HFBoards

Go Back   HFBoards > NHL Eastern Conference > Metropolitan Division > Philadelphia Flyers
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
Notices

Flyers’ Meszaros Quietly Having Banner Year

Reply
 
Thread Tools
Old
01-14-2011, 07:49 PM
  #276
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by Jester View Post
Yeah, their opinions are largely worthless on anything past basic algebra, too.



To be perfectly honest, I do not know how any part of this paragraph addressed the point I made.

Just because a bunch of people that don't understand statistics in the slightest don't pay attention to a stat, does not mean that a stat is without merit and/or value.

CORSI is a relatively blunt statistic in the grand scheme of things. However, it does do a good job of tracking puck possession. If you don't think puck possession is an important factor in the course of a hockey game, I don't know what to tell you.
You don't think these guys don't know what they are? They do have other numbers like attack time and such. Shots themselves are too volatile to base a theory around it. Coaches know on the bench whenever they're never having the puck, and how much they're supposed to have the puck because of how their systems are designed.

Quote:
Originally Posted by Jester View Post
BTW, the Corsi statistic developed within the Sabres organization... so it's unclear to me why you would assume hockey people necessarily look down on it.
And a lot of good that's done them. Jim Corsi is also a goaltending coach. Goaltenders aren't decision makers.

__________________
Philadelphia's Real Alternative
(ynotradio.net)

Stop Feeding the Rumor-Monger

"I wonder if Norstrom has Forsberg's spleen mounted on his wall." - KINGS17

My 50 Favorite Albums of 2014 (sorry it's late)
GKJ is offline   Reply With Quote
Old
01-14-2011, 07:51 PM
  #277
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Quote:
Originally Posted by go kim johnsson 514 View Post
You don't think these guys don't know what they are? They do have other numbers like attack time and such. Shots themselves are too volatile to base a theory around it. Coaches know on the bench whenever they're never having the puck, and how much they're supposed to have the puck because of how their systems are designed.
Seriously? SERIOUSLY?

Quote:
And a lot of good that's done them. Jim Corsi is also a goaltending coach. Goaltenders aren't decision makers.
Whiskey. Tango. Foxtrot.

Jester is offline   Reply With Quote
Old
01-14-2011, 07:59 PM
  #278
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by Jester View Post
Seriously? SERIOUSLY?



Whiskey. Tango. Foxtrot.
And that's where I started when someone posted an article from some idiot who thinks he's a genius because he reads Corsi ratings and decides he knows who is good and who is bad.

The Caps have been good with it, but how's it worked for them in the playoffs?

It's only a good tool because someone said it is, just like it's not a good tool because others say it's not. It doesn't take into account deflections, missed shots, and blocked shots, therefore you're only using a partial variable. If you have a team that keeps missing the nets with their shots, your Corsi rating is going to look really good. Just like when you have defensive assignments and allow only perimeter shots that your goalie sees, it won't look good. The law of averages don't narrow it down enough.

GKJ is offline   Reply With Quote
Old
01-14-2011, 08:07 PM
  #279
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Quote:
Originally Posted by go kim johnsson 514 View Post
And that's where I started when someone posted an article from some idiot who thinks he's a genius because he reads Corsi ratings and decides he knows who is good and who is bad.

The Caps have been good with it, but how's it worked for them in the playoffs?

It's only a good tool because someone said it is, just like it's not a good tool because others say it's not. It doesn't take into account deflections, missed shots, and blocked shots, therefore you're only using a partial variable. If you have a team that keeps missing the nets with their shots, your Corsi rating is going to look really good. Just like when you have defensive assignments and allow only perimeter shots that your goalie sees, it won't look good. The law of averages don't narrow it down enough.
GKJ, a moment ago you used "margin of error" to make some bizarre insinuation relative to averaged shots. You realize "margin of error" is used to make a statement about the flexibility of a prediction... not so you can somehow include a multitude of "outliers" or whatever else.

Folks like CORSI because there is some empirical evidence of correlation between CORSI and success and failure on the ice, not because someone said it is a good tool. Is it the end all be all? Nope, no one statistic is going to be. On top of that, a stat like CORSI is a model of success and failure, and all the erratic outliers and whatnot will go against it.

CORSI will not be 100% accurate by its very nature. If you want to make an argument against CORSI, you need to breakdown its correlation to what it claims it's telling you. Not cite a bunch of nameless and faceless "professionals with NHL connections" who may or may not understand the first thing about a statistical argument.

Because, honestly, who the F cares what they think if they don't understand the statistical argument that is being made in the first place? I don't, and you shouldn't either because no matter how much they may know about hockey, they don't necessarily have the first clue how to assess a statistical model.

Jester is offline   Reply With Quote
Old
01-14-2011, 08:22 PM
  #280
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Just so we have a basis, here's an example of adjusted CORSI for this year:



Difference between it and straight CORSI:

Quote:
only even strength events with the score close (when the score margin is one or zero in the first two periods or tied in the third period and overtime) are included

a strength of schedule correction is made with reference to both game location and oppositional strength
http://objectivenhl.blogspot.com/201...goalposts.html

Now, bear in mind that part of the argument about the application of CORSI is that it does not account for the level of goaltending a team receives. Obviously, if you're doing everything right but your goalie is a sieve, it isn't going to matter. Conversely, stellar goaltending can push you ahead.

In any event, it cannot account for lucky bounces for and against.

Jester is offline   Reply With Quote
Old
01-14-2011, 08:23 PM
  #281
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by Jester View Post
GKJ, a moment ago you used "margin of error" to make some bizarre insinuation relative to averaged shots. You realize "margin of error" is used to make a statement about the flexibility of a prediction... not so you can somehow include a multitude of "outliers" or whatever else.

Folks like CORSI because there is some empirical evidence of correlation between CORSI and success and failure on the ice, not because someone said it is a good tool. Is it the end all be all? Nope, no one statistic is going to be. On top of that, a stat like CORSI is a model of success and failure, and all the erratic outliers and whatnot will go against it.

CORSI will not be 100% accurate by its very nature. If you want to make an argument against CORSI, you need to breakdown its correlation to what it claims it's telling you. Not cite a bunch of nameless and faceless "professionals with NHL connections" who may or may not understand the first thing about a statistical argument.

Because, honestly, who the F cares what they think if they don't understand the statistical argument that is being made in the first place? I don't, and you shouldn't either because no matter how much they may know about hockey, they don't necessarily have the first clue how to assess a statistical model.

If Corsi isn't 100% accurate, you're only showing trends. Trends change, and trends change back, doing so at undefined rates. There's no reason for anyone to base a whole scientific theory on it when it's very definition makes it a volatile variable, where the "erratic" outliers have too great of an influence on the averages.

GKJ is offline   Reply With Quote
Old
01-14-2011, 08:25 PM
  #282
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Quote:
Originally Posted by go kim johnsson 514 View Post
If Corsi isn't 100% accurate, you're only showing trends. Trends change, and trends change back, doing so at undefined rates. There's no reason for anyone to base a whole scientific theory on it when it's very definition makes it a volatile variable, where the "erratic" outliers have too great of an influence on the averages.
In all seriousness, all your proving is that you're using words you don't really understand.

No statistical model is 100% accurate. Not one.

This isn't a "scientific theory" either. Outliers only have "too great" an influence on averages if you're focusing on small samples. An individual year (players and teams have down years all the time)... so on and so forth.

Jester is offline   Reply With Quote
Old
01-14-2011, 08:43 PM
  #283
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by Jester View Post
In all seriousness, all your proving is that you're using words you don't really understand.

No statistical model is 100% accurate. Not one.

This isn't a "scientific theory" either. Outliers only have "too great" an influence on averages if you're focusing on small samples. An individual year (players and teams have down years all the time)... so on and so forth.
My argument is that it's not accurate enough. Blocked shots, and missed shots, are not small samples (if that's what you mean by it). It's not like I'm talking about hitting the post and goaltenders looking at jumbotrons. These are things that happen with regularity, if not with more regularity at times when they're put together, and are not part of the Corsi formula (unless I am mistaken and/or it was changed, I'm sure you'll correct me if so).

GKJ is offline   Reply With Quote
Old
01-14-2011, 08:59 PM
  #284
FlyHigh
Registered User
 
FlyHigh's Avatar
 
Join Date: Jan 2004
Posts: 28,156
vCash: 500
Send a message via AIM to FlyHigh Send a message via MSN to FlyHigh
LMAO at turning the top-10 into a top-11 just so you can include Dallas to help your argument.

Classic.

EDIT: Even more LMAO at calling a 25% deviation "within the margin of error"

SERIOUSLY?

FlyHigh is offline   Reply With Quote
Old
01-14-2011, 09:03 PM
  #285
FlyHigh
Registered User
 
FlyHigh's Avatar
 
Join Date: Jan 2004
Posts: 28,156
vCash: 500
Send a message via AIM to FlyHigh Send a message via MSN to FlyHigh
Quote:
Originally Posted by go kim johnsson 514 View Post
If Corsi isn't 100% accurate, you're only showing trends. Trends change, and trends change back, doing so at undefined rates. There's no reason for anyone to base a whole scientific theory on it when it's very definition makes it a volatile variable, where the "erratic" outliers have too great of an influence on the averages.
Please tell me this is some kind of sarcasm.

FlyHigh is offline   Reply With Quote
Old
01-14-2011, 09:12 PM
  #286
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by FlyHigh View Post
LMAO at turning the top-10 into a top-11 just so you can include Dallas to help your argument.

Classic.

EDIT: Even more LMAO at calling a 25% deviation "within the margin of error"

SERIOUSLY?
LMAO THE DIFFERENCE WAS .1 PER GAME, LOLLERSKATES

I'm not sure if you're doing that so you can act cool on a message board because Jester said "seriously,?" but the reasoning is the difference was .1 shot per game. So, yes, seriously. The difference over 82 games was 10 shots.

The team that won the Cup last year allowed the fewest shots, the Flyers were 5th. If you went that route I probably wouldn't have said anything (I go defense first anyways).

Quote:
Originally Posted by FlyHigh View Post
Please tell me this is some kind of sarcasm.
Why don't you just make a counterpoint to it like Jester is doing, instead of acting all high and holy about it? That is, assuming you have one instead of acting like you can't believe someone argues against it.

GKJ is offline   Reply With Quote
Old
01-14-2011, 09:22 PM
  #287
FlyHigh
Registered User
 
FlyHigh's Avatar
 
Join Date: Jan 2004
Posts: 28,156
vCash: 500
Send a message via AIM to FlyHigh Send a message via MSN to FlyHigh
Quote:
Originally Posted by go kim johnsson 514
Why don't you just make a counterpoint to it like Jester is doing, instead of acting all high and holy about it?
Because Jester's already doing it, so why would I do it too? It's still funny to watch.

FlyHigh is offline   Reply With Quote
Old
01-14-2011, 09:32 PM
  #288
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by FlyHigh View Post
Because Jester's already doing it, so why would I do it too? It's still funny to watch.
Well, I'm confused about something, then. He said it's not a 100% accurate. Why is it funny that I essentially agreed with him? I only repeated what he said. Someone's keeping track of it, so it's information. My saying that it's a trend is an elaboration of my disagreement on the reliability of Corsi ratings. Otherwise, if you're just going the "lol I'm smarter than the moderator," then congratulations, you're that you're smarter than someone on an internet message board (as someone who hasn't completed college, I won't argue). Sorry though, I've run out of gold stars.

GKJ is offline   Reply With Quote
Old
01-14-2011, 09:39 PM
  #289
FlyHigh
Registered User
 
FlyHigh's Avatar
 
Join Date: Jan 2004
Posts: 28,156
vCash: 500
Send a message via AIM to FlyHigh Send a message via MSN to FlyHigh
Well, statistics is essentially the study of trends, there is no such thing as a "100% accurate" statistic, you are always going to find data points that may not fit the curve. The question is whether there are enough of these data points (outliers) to disprove the statistic.

But you can't just find one exception in a Corsi stat and say that it automatically devalues the statistic.

FlyHigh is offline   Reply With Quote
Old
01-14-2011, 09:46 PM
  #290
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by FlyHigh View Post
Well, statistics is essentially the study of trends, there is no such thing as a "100% accurate" statistic, you are always going to find data points that may not fit the curve. The question is whether there are enough of these data points (outliers) to disprove the statistic.

But you can't just find one exception in a Corsi stat and say that it automatically devalues the statistic.
My issue is that there is not enough in Corsi ratings to make a true enough determination. You rely on it to a degree less so than I. I disagree there is enough information in the rating itself. It excludes missed shots and blocked shots, no? They happen on a regular basis, but are not included in those numbers. I don't know how you can essentially ignore them.

GKJ is offline   Reply With Quote
Old
01-14-2011, 09:59 PM
  #291
FlyHigh
Registered User
 
FlyHigh's Avatar
 
Join Date: Jan 2004
Posts: 28,156
vCash: 500
Send a message via AIM to FlyHigh Send a message via MSN to FlyHigh
Quote:
Originally Posted by go kim johnsson 514 View Post
My issue is that there is not enough in Corsi ratings to make a true enough determination. You rely on it to a degree less so than I. I disagree there is enough information in the rating itself. It excludes missed shots and blocked shots, no? They happen on a regular basis, but are not included in those numbers. I don't know how you can essentially ignore them.
Of course it's not perfect. I've pointed out MANY times on this board that no QUALCOMP stat is perfect.

Your original point regarding this stat was...

Quote:
Completely useless, and most writers with connections to NHL people will tell you the same thing.
I think that's a dumb assessment tbh. There's a long gap between "completely useless" and "perfect."

Corsi parses a lot more data than QUALCOMP which makes it superior IMO, it's not perfect, but it's far from useless. If hockey professionals don't embrace it, it's because hockey is pretty far behind the advanced stats revolution (see Norris trophy voting last year).

FlyHigh is offline   Reply With Quote
Old
01-14-2011, 10:07 PM
  #292
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Well, I only said that because they get to talk to those people about it (that's why they wrote about it back when it was a new concept). One guy came up with it, there's a lot who don't bother. If coaches and front office people used it (beyond a very select few), they'd all be talking about it acting like they invented it. It's not the whole sabremetrics philosophy, it's only that one thing they're/I'm against for all the reasons I stated. I just don't see it as close enough for it to be practical because there's not enough in it.

GKJ is offline   Reply With Quote
Old
01-14-2011, 11:21 PM
  #293
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Quote:
Originally Posted by go kim johnsson 514 View Post
Well, I only said that because they get to talk to those people about it (that's why they wrote about it back when it was a new concept). One guy came up with it, there's a lot who don't bother. If coaches and front office people used it (beyond a very select few), they'd all be talking about it acting like they invented it. It's not the whole sabremetrics philosophy, it's only that one thing they're/I'm against for all the reasons I stated. I just don't see it as close enough for it to be practical because there's not enough in it.
Except you don't have a single bit of supporting evidence to back up that statement. Unless you're willing to provide some data that attacks the correlation rate for CORSI to what it claims to be stating?

Saying "Oh, but Pronger shoots it on net!" is not constructive in the slightest. In fact, that's a highly ignorant critique of a statistical model. As said--and as has been noticeable in the past when you've engaged in statistical discussions--you are REALLY out of your element and don't know what you're talking about. You've thrown around "margin of error" in here in an entirely incorrect way, for example.

Whether Joe Schmoe in whatever NHL organization you want to cite embraces CORSI or not is in no way indicative that the claims of the statistic are accurate or not. So, you really need to give up that tangent. The vast majority of people in hockey really have no clue on this stuff. The majority of the players never spent any time in college, and a lot of the professional folks followed similar routes to get into the profession.

Simply put, they have zero training or experience that would lead to them having any sort of useful opinion on a statistical model. Moreover, advanced statistics are relatively new in hockey and we're just at the beginning of them having any sort of mainstream outlet. It took *ing years before baseball began to embrace it, and we are less than a decade removed from Moneyball... which was a *ing bomb dropped on the baseball universe (and what that book outlines is an incredibly simple concept, nowhere near some of the stuff we talk about when it comes to really complex stats).

Jester is offline   Reply With Quote
Old
01-15-2011, 12:58 AM
  #294
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by Jester View Post
Except you don't have a single bit of supporting evidence to back up that statement. Unless you're willing to provide some data that attacks the correlation rate for CORSI to what it claims to be stating?

Saying "Oh, but Pronger shoots it on net!" is not constructive in the slightest. In fact, that's a highly ignorant critique of a statistical model. As said--and as has been noticeable in the past when you've engaged in statistical discussions--you are REALLY out of your element and don't know what you're talking about. You've thrown around "margin of error" in here in an entirely incorrect way, for example.

Whether Joe Schmoe in whatever NHL organization you want to cite embraces CORSI or not is in no way indicative that the claims of the statistic are accurate or not. So, you really need to give up that tangent. The vast majority of people in hockey really have no clue on this stuff. The majority of the players never spent any time in college, and a lot of the professional folks followed similar routes to get into the profession.

Simply put, they have zero training or experience that would lead to them having any sort of useful opinion on a statistical model. Moreover, advanced statistics are relatively new in hockey and we're just at the beginning of them having any sort of mainstream outlet. It took *ing years before baseball began to embrace it, and we are less than a decade removed from Moneyball... which was a *ing bomb dropped on the baseball universe (and what that book outlines is an incredibly simple concept, nowhere near some of the stuff we talk about when it comes to really complex stats).
I don't have real data, that's why you're not accepting of my rejection of the Corsi rating. I know you're a guy who wants hard data, so that's not going to fly. And I'm not a math person, so there's that too. We (and others) have similar opinions on a fair number of players on the team (at least), and you're looking at the numbers a lot more than most others here do. I don't know every number that guy has lined up, but I've always known what a Corsi rating is. I can buy into other sabremetrics, but in my opinion Corsi ratings don't do anything. I feel too much happens in a game that it the Corsi rating itself doesn't give enough accuracy. I can't quantify that. It's almost impossible to. Missed shots, blocked shots, systematical/strategical changes...just too many things that happen outside of shots on goal (and not rare occasions like goaltenders looking at jumbotrons, and sticks shattering on breakaways), and then say "this, that, and the other thing is a fact because the Corsi rating says so." But one can't put a value on that. Well, maybe if they were doing a college thesis or something on it, but I don't think that was the intention of any sabremetric.

Teams win this league win enough games as it is when they get out-shot, so the difference can't be a negligible gap between Corsi ratings and "100% accuracy." I've written too much, probably.

GKJ is offline   Reply With Quote
Old
01-15-2011, 02:02 AM
  #295
Giroux tha Damaja
Registered User
 
Giroux tha Damaja's Avatar
 
Join Date: Apr 2009
Location: Mount Holly, NJ
Country: United States
Posts: 9,234
vCash: 500
Send a message via AIM to Giroux tha Damaja
The amount of useful information gleaned from the gathering and investigation of complex statistics in baseball is not a benefit that can be similarly enjoyed in hockey IMO. Baseball is a painfully simple game. The options before you as a player at any point in the game are extremely restricted, and because of this the game is easy to track and model. That simplicity has also allowed teams to get the sport damn near to a science . . . and made the smallest edges worth pursuing.

Hockey is an entirely different animal IMO. Everything about the game is constantly in flux and chaotic. The derivation of complex statistics and subsequent examining of the correlations between them seems like a pretty ****** way to analyze/quantify the dynamics present with in the game, particularly when we already understand it pretty well. It's like an IT guy taking apart a ten year old PC with a stone hammer so he can reverse engineer it: an absolutely piss poor way to research the workings of something you (presumably) already understand.

I don't mind looking at hockey stats to quantify some simple things, because human memory is influenced largely by the emotions of the game. But looking at a correlation between two esoteric statistics derived from other statistics that are subjectively compiled anyway....I've got no use for it.

Giroux tha Damaja is offline   Reply With Quote
Old
01-15-2011, 02:19 AM
  #296
Valhoun*
 
Join Date: Sep 2006
Location: West Chester, PA
Posts: 10,311
vCash: 500
Send a message via AIM to Valhoun*
Quote:
Originally Posted by Giroux tha Damaja View Post
The amount of useful information gleaned from the gathering and investigation of complex statistics in baseball is not a benefit that can be similarly enjoyed in hockey IMO. Baseball is a painfully simple game. The options before you as a player at any point in the game are extremely restricted, and because of this the game is easy to track and model. That simplicity has also allowed teams to get the sport damn near to a science . . . and made the smallest edges worth pursuing.

Hockey is an entirely different animal IMO. Everything about the game is constantly in flux and chaotic. The derivation of complex statistics and subsequent examining of the correlations between them seems like a pretty ****** way to analyze/quantify the dynamics present with in the game, particularly when we already understand it pretty well. It's like an IT guy taking apart a ten year old PC with a stone hammer so he can reverse engineer it: an absolutely piss poor way to research the workings of something you (presumably) already understand.

I don't mind looking at hockey stats to quantify some simple things, because human memory is influenced largely by the emotions of the game. But looking at a correlation between two esoteric statistics derived from other statistics that are subjectively compiled anyway....I've got no use for it.
Believe it or not, your arguments are fairly close to a lot of the arguments made by those that resisted Sabrmetrics in baseball. Baseball is a superstitious sport, likely even moreso than hockey. It was hard for proponents of advanced metrics to overcome the "go with your gut" and "tradition says to play this way" attitudes that were prevalent throughout the league. Hell, superstition and tradition still play a huge role in baseball. Just look at every team that only uses its closer in the final inning instead of when the team really needs to get through an inning to preserve the lead. It absolutely defies all logic.

Advanced statistics in hockey are still in their infancy. It will be another 10 or 20 years before they are fully accepted. Some will go away, others will be created, and many will continue to be tweaked.

Hockey does seem like orders of magnitude more complex than baseball but there are still a fairly small number of things that any given player can be doing at any given time. The puck can only be in a finite number of zones, FOs can only won or lost.

The Xs and Os of hockey are often overlooked. But just see what Bylsma does for the Pens versus Boudreau's more emotional style.

Valhoun* is offline   Reply With Quote
Old
01-15-2011, 02:22 AM
  #297
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Quote:
Originally Posted by go kim johnsson 514 View Post
I don't have real data, that's why you're not accepting of my rejection of the Corsi rating. I know you're a guy who wants hard data, so that's not going to fly. And I'm not a math person, so there's that too. We (and others) have similar opinions on a fair number of players on the team (at least), and you're looking at the numbers a lot more than most others here do. I don't know every number that guy has lined up, but I've always known what a Corsi rating is. I can buy into other sabremetrics, but in my opinion Corsi ratings don't do anything. I feel too much happens in a game that it the Corsi rating itself doesn't give enough accuracy. I can't quantify that. It's almost impossible to. Missed shots, blocked shots, systematical/strategical changes...just too many things that happen outside of shots on goal (and not rare occasions like goaltenders looking at jumbotrons, and sticks shattering on breakaways), and then say "this, that, and the other thing is a fact because the Corsi rating says so." But one can't put a value on that. Well, maybe if they were doing a college thesis or something on it, but I don't think that was the intention of any sabremetric.

Teams win this league win enough games as it is when they get out-shot, so the difference can't be a negligible gap between Corsi ratings and "100% accuracy." I've written too much, probably.
Well, CORSI can and does include missed shots and blocked shots... so maybe you don't know CORSI as well as you think you do. The adjusted CORSI I noted above normalizes for even strength and games that are close in score (so not a blow out where stupid things are happening, or a team is forced to really open up and play different than they want to).

So, suddenly maybe the CORSI statistical value is more robust than you thought, huh?

The issue here isn't that I want hard facts, or your not a math guy. You're simply talking about crap that you don't really understand. It's clear you haven't really looked into what CORSI is, you've probably read someone criticizing it that didn't really understand it themselves, and you haven't read any in depth discussion of the strengths and weaknesses of CORSI as an analytical tool.

And teams certainly win when they get outshot. The Flyers were a case in point a couple years ago. However, what do we know about that team? They struggled desperately to get puck possession, and, therefore, rarely "controlled" the flow of play. They also relied on finishing chances at a very efficient rate, and received good goaltending from Biron.

CORSI was quite accurate about that team... didn't mean they couldn't win games, just other parameters had to be in the right spot in order for that to play out successfully for 'em.

Quote:
Originally Posted by Giroux tha Damaja View Post
The amount of useful information gleaned from the gathering and investigation of complex statistics in baseball is not a benefit that can be similarly enjoyed in hockey IMO. Baseball is a painfully simple game. The options before you as a player at any point in the game are extremely restricted, and because of this the game is easy to track and model. That simplicity has also allowed teams to get the sport damn near to a science . . . and made the smallest edges worth pursuing.

Hockey is an entirely different animal IMO. Everything about the game is constantly in flux and chaotic. The derivation of complex statistics and subsequent examining of the correlations between them seems like a pretty ****** way to analyze/quantify the dynamics present with in the game, particularly when we already understand it pretty well. It's like an IT guy taking apart a ten year old PC with a stone hammer so he can reverse engineer it: an absolutely piss poor way to research the workings of something you (presumably) already understand.

I don't mind looking at hockey stats to quantify some simple things, because human memory is influenced largely by the emotions of the game. But looking at a correlation between two esoteric statistics derived from other statistics that are subjectively compiled anyway....I've got no use for it.
Anything can be modeled statistically given sufficient data markers. Anything. And, importantly, far more complicated things get modeled than the sport of hockey.

How accurate a model is as a predictive tool is another question, however. Moreover, the use of statistics only gets you so far. They represent a "best guess" and nothing more. As noted above, no statistical model is 100% accurate in its predictions.

Jester is offline   Reply With Quote
Old
01-15-2011, 03:10 AM
  #298
GKJ
Global Moderator
Entertainment
 
GKJ's Avatar
 
Join Date: Feb 2002
Location: Do not trade plz
Country: United States
Posts: 113,345
vCash: 50
Quote:
Originally Posted by Jester View Post
Well, CORSI can and does include missed shots and blocked shots... so maybe you don't know CORSI as well as you think you do. The adjusted CORSI I noted above normalizes for even strength and games that are close in score (so not a blow out where stupid things are happening, or a team is forced to really open up and play different than they want to).

So, suddenly maybe the CORSI statistical value is more robust than you thought, huh?

The issue here isn't that I want hard facts, or your not a math guy. You're simply talking about crap that you don't really understand. It's clear you haven't really looked into what CORSI is, you've probably read someone criticizing it that didn't really understand it themselves, and you haven't read any in depth discussion of the strengths and weaknesses of CORSI as an analytical tool.

And teams certainly win when they get outshot. The Flyers were a case in point a couple years ago. However, what do we know about that team? They struggled desperately to get puck possession, and, therefore, rarely "controlled" the flow of play. They also relied on finishing chances at a very efficient rate, and received good goaltending from Biron.

CORSI was quite accurate about that team... didn't mean they couldn't win games, just other parameters had to be in the right spot in order for that to play out successfully for 'em.
Corsi ratings, originally as I recall (and not from some dude making it up, it was the definition he was given when asked) is that it's shots on net, unless that was adjusted over time (each type is kept separate as it is, that's what I tend to look at). In fact, that was the reason a lot of people just threw it out, it was not a full reading.

The Flyers are a case in point right now. A team that stresses puck possession is the 2nd best team in the league when being outshot. In fact, teams across the league are actually better when they being outshot. There's nothing for the quality of shots/chances (would be nearly impossible, that's why teams review tape).

I'd be more interested in attack time/60 and scoring chances (and quality of) because it still looks good if someone gets a +10 for it, but two bad giveaways/lapses end up in your net, and you're in the dog house. People talked about this when Randy Jones was here as everyone blamed him, even though he had favorable numbers.

GKJ is offline   Reply With Quote
Old
01-15-2011, 08:50 AM
  #299
FlyHigh
Registered User
 
FlyHigh's Avatar
 
Join Date: Jan 2004
Posts: 28,156
vCash: 500
Send a message via AIM to FlyHigh Send a message via MSN to FlyHigh
Quote:
Originally Posted by Giroux tha Damaja View Post
The amount of useful information gleaned from the gathering and investigation of complex statistics in baseball is not a benefit that can be similarly enjoyed in hockey IMO. Baseball is a painfully simple game. The options before you as a player at any point in the game are extremely restricted, and because of this the game is easy to track and model. That simplicity has also allowed teams to get the sport damn near to a science . . . and made the smallest edges worth pursuing.

Hockey is an entirely different animal IMO. Everything about the game is constantly in flux and chaotic. The derivation of complex statistics and subsequent examining of the correlations between them seems like a pretty ****** way to analyze/quantify the dynamics present with in the game, particularly when we already understand it pretty well. It's like an IT guy taking apart a ten year old PC with a stone hammer so he can reverse engineer it: an absolutely piss poor way to research the workings of something you (presumably) already understand.

I don't mind looking at hockey stats to quantify some simple things, because human memory is influenced largely by the emotions of the game. But looking at a correlation between two esoteric statistics derived from other statistics that are subjectively compiled anyway....I've got no use for it.
You also have to understand that the "advanced hockey stats" are much less advanced than the baseball ones.

I have zero idea what the formulas for xFip and UZR are, but it probably takes a lot of observation and recording of data.

But stats like GFON/GFOFF and GAON/GAOFF are really pretty simple. Obviously you need to still watch the games (Carle had an outstanding GAON/60 last year because of Pronger), but they still have a lot of use.

I do tend to be fairly mistrustful of the QUALCOMP stats, even Corsi, and I don't know if there's a better way to design that kind of stat, but I don't think you can simply discount them.

FlyHigh is offline   Reply With Quote
Old
01-15-2011, 02:35 PM
  #300
Jester
Registered User
 
Jester's Avatar
 
Join Date: Jul 2004
Location: St. Andrews
Country: Scotland
Posts: 34,075
vCash: 500
Quote:
Originally Posted by go kim johnsson 514 View Post
Corsi ratings, originally as I recall (and not from some dude making it up, it was the definition he was given when asked) is that it's shots on net, unless that was adjusted over time (each type is kept separate as it is, that's what I tend to look at). In fact, that was the reason a lot of people just threw it out, it was not a full reading.
Well, that's not really what CORSI is... or how it can be applied if you place more nuance into it.

Quote:
The Flyers are a case in point right now. A team that stresses puck possession is the 2nd best team in the league when being outshot. In fact, teams across the league are actually better when they being outshot. There's nothing for the quality of shots/chances (would be nearly impossible, that's why teams review tape).
Actually, this isn't true. I've done it in the past (specifically to cite the problems with the Flyers a couple years ago under Stevens) and there's a pretty strong correlation between positive shot diff and being a winning team.

However, I think you're focusing on Win %... which is the wrong way to look at it. For example, Chicago had a .629 WinPct last year when they outshot their opponent... and a .667 rate when they were outshot. Aha! You'd say. Well, that ignores that the Hawks had 44 wins when outshooting their opponents, to 8 when they were outshot. The sample size is grossly out of whack if you're just looking at when they won games whilst being outshot.

16 teams had more wins when being outshot last year.

SJ
PHX

STL
NYR
CGY
BUF
MTL

NYI
CAR
MIN
COL
CBJ
TBL
ATL
FLA
EDM

The 5 bolded made the playoffs. What do we know about those teams?

SJ - 3rd in the league
PHX - 3rd in the league
BUF - 1st in the league
MTL - 5th in the league
COL - 10th in the league (and COL was viewed as a "lucky" team in CORSI evaluation)

CORSI claims to be roughly 75% of the story in evaluating a team. I'll let you ponder what they say the other 25% is.

Here is the run for this year:

MIN
ANA
DAL
PHX
NYR
COL
PHI
CAR
ATL
EDM
NYI
BOS
FLA
TOR
STL
OTT
NJD

17 teams with more wins whilst being outshot. So, as of now, you have 8 teams in the playoffs (though, Anaheim probably wouldn't be there with games in hand played out).

Boston 1st (by a ridiculous margin)
Anaheim 4th
New York 4th
Dallas 8th
PHX 10th
PHI 11th
ATL 13th (of course, this is deceiving given the season Pavelec is having)
Colorado 26th

So, everyone that is getting that done is receiving top half of the league goaltending outside of Colorado. The teams outside of the top 10 in goaltending are 1st, 4th, and 7th in scoring. Goalies, for example, are posting an .895 SVPCT against the Flyers... which is a credit to the level of their finishers, and well below league average for goalies.

It's also early. 40ish games into the season still weighs heavily towards the first 20 game when teams are getting situated and everything is goofy. However, every year the teams that have the most success tend to get more shots on goal than they give up.

Top 5 in outshooting wins '08-'09: DET, NJD, SJS, CHI, WSH
... '07-'08: DET, SJS, NYR, CAR, NJD
... '06-'07: DET, SJS, TOR, ANA, DAL
... '05-'06: DET, OTT, DAL, SJS, BUF

Ever wonder why Detroit is always so good? 18 of 20 teams on that list made the playoffs.

Quote:
I'd be more interested in attack time/60 and scoring chances (and quality of) because it still looks good if someone gets a +10 for it, but two bad giveaways/lapses end up in your net, and you're in the dog house. People talked about this when Randy Jones was here as everyone blamed him, even though he had favorable numbers.
Whether you'd prefer A, B, or C is not a relevant criticism of CORSI... like, not at all. Scoring chances are also a terrible statistic, because they are not objective in the slightest. Every team has different ways they define scoring chances. Something to keep in mind when you see folks attempting to breakdown the "threat" of a specific shot given abstracted data. For example, a shot from the point with no one in front of the net isn't dangerous at all... put a couple of bodies in there and it is suddenly a very dangerous play potentially.

Jester is offline   Reply With Quote
Reply

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 09:50 AM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"
Contact Us - HFBoards - Archive - Privacy Statement - Terms of Use - Advertise - Top - AdChoices

vBulletin Copyright ©2000 - 2015, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. ©2015 All Rights Reserved.