Calgarypuck Forums - The Unofficial Calgary Flames Fan Community

Go Back   Calgarypuck Forums - The Unofficial Calgary Flames Fan Community > Main Forums > Fire on Ice: The Calgary Flames Forum
Register Forum Rules FAQ Community Calendar Today's Posts Search

Reply
 
Thread Tools Search this Thread
Old 08-17-2013, 11:53 AM   #601
Lanny_McDonald
Franchise Player
 
Lanny_McDonald's Avatar
 
Join Date: Mar 2013
Exp:
Default

Quote:
Originally Posted by edslunch View Post
I'm not getting this obsession with NHL-E.
NHL-E is a horribly unreliable statistic trying to project the unprojectable. There are too many variables that can impact the output of a player. What this stat really says is that if player X's team were playing in the NHL this is the expectation of his output based on the performance in their current league. What this doesn't tell you is what his probable output would be in a different scenario, say like playing on BC or another player playing at Providence. According to the model Gaudreau's NHL-E should be the same playing on Providence. I think we all can agree that Gaudreau playing at Providence is going to affect his production and his numbers are going to drop, dramatically. Statistics like NHL-E are useless because they can't even predict changes in scoring for players moved in the same league let alone the transition to higher or lower leagues.
Lanny_McDonald is offline   Reply With Quote
Old 08-17-2013, 11:57 AM   #602
Flames Draft Watcher
In the Sin Bin
 
Join Date: Oct 2001
Exp:
Default

Quote:
Originally Posted by edslunch View Post
I'm not getting this obsession with NHL-E.
People like to read more into statistics then they should.

There's no substitute for watching the player develop. Stats only tell you a part of reality.
Flames Draft Watcher is offline   Reply With Quote
Old 08-17-2013, 12:00 PM   #603
BurningSteel
Powerplay Quarterback
 
Join Date: Oct 2011
Exp:
Default

Dear god please make summer fly by so hockey can start instead of talking math.
BurningSteel is offline   Reply With Quote
Old 08-17-2013, 12:03 PM   #604
Rubicant
First Line Centre
 
Rubicant's Avatar
 
Join Date: Jul 2008
Location: Peterborough, ON
Exp:
Default

Statistics should be used in the same manner as a drunk uses a light post - for support rather than illumination.
Rubicant is offline   Reply With Quote
The Following 7 Users Say Thank You to Rubicant For This Useful Post:
Old 08-17-2013, 12:22 PM   #605
kirant
Franchise Player
 
kirant's Avatar
 
Join Date: Apr 2009
Exp:
Default

Quote:
Originally Posted by edslunch View Post
I'm not getting this obsession with NHL-E
Self amusement and boredom for the most part. There's isn't great theory behind it, using simply a single variable and all, but it's easy enough that anybody can use while being able to compare two different leagues.
__________________
kirant is offline   Reply With Quote
Old 08-17-2013, 01:07 PM   #606
Street Pharmacist
Franchise Player
 
Street Pharmacist's Avatar
 
Join Date: Nov 2006
Location: Salmon with Arms
Exp:
Default

Quote:
Originally Posted by Rubicant View Post
Statistics should be used in the same manner as a drunk uses a light post - for support rather than illumination.
Haha. Post of the thread so far
Street Pharmacist is offline   Reply With Quote
Old 08-17-2013, 01:54 PM   #607
IamNotKenKing
#1 Goaltender
 
Join Date: Nov 2006
Exp:
Default

Quote:
Originally Posted by BurningSteel View Post
Dear god please make summer fly by so hockey can start instead of talking math.
On first glance I thought this said taking meth.
IamNotKenKing is offline   Reply With Quote
Old 08-17-2013, 03:23 PM   #608
Oil Stain
Franchise Player
 
Oil Stain's Avatar
 
Join Date: Feb 2006
Exp:
Default

Quote:
Originally Posted by Flames Draft Watcher View Post
People like to read more into statistics then they should.

There's no substitute for watching the player develop. Stats only tell you a part of reality.
I think it goes both ways. It seems pretty likely that a lot of people that solely watch and don't use any stats to back it up suffer from bias, feelings, impressions, and fuzzy memories that cloud and distort their perceptions of a player.

If a fan likes a particular player they are going to be more likely to remember his good deeds and gloss over his mistakes. That's just human nature.


Phaneuf is one example. MC79 posted in a thread on Calgarypuck during Phaneuf's rookie or sophmore season using advanced stats. He argued that Phanuef was not one of the league's best defenders, and not Norris worthy. He argued that he was sheltered on a deep D-core and relied on the PP for points. Stating something like there were probably something like 40-50 NHL defencemen that were better then him. Well, he got shouted down by the masses for relying too much on the stats, not watching him, play, called all kinds of names, etc.. Clearly ANYONE could SEE that Phaneuf was a great defender headed towards even higher greatness.

In hindsight, I think there would be a sizable segment of the fanbase that is now leaning more towards MC79's side of the arguement.


Schremp is another example. Stats guys were some of the first fans that rang the alarm about Rob Schremp. There were a bunch of London fans and frequent watchers of the OHL pumping up Oilers fans about how Schremp was going to be an NHL star after his big season. A few stats guys were pointing out his reliance on the PP for 75% of his points and how that probably wouldn't translate to the NHL.

With more and more NHL teams getting into analytics, I think its clear that the guys at the top believe there is some kind of value in using more advanced statistics. Of course stats aren't everything, but the guys that say that math doesn't belong in hockey are starting to look like dinosaurs.

Last edited by Oil Stain; 08-17-2013 at 03:27 PM.
Oil Stain is offline   Reply With Quote
The Following 5 Users Say Thank You to Oil Stain For This Useful Post:
Old 08-17-2013, 03:41 PM   #609
Day Tripper
Powerplay Quarterback
 
Join Date: Nov 2009
Location: Chair
Exp:
Default

Quote:
Originally Posted by Oil Stain View Post
I think it goes both ways. It seems pretty likely that a lot of people that solely watch and don't use any stats to back it up suffer from bias, feelings, impressions, and fuzzy memories that cloud and distort their perceptions of a player.

If a fan likes a particular player they are going to be more likely to remember his good deeds and gloss over his mistakes. That's just human nature.


Phaneuf is one example. MC79 posted in a thread on Calgarypuck during Phaneuf's rookie or sophmore season using advanced stats. He argued that Phanuef was not one of the league's best defenders, and not Norris worthy. He argued that he was sheltered on a deep D-core and relied on the PP for points. Stating something like there were probably something like 40-50 NHL defencemen that were better then him. Well, he got shouted down by the masses for relying too much on the stats, not watching him, play, called all kinds of names, etc.. Clearly ANYONE could SEE that Phaneuf was a great defender headed towards even higher greatness.

In hindsight, I think there would be a sizable segment of the fanbase that is now leaning more towards MC79's side of the arguement.
In all fairness, wasn't he the guy who argued that Ty Conklin was better than Martin Brodeur?
Day Tripper is offline   Reply With Quote
Old 08-17-2013, 03:48 PM   #610
SebC
tromboner
 
SebC's Avatar
 
Join Date: Mar 2006
Location: where the lattes are
Exp:
Default

Quote:
Originally Posted by edslunch View Post
I'm not getting this obsession with NHL-E. So if Jankowski played in the NHL last year he would have had seven points...but he didn't and never would have.
NHL-E is also predictive. A prospect's draft-1, draft, draft+1, etc. NHL-Es can tell us the type of player he is likely to become.

Quote:
Originally Posted by New Era View Post
Statistics like NHL-E are useless because they can't even predict changes in scoring for players moved in the same league let alone the transition to higher or lower leagues.
I don't think you understand the difference between "imperfect" and "useless".
SebC is offline   Reply With Quote
The Following 2 Users Say Thank You to SebC For This Useful Post:
Old 08-17-2013, 03:51 PM   #611
Oil Stain
Franchise Player
 
Oil Stain's Avatar
 
Join Date: Feb 2006
Exp:
Default

Quote:
Originally Posted by Day Tripper View Post
In all fairness, wasn't he the guy who argued that Ty Conklin was better than Martin Brodeur?
I'm not sure he ever made that arguement. Fans here might have represented it that way. I don't recall much about that article in question. I think he was saying Conklin might be a good goalie based on ES save %.

In looking for that article, I found this old post from 2005 when he predicts a list of goalies based on AHL performance. Does fairly well.

http://hfboards.hockeysfuture.com/sh...d.php?t=139896

Either way. Anyone that makes a lot of predictions is going to be wrong a lot of the time. Stats guys aren't infallible. I just think they make for a more interesting conversation as there is real evidence to discuss, rather then someone talking about their gut feelings.
Oil Stain is offline   Reply With Quote
The Following User Says Thank You to Oil Stain For This Useful Post:
Old 08-17-2013, 04:22 PM   #612
BACKCHECK!!!
First Line Centre
 
BACKCHECK!!!'s Avatar
 
Join Date: Nov 2009
Location: TEXAS!!
Exp:
Default

Quote:
Originally Posted by New Era View Post
What this stat really says is that if player X's team were playing in the NHL this is the expectation of his output based on the performance in their current league.
This is not correct at all.

NHL-E is an attempt to estimate, based on your output this year, how you would produce in the NHL if you were promoted next year.

It's saying "The kids who score 50 points in League X, can generally be expected to score around 22 points or so if they jump to the NHL the following year."

For feeders leagues the NHL-E accounts for aging and development, because every player that contributed to the average got older and improved in the year between the two numbers being looked at.

It is also age-dependent. For example, there will be a different NHL-E for guys who are young up-and-comers transitioning to the NHL, as opposed to old guys who are washing out to a lower league.

Just because you don't understand what it's supposed to be used for, that doesn't make it useless.
__________________
I am a lunatic whose world revolves around hockey and Oilers hate.
BACKCHECK!!! is offline   Reply With Quote
The Following 2 Users Say Thank You to BACKCHECK!!! For This Useful Post:
Old 08-17-2013, 07:17 PM   #613
Lanny_McDonald
Franchise Player
 
Lanny_McDonald's Avatar
 
Join Date: Mar 2013
Exp:
Default

Quote:
Originally Posted by SebC View Post
I don't think you understand the difference between "imperfect" and "useless".
Let me ask you this, since you seem to believe this measure being useful but imperfect. When you run an experiment and you allow the controls to change, affecting the variable, what do you think happens to your data? Is it reliable and verifiable? What happens when you change the make up of a team from year-to-year, or change the system under which a team plays compared to others in the league? Is this reflected in the NHL-E model?
Lanny_McDonald is offline   Reply With Quote
Old 08-17-2013, 07:38 PM   #614
BACKCHECK!!!
First Line Centre
 
BACKCHECK!!!'s Avatar
 
Join Date: Nov 2009
Location: TEXAS!!
Exp:
Default

Quote:
Originally Posted by New Era View Post
Let me ask you this, since you seem to believe this measure being useful but imperfect. When you run an experiment and you allow the controls to change, affecting the variable, what do you think happens to your data? Is it reliable and verifiable? What happens when you change the make up of a team from year-to-year, or change the system under which a team plays compared to others in the league? Is this reflected in the NHL-E model?
They are explicitly reflected by the standard deviations and calculated uncertainty associated with each conversion factor. As are changes in usage as a result of being a young guy in a new league, and the learning curve associated with a new system and coach.


So the answer is yes.


You are attacking a straw man. NHL-E is not meant to predict within 1 point exactly the number of points a guy will get in the NHL. It gives you a numerical indication of how big of a jump there is between each league, and an estimate of a player's likely production after such a jump.

There is a huge evidence base for this, you just can't be bothered following the links to look at it.
__________________
I am a lunatic whose world revolves around hockey and Oilers hate.
BACKCHECK!!! is offline   Reply With Quote
Old 08-17-2013, 08:52 PM   #615
Lanny_McDonald
Franchise Player
 
Lanny_McDonald's Avatar
 
Join Date: Mar 2013
Exp:
Default

Quote:
Originally Posted by BACKCHECK!!! View Post
This is not correct at all.

NHL-E is an attempt to estimate, based on your output this year, how you would produce in the NHL if you were promoted next year.

It's saying "The kids who score 50 points in League X, can generally be expected to score around 22 points or so if they jump to the NHL the following year."

For feeders leagues the NHL-E accounts for aging and development, because every player that contributed to the average got older and improved in the year between the two numbers being looked at.

It is also age-dependent. For example, there will be a different NHL-E for guys who are young up-and-comers transitioning to the NHL, as opposed to old guys who are washing out to a lower league.

Just because you don't understand what it's supposed to be used for, that doesn't make it useless.
I know exactly what the stat is supposed to be used for and how it is applied. What I described is what really happens with it. There are too many variables and lack of controls to properly build a statistical model to project hockey. The stat is junk science.

Quote:
Originally Posted by BACKCHECK!!! View Post
There is a huge evidence base for this, you just can't be bothered following the links to look at it.
No, there is nothing to it. There is a reason why real statisticians do not bother trying to model complex systems like team sports. If they were really capable of generating a predictive model they would do so and take Vegas for every cent they have. There isn't enough reliability in these models so they have no true value. If there was any value to them teams would be firing their scouting staffs and hiring rooms full of stat nerds.

Last edited by Lanny_McDonald; 08-17-2013 at 09:01 PM. Reason: add second response.
Lanny_McDonald is offline   Reply With Quote
The Following User Says Thank You to Lanny_McDonald For This Useful Post:
Old 08-17-2013, 10:56 PM   #616
BACKCHECK!!!
First Line Centre
 
BACKCHECK!!!'s Avatar
 
Join Date: Nov 2009
Location: TEXAS!!
Exp:
Default

Quote:
Originally Posted by New Era View Post
I know exactly what the stat is supposed to be used for and how it is applied. What I described is what really happens with it.
Lol. "I know that my explanation was actually completely wrong. I was just trying to illustrate a common misconception. That's why what I said was factually incorrect."

Quote:
No, there is nothing to it. There is a reason why real statisticians do not bother trying to model complex systems like team sports. If they were really capable of generating a predictive model they would do so and take Vegas for every cent they have. There isn't enough reliability in these models so they have no true value. If there was any value to them teams would be firing their scouting staffs and hiring rooms full of stat nerds.
Every single sentence in this paragraph is factually incorrect. There are hundreds of books, websites, and people being employed doing statistical analysis of sports, and every major league sports team employs people who use statistical analysts. Their accuracy and limitstions have been clearly quantified and delineated. Your statement that statisticians don't even try to model sports is just plain ######ed. There are statisticians who do nothing but model sports. For sports teams.

You want to convince people that this is voodoo? Show us that there is no statistical correlation between NHL-E and a player's point production in the NHL. Or better yet, don't bother, because it gets done by many people every year, and SPOILER ALERT: there is a correlation.
__________________
I am a lunatic whose world revolves around hockey and Oilers hate.

Last edited by BACKCHECK!!!; 08-17-2013 at 11:00 PM.
BACKCHECK!!! is offline   Reply With Quote
The Following 4 Users Say Thank You to BACKCHECK!!! For This Useful Post:
Old 08-17-2013, 11:07 PM   #617
MrMastodonFarm
Lifetime Suspension
 
MrMastodonFarm's Avatar
 
Join Date: Oct 2001
Exp:
Default

Quote:
Originally Posted by BACKCHECK!!! View Post
You want to convince people that this is voodoo? Show us that there is no statistical correlation between NHL-E and a player's point production in the NHL. Or better yet, don't bother, because it gets done by many people every year, and SPOILER ALERT: there is a correlation.
Except no team (hockey) is solely using stat based scouting. Teams have stat guys and the traditional scouting methods. I think that's his point. There are a couple guys in this thread who are drawing conclusions based only on stat sheets, that will lead you down a faulty path. The key is to mix them with actual watching.

I kind of cringe when I see SebC pump up Gaudreau using only NHL-E stats. Gaudreau can score all the points he can in College but the NHL is a different game and he has to show he can transition his game as a small man to the much bigger NHL.

Thankfully, if you've seen him play, you know that he has the amazing ability to avoid damage and be incredibly shifty and heads up.
MrMastodonFarm is offline   Reply With Quote
Old 08-17-2013, 11:12 PM   #618
MrMastodonFarm
Lifetime Suspension
 
MrMastodonFarm's Avatar
 
Join Date: Oct 2001
Exp:
Default

Quote:
Originally Posted by Oil Stain View Post
Phaneuf is one example. MC79 posted in a thread on Calgarypuck during Phaneuf's rookie or sophmore season using advanced stats. He argued that Phanuef was not one of the league's best defenders, and not Norris worthy. He argued that he was sheltered on a deep D-core and relied on the PP for points. Stating something like there were probably something like 40-50 NHL defencemen that were better then him. Well, he got shouted down by the masses for relying too much on the stats, not watching him, play, called all kinds of names, etc.. Clearly ANYONE could SEE that Phaneuf was a great defender headed towards even higher greatness.
I'd like to see that thread, think you can dig it up?

There was a plethora of guys here, and media types who were disgusted with Phaneuf's play on his side of the blueline. Similar to PK Subban and Kris Lateng this season. I would imagine the backlash was more a typical homer backlash to an Oiler guy bringing down a Flames guy on a Flames forum.



Quote:
Schremp is another example. Stats guys were some of the first fans that rang the alarm about Rob Schremp. There were a bunch of London fans and frequent watchers of the OHL pumping up Oilers fans about how Schremp was going to be an NHL star after his big season. A few stats guys were pointing out his reliance on the PP for 75% of his points and how that probably wouldn't translate to the NHL.
Stats guys were some of the first to rang the alarm bell about Schremp?

That's an interesting take. Anyone who ever saw him skate knew it would be a tough road ahead for Schremp. Then once you knew his poor attitude it was written on the wall. You didn't need to be a stat guy to know that.
MrMastodonFarm is offline   Reply With Quote
The Following User Says Thank You to MrMastodonFarm For This Useful Post:
Old 08-17-2013, 11:35 PM   #619
MrMastodonFarm
Lifetime Suspension
 
MrMastodonFarm's Avatar
 
Join Date: Oct 2001
Exp:
Default

So, I think this is the thread you're talking about, Stain. It played out a little different then you recall.

Mudcrutch79 made the claim there was 79 guys in the NHL better then Phaneuf at the time, not 40-50. He also claimed in that thread Horcoff was a 10 ten center at the time. The backlash you seem to recall just isn't there. The main guys responding to him agree he isn't a top guy but put him at a more respectable top 30 (Kootenay Flames Fan). Not the cave man response you try to paint the board with. Time has made him look even worse in that thread, not better IMO.

It's a perfect example of how he went overboard with his solely stat based analysis. If mudcrutch79 had actually watched some games he would have looked a heck of a lot better.

Last edited by MrMastodonFarm; 08-17-2013 at 11:41 PM. Reason: sorry for the many edits, a little drunk, grammar/spelling
MrMastodonFarm is offline   Reply With Quote
The Following 2 Users Say Thank You to MrMastodonFarm For This Useful Post:
Old 08-18-2013, 12:36 AM   #620
dying4acup
First Line Centre
 
Join Date: Oct 2008
Location: Houston, TX
Exp:
Default

Jankowski pick was a boom or bust HS pick. To equate to the NBA, he could be Kevin Garnett, or he could be kwame brown. 1 year after the draft told us nothing. 2 years after the draft told us little. 3 yrs after the draft told us a lot.
dying4acup is offline   Reply With Quote
Reply


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

Forum Jump


All times are GMT -6. The time now is 02:24 PM.

Calgary Flames
2024-25




Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2025, Jelsoft Enterprises Ltd.
Copyright Calgarypuck 2021 | See Our Privacy Policy