Today I am continuing my summer sabermetrics series by posting the 20 worst players by team adjusted Corsi in 2015/16. A couple days ago I posted the top 20 and these are the players who most struggled in term of puck possession.
Corsi is the difference in attempted shots taken by a team and its opponents while a given player is on the ice in 5 on 5 situations. This number is then team adjusted. A baseline is found for each team that is used as a zero point for comparison between different teams.
In order to make sure the adjustment is sensible, I have limited this list to players who played 50 or more games with only one team in 2015/16.
Here are the 20 worst players by team adjusted Corsi:
Yesterday I posted the top 20 players by team adjusted Corsi. Atop this list is Drew Doughty of the Los Angeles Kings but since he led the league in raw Corsi as well I have already written about him. I thought I would take this time instead to look at the top forward by team adjusted Corsi. Since defencemen get more playing time than forwards, it is harder for a forward to lead in any counting stat. Patrice Bergeron of the Boston Bruins having the fourth highest team adjusted Corsi of +250.2 is quite an achievement.
Corsi is a puck possession stat. It measures the difference between attempted shots between a team and their opponents when a given player is on the ice in 5 on 5 situations. This number is team adjusted much like I adjust +/-. Basically a team "zero point" is found so that players on differing teams can be better compared.
Bergeron is well-established as a top puck possession player. For example here is a post I wrote about his puck possession a few years ago.
Today I am continuing my summer sabermetrics series by looking at the top 20 players in 2015/16 by team adjusted Corsi. A Corsi rating is the difference between attempted shots for a player's team and his opponent when a player is on the ice in 5 on 5 situations. It is a measure of a player's puck possession. In order to make it more of an individualized rating that can be compared between teams a baseline for each team is calculated and used as a zero point. I have only considered players who played 50 or more games in 2015/16 and played with only one team in order to make the team adjustment meaningful. Otherwise a player who plays one shift with the Colorado Avalanche is better than a +100 and doesn't play enough time to have a rating much different from zero.
Here are the top 20 players in 2015/16 by team adjusted Corsi:
A few days ago I posted the team Corsi ratings. This is a puck possession measure which is the difference in attempted shots between a team and their opponents in 5 on 5 situations. It has significant predictive power - for example the top nine teams were all playoff teams. However there is a statistical fluke that could be used as a sound bite to show that Corsi results are irrelevant. Last season the last place Toronto Maple Leafs had a better team Corsi than the first place Washington Capitals. Toronto finished 12th in the league with a +197 Corsi rating and Washington was 14th with a +147 Corsi rating. To a doubter of analytics this is a sign that stats say Toronto is better than Washington.
In reality this is giving a lot of significance to one result that you cherry picked to look silly and ignoring the many other results that seem more reasonable. Nevertheless there is an explanation for how the worst team in the league can have a slightly worse Corsi than the best team.
The most likely attempt to explain a fluke like this might be to look at the situations in which a team plays. A team like Washington is often ahead in games and may tend to sit on their lead. A team like Toronto is often behind in games and benefits from their opponents sitting on their lead defensively. This can be shown by looking at Corsi ratings only in close situations. However that doesn't explain anything. In close situations Toronto is a +70, while Washington is +51. Toronto stays ahead here.
A couple days ago I posted the team Corsi ratings for 2015/16. Corsi is a puck possession measure which is the difference in attempted shots by the team and their opponents in 5 on 5 situations. I wrote yesterday about the Los Angeles Kings who had the league best Corsi. Today I want to write about the league worst team by Corsi in the Colorado Avalanche. They posted a -839 team Corsi. This was almost 400 points worse than the second worst New Jersey Devils.
Colorado posted a 39-43 record with four regulation tie points. This was good enough for 82 points and the tenth worst record in the NHL. They were 13 points ahead of last place Toronto. The question is how can such a poor puck possession team finish that far ahead of last place?
Colorado had the fewest regulation tie points of any team in the NHL. Thus they probably should have had even more points for that reason alone. This makes the explanation of why they did better than their Corsi suggests even harder.
Yesterday I posted the team Corsi ratings for 2015/16. The Los Angeles Kings led the league significantly. They posted a +932 team Corsi which gave them a large lead of second place Dallas who was +392. This means that in 5 on 5 situations the LA Kings took 932 more attempted shots than their opponents. This shows that Los Angeles had a significant lead over any team in the league in terms of puck possession. I think the question to ask is that if LA was so much better than the rest of the league in this aspect of the game why were they not the first place team in the league?
Los Angeles finished eighth in the league with a record of 48-34 with six regulation tie points for 102 points. Every team that finished above Los Angeles had more regulation tie points. Merely having three more regulation tie points, which would put LA at essentially the league average would have placed the Kings in fourth place overall. So the NHL's point scheme worked against LA and they were a better team than their record showed.
Today I am continuing my sabermetrics and hockey series by looking at the team Corsi ratings for each of the 30 teams in the 2015/16 season. This is the difference between attempted shots in 5 on 5 situations for that team and their opponents. This number is a measure of puck possession and measures a repeatable part of team performance. This number is not the only measure of a team. It does not include such things as special teams, goaltending, shooting and luck but it should correlate with the team's success. These numbers are also useful on the individual level because they can be used to establish a baseline for each team in order to adjust Corsi ratings for the team on which they are achieved.
Here are the team Cori ratings for 2015/16:
This is the second of these similar hockey sabermetrics posts where there is a player who seems to have contradictory information from his various puck possession stats. So far this summer I have looked at Corsi or the difference between attempted shots for a player's team and his opposition while a given player is on the ice in 5 on 5 situations and adjusted +/- which is the difference between goals for and goals against in similar situations corrected to try to remove team effects. Nick Shore of the Los Angeles Kings is a player who has a high Corsi and a low adjusted +/- and I wrote about this. Adam Larsson then of the New Jersey Devils was the opposite case last year. He posted an adjusted +/- that was 10th best in the league last year at +21.0 and a Corsi of -230 which was 12th worst in the league. How it is possible in Larsson's case that various puck possession stats can give such different results.
Larsson is an especially interesting player this summer because he was involved in a high profile trade this summer. Edmonton acquired him for Taylor Hall. This trade is widely seen as a poor move by the Oilers. In order to better analyze this deal it is necessary to have a reasonable valuation of Larsson, thus we must make sense of his contradictory puck possession numbers.
One of the biggest differences between Corsi and +/- stats are that +/- being based upon goals scored has a much lower frequency of events. Thus it is much more likely to be affected by statistical flukes. These statistical flukes can be measured in the form of PDO. This is the sum of the saves and shooting percentages when a given player is on the ice in 5 on 5 situations. A given player has very little ability to influence this number. Thus a player with a particularly high PDO is almost certainly lucky and one with a particularly low one is unlucky. Larsson's PDO last year was 1027 which is relatively high compared to the defined average of 1000. It is the second highest PDO among New Jersey players who played 50 or more games played.
A few days ago I posted the worst 20 players by raw Corsi in the 2015/16 NHL season. Last in the league was Dan Girardi of the New York Rangers. He posted a -389 Corsi. This shows puck possession problems while Girardi is on the ice. Opposing teams took 389 more attempted shots than the Rangers did in 5 on 5 situations with Girardi on the ice.
Girardi is a 32 year old defenceman who had played with the Ramgers since 2006. He posted 17 points in his 74 games played. This is his worst offensive total in his NHL career when he spent a full non-lockout season spent in the league. That is a sign Girardi is in decline. Girardi is one of those players who has a reputation as a hard working team player who is selfless and willing to block a shot and put his body on the line. The problem is players like this tend to be tough to take out of the line-up even when they are hurting the team. Usually this is a reasonable description of the players that I chose as the worst player in the league in any given season. The problem is Girardi is not playing well for the Rangers at he seems to be getting worse each season.
Today the Las Vegas expansion team which remains without a nickname picked their first general manager. The selected former Washington Capitals GM George McPhee. McPhee had been the Caps GM from 1997-2014. He was responsible for acquiring much of the current core of the Capitals who won the Presidents Trophy this year. He has a good record as a GM. There is one major blemish on it in that he traded Filip Forsberg to the Nashville Predators for Martin Erat but despite that bad trade he was successful in Washington. He likely will be a good GM in Las Vegas as well.
McPhee has a history of drafting well. Alex Ovechkin, Nicklas Backstrom, John Carlson, Braden Holtby and Mike Green are among the players the Caps drafted in his tenure. That skill will be valuable for an expansion team. I think Las Vegas did pretty well with their first GM selection.
About The Puck Stops Here
Who am I? A diehard hockey fan.
Why am I blogging? I want to.
Why are you reading it? ???