What Corsi really translates to (in numbers that are easy to grasp)

[Ol' Jase]

Because that's how hockey works.

40% of all shifts have a zone start.

100% of shifts have corsi.

We can end with this. If this is the foundation of your "proof", I'm good with saying, no, you have proved absolutely nothing. Basic mathematics have my back on this one.

 

[Machinehead]

We can end with this. If this is the foundation of your "proof", I'm good with saying, no, you have proved absolutely nothing. Basic mathematics have my back on this one.

You're suggesting that I apply zone start data to situations where zone starts don't exist. I'm not sure where to even go with that. I couldn't if I wanted to.

 

[RageQuit77]

o_o

It would be indeed nice to know how things differ generally between players also divided by how they start their shifts.

It's obvious that there might be big differences (especially in individual cases of dedicated, good starters) as good face-off players probably influence relatively more to a team's shooting differentials by the fact they have directly involved to it in all zones, while folks that start on bench have nil to that start.

On the other hand player jumping to the rink in middle of shift doesn't get influenced by opposing team's shoots regardless how many his own team takes (during ongoing shift started at a Zone), and in best cases that guy shoots once and score goal once before his shift is over.

If there are stats for corsi of Zone starts for starting players, and corsi for players who begin their shifts after face-off, one of course doesn't apply Zone-start stats to a player who wasn't there, and visa versa. Otherwise such stats are meaningless for individual player. Everyway possible.

As it would mean that Zone starter players' that have already - grinded for example 2 minutes - accumulated shooting differentials would be just added to the player's stats who join to the group for it's last 10 seconds on the ice.

 

[Ol' Jase]

o_o

It would be indeed nice to know how things differ generally between players also divided by how they start their shifts.

It's obvious that there might be big differences (especially in individual cases of dedicated, good starters) as good face-off players probably influence relatively more to a team's shooting differentials by the fact they have directly involved to it in all zones, while folks that start on bench have nil to that start.

On the other hand player jumping to the rink in middle of shift doesn't get influenced by opposing team's shoots regardless how many his own team takes (during ongoing shift started at a Zone), and in best cases that guy shoots once and score goal once before his shift is over.

The ONLY way to prove that Zone Starts, as they are defined, have no affect on Corsi is to segregate the situational Corsi measurement as it applies to the Zone Start. Basically, the measured Corsi when a player is on the ice on a faceoff until he leaves the ice.

It is the only way to truly measure whether or not Zone Starts affect Corsi. Saying otherwise isn't statistically applicable at its root.

Not arguing with you, I think we are on the same page.

 

[eternalbedhead]

Maybe I'm mistaken (sorry if that's the case), but I thought your view on Fowler is that he hurts his team more than he helps?

Now, if that's what you are saying, is it your belief that Fowler having a +20 goal differential during the last four seasons (including playoffs) is an anamoly, and that the luck will change any day now?

I hate using +/- because it's so dependent on the team you play on. An absolute train wreck of a team might not have a single plus player, even though their best defender might be better than the #7 on a championship squad, who might be +2 or +3 in the same stat.

That being said, Corsi is misconstrued and too many people recite stats off a page without knowing anything about the player. What does a bad CF% rating say? Does the player have missed assignments aplenty in his own end? Or does he continually make dumbass turnovers? (Exhibit A: Kevin Bieksa)


Now, there's a problem here as well: let's say that this horrible defender (cough cough Bieksa) is paired with a good player. This unnamed horrible player can't contribute to the transition game, he is constantly caught up the ice with his pants down on the rush back, he's a negative player offensively with only a decent slapshot to contribute to offensive zone play, and his defensive coverage is nothing short of horrible. Obviously he just might drag down his partner, right? His partner is good (albeit inconsistent at times) defensively, he's a fantastic skater and very good on the transition, and his offensive zone play is very good and is complemented with a much-improved shot. A solid #2.


Now, Fowler isn't a shutdown defender, regardless of what Boudreau tried to make him into. Throughout his career, he's shined his brightest in an offensive role. Though his offseason training and his changes to his playing style deserve much of the credit for this, his usage by RC is one of the big reasons he's having an offensive renaissance. He's seeing plenty of PP time, he's being more aggressive offensively, and he's been employed in a much more offensive role. He's not a shutdown defender, and trying to turn him into one is likely not going to aid his development any. He's not as bad defensively as people seem to think; from Lovejoy to Bieksa to now Vatanen, he's never been paired with defensive stalwarts. In fact, when he was paired with Manson, his Corsi stats looked a lot better. Now, Manson is a good #4 who has excelled as a complement to Lindholm, but he's not good enough where he can carry defenders. For Fowler's Corsi stats to jump up like they did with a legitimate shutdown DFD such as Manson means less that Manson is carrying the pairing than Fowler is finally playing with a good player. Fowler's now paired with Vatanen -- both OFDs who love to push the play. Fowler's offensive style has dominated over Vatanen and Vatanen is then being forced into a more defensive role where he's really struggled. (I might note that Vatanen has been mostly sheltered throughout his career yet stats say that he's better than Fowler who hasn't been sheltered at all over the past few years)


My verdict on Fowler: His defensive game isn't his strong suit. He's good there and he unfairly gets **** on due to having defensively lackluster partners, but he shines when he's able to move the puck up the ice and make things happen in the offensive zone, not when he's being coached into being a DFD while being paired with one of the worst players in the league. According to the metrics, Fowler is one of the best transitional defenders in the league, but this gets ignored in favor of more popular stats like Corsi which don't exactly portray him in an endearing light.


And my verdict on Corsi: Better used as a metric for evaluating line/team performance than for individual performance/skill. There's just too many independent variables in what affects the metric to use it reliably for an individual player. A player's chemistry with his partner/linemates, his partner's or linemates' talent, his usage (zone starts and strength of opposition), and the system in which he plays can all have profound affects on a player's Corsi rating, as well as several other variables. (such as coaching skill and the strength of the surrounding team) Attempting to use it as a tell-all for overall skill is like stepping into a poorly-insulated refrigerator and attempting to tell the temperature outside. You just can't do it because that darned refrigeration system is getting in the way and marring your results. If this poorly-illustrated example doesn't do the job for you, let me put it this way: all scientific experiments have a potential to fail if there are significant variables affecting it, which is why having a controlled environment for experiments you conduct is very important. Corsi is no different, but a controlled environment is unfortunately impossible. Now, if you're using Corsi for the evaluation of a line's performance (not necessarily skill), you can eliminate the first two variables. Not enough to make it a tell-all, but no statistic is perfect. If you use Corsi for the evaluation of a team's performance, you can knock off usage for the most part, as well as obviously any negative or positive influences from overall team skill.

 

[Machinehead]

The ONLY way to prove that Zone Starts, as they are defined, have no affect on Corsi is to segregate the situational Corsi measurement as it applies to the Zone Start. Basically, the measured Corsi when a player is on the ice on a faceoff until he leaves the ice.

It is the only way to truly measure whether or not Zone Starts affect Corsi. Saying otherwise isn't statistically applicable at its root.

Not arguing with you, I think we are on the same page.

This has been done already.

Situation 1 - player takes a faceoff in the defensive zone; average CF% - 37.7%

Situation 2 - player takes a faceoff in the offensive zone; average CF% - 59.8%

Situation 3 - player takes a faceoff in the neutral zone; average CF% - 47.7%

Situation 4 - player jumps over the boards; average CF% - 50.9%

Those averages are from faceoff until leaving the ice or a whistle.

https://puckplusplus.com/2015/01/20...-lot-on-their-own-not-that-much-in-aggregate/

The isolated effects of situations 1 and 2 are massive, but situations 1 and 2 only occur in isolation about 8% of the time.

You can't just ignore 92% of the game.

 

[Elvs]

I hate using +/- because it's so dependent on the team you play on. An absolute train wreck of a team might not have a single plus player, even though their best defender might be better than the #7 on a championship squad, who might be +2 or +3 in the same stat.

That being said, Corsi is misconstrued and too many people recite stats off a page without knowing anything about the player. What does a bad CF% rating say? Does the player have missed assignments aplenty in his own end? Or does he continually make dumbass turnovers? (Exhibit A: Kevin Bieksa)


Now, there's a problem here as well: let's say that this horrible defender (cough cough Bieksa) is paired with a good player. This unnamed horrible player can't contribute to the transition game, he is constantly caught up the ice with his pants down on the rush back, he's a negative player offensively with only a decent slapshot to contribute to offensive zone play, and his defensive coverage is nothing short of horrible. Obviously he just might drag down his partner, right? His partner is good (albeit inconsistent at times) defensively, he's a fantastic skater and very good on the transition, and his offensive zone play is very good and is complemented with a much-improved shot. A solid #2.


Now, Fowler isn't a shutdown defender, regardless of what Boudreau tried to make him into. Throughout his career, he's shined his brightest in an offensive role. Though his offseason training and his changes to his playing style deserve much of the credit for this, his usage by RC is one of the big reasons he's having an offensive renaissance. He's seeing plenty of PP time, he's being more aggressive offensively, and he's been employed in a much more offensive role. He's not a shutdown defender, and trying to turn him into one is likely not going to aid his development any. He's not as bad defensively as people seem to think; from Lovejoy to Bieksa to now Vatanen, he's never been paired with defensive stalwarts. In fact, when he was paired with Manson, his Corsi stats looked a lot better. Now, Manson is a good #4 who has excelled as a complement to Lindholm, but he's not good enough where he can carry defenders. For Fowler's Corsi stats to jump up like they did with a legitimate shutdown DFD such as Manson means less that Manson is carrying the pairing than Fowler is finally playing with a good player. Fowler's now paired with Vatanen -- both OFDs who love to push the play. Fowler's offensive style has dominated over Vatanen and Vatanen is then being forced into a more defensive role where he's really struggled. (I might note that Vatanen has been mostly sheltered throughout his career yet stats say that he's better than Fowler who hasn't been sheltered at all over the past few years)


My verdict on Fowler: His defensive game isn't his strong suit. He's good there and he unfairly gets **** on due to having defensively lackluster partners, but he shines when he's able to move the puck up the ice and make things happen in the offensive zone, not when he's being coached into being a DFD while being paired with one of the worst players in the league. According to the metrics, Fowler is one of the best transitional defenders in the league, but this gets ignored in favor of more popular stats like Corsi which don't exactly portray him in an endearing light.


And my verdict on Corsi: Better used as a metric for evaluating line/team performance than for individual performance/skill. There's just too many independent variables in what affects the metric to use it reliably for an individual player. A player's chemistry with his partner/linemates, his partner's or linemates' talent, his usage (zone starts and strength of opposition), and the system in which he plays can all have profound affects on a player's Corsi rating, as well as several other variables. (such as coaching skill and the strength of the surrounding team) Attempting to use it as a tell-all for overall skill is like stepping into a poorly-insulated refrigerator and attempting to tell the temperature outside. You just can't do it because that darned refrigeration system is getting in the way and marring your results. If this poorly-illustrated experiment doesn't do the job for you, let me put it this way: all scientific experiments have a potential to fail if there are significant variables affecting it, which is why having a controlled environment for experiments you conduct is very important. Corsi is no different, but a controlled environment is unfortunately impossible. Now, if you're using Corsi for the evaluation of a line's performance (not necessarily skill), you can eliminate the first two variables. Not enough to make it a tell-all, but no statistic is perfect. If you use Corsi for the evaluation of a team's performance, you can knock off usage for the most part, as well as obviously any negative or positive influences from overall team skill.

Thank you. Great post!

I agreed with every single word, which is impressive considering how long it was.

 

[Ol' Jase]

This has been done already.

Situation 1 - player takes a faceoff in the defensive zone; average CF% - 37.7%

Situation 2 - player takes a faceoff in the offensive zone; average CF% - 59.8%

Situation 3 - player takes a faceoff in the neutral zone; average CF% - 47.7%

Situation 4 - player jumps over the boards; average CF% - 50.9%

Those averages are from faceoff until leaving the ice or a whistle.

https://puckplusplus.com/2015/01/20...-lot-on-their-own-not-that-much-in-aggregate/

The isolated effects of situations 1 and 2 are massive, but situations 1 and 2 only occur in isolation about 8% of the time.

You can't just ignore 92% of the game.

But, as you have said many times, situation 4 doesn't exist. That is not a Zone Start.

Corsi numbers, as you have just presented them, dramatically swing depending on measured Zone Start.

Now, if you are of a mind that 82 game trends have more of a relevance to winning a single game that single game metrics do, I can't help you there.

Again, games are played one at a time. Something else "you guys" always seem to forget.

 

[Machinehead]

But, as you have said many times, situation 4 doesn't exist. That is not a Zone Start.

Oh it exists. It's just not a zone start.

That seems to be where we're not understanding each other. It's not a zone start but it's still a shift, and a player is going to take that shift and be part of corsi events.

Those corsi events still count towards his season total, and not towards his zone starts.

We don't throw that shift out because it's not a zone start.

Like if a player has 4 goals and scores a shorthanded goal, he now has 5 goals, but he still has 4 at even strength.

 

[Ol' Jase]

Oh it exists. It's just not a zone start.

That seems to be where we're not understanding each other. It's not a zone start but it's still a shift, and a player is going to take that shift and be part of corsi events.

Those corsi events still count towards his season total, and not towards his zone starts.

We don't throw that shift out because it's not a zone start.

Like if a player has 4 goals and scores a shorthanded goal, he now has 5 goals, but he still has 4 at even strength.

It's like I'm talking to a wall.

If you claim that it is irrefutable that one variable has no correlation with another variable and you are using 60% of one said variable and 100% of the other said variable, you have proved absolutely nothing due to the discrepancy in dependent variables.

Seriously, this is very basic stuff we are talking about. It doesn't make a goddamn difference if it's hockey or not. You have proved absolutely nothing based on the hypothesis you have put forward.

Again, I am floored I'm having this conversation with someone who claims to represent a community based entirely on statistical models.

 

[Machinehead]

It's like I'm talking to a wall.

If you claim that it is irrefutable that one variable has no correlation with another variable and you are using 60% of one said variable and 100% of the other said variable, you have proved absolutely nothing due to the discrepancy in dependent variables.

Seriously, this is very basic stuff we are talking about. It doesn't make a goddamn difference if it's hockey or not. You have proved absolutely nothing based on the hypothesis you have put forward.

Again, I am floored I'm having this conversation with someone who claims to represent a community based entirely on statistical models.

Because corsi happens 100% of the time and zone starts happen 40% of the time.

That's not my choice.

 

[Ol' Jase]

Because corsi happens 100% of the time and zone starts happen 40% of the time.

That's not my choice.

One more time. Statistically, you have proved absolutely nothing.

One of two things, you don't understand math or you think you can change the basic rules of mathematical applications.

Either way, holy ****, you haven't proved a damn thing.

 

[Machinehead]

I've proved that some concepts are not as simple as I thought they were.

 

[SladeWilson23]

Because corsi happens 100% of the time and zone starts happen 40% of the time.

That's not my choice.

He wants you to take the Corsi events from the 40% situations and split it up. What is a player's CF% just on o-zone shifts, and what is a player's cf% on d-zone shifts?

When trying to show how much zone starts affect corsi, you have to eliminate the non zone start Corsi, and only use the 40% data which is the zone start data.

 

[Machinehead]

He wants you to take the Corsi events from the 40% situations and split it up. What is a player's CF% just on o-zone shifts, and what is a player's cf% on d-zone shifts?

When trying to show how much zone starts affect corsi, you have to eliminate the non zone start Corsi, and only use the 40% data which is the zone start data.

I did that.

Average O-zone start - 59% CF

Average D-zone start - 37% CF

The effects in those isolated situations are profound. It's just that those isolated situations are actually extremely rare when you consider that there's over the boards, plus two other zones to start in.

 

[SladeWilson23]

I did that.

Average O-zone start - 59% CF

Average D-zone start - 37% CF

The effects in those isolated situations are profound. It's just that those isolated situations are actually extremely rare when you consider that there's over the boards, plus two other zones to start in.

IMO, neutral zone start should only be defined as face-offs from the center ice circle. The red line should dictate if o-zone vs d-zone start. But I digress.

 

[Machinehead]

IMO, neutral zone start should only be defined as face-offs from the center ice circle. The red line should dictate if o-zone vs d-zone start. But I digress.

I'd definitely be interested in how the numbers differ between actually in the zone, and in between the red and blue lines.

Don't think anyone has done that.

 

[nyr__1994]

I did that.

Average O-zone start - 59% CF

Average D-zone start - 37% CF

The effects in those isolated situations are profound. It's just that those isolated situations are actually extremely rare when you consider that there's over the boards, plus two other zones to start in.

I dont think extremely rare is the proper way to describe this. Your o-zone and d-zone are roughly 25% of your starts. That is a quarter of your shifts that are either starting in the o-zone with a 10% bump in CF% or the d-zone with a 10% reduction in CF%.

The difference in O vs D zone on 25% of the shifts should be enough to sway a players CF% 2-4%. a 4% swing in a players CF% according to "the community" is the difference between Tanner Glass and Patrice Bergeron (hyperbolic example).

 

[boopronger]

Again 22 out of the last 24 Cup winners were top 10 in shot differential. Not sure what part of that people can't grasp.

22 out of last 24 cup winners have an A in their name. A smart owner would name their team with an A in it to increase odds of success.

 

[projexns]

I dont think extremely rare is the proper way to describe this. Your o-zone and d-zone are roughly 25% of your starts. That is a quarter of your shifts that are either starting in the o-zone with a 10% bump in CF% or the d-zone with a 10% reduction in CF%.

The difference in O vs D zone on 25% of the shifts should be enough to sway a players CF% 2-4%. a 4% swing in a players CF% according to "the community" is the difference between Tanner Glass and Patrice Bergeron (hyperbolic example).

Here's a link to Bergeron's 3-year charts:

http://hockeyviz.com/static/img/player/cards/playerCard-bergepa85.png

Per 1,000 shifts this works out to:

500 starts on the fly (Corsi neutral)
220 face-offs in the neutral zone (Corsi neutral)
130 face-offs in the offensive zone (76 wins @ 58.5% FO)
150 face-offs in the defensive zone (62 losses @ 58.5% FO)

As per the bolded numbers, Bergeron had a Corsi advantage on 1.4% of his shifts, 14 shifts out of 1000 based just on zone-starts (and because of his face-off ability.)

Assuming 50% face-offs, he would've had a Corsi disadvantage on 1.0% of his shifts, 10 shifts out of 1,000.

 

[Bedards Dad]

22 out of last 24 cup winners have an A in their name. A smart owner would name their team with an A in it to increase odds of success.

The good old, "I don't have a leg to stand on so I'll make ano outrageous post" tactic. Well played.

 

[nyr__1994]

Here's a link to Bergeron's 3-year charts:

http://hockeyviz.com/static/img/player/cards/playerCard-bergepa85.png

Per 1,000 shifts this works out to:

500 starts on the fly (Corsi neutral)
220 face-offs in the neutral zone (Corsi neutral)
130 face-offs in the offensive zone (76 wins @ 58.5% FO)
150 face-offs in the defensive zone (62 losses @ 58.5% FO)

As per the bolded numbers, Bergeron had a Corsi advantage on 1.4% of his shifts, 14 shifts out of 1000 based just on zone-starts (and because of his face-off ability.)

Assuming 50% face-offs, he would've had a Corsi disadvantage on 1.0% of his shifts, 10 shifts out of 1,000.

I believe the CF% advantage/disadvantage was for zone starts, not sone starts with a face-off win or loss.

You essentially have a 20% swing in CF% based on O-zone starts and D-zone starts. While a minority of shifts start in one of these two zones, I believe that the CF% difference is enough to sway a players CF% a couple percentage points if they are getting heavy O-zone or D-zone starts.

Bergeron is pretty much 50/50 in his zone starts, but if you look a player that is 60-40 or so, and that workload difference may skew his CF enough to make a good player look bad.

 

[urho]

Again 22 out of the last 24 Cup winners were top 10 in shot differential. Not sure what part of that people can't grasp.

But isn't this quite a banal statement? A good team usually has better shot differential than its opponent most of the time. And more goals. I think it's quite obvious and not all that revealing...

 

[projexns]

I believe the CF% advantage/disadvantage was for zone starts, not sone starts with a face-off win or loss.

When you lose a face-off in the offensive zone, you're in the right place to generate a Corsi event but you don't have the puck. Zone start advantages/disadvantages are cut in half because of the 50/50 nature of face-offs.

You essentially have a 20% swing in CF% based on O-zone starts and D-zone starts. While a minority of shifts start in one of these two zones, I believe that the CF% difference is enough to sway a players CF% a couple percentage points if they are getting heavy O-zone or D-zone starts.

Numerically, the 20% swing is happening on only 20% of the events. 20% of 20% is 4%, swinging a 50/50 Corsi to 52/48.

Rarely does the difference seem to be that much. Here's two charts for the Blackhawks over three years, the first one is Corsi the 2nd one is zone-start adjusted Corsi. The differences are minimal except for the guys with extreme deployments (Kruger defensively, Teravainen offensively):

http://stats.hockeyanalysis.com/rat...rds&minutes=1000&disp=1&sort=PCT&sortdir=DESC

http://stats.hockeyanalysis.com/rat...rds&minutes=1000&disp=1&sort=PCT&sortdir=DESC

Bergeron is pretty much 50/50 in his zone starts, but if you look a player that is 60-40 or so, and that workload difference may skew his CF enough to make a good player look bad.

Here's a link to Patrick Kane's 3-year charts:

http://hockeyviz.com/static/img/player/cards/playerCard-kanexpa88.png

Unlike Bergeron who you want taking face-offs as much as possible, you can change Kane on the fly more often.

Per 1,000 shifts this works out to:

620 starts on the fly (Corsi neutral)
160 face-offs in the neutral zone (Corsi neutral)
130 face-offs in the offensive zone (65 wins @ 50% FO)
90 face-offs in the defensive zone (45 losses @ 50% FO)

As per the bolded numbers, Kane had a Corsi advantage on 2.0% of his shifts, 20 shifts out of 1000 based on zone-starts alone.

 

[ps241]

I hate using +/- because it's so dependent on the team you play on. An absolute train wreck of a team might not have a single plus player, even though their best defender might be better than the #7 on a championship squad, who might be +2 or +3 in the same stat.

That being said, Corsi is misconstrued and too many people recite stats off a page without knowing anything about the player. What does a bad CF% rating say? Does the player have missed assignments aplenty in his own end? Or does he continually make dumbass turnovers? (Exhibit A: Kevin Bieksa)


Now, there's a problem here as well: let's say that this horrible defender (cough cough Bieksa) is paired with a good player. This unnamed horrible player can't contribute to the transition game, he is constantly caught up the ice with his pants down on the rush back, he's a negative player offensively with only a decent slapshot to contribute to offensive zone play, and his defensive coverage is nothing short of horrible. Obviously he just might drag down his partner, right? His partner is good (albeit inconsistent at times) defensively, he's a fantastic skater and very good on the transition, and his offensive zone play is very good and is complemented with a much-improved shot. A solid #2.


Now, Fowler isn't a shutdown defender, regardless of what Boudreau tried to make him into. Throughout his career, he's shined his brightest in an offensive role. Though his offseason training and his changes to his playing style deserve much of the credit for this, his usage by RC is one of the big reasons he's having an offensive renaissance. He's seeing plenty of PP time, he's being more aggressive offensively, and he's been employed in a much more offensive role. He's not a shutdown defender, and trying to turn him into one is likely not going to aid his development any. He's not as bad defensively as people seem to think; from Lovejoy to Bieksa to now Vatanen, he's never been paired with defensive stalwarts. In fact, when he was paired with Manson, his Corsi stats looked a lot better. Now, Manson is a good #4 who has excelled as a complement to Lindholm, but he's not good enough where he can carry defenders. For Fowler's Corsi stats to jump up like they did with a legitimate shutdown DFD such as Manson means less that Manson is carrying the pairing than Fowler is finally playing with a good player. Fowler's now paired with Vatanen -- both OFDs who love to push the play. Fowler's offensive style has dominated over Vatanen and Vatanen is then being forced into a more defensive role where he's really struggled. (I might note that Vatanen has been mostly sheltered throughout his career yet stats say that he's better than Fowler who hasn't been sheltered at all over the past few years)


My verdict on Fowler: His defensive game isn't his strong suit. He's good there and he unfairly gets **** on due to having defensively lackluster partners, but he shines when he's able to move the puck up the ice and make things happen in the offensive zone, not when he's being coached into being a DFD while being paired with one of the worst players in the league. According to the metrics, Fowler is one of the best transitional defenders in the league, but this gets ignored in favor of more popular stats like Corsi which don't exactly portray him in an endearing light.


And my verdict on Corsi: Better used as a metric for evaluating line/team performance than for individual performance/skill. There's just too many independent variables in what affects the metric to use it reliably for an individual player. A player's chemistry with his partner/linemates, his partner's or linemates' talent, his usage (zone starts and strength of opposition), and the system in which he plays can all have profound affects on a player's Corsi rating, as well as several other variables. (such as coaching skill and the strength of the surrounding team) Attempting to use it as a tell-all for overall skill is like stepping into a poorly-insulated refrigerator and attempting to tell the temperature outside. You just can't do it because that darned refrigeration system is getting in the way and marring your results. If this poorly-illustrated example doesn't do the job for you, let me put it this way: all scientific experiments have a potential to fail if there are significant variables affecting it, which is why having a controlled environment for experiments you conduct is very important. Corsi is no different, but a controlled environment is unfortunately impossible. Now, if you're using Corsi for the evaluation of a line's performance (not necessarily skill), you can eliminate the first two variables. Not enough to make it a tell-all, but no statistic is perfect. If you use Corsi for the evaluation of a team's performance, you can knock off usage for the most part, as well as obviously any negative or positive influences from overall team skill.

Thank you for this well thought out post.

One of the things that really hurt Bogosian's development in Winnipeg is Huddy and Noel kept trying to force Zach into a shutdown defenders role and that was not playing to his strengths at all.