Corsi Numbers

[hobarth]

I question the validity of this, basically it's the summation of shot attempts, the differential.

I remember TO having Jason Blake, he was a shot attempt machine, I believe he even took 400 shots one year so how many attempts would he have taken as well? Anyone playing with this human shot machine would undoubtedly sport a very impressive corsi so in a corsi measured world Blake would be a star, he wasn't a star, and TO did well to move him ASAP.

This stat to me has never made practical sense much the same as +/- because it doesn't truly take into account the quality of the opposition, the quality of whoever is being measured icemates, the difference between the quality of the opposition's players as opposed to the quality of the players who are playing with the player being measured. The possibilities for this stat being misrepresented are endless yet we are being told that Corsi is probably being used as measuring device in arbitration.

 

[Kin]

For people actually interested in the numbers Blake's CORSI's are ok but they're nothing to lose your mind over. The guys playing with him had good ones as well but that's because he mainly played with players who were good possession guys as well. Sundin was a pretty good hockey player, Ponikarovsky was very effective at the cycle game, etc.

There's really nothing to suggest that Blake, who never took 400 shots in a season or really anything close to it, is some kind of crazy outlier creating a ton of phantom value with his shot attempts. His on ice CORSI doesn't make him a "star". In his second year with the team, so away from Sundin, he ranked 102nd in the league among players who played 50+ games. With Sundin he was 41st. Again, good, but not crazy.

What the numbers do show that he was a reasonably effective player in his time here which is more or less true. His goal totals weren't great in his first year here but that seems to more be a fluke of shooting percentages than anything else(he shot 13% in his 40 goal year, 4% in his first year with Toronto and about 8 or 9 percent every other year where he was a pretty consistent 20-25 goal scorer).

That took me, for the record, about 8 minutes of research.

 

[bustaheims]

We get it already. You're stuck in the past, struggle with fact, progressive ideas, as well as the current state of the team, league and, well, the game itself. It's time for a new act. This one won't play in Peoria anymore.

 

[Heroic Shrimp]

That took me, for the record, about 8 minutes of research.

Geez, who has 8 minutes to spare?  You must be retired or something.

 

[hobarth]

I'm retired but I don't know where to get this kind of info.

"With Sundin he was 41st. Again, good, but not crazy. "

He was 41st because of who he played with and maybe who he played against but I would never accept he was the 41st corsi player of any time, his 40 goals were a fluke but his 20 goal seasons were closer to his norm, a Clarkson anomaly, to accept him as a star is too be blind as to what actually happens on the ice, corsi is blind.

There is room for advanced stats as +/- was when it was introduced but I think corsi is a reach, in no universe is games won by shot attempts nor are games won by the number of pass attempts but games are probably won by the # of successful passes as a measure of possession.

LA which I believe had one of the most successful team corsi this last season didn't make the playoffs, this is hard to understand especially considering the quality of the players on LA, NJ the previous year also had one of the leagues best corsi stats but missed the playoffs, more understandable that they missed the playoffs because of the quality of their players but confusing that such a player poor team could rank highly in corsi stats.

We get it already. You're stuck in the past, struggle with fact, progressive ideas, as well as the current state of the team, league and, well, the game itself. It's time for a new act. This one won't play in Peoria anymore.

Just because Terminator doesn't play in Peoria any more doesn't mean that it isn't superior to whatever version is in theaters now.

I agree that progressive stats should have a place in the NHL and I'm anxiously awaiting some that actually make sense, just because it's shiny and new doesn't necessarily mean it's better, just different.

 

[Kin]

He was 41st because of who he played with and maybe who he played against but I would never accept he was the 41st corsi player of any time

It's a fact. It actually doesn't matter whether or not you accept it.

his 40 goals were a fluke but his 20 goal seasons were closer to his norm, a Clarkson anomaly, to accept him as a star is too be blind as to what actually happens on the ice, corsi is blind.

Corsi doesn't "accept him as a star"(whatever that means). He had one year of a pretty high Corsi much like he had one year with a pretty high goal total. That doesn't invalidate Corsi as a measurement any more than it would invalidate Goals.

 

[hobarth]

Why do you Nik Pik rather than address the bulk of an argument?

If he's the 41st best corsi player of any year is he not a star according to corsi at least for that year, the 41st best player in approximately 450 other players, that's a star in my books. I don't think that anyone would accept that Blake was hockey's 41st best player of any year except perhaps the creator of corsi and his/corsi disciplines.

 

[Kin]

Why do you Nik Pik rather than address the bulk of an argument?

I have addressed the bulk of your argument. The bulk of your argument was that Jason Blake's shot totals would artificially inflate both his and his linemates Corsi numbers to a significant amount. This was based on nothing, no research and no actual knowledge of the facts. Pointing out that the facts say that's not true, that Blake's Corsi and that of his linemates was never exceptionally high aside from a year when it's more likely that Blake benefitted from playing with Sundin then vice-versa, contradicts your argument on a fundamental level.

If he's the 41st best corsi player of any year is he not a star according to corsi at least for that year, the 41st best player in approximately 450 other players, that's a star in my books. I don't think that anyone would accept that Blake was hockey's 41st best player of any year except perhaps the creator of corsi and his/corsi disciplines.

That just shows a complete misunderstanding of the subject though. Any statistic has inherent flaws and contributing factors that can, in a short time frame, give a distorted picture. The year before that, Blake scored the 10th most goals. Did that make him the 10th best goal scorer in the league? Or was it, as you said, "a fluke"? If it's the latter, why can Goals be an important measurement of a player's abilities despite one year flukes but not Corsi? His goals were a fluke one year, why can't his Corsi be a fluke the next year?

The standard you're holding Corsi to would fail every statistic. So if it's meaningless by that standard, so are goals and points and save percentage and on and on. A basic understanding of statistics would tell you that there can be unusual results that deviate greatly from the expected norm but, so long as they don't repeat, are not particularly relevant to the larger data you're looking at.

 

[WhatIfGodWasALeaf]

You can join Simpson Nik.

 

[Leaflifer]

Why do you Nik Pik rather than address the bulk of an argument?

I have addressed the bulk of your argument. The bulk of your argument was that Jason Blake's shot totals would artificially inflate both his and his linemates Corsi numbers to a significant amount. This was based on nothing, no research and no actual knowledge of the facts. Pointing out that the facts say that's not true, that Blake's Corsi and that of his linemates was never exceptionally high aside from a year when it's more likely that Blake benefitted from playing with Sundin then vice-versa, contradicts your argument on a fundamental level.

If he's the 41st best corsi player of any year is he not a star according to corsi at least for that year, the 41st best player in approximately 450 other players, that's a star in my books. I don't think that anyone would accept that Blake was hockey's 41st best player of any year except perhaps the creator of corsi and his/corsi disciplines.

That just shows a complete misunderstanding of the subject though. Any statistic has inherent flaws and contributing factors that can, in a short time frame, give a distorted picture. The year before that, Blake scored the 10th most goals. Did that make him the 10th best goal scorer in the league? Or was it, as you said, "a fluke"? If it's the latter, why can Goals be an important measurement of a player's abilities despite one year flukes but not Corsi? His goals were a fluke one year, why can't his Corsi be a fluke the next year?

The standard you're holding Corsi to would fail every statistic. So if it's meaningless by that standard, so are goals and points and save percentage and on and on. A basic understanding of statistics would tell you that there can be unusual results that deviate greatly from the expected norm but, so long as they don't repeat, are not particularly relevant to the larger data you're looking at.

I'm gonna regret this, but I don't recall Hobarth saying goals are a good measure and Corsi isn't. You're building some sort of person to argue with, using straw perhaps.

 

[Kin]

I'm gonna regret this, but I don't recall Hobarth saying goals are a good measure and Corsi isn't. You're building some sort of person to argue with, using straw perhaps.

Not really. He said Corsi isn't a good measurement repeatedly in this thread and you're right that I'm assuming he thinks that goals scored is a good measurement of a player based on the fact that just about everybody does.

But heck, if he wants to show up and say that how many goals a player scores does not reflect on how good a goal scorer he is then, well, I'll concede that point to you.

 

[hobarth]

Actually I did say that corsi is a poor stat to substantiate, what, which is the gist of my argument against using corsi as a measurement of anything other than shots for and against. We already have that but corsi has concentrated on single players which we never had before but has also been expanded to include team corsi. I personally have always endorsed the notion that teams having a significantly greater volume of shots for is usually the superior team but I also accept that that isn't always the case. I can remember TO during the Quinn era winning games while being often outshot, sometimes significantly, but I also remember TO being the worst team in the NHL but often outshooting the opposition.

TO won during the Quinn era because they had the better team, they lost even tho outshooting the opposition when they were bad because they were an inferior team. The corsi results might tell us the opposite as far as both teams are concerned but the low shot superior team was the better possession team because they held the puck toward the goal of scoring, the greater shot attempt inferior team because of lack of talent shot often and whenever because they couldn't control the puck, took a significant volume of shots instead of quality shots.

Jason Blake was such a player as TO's inferior team, he had a brief success but essentially he wasn't a possession player, actually he was counter productive because he shot, didn't utilize his team mates, shot from anywhere at any time, the play often died because of his propensity.

A more telling stat I would think is the scoring chances that the commentators of many games mention, rarely do the commentators attribute the scoring chances to specific players which I would think would be the actual possession drivers rather than some mindless shot machine.

I don't understand how shot attempts somehow bestows some sort of credence to a player. I think actual shots on goal are far more important. We already have and have had for many years, had the shots on net stats for players and teams, corsi is merely overkill and is far less telling of a players value than +/-.

Nit you haven't dealt with the bulk of my contention you've simply dismissed it and then proceeded to tell me that of coarse there are always outside of the norm situations.

I'm gonna regret this, but I don't recall Hobarth saying goals are a good measure and Corsi isn't. You're building some sort of person to argue with, using straw perhaps.

Not really. He said Corsi isn't a good measurement repeatedly in this thread and you're right that I'm assuming he thinks that goals scored is a good measurement of a player based on the fact that just about everybody does.

But heck, if he wants to show up and say that how many goals a player scores does not reflect on how good a goal scorer he is then, well, I'll concede that point to you.

This is just nonsense from a nonsensical person, a troll, I understand you have a difficult time following arguments but stooping to this level simply exposes you as a Nit Piking bully rather than a rational person. Buck up you can do better, I hope?

 

[Kin]

Jason Blake was such a player as TO's inferior team, he had a brief success but essentially he wasn't a possession player, actually he was counter productive because he shot, didn't utilize his team mates, shot from anywhere at any time, the play often died because of his propensity.

I don't understand how shot attempts somehow bestows some sort of credence to a player. I think actual shots on goal are far more important. We already have and have had for many years, had the shots on net stats for players and teams, corsi is merely overkill and is far less telling of a players value than +/-.

Yeah, ok.

 

[hobarth]

Thru this all I understand the need of an analytics dept., but I just don't understand the value of corsi or how commonly obtainable info can give a team an upper hand. Nit would know if Dubas has mentioned corsi per say as a source of outstanding information but I can't remember him ever referencing corsi as being TO's bible or tool as an essential player evaluation tool, thank God, but he has frequently extolled the virtue of analytics.

I think an analytics dept. should be designed to give teams a greater insight into a player's usefulness outside of commonly available info since these depts. are becoming the staple of most teams. I have no idea as to how TO might compile their conception of useful info but I would think that with TO's wallet they should have the best and the brightest devising just such a master plan.

 

[herman]

Corsi needs to be looked at over a larger span of time, rather than game-by-game and individual basis. Hockey, more than most athletic sports, is luck-based, because of all the moving parts in play. Corsi/Fenwick, measured over a long period of time, has a strong correlation to team success. This doesn't always apply on a game-by-game basis as factors like PDO come into play (like the Leafs defeating Chicago in Nov 2014 when Reimer had to stand on his head to eke out a 3-2 victory).

Shot-attempt metrics take out the luck-factor that contributes to pucks either going into or staying out of the net. The stats show that if a team can continually generate more shots than the opponent on a regular basis, over multiple seasons, they will win more than they lose. It's a stat (on the individual level) that allowed Dubas to pick gems out of the garbage pile like Winnik (market inefficiency).

I was not uncomfortable with the concept of 'advanced stats' before but I didn't know what they meant until reading up on them a bit more. This post really helped me start to understand what everyone was talking about: http://www.pensionplanpuppets.com/2012/7/23/3173579/what-is-corsi-how-do-you-use-corsi-who-is-corsi-don-cherry-hates-corsi

 

[Heroic Shrimp]

Corsi needs to be looked at over a larger span of time, rather than game-by-game and individual basis. Hockey, more than most athletic sports, is luck-based, because of all the moving parts in play. Corsi/Fenwick, measured over a long period of time, has a strong correlation to team success. This doesn't always apply on a game-by-game basis as factors like PDO come into play (like the Leafs defeating Chicago in Nov 2014 when Reimer had to stand on his head to eke out a 3-2 victory).

Shot-attempt metrics take out the luck-factor that contributes to pucks either going into or staying out of the net. The stats show that if a team can continually generate more shots than the opponent on a regular basis, over multiple seasons, they will win more than they lose. It's a stat (on the individual level) that allowed Dubas to pick gems out of the garbage pile like Winnik (market inefficiency).

I was not uncomfortable with the concept of 'advanced stats' before but I didn't know what they meant until reading up on them a bit more. This post really helped me start to understand what everyone was talking about: http://www.pensionplanpuppets.com/2012/7/23/3173579/what-is-corsi-how-do-you-use-corsi-who-is-corsi-don-cherry-hates-corsi

Well put.  I'd add that while you correctly say that "Corsi needs to be looked at over a larger span of time", that statement applies to all statistics in sports.  But even in a short period of time, Corsi can be way more telling than primitive +/-.  You can have a 1-0 game in which the one goal was a power play goal.  Every player on both teams will be a +/- of 0.  The Corsi evaluations of individual player performance will be vastly different if the shot tallies were for 27-27 game vs. a 45-18 game.  But, to your point, there's no question that the more games that get taken into statistical consideration, the better.

All sports statistics rely on identifiable events within a game and over multiple games.  The more events you have, the more data you have to work with.  More data makes for better and more predictable statistics.  Of course we want to know how many goals are scored by players and teams and what it tells us about the players on the ice, but we need to go beyond that in statistically identifying players who contribute or don't contribute in ways that are reflected on the score sheet.  Corsi takes into account a vastly higher number of game events than goals.  It's not a perfect statistic, and NOTHING IS in sports, but it's very useful.

The Jason Blake example isn't much of an example at all, as Corsi isn't remotely a simple measurement of how many shots an individual player takes.  A player like Blake who takes a lot of shots but doesn't utilize his teammates and creates turnovers will simply not be identified as a star or superstar in advanced metrics.

 

[herman]

Corsi needs to be looked at over a larger span of time, rather than game-by-game and individual basis. Hockey, more than most athletic sports, is luck-based, because of all the moving parts in play. Corsi/Fenwick, measured over a long period of time, has a strong correlation to team success. This doesn't always apply on a game-by-game basis as factors like PDO come into play (like the Leafs defeating Chicago in Nov 2014 when Reimer had to stand on his head to eke out a 3-2 victory).

Shot-attempt metrics take out the luck-factor that contributes to pucks either going into or staying out of the net. The stats show that if a team can continually generate more shots than the opponent on a regular basis, over multiple seasons, they will win more than they lose. It's a stat (on the individual level) that allowed Dubas to pick gems out of the garbage pile like Winnik (market inefficiency).

I was not uncomfortable with the concept of 'advanced stats' before but I didn't know what they meant until reading up on them a bit more. This post really helped me start to understand what everyone was talking about: http://www.pensionplanpuppets.com/2012/7/23/3173579/what-is-corsi-how-do-you-use-corsi-who-is-corsi-don-cherry-hates-corsi

Well put.  I'd add that while you correctly say that "Corsi needs to be looked at over a larger span of time", that statement applies to all statistics in sports.  But even in a short period of time, Corsi can be way more telling than primitive +/-.  You can have a 1-0 game in which the one goal was a power play goal.  Every player on both teams will be a +/- of 0.  The Corsi evaluations of individual player performance will be vastly different if the shot tallies were for 27-27 game vs. a 45-18 game.  But, to your point, there's no question that the more games that get taken into statistical consideration, the better.

All sports statistics rely on identifiable events within a game and over multiple games.  The more events you have, the more data you have to work with.  More data makes for better and more predictable statistics.  Of course we want to know how many goals are scored by players and teams and what it tells us about the players on the ice, but we need to go beyond that in statistically identifying players who contribute or don't contribute in ways that are reflected on the score sheet.  Corsi takes into account a vastly higher number of game events than goals.  It's not a perfect statistic, and NOTHING IS in sports, but it's very useful.

The Jason Blake example isn't much of an example at all, as Corsi isn't remotely a simple measurement of how many shots an individual player takes.  A player like Blake who takes a lot of shots but doesn't utilize his teammates and creates turnovers will simply not be identified as a star or superstar in advanced metrics.

Thanks for that additional clarification, Heroic Shrimp.

To your point about sports statistics relying on identifiable events, this is the reason why I'm quite excited about the new stats and technology partnerships the NHL announced earlier this season. All player jerseys will be tagged with RFID tags, as will the pucks, which will yield quantifiable measures of actual puck possession time. Coupled with the GoPros circling the rink, the NHL is going to have a huge database of numbers (and videos) that can be pivoted to yield more useful and accurate measures of success.

I just don't understand the value of corsi or how commonly obtainable info can give a team an upper hand.

I forgot to talk about this point. Not every team believes in the value of these statistics, case in point the Leafs BD (Before Dubas). Some people put a greater emphasis on the results (wins, points) rather than the driving processes that contribute to success (puck possession, skills development), and it allows teams that identify a lesser-known metric (that is statistically confirmed to be positive) and exploit the benefits with minimal competition (Winnik, trading down in the draft when the values of picks flattens out).

No market inefficiency lasts forever as the game and how it's played changes (especially if others see the success you are having with your methods).

 

[Heroic Shrimp]

As for +/-, it is beyond outdated as a statistic.  7 of the top 10 scorers in the NHL in 2014-15 had a +/- of +5 or less. 
In 2014-15 Hart voting among skaters, 2 of the top 3 vote getters (Tavares and Crosby) had a +/- of merely +5, and Ovechkin (#1 skater vote getter) was a +10.
In 2014-15 Hart voting, 6 of the top 10 vote getters had a +/- of +10 or less, and 4 of the top 10 had a +/- of +5 or less.

If a person is attached to +/- as a measure of anything, I would assume that they wouldn't want to acquire any minus players from around the league for their team, such as, from 2014-15:

Tyler Seguin -1
Claude Giroux -3
Ryan Johansen -6
Evgeni Malkin -2
Logan Couture -6
Henrik Zetterberg -6
Joe Thornton -4
Anze Kopitar -2
Jason Spezza -7

Damn, there's sure a lot of #1 centres to pass up on because of their poor +/-.  Please join the modern age and forget about +/-.

 

[Frank E]

I think whenever you deal with data that is a new and a little more complicated, you're going to get some push back from those that can't compartmentalize the information within their traditional judgment parameters. 

And it's not as though some new aspect of the game has been created in the past few years.  Corsi/Fenwick and PDO provide a more detailed way of measuring some of the performance on the ice than the typical score-sheet, and teams can extrapolate some indication of future results and coach for better performance.   

 

[Frank E]

Is it fair to place more value in unblocked shot attempts close vs. SAT close?