In my summer look at sabermetrics and hockey, I have recently looked at team Corsi ratings. This is the difference between a team's attempted shots (shots on goal, blocked shots and missed shots) and those of their opponents in 5 on 5 situations. This number is interesting because it shows team's puck possession ability. A team that has the puck will have a positive Corsi and a team that plays without the puck will have a negative one. The number correlates with the success of each individual team, but it is an imperfect method to rank teams because it neglects several important facets of hockey including special teams, goaltending quality and shot quality.
The NHL's leader in the 2013 season by team Corsi is the Los Angeles Kings. They posted a +495 team Corsi. In a 48 game schedule, this means that they attempted more than ten extra shots than their opponents at even strength per game. That is a significant difference in puck possession.
While doing this, the Kings managed to finish fifth in the West Conference with 59 points. They made their way to the semi-finals before losing to the eventual champion Chicago Blackhawks.
This summer I have been looking at sabermetrics and hockey. Specifically, I have been looking at the Corsi rating, which is the difference between shots attempted by a team and their opponents in 5 on 5 situations. This is a very good proxy for puck possession, as teams cannot attempt shots without possession of the puck. I am looking at these numbers for individual players and I have posted the top 20 and worst 20 players by raw (unadjusted) Corsi numbers. What we have seen is that these rankings are context dependent. They select players from a handful of teams in the top group and a different handful of teams in the bottom group. The top and bottom Corsi players play on the top and bottom puck possession teams and often play offensive roles in the top group and defensive ones in the bottom group.
In fact the top player in the league in 2013 was Justin Williams. He played the most regular offensive role on the top puck possession team in the NHL - the Los Angeles Kings. The worst player in the league in 2013 was Jay McClement. He played the most regular defensive role on the worst puck possession team in the NHL - the Toronto Maple Leafs. We have learned a lot about the context of how these two (and other) players play from our Corsi analysis already. The biggest selection effect right now is a team effect. In order to accommodate that I need, I need to know the overall Corsi ratings for the various teams.
Here they are:
Yesterday I continued my look at sabermetrics and hockey but posting the worst 20 players in the 2013 season by their raw Corsi rating. The worst player last season was Jay McClement of the Toronto Maple Leafs. He posted a -286 Corsi in the season. That means his opponents attempted 286 more shots than Toronto did at even strength while McClement was on the ice. If McClement was on the ice the other team likely had the puck.
Jay McClement has been willing to play a difficult thankless role for several years in the NHL. He plays the toughest defensive shifts and rarely starts his shifts in the offensive zone. It has taken a toll on his career. McClement posted the second worst +/- rating in the NHL in between the 2005 and 2012 lockouts. This is largely a sign of how tough a role McClement takes on, although he doesn't help himself out with much in the way of offensive skill.
After adjusting for team and his high defensive zone starts, it is clear that Jay McClement is not the worst player in the NHL. He plays a very tough role and that is the major factor in his poor puck possession numbers. Contextually adjusting his Corsi will show that he is not in the running to be the worst player in the NHL despite a league worst raw Corsi and a perennially bad +/- rating.
Today I continue my summer look at sabermetrics and hockey with a look at the 20 worst players in the 2013 by their raw Corsi rating. A while ago I posted the top 20 players. Today's group of players had the worst puck possession while they were on the ice in 5 on 5 situations. The number of attempted shots (shots on goal, blocked and missed) by their opponents far exceeded that of their own team. This is a group of players who either had poor seasons contributing little or played in tough defensive situations on weak teams or both. With some future analysis we will be able to have a better idea as to who falls into which case.
Here are the worst 20 players in the 2013 season by their raw Corsi number:
It is strong evidence to those who do not understand what the numbers mean that Corsi is a fraud when I posted the top 20 players in the 2013 season by Raw Corsi and Justin Williams of the Los Angeles Kings lead the NHL. He posted a +339 difference is shots attempted by his team in 5 on 5 situations than his opponents took. That must be wrong because nobody claims Williams is the best player in the NHL. In fact nobody seriously considers him an All Star player. Therefore those who do not understand Corsi think it is all disproven and the idea should go away never to be heard from again. There is some kind of false idea that posting the leader in some statistical area has to produce the name of a player who is at least arguably the top player in the NHL or else that stat proves nothing. That idea is beyond silly if you think about it for any longer than a second.
In fact the case of Justin Williams posting a top Corsi is not new. I wrote about him last year here when he posted the 4th best raw Corsi in the league in 2011/12 and he posted the 17th best raw Corsi in 2010/11. Thus it is not a fluke that Williams posts a top Corsi. He is a good example of the context dependent information that comes from Corsi ratings and the fact that analysis is necessary to raw numbers.
The easiest example of a context dependent conclusion is that Williams gets an ever improving raw Corsi rating as he plays on an ever improving team. Team context is a big part of any Corsi number. A player on a good team should post a better Corsi rating than one on a bad team. Of course Corsi ratings cannot be directly compared between different teams. In fact they cannot be directly compared between players on the same team. It would be a silly conclusion that Justin Williams is the best player on the Los Angeles Kings because he has the best Corsi on the team. It doesn't pass the "eye test" where I watch hockey and run any conclusions the numbers may show past what I see with my eyes.
As I continue my summertime look at sabermetrics and hockey, today I look a Corsi ratings. A Corsi rating is a very valuable sabermetric tool to evaluate players. It is the difference between shots directed on goal by a team in 5 on 5 situations and those allowed by a team. Included are shots on goal, blocked shots and missed shots. It is a strong gage of puck possession. If one team has the puck a lot more than another, they will have a good Corsi rating and are most likely a strong team. Corsi ratings can be used to evaluate individual players as well. The problem is finding the right context for numbers. Players on good teams will tend to have better Corsi ratings than those on bad teams and this is not evidence that a player on a good team is necessarily better. Similarly players who play with strong linemates, or against weaker opponents or who start shifts in their offensive zone will do better than those who don't. This must all be taken into account when rating players by Corsi ratings.
Today I will post the top 20 players by raw (unadjusted) Corsi ratings in the 2013 season. It should be obvious that team effects are significant and will be the first thing I adjust for in the future.
Here are the top 20 players in 2013 by raw Corsi rating:
It was late June when I wrote the sabermetrics and hockey post listing the worst 20 players in the 2013 season by adjusted +/-. The worst player was Tanner Glass of the Pittsburgh Penguins. Pittsburgh also had the top players in the league in Sidney Crosby's linemates Pascal Dupuis and Chris Kunitz.
Tanner Glass is one player who doesn't belong on a team as good as Pittsburgh. He managed only two points all season but played every regular season game. Come playoff time, he was scratched in all but five playoff games. When the games got important, Pittsburgh realized that he was a liability. Despite that he is signed for next season. It will be interesting if Pittsburgh plays him regularly next season. Glass is your typical example of a player who is ineffective and still playing. He is a hard working, fan favorite that his teammates like who has been totally ineffective. It is hard to scratch popular player who works hard regardless of how much of a liability he is. A player like Tanner Glass shouldn't be in the NHL. Even in his best season, he had never done better than five goals and 16 points. The typical "replacement player" would be more valuable than Tanner Glass.
This is the third of my Hall of Fame cases for the male players in the class of 2013. I have already written about Chris Chelios and Scott Niedermayer. Today I turn my attention to Brendan Shanahan. Here is what I wrote when I first considered him a Hall of Famer and here is what I wrote when he retired. In order to make his Hall of Fame case I will use the Keltner List, which was developed for baseball but was easily adapted to hockey.
1.Was he ever regarded as the best player in baseball? Did anybody, while he was active, ever suggest that he was the best player in baseball? I can't think of any reasonable argument that anyone ever made to suggest Brendan Shanahan was the best player in hockey.
2.Was he the best player on his team? Perhaps the only time Shanahan was the best player on his team was during his season in Hartford. In Detroit, Steve Yzerman and Nicklas Lidstrom were the best players on the team. In St Louis, Brett Hull was the best player on the team. It is a bit unfair to say that Shanahan had to play on a weak team like Hartford to be the best player on his team, but that is what happened. In Detroit where he played his prime seasons there were better players. He might have been the best player on a more average team, but he was not the best player on the strong team he played with.
I am writing about the Hall of Fame cases for the three male players inducted into the Hockey Hall of Fame in the 2013 class. I have already written about Chris Chelios. Today is Scott Niedermayer's turn and Brendan Shanahan is still to come. Here is the post I wrote when Niedermayer retired and here is my post when I first considered him a Hall of Famer. In order to make my case I will use the Keltner List, which was developed for baseball but is easily transferable to hockey.
1.Was he ever regarded as the best player in baseball? Did anybody, while he was active, ever suggest that he was the best player in baseball? No. There is no clear argument that Scott Niedermayer was the best player in hockey at any point.
2.Was he the best player on his team? I would argue that he was but not for a long period of time. In New Jersey both Martin Brodeur and Scott Stevens were better players. In Anaheim I would argue that Chris Pronger was a better player when he was on the team, but in the years Pronger was in Edmonton before coming to Anaheim and Philadelphia after leaving Anaheim, I think Niedermayer was their best player.
Probably the biggest shock of the off-season is the retirement announcement of Ilya Kovalchuk. He is 30 years old and his contract with the New Jersey Devils runs until 2025. The New Jersey Devils have voided his contract with his signing of retirement papers. This allows him to play in any league outside the NHL. He is expected to sign with St Petersburg in the KHL - where he played during the lockout.
This is a surprise move but I will suggest motivations for why this may have occurred. Kovalchuk enjoyed his time in the KHL and was not enjoying playing in New Jersey. They are a non-playoff team and a lot of the blame for that falls upon his shoulders as their highest paid player. He leaves behind a significant amount of money by getting out of his contract. There is $77 million that he is due to still be paid. Obviously this is not a purely financial decision. Kovalchuk is already extremely rich from money he has already made playing in the NHL and he will be well paid in the KHL. There is an expectation that he may be paid as much as $20 million this year to play in the KHL and be taxed at a lower rate. In fact he is looking at a short term pay raise. Likely he will lose money longterm. I cannot imagine his salary remaining this high for well over a decade.
About The Puck Stops Here
Who am I? A diehard hockey fan.
Why am I blogging? I want to.
Why are you reading it? ???