HFBoards Quality of Competition
 Register FAQ Members List Search Today's Posts Mark Forums Read
 Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
 Notices Please do not post or solicit links to illegal game streams.

 By The Numbers Hockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

Quality of Competition

 09-01-2012, 04:17 PM #26 Lonewolfe2015 Registered User     Join Date: Dec 2007 Country: Posts: 14,806 vCash: 413 http://hockeyanalysis.com/2012/01/25...tionteammates/ One of my favorite readings on the QoC/QoT stat. It's not often the stat actually offers any sort of special insight into the players between compared in my opinion.
09-01-2012, 07:10 PM
#27
sjaustin77
Registered User

Join Date: Oct 2007
Country:
Posts: 754
vCash: 500
Quote:
 Originally Posted by seventieslord Yes, it does mean that. If your goal is to demonstrate who is the best at recording points per unit of ice time, it is absolutely important to know what situations they played in. If you separate it into two situations, fine, but don't try to have a "catch all" stat that pretends that ES and PP time are the same thing. (offensive GVT, for example, is a "catch all" stat that determines how many points a player "should" have, based on how much ES and PP icetime they received, not just a total number) player A: 800 ES minutes, 300 PP minutes, 40 ESP, 30 PPP player B: 800 ES minutes, 0 PP minutes, 50 ESP who is the better producer? Player A has 3.81 points per 60 minutes, and player B has 3.75. Player A appears better, or, if we're not splitting hairs, they are about equal. And certainly, from a "hockey card stats" perspective, player A looks much better. But if the only thing making player A the better "per minute" producer is his PP opportunities, what good are these 3.81 and 3.75 numbers? Why would we not want to get to the bottom of who was outperforming who? Those two numbers might shed light on who was better, or they might just shed light on who was playing on the PP and who wasn't. Most players score at about double the rate on the PP. Player A did. Player B would also likely see a rate of production on the PP that was close to double his ES scoring rate. But because that wasn't how his coach used him, did that make him a lesser offensive player? This isn't an absurd extreme example, either. This happens.
1. There are no catch all stats. The ones available leave out a lot of stats and context.
2. I have never pretended that ES & PP are the same thing. In fact I know they are not and that is why I said they would be weighted differently. If I have to use one number to compare players it is ES Pts/60 right now, but I don't do it without looking at the context. QOC, QOT, Zone starts, Corsi, etc.

If you weight ES & PP to reflect that only about 9% of ice time is on the Powerplay than you can come up with 1 number that is pretty accurate. Production is better than PPG and PPG is better than Pts. Production doesn't take into account the situational ice time differences though.

Seguin had a better Pts/60 at both ES and PP than Kessel but Kessel had better overall production because he played more on the powerplay where the rate is higher than ES. My number will accurately reflect this type of discrepancy and show that Seguin was the better producer based on rates instead of who got more PP opportunity. I think my method provides a very accurate list of the best players. It isn't perfect, it isn't catch all and it isn't predictive but it works for what it does and is better than Pts, PPG, or Production.

Your example is absurd and is certainly extreme. Only 9 forwards played more than 300 minutes on the PP and only 8 with any regular time had better than 6 Pts/60 on the powerplay. Giroux is the only one common to both lists. Only Crosby and Malkin had the overall per minute stats that you posted. If you had a player who put up these numbers as only Crosby and Malkin do than they are going to play on the powerplay.

You have basically compared an Eberle at ES to a John Scott on the powerplay who somehow has Malkin like abilities at ES. Your scenario does not happen. There are no players that good at ES or even half that good who get no powerplay time. If that happened there would be a reason for it such as Player B probably put up those points against weak competition so he won't see time on the powerplay and how good he is should be adjusted accordingly. There are defensive players who see little powerplay time such as a Chris Kelly but he doesn't put up a 3.75 ES Pts/60

My formula is a little more complex than just ES & PP Pts/60, but for simplicity using just those parts from your absurd example it still shows player B to be better but not by as much as just even strength shows.

Powerplay should be included and can be accurately included. It won't be perfect, nothing is. Pts aren't, Corsi, QOC, GVT, Point Shares. None are perfect or are really catch all numbers.

And back to the QOC/QOT measure. I feel using some type of production based measurement will be more accurate than shots or +/- based and it is not irrelevant.

If I can ever get the data I need in a usable form and the time than I will make my work available and I would put it up against anything out there now as a better measure of production/overall play and in turn using it for QOC and QOT that is more accurate and makes more sense to compare players across the league.

09-01-2012, 07:27 PM
#28
sjaustin77
Registered User

Join Date: Oct 2007
Country:
Posts: 754
vCash: 500
Quote:
 Originally Posted by Lonewolfe2015 http://hockeyanalysis.com/2012/01/25...tionteammates/ One of my favorite readings on the QoC/QoT stat. It's not often the stat actually offers any sort of special insight into the players between compared in my opinion.
That looks like the same article already posted. All that says is the average player plays the average player. There are a lot of well above and well below average players and they usually get matched up vs each other making a large difference in QOC.

For the most part 1st lines play 1st lines and 4th lines play 4th lines. Rarely are they out 1st vs 4th. You can't avoid having your offensive stars play the best D on the other teams unless you just don't play them. With how fluid the game is, changing of possessions, and shifts on the fly, if a coach wants his shutdown D out vs Malkin he is going to get him out there 80+% of the game even on the road.

09-02-2012, 02:59 PM
#29
Registered User

Join Date: Aug 2006
Location: Brooklyn
Country:
Posts: 45,775
vCash: 500
Quote:
 Originally Posted by sjaustin77 That looks like the same article already posted. All that says is the average player plays the average player. There are a lot of well above and well below average players and they usually get matched up vs each other making a large difference in QOC. For the most part 1st lines play 1st lines and 4th lines play 4th lines. Rarely are they out 1st vs 4th. You can't avoid having your offensive stars play the best D on the other teams unless you just don't play them. With how fluid the game is, changing of possessions, and shifts on the fly, if a coach wants his shutdown D out vs Malkin he is going to get him out there 80+% of the game even on the road.
Agreed. The "coaches always get the matchups they want at home" is one is one of the most absurd things I've read from a stats guy in a long time, and that's saying a lot

09-02-2012, 11:19 PM
#30
Iain Fyffe
Hockey fact-checker

Join Date: Feb 2009
Location: Fredericton, NB
Country:
Posts: 3,509
vCash: 500
Quote:
 Originally Posted by TheDevilMadeMe one is one of the most absurd things I've read from a stats guy in a long time, and that's saying a lot
Is this really necessary?

09-03-2012, 02:01 AM
#31
Master_Of_Districts
Registered User

Join Date: Apr 2007
Location: Black Ruthenia
Country:
Posts: 1,744
vCash: 500
Quote:
 Originally Posted by TheDevilMadeMe Agreed. The "coaches always get the matchups they want at home" is one is one of the most absurd things I've read from a stats guy in a long time, and that's saying a lot

You were klinging to the old "save percentage is inversely correlated to shots against" theorem until the "stats guys" discovered that N.J was undercounting.

The "stats guys" discovered that there's no relationship between shots gainst and save percentage.

And it was the "stats guys" that discovered that N.J was undercounting shots.

You should choose your words more carefully.

09-03-2012, 08:54 AM
#32
Registered User

Join Date: Aug 2006
Location: Brooklyn
Country:
Posts: 45,775
vCash: 500
Quote:
 Originally Posted by Master_Of_Districts Haha. Please. You were klinging to the old "save percentage is inversely correlated to shots against" theorem until the "stats guys" discovered that N.J was undercounting. The "stats guys" discovered that there's no relationship between shots gainst and save percentage. And it was the "stats guys" that discovered that N.J was undercounting shots. You should choose your words more carefully.
I don't think this is on topic but...

Everyone, right down to the TV announcers, knew NJ was undercounting shots for years before anyone did the statistical work.

09-03-2012, 08:58 AM
#33
Registered User

Join Date: Aug 2006
Location: Brooklyn
Country:
Posts: 45,775
vCash: 500
Quote:
 Originally Posted by Iain Fyffe Is this really necessary?
Perhaps my words sounded like I was painting too broad a brush. I think a lot of the work you guys do is valuable.

But to me "quality of competition doesn't matter" sounds an awful lot like "faceoffs are irrelevant to winning" or "on ice shooting percentages (for and against) are entirely based off luck."

Last edited by TheDevilMadeMe: 09-03-2012 at 09:04 AM.

 09-03-2012, 05:20 PM #34 Verviticus Registered User   Join Date: Jul 2010 Posts: 11,762 vCash: 50 eric t pretty much nailed the issue here and came up with a much better (also easier to sell to idiots) metric on the issue http://nhlnumbers.com/2012/7/23/the-...of-competition there's a followup post that has every team on a chart here http://nhlnumbers.com/2012/7/23/the-...of-competition
 09-04-2012, 02:40 AM #35 Pietraneglo222     Join Date: Jul 2009 Location: Gatineau Country: Posts: 3,271 vCash: 500 delete me Last edited by Pietraneglo222: 09-14-2012 at 10:34 AM.
09-04-2012, 05:48 AM
#36
Verviticus
Registered User

Join Date: Jul 2010
Posts: 11,762
vCash: 50
Quote:
 Originally Posted by Immanuel What metric? He says that QoC is not worth accounting for, which is precisely what I said in the OP. The shift level analysis is interesting, though, so it was a nice read for that. That said I have yet to see a rational objection to my findings that QoC doesn't vary meaningfully from player to player. I independently came to the same conclusion as the two other researchers kindly reported in this thread so I think we can consider this a solved problem in Hockey Analytics. At least until a credible objection comes up.
qoc measured through TOI? did you not read the rest of the article

what you said is correct. QoC measured via corsi is not very useful

 09-04-2012, 12:25 PM #37 Pietraneglo222     Join Date: Jul 2009 Location: Gatineau Country: Posts: 3,271 vCash: 500 delete me Last edited by Pietraneglo222: 09-14-2012 at 10:35 AM.
09-04-2012, 12:33 PM
#38
Iain Fyffe
Hockey fact-checker

Join Date: Feb 2009
Location: Fredericton, NB
Country:
Posts: 3,509
vCash: 500
Quote:
 Originally Posted by TheDevilMadeMe Perhaps my words sounded like I was painting too broad a brush. I think a lot of the work you guys do is valuable. But to me "quality of competition doesn't matter" sounds an awful lot like "faceoffs are irrelevant to winning" or "on ice shooting percentages (for and against) are entirely based off luck."
To be clear, I don't really consider myself one of "those guys". I do very little with "advanced" stats, though I do monitor the work.

And those two comments you provide sound an awful lot like things that aren't really said, or at least are said with more nuance than you are presenting here.

09-04-2012, 12:36 PM
#39
Iain Fyffe
Hockey fact-checker

Join Date: Feb 2009
Location: Fredericton, NB
Country:
Posts: 3,509
vCash: 500
Quote:
 Originally Posted by TheDevilMadeMe Everyone, right down to the TV announcers, knew NJ was undercounting shots for years before anyone did the statistical work.
Can you demonstrate this? Did everyone know it or did some suspect it? I certainly recall many claiming that the quality of shot Brodeur faced was very high, but not much about this.

09-05-2012, 10:03 AM
#40
seventieslord
Student Of The Game

Join Date: Mar 2006
Location: Regina, SK
Country:
Posts: 30,929
vCash: 500
Quote:
 Originally Posted by sjaustin77 Your example is absurd and is certainly extreme. Only 9 forwards played more than 300 minutes on the PP and only 8 with any regular time had better than 6 Pts/60 on the powerplay. Giroux is the only one common to both lists. Only Crosby and Malkin had the overall per minute stats that you posted. If you had a player who put up these numbers as only Crosby and Malkin do than they are going to play on the powerplay.
No. Forget the raw numbers themselves (they were only for illustration), and look at their relation to eachother. Yes, there are plenty of players who otherwise produce at the same rate as others who get PP time but their hockey card stats suffer by not getting the PP boost.

09-06-2012, 10:17 AM
#41
Registered User

Join Date: Aug 2006
Location: Brooklyn
Country:
Posts: 45,775
vCash: 500
Quote:
 Originally Posted by Iain Fyffe Can you demonstrate this? Did everyone know it or did some suspect it? I certainly recall many claiming that the quality of shot Brodeur faced was very high, but not much about this.
Everyone (meaning everyone who actually paid attention to such things) who routinely watched Devils broadcasts knew.

This was a very typical situation:

1) Flurry of shots on one of the goals after a long period of inactivity
2) A little while later, Doc Emrick would give the officially game stats: shots on goal would be listed as 1-1.
3) Doc would laugh and say "well it certainly seemed like more than that to us!"

While I certainly don't blame the mainstream stats people for not taking a bunch of potentially homer Devila fans at their word, it should not have taken so many years to test whether different arenas record shots differently, considering modern goaltending analysis is based on the assumption that shots are recorded the same.

09-06-2012, 10:23 AM
#42
Registered User

Join Date: Aug 2006
Location: Brooklyn
Country:
Posts: 45,775
vCash: 500
Quote:
 Originally Posted by Immanuel I read that part:
Correct me if I'm wrong, but...

1) the article is only referring to QualComp related to shot-based metrics like Corsi/Fenwick. Goal-based metrics that incorporate shooting percentages are unaffected.

2) there is no adjustment for the fact that good players tend to play good players and bad players tend to play bad players, dragging everyone's statistical QualComp artificially towards the mean.*

*Similar to how the best faceoff men usually match up against each other since only some faceoffs are important - this drags faceoff numbers of the best faceoff men artificially towards 50%

09-06-2012, 12:22 PM
#43
barneyg
Registered User

Join Date: Apr 2007
Posts: 2,383
vCash: 500
Quote:
 Originally Posted by Immanuel I read that part:
A few weeks ago, I discussed this article in another thread. The conclusion that "QoC doesn't matter" is largerly unjustified. Everything you bolded represents the author's interpretation of a rather simple calculation, and it can be reduced to "I believe the range of expected Corsi values is numerically small, therefore I believe such an adjustment isn't worth the effort." That's not evidence that QoC doesn't matter; at most, it's a rationale not to reject the null hypothesis before the actual tests are carried out, the problem being that no actual tests are carried out afterwards.

09-06-2012, 03:29 PM
#44
seventieslord
Student Of The Game

Join Date: Mar 2006
Location: Regina, SK
Country:
Posts: 30,929
vCash: 500
Quote:
 Originally Posted by TheDevilMadeMe 2) there is no adjustment for the fact that good players tend to play good players and bad players tend to play bad players, dragging everyone's statistical QualComp artificially towards the mean.* *Similar to how the best faceoff men usually match up against each other since only some faceoffs are important - this drags faceoff numbers of the best faceoff men artificially towards 50%
You are right about both things, but although these numbers are dragged towards the mean, they aren't dragged all the way there, or even that close. there is still great variance and the results pass the smell test (i.e. look at the faceoff leaders or QoC leaders)

I believe there is a better faceoff percentage out there now that factors the competion, situation and a couple other factors into it. I think hockey prospectus had it.

Also, if what you're saying is true (and I believe it is), then the effects of QoC are very much understated. Which would explain why even after including QoC adjustments and running formulas the "best defensive defenseman" lists still look a little odd. This number ought to be dragged away from the mean in order to understand its true impact.

 09-06-2012, 09:39 PM #45 Pietraneglo222     Join Date: Jul 2009 Location: Gatineau Country: Posts: 3,271 vCash: 500 delete mi Last edited by Pietraneglo222: 09-14-2012 at 10:35 AM.
09-07-2012, 09:30 AM
#46
overpass
Registered User

Join Date: Jun 2007
Posts: 3,941
vCash: 500
Quote:
 Originally Posted by Immanuel It's not "i believe". I've mathematically and unequivocally shown that QoT varies by an order of magnitude more than QoC. Making QoC "largely irrelevant", at least in relation to QoT. (Please keep that wording and avoid the obvious strawman you just made in your last post.) I don't know why you're talking about "actual tests" when we're not talking about theory, we're directly observing the profession as it happens with real data. If it shows up on the data, it happened on the ice and is therefore "an actual test".
More precisely you have presented some numbers that show that single-iteration Corsi QoC had a smaller range than single-iteration Corsi QoT for the 2011-12 regular season.

How does this change when multiple iterations are used to calculate QoC and QoT? Taking into consideration that if good players tend to play against good players, single iteration QoC will underestimate the true range of QoC, and if good players tend to play with good players, single iteration QoT will overestimate the true range of QoT.

How does this change when looking at plus-minus numbers instead of Corsi numbers?

How does this change when looking at playoff numbers instead of regular season?

How does this change when looking at other seasons than 2011-12?

09-07-2012, 07:58 PM
#47
Verviticus
Registered User

Join Date: Jul 2010
Posts: 11,762
vCash: 50
Quote:
 Originally Posted by Immanuel I read that part:
oh holy **** i clearly posted the same link twice, lmao

http://nhlnumbers.com/2012/8/16/a-co...ed-on-ice-time

there we go! sorry

 09-07-2012, 08:02 PM #48 Pietraneglo222     Join Date: Jul 2009 Location: Gatineau Country: Posts: 3,271 vCash: 500 fugu please delete me Last edited by Pietraneglo222: 09-14-2012 at 10:36 AM.
09-07-2012, 08:17 PM
#49
Verviticus
Registered User

Join Date: Jul 2010
Posts: 11,762
vCash: 50
Quote:
 Originally Posted by Immanuel I don't know man. I don't agree with the premise: "It makes sense -- a player's ice time is a direct reflection of the coach's opinion of the player, and at this relatively early stage in the evolution of analytics, the coach's opinion is more accurate than any one individual statistic." Not a step in the right direction IMO.
the statement is as literally true as you can get. read the results there and here, they're excellent - denying the stat because you think its not parallel with the Number Revolution is pointlessly ignoring a useful tool

09-07-2012, 08:56 PM
#50
overpass
Registered User

Join Date: Jun 2007
Posts: 3,941
vCash: 500
Quote:
 Originally Posted by Immanuel 1) Prove that these questions are important and relevant. 2) Answer the questions yourself. 3) Explain clearly if the results matter. Until you do that, I'm not really interested. Sorry. I don't know man. I don't agree with the premise: "It makes sense -- a player's ice time is a direct reflection of the coach's opinion of the player, and at this relatively early stage in the evolution of analytics, the coach's opinion is more accurate than any one individual statistic." Not a step in the right direction IMO.
You're trying to post the final word on a topic and end discussion, bud. The burden of proof is on you.

Forum Jump