halak (.777, 34th out of 44 goalies playing at least 1,000 minutes) and Thomas Greiss (.667, 44th) create with their inability to bail out their defensemen by stopping high-quality chances.
Does this mean we should judge players solely on their ability to produce and prevent scoring chances? Of course not, but it provides much more quality information when evaluating teams and players than Corsi does, making it a significant step forward in advanced analytical analysis of the sport and essentially rendering Corsi obsolete.