As advanced metrics have slowly begun to take over the baseball world, one wonders if hockey has the same type of metrics that may one day become widely accepted as the norm. The answer is yes.
This purpose of this article is two fold. It will introduce (or re-introduce) advanced hockey metrics and then apply them to the current crop of Phoenix Coyotes. In this article you will find out why Oliver Ekman-Larsson's season is more impressive than it appears on the surface and why Steve Sullivan may be on the verge of a turnaround.
**Please note that all stats are 5-on-5 stats with a minimum of 20 games played as of 3/26.**
The first statistic we will look at is PDO. Hockey Prospectus defines PDO as, "PDO is the sum of a player's on-ice save percentage and on-ice shooting percentage." In other words, when a player is on the ice, the save percentage of his team added to the team's shooting percentage. The league average is 1000 (sometimes shown as 100). PDO regresses heavily to the mean of 1000 over time. Players well over 1000 have been a bit "lucky" while players well below 1000 have been "unlucky." Just like BABIP in baseball, PDO is an indicator of future regressions or turnarounds.
Coyotes: Ekman-Larsson leads the Coyotes with a PDO of 1025. That number suggests a regression in scoring as his 11.57 shooting percentage is well above league average.
Sullivan has the worst PDO on the team as it is 941. That number is well below 1000 and signals that he should experience a turnaround soon. His shooting percentage is well below league average but that is not all. The save percentage when he is on the ice is .879, the worst on the team. Defense appears top be the problem here as the Coyotes have a +/- of -1.49 per 60 minutes when Sullivan is on the ice.
As of March 11 (and before the losing streak), Phoenix had a team PDO of 991, suggesting that the team should sustain the successes it had previously. As the number drops after the losing streak, a small re-bound should be in store.
The next stat we will look at is Corsi and all its uses. According to Hockey Prospectus, "Corsi is essentially a plus-minus statistic that measures shot attempts. A player receives a plus for any shot attempt (on net, missed, or blocked) that his team directs at the opponent's net, and a minus for any shot attempt against his own net."
On-ice Corsi is first up and it measures a player's Corsi as a rate per 60 minutes of play.
Coyotes: Derek Morris leads the Yotes with an on-ice Corsi of 12.2. Meaning, while Morris is on the ice, the team directs an average of 12.2 more shots per game toward the opposing net than shots allowed on their own net. Keith Yandle is second on the team with an on-ice Corsi of 10.91.
Rusty Klesla has the lowest on-ice Corsi with a -4.5. Four and a half more shots go toward the Phoenix net than toward the opposing net while Klesla is on the ice.
Next up is Relative Corsi (Corsi Rel). Hockey Prospectus defines Relative Corsi as, "A player's Corsi value in comparison to his teammates. Relative Corsi is expressed as the player's Corsi minus the team Corsi rate. A positive value indicates a player who is better than the team average and a negative number is a player who is worse than the team average." In other words, a player's on-ice Corsi minus his off-ice Corsi.
Coyotes: Yandle leads the team in this stat as well with a Corsi Rel of 10.8. We already know that his on-ice Corsi is 10.91, that means his off-ice Corsi is 0.11. Phoenix barely out-shoots (on net, missed, or blocked) the opposition while Yandle is off the ice as opposed to vastly outshooting (on net, missed, or blocked) other teams when he is on it.
Klesla is once again the worst on the team with a Corsi Rel of -12. He has an on-ice Corsi of -4.5. His off-ice Corsi is +7.5. Meaning the team has significantly more shots (on net, missed, or blocked) and possession while Klesla is off the ice.
Ekman-Larsson is 10th on the team with a Corsi Rel of -1.8. His on-ice Corsi is 3.18 but his off-ice Corsi is +4.88. Why would such a good player have that stat? The final Corsi stat of the day should explain it.
CORSI REL QoC
Relative Corsi Quality of Competition. What is the quality of completion a player faces when he is on the ice? Using Relative Corsi, the stat Corsi Rel QoC quantifies the average Corsi Rel of the opponents a player faces on the ice.
Coyotes: Ekman-Larsson leads the NHL with a Corsi Rel QoC of 2.794. This means two things: one, OEL faces the hardest completion of any skater in the NHL, two that the average opponent outshoots (on net, missed, or blocked) their competition by 2.794 shots per game. This explains why OEL's Corsi Rel is so low because his stats are competing with the best competition in the league.
On the flip side, Michael Stone had faced the easiest competition on the Coyotes with a Corsi Rel QoC of -0.528.
Just as the quality of competition explained OEL's low Corsi Rel numbers, it also explains Yandle's high numbers.
Yandle faced the third easiest quality of competition on the team (second easiest of any defenseman). His Corsi Rel QoC is -0.426.
Food For Thought
Advances metrics serve many purposes but two key ones: future indicators and quantifying what is seen on the ice. There is no guarantee that Sullivan has a great final 5 weeks or that OEL's scoring may slightly drop. They are just indicators of what is most likely to occur given years of research and past circumstances. Please read a multitude of stats because the goal is to get the most complete picture of what is happening on the ice. Final note, please understand that advanced metrics (in any sport) normalize in large sample sizes and are not meant to be used for game-by-game analysis.
Here are the NHL leaders (as of 3/20) in the categories listed above:
PDO Team: Anaheim Ducks - 1045
PDO Player: Mark Fraser (TOR) - 1101
On-Ice Corsi: Justin Williams (LA) - 36.99
Corsi Rel: Justin Williams (LA) - 29.2
Corsi Rel QoC: Oliver Ekman-Larsson (PHX) - 2.474