Do you ever actually watch hockey? Vermette’s an excellent defensive forward and one of the better faceoff men in the league. But you literally ALWAYS stare yourself blind at stats.

Posted by Dakkster from Southern Sweden on 09/05/10 at 11:37 PM ET

Yes I watch hockey.  Hundreds of games a year. 

Vermette isn’t even the best defensive forward on his team.  I would give that position to Sami Pahlsson.  And the second (or so) best defensive forward on a weak defensive team like Columbus is not the 17th best defensive forward in the NHL.  Its a statistical anomaly that gets him ranked that highly.

Posted by PuckStopsHere on 09/05/10 at 11:43 PM ET

Jesus you’re thick… you are literally blinded by stats. You are letting the combined statistical effort of an entire team weigh down one individual player. Start watching Vermette play, then write. Until then, you have exactly no credibility whatsoever. And even IF you wanted stats as the sole leg to stand on, your use of them is flawed.

Posted by Dakkster from Southern Sweden on 09/06/10 at 12:02 AM ET

Let’s see if i understand your point.  I am saying that this statistical system is flawed and Antoine Vermette is a great case study that shows it.  I am saying the statistics are wrong.

And you respond by telling me that I am wrong because the statistics are wrong.

Is that close to you point?

Posted by PuckStopsHere on 09/06/10 at 12:06 AM ET

As long as I understand that you’re saying that goals-versus-threshold ratings are flawed and mostly useless, then we have a good understanding.  There have been noticeable problems in every one of the GVT ratings so far, many of which you’ve taken the time to address.  I think the overarching lesson here is that GVT ratings are too untrustworthy to be used consistently.

Posted by J.J. from Kansas on 09/06/10 at 08:57 AM ET

That’s exactly it J.J.

Everytime you post one of these you have to do a follow up to explain a few people here and there and why they’re even on the list and that the stats get askewed because they’re based on other stats which you believe are askewed to begin with.  So people start realizing that you’re giving us new stats that are supposedly superior because they take other factors into account but they’re based redolantly on the stats that you’re seeking to replace.  So if you’re building these stats whatsoever off of +/- how are they supposed to take you seriously if you’re still correcting “anomolies” everytime you post. 

One other note, I do notice that you pretty much only blog about your stats you do and not really hockey.  You may watch games but I don’t think you WATCH the game.  Akin to constantly telling me about the velocity a pair a hooter can attain at a 5 m.p.h. jog with calculations such as cup size and bust and stride and yadda yadda, but when it comes down to it, all you have to do is watch them to know if they’re nice or not, yeah?  Quit being a Sherminator and actually talk about the game once in a while to let us know you’re still part human

Posted by HockeytownOverhaul on 09/06/10 at 03:20 PM ET

Everytime you post one of these you have to do a follow up to explain a few people here and there and why they’re even on the list and that the stats get askewed because they’re based on other stats which you believe are askewed to begin with.  So people start realizing that you’re giving us new stats that are supposedly superior because they take other factors into account but they’re based redolantly on the stats that you’re seeking to replace. 

I am showing you the state of the art in hockey sabermetrics.  Somethings work very well and others do not work so well.  Nobody (aside from you) claims I am showing systems that are supposedly superior to anything.  Goals versus threshold is the best attempt I have seen to date to assign every player a single number for his value.  So far i have looked at three portions of it in detail.  The offensive portion, which usually provides good results, afterall we understand offensive statistics quite well, but nevertheless there are certain players in misrates by sizeable margins.  The goaltending portion is the strongest part of the system.  The defensive portion is theoretically flawed from the beginning because all that is measured is team defence and +/-.  Nevertheless it does surprisingly well, because it captures the players with the most ice time and good +/- ratings on good defensive teams and that is usually a reasonable estimate of the better defensive players in hockey.  However, there are cases where it misses the mark significantly.  It is worth looking at some of the more interesting ones.  I may not immediately have a solution to them, but somebody reading this might.

So if you’re building these stats whatsoever off of +/- how are they supposed to take you seriously if you’re still correcting “anomolies” everytime you post. 

How good can a stat be if it is built (in part) off +/-?  Good question.  I think in principle it can be better than this, but it isn’t the best gage of defensive play.  Nevertheless it is interesting because this system is relatively commonly used.  People have tried to use it to debate things in my comments in the past.  It is worthwhile to profile it to see its strengths and limitations.

One other note, I do notice that you pretty much only blog about your stats you do and not really hockey.

In the summer, when there is no hockey to watch, I dive into the statistics.  I think that is an interesting thing to do.  If you do not, go read another blog.

Of course with a strong statistical background, when the season is on, I may argume in terms of some of the more reliable of these numbers when it is appropriate and if you do not know what that number represents, you will be lost.

However, any claim that I do not watch hockey is factually wrong.  It is not consistent with reality in the slightest.

The rest of your comment goes off into gibberish in my opinion.

Posted by PuckStopsHere on 09/06/10 at 04:36 PM ET

I am showing you the state of the art in hockey sabermetrics.

I think that shows that the state of the art of hockey sabermetrics still needs quite a bit of work to try to create a unified ranking system for hockey value.

I’m glad that statistics junkies like yourself are still working on it, but there are too many flaws right now.

Posted by J.J. from Kansas on 09/06/10 at 05:20 PM ET

“That can’t possibly be accurate because the Colombia Blue Jackets suck. Everybody knows that.”

I’ve said it before and I’ll say it again - it’ll take a second Stanley Cup for anyone here to ever get anything even distantly approximating respect.

Posted by Viqsi on 09/06/10 at 09:39 PM ET

See, yo’ure still not getting it.  I never said you didn’t watch hockey.  Dont put words in my mouth even if they are jibberish. 

All I was saying is you may want to square away your stats before posting your blog.  If the stats don’t work then why would we want to read them?  You’re like a freakin’ robot so no one is getting anything else but cold hard math from you, but the math is wrong consistently.  A bunch of half truths doesn’t make a truth. 

Take my critique for what it’s worth, I’m a know-nothing fan giving you my honest feedback about reading your blogs hoping that one time I might actually be wowwed by something, but habitually I’m let down because it doesn’t correct anything, it’s just a different set of flawed data and rankings.  So in the end I don’t really get anything out of it.  Maybe more people should stop reading your blogs, you’re right buddy.

The end of my previous statement probably sounded like jibberish because I used a term or scenerio unfamiliar to you.  Sorry about that.

Posted by HockeytownOverhaul on 09/07/10 at 03:31 PM ET