Advanced Stats: Corsi, QoC, etc

[tmg]

Does anyone think some team/managers are using Sabermetric-type analysis extensively in their player evaluation and acquisitions?

I've had my suspicions about Tallon, with how many 'diamonds in the rough' he brought in to Chicago, and now seemingly doing the same in Florida. It really makes me wonder if his management/scouting approach does something that other teams simply don't do (?as much?), and the first thing that came to mind is something in this realm of statistical analysis.

 

[Les Wynan*]

Does anyone think some team/managers are using Sabremetric-type analysis extensively in their player evaluation and acquisitions?

I've had my suspicions about Tallon, with how many 'diamonds in the rough' he brought in to Chicago, and now seemingly doing the same in Florida. It really makes me wonder if his management/scouting approach does something that other teams simply don't do (?as much?), and the first thing that came to mind is something in this realm of statistical analysis.

The good ones undoubtedly do but they'll never talk to the media or anyone else about it.

It's pretty obvious that Tallon does not.

 

[Erick*]

Like pretty much any stat, you can take them out of context of other related stats. And they are statistics; They dont predict the future, but tell you about the past.

I like the sabremetric stuff but it should not be a major discussion point in hockey. Understanding how plays develop and how a player is supposed to make a decision is more interesting and useful to understand. That is scouting and coaching mostly. Stats are quantifications of that.

They're not perfect, but advanced statistics can certainly predict the future somewhat. It's why they're used. No one can watch every play of every game. Advanced stats can give a more accurate assessment of how good or bad/how lucky a player/team has been.

 

[Erick*]

hockeyprospectus' GVT is free to view.

I don't think it's updated daily unless you pay.

 

[Erick*]

The good ones undoubtedly do but they'll never talk to the media or anyone else about it.

It's pretty obvious that Tallon does not.

I don't think using one example of selling low on a player is a telling example that he definitely has no interest in it, whatsoever.

Tallon does tend to overpay his guys though, so that might be more telling.

I think the Panthers do use it somewhat (just opinion here) because Mike Santos, our assistant GM/used to work for the Predators, loves to talk about the little things in the game like faceoffs, etc. and he played a part in getting us some of our more cost-effecient players (for example, Marcel Goc this year/Mike Santorelli last year).

The Panthers, despite some overpayments, have gotten players that have been effective that no one else seemed to want. One of Tallon's first moves was getting Mike Weaver and he's pretty undervalued.

 

[Landeskog]

there is a book Hockeynomics
http://www.amazon.ca/Hockeynomics-What-Stats-Really-Reveal/dp/1897277458/ref=sr_1_sc_1?ie=UTF8&qid=1328383791&sr=8-1-spell

it was fun to read and play around with. During the book you picking the best player for there value. I picked Datsyuk and Hartnell as the who i thought it was going to be. Also talks about drafting and is Marty B as good as he seems by using stats.

 

[GKJ]

behindthenet.ca, timeonice.com, arcticicehockey and any sites affiliated with it.

Arctic Ice Hockey and Behind the Net are essentially the same thing. BTN has the stats, but the analysis is part of AIH now.


If someone wants to pull up James Mirtle's pieces from pre-season, he lists a few teams that use advanced numbers. Penguins are one of them, one reason they did so well last year. Somehow, the Flames are another.


I used to be vehemently against it, but now I am definitely for it. No one can watch 1230, and those who study them don't claim it's 100% truth. Not that this is absolute proof, but I'm doing much better in fantasy leagues basing some of the things I've done on advanced numbers.


To buy into the concept is to allow yourself to reinvent your line and process of thinking. Making or wanting your brain to re-calibrate itself is not easy, and thus it's different for everyone.


Someone brought up GVT, I use that to rate most players around the league.

 

[Erick*]

Arctic Ice Hockey and Behind the Net are essentially the same thing. BTN has the stats, but the analysis is part of AIH now.


If someone wants to pull up James Mirtle's pieces from pre-season, he lists a few teams that use advanced numbers. Penguins are one of them, one reason they did so well last year. Somehow, the Flames are another.


I used to be vehemently against it, but now I am definitely for it. No one can watch 1230, and those who study them don't claim it's 100% truth. Not that this is absolute proof, but I'm doing much better in fantasy leagues basing some of the things I've done on advanced numbers.


To buy into the concept is to allow yourself to reinvent your line and process of thinking. Making or wanting your brain to re-calibrate itself is not easy, and thus it's different for everyone.


Someone brought up GVT, I use that to rate most players around the league.

Which ones do you think are most important?

I completely agree with the whole "reinvent your line and process of thinking."

 

[GKJ]

GVT seems rather similar to WAR in baseball.

http://hockeyprospectus.com/article.php?articleid=233

I think it's more like VORP.

 

[rye&ginger]

The studies I have seen shown an
r = .62
rsquared correlation = .62x.62 = 0.384

Fenwick Road Tied is the best measure that exists to describe team skill but even so it fails to explain over 60% why teams win.

This is what Im getting at. Stats only give us about half the picture. No doubt they are useful, but I use them along with observations to make predictions. Using just stats wont get you as far.

 

[MasterofGrond]

This is what Im getting at. Stats only give us about half the picture. No doubt they are useful, but I use them along with observations to make predictions. Using just stats wont get you as far.

They give you about half the picture.

Luck is most of the rest.

 

[Erick*]

I think it's more like VORP.

I'm a bit new to it obviously.
If it's more like VORP, it should only be used to compare players who play the same position, right?

Kinda confusing because hockeyprospectus has them listed all together.

Seems like goalies rate higher than forwards/defensemen, as well.

 

[GKJ]

Which ones do you think are most important?

I completely agree with the whole "reinvent your line and process of thinking."

http://www.theglobeandmail.com/sports/hockey/hockeys-new-numbers/article2178781/

That's what I was looking for.

Offensive zone start and finish percentages are a good one too, especially if you're comparing your depth players.


Gabe Desjardins actually answered a couple of tweets of mine this week, he said that Corsi is best league-wide, but comparing players on a given team QUALCOMP is better (now listed as +- QoC)

Pass It To Bulis wrote about Manny Malholtra being an enabler to Ryan Kesler is the turning point of where I fully bought in.

http://passittobulis.blogspot.com/2011/02/manny-malhotra-is-enabler_04.html

 

[MasterofGrond]

I'm a bit new to it obviously.
If it's more like VORP, it should only be used to compare players who play the same position, right?

Kinda confusing because hockeyprospectus has them listed all together.

Seems like goalies rate higher than forwards/defensemen, as well.

Yeah, I wouldn't use it to compare goalies to forwards or defensemen.

 

[GKJ]

I'm a bit new to it obviously.
If it's more like VORP, it should only be used to compare players who play the same position, right?

Kinda confusing because hockeyprospectus has them listed all together.

Seems like goalies rate higher than forwards/defensemen, as well.

You could do that, or a given role on the team. GVT for goalies is highly volatile no matter which way you look at it. Consider that last year Brian Elliott had the worst GVT in the league and now he's in the Top 10.

 

[Loner]

I actually like this site for habs statistics: http://www.boucherscouting.com/ I'm not sure what's the value of those stats but it's an interesting method.

It is really time consuming to do those kind of stats, you needs to analyze each players and note each events(positive and negative) he product on the ice.

 

[Les Wynan*]

Gabe Desjardins actually answered a couple of tweets of mine this week, he said that Corsi is best league-wide, but comparing players on a given team QUALCOMP is better (now listed as +- QoC)

Link? I highly doubt he said this.

 

[GKJ]

Link? I highly doubt he said this.

This is what he said on QualComp

"I like it when they match up on a given team. League-wide, regular Corsi is usually the most telling. Need a multi-year sample"

 

[Les Wynan*]

This is what he said on QualComp

"I like it when they match up on a given team. League-wide, regular Corsi is usually the most telling. Need a multi-year sample"

Where did he say +/- QoC is best for measuring within a team though? Since it's based on +/- it really doesn't have a ton of value. IMO Corsi Rel QoC matches the eye test within a team to the greatest degree.

 

[GKJ]

Where did he say +/- QoC is best for measuring within a team though? Since it's based on +/- it really doesn't have a ton of value. IMO Corsi Rel QoC matches the eye test within a team to the greatest degree.

That was in the mix, he was just responding to something I said specifically about +/- QoC. I guess I shouldn't have said "best," more like "best used as."

 

[Shrimper]

Somebody please explain what Sabremetrics are?

 

[Les Wynan*]

Somebody please explain what Sabremetrics are?

To be honest, it's kind of a misnomer to apply it to hockey but I guess it's evolved to mean any type of statistical analysis in sports. It was originally used to refer to the work being done at Bill James' Society for American Baseball Research, hence SABRmetrics or sabremetrics.

 

[Talks to Goalposts]

GVT seems rather similar to WAR in baseball.

http://hockeyprospectus.com/article.php?articleid=233

GVT is pretty heavily flawed as a unitary measure though. It makes no allowance for context based things like competition/teammate strength and zone start. Also its predictive value gets heavily effected by puck luck stuff like on ice shooting percentage and save percentage. Guys like Ville Leino last season tend to get heavily over-rated by it while the contributions of guys like Malhotra and Bolland get missed.

Its really the infant version of a future NHL stat that might get pretty close to what WAR or even VORP is in baseball. The hockeyprospectus guys tend to have a somewhat inflated opinion of how useful it is.

 

[Les Wynan*]

GVT is pretty heavily flawed as a unitary measure though. It makes no allowance for context based things like competition/teammate strength and zone start. Also its predictive value gets heavily effected by puck luck stuff like on ice shooting percentage and save percentage. Guys like Ville Leino last season tend to get heavily over-rated by it while the contributions of guys like Malhotra and Bolland get missed.

Its really the infant version of a future NHL stat that might get pretty close to what WAR or even VORP is in baseball. The hockeyprospectus guys tend to have a somewhat inflated opinion of how useful it is.

Terrifically put and agreed on all points. It's really only useful when comparing players used in similar situations or when averaging several years of a player's GVT, although even then it ignores context.

 

[GKJ]

GVT is best used as a guide rather than absolute. For instance, when we get a fool or two who say Kimmo Timonen is overrated, old, and broken down, when he's been running among the top 5 among defensemen in GVT. But I don't think people would rather have Anthony Stewart over Jeff Carter, nor Ian White over Nicklas Lidstrom.