Michalek And Corsi Stats

by GoPens! on Jul 10, 2010 1:32 AM EDT

In my looking at Zbynek Michalek article, I used QCOMP, QTEAM, Corsi QoC, and Corsi QoT as the primary statistics to assess the quality of competition and teammates. I think that most people on this site are familiar with the formula for QCOMP and QTEAM. For those who would like a refresher on Corsi QoC and Corsi QoT, here's my description:

Corsi QoC measures comeptition just like QCOMP, but instead of using adjusted +/- as a base, it uses the Corsi ON statistic. What this is measuring is how well your competition out-attempted opponents. High Corsi QoC's mean that your competition was out-attempting opponents and controlling possession, making them very formidable players. Corsi QoT uses the same logic, but applies it to your teammates. If you're Corsi QoT is low, then you're on the ice with a bunch of guys who get out-attempted, which makes your job more difficult.

The reason I'm bringing this up is that stoopidful remarked in a comment that he or she thought the Penguins could have re-signed Jordan Leopold at around $2M a year before July 1st. I went through the stats again in my article and noticed that Leopold and Michalek weren't too far apart; in some measures, Leopold was better, while in others Michalek prevailed. But was there a $2M difference between these two players?

I've thought all along that one of the biggest upsides to Michalek's play has been that he's succeeded against really tough competition while dragging around terrible line mates. His QTEAM is ridiculously low and his QCOMP is really high. Leopold has a similar QCOMP but a much better QTEAM. Advantage: Michalek. But when you look at Corsi QoC and Corsi QoT, Leopold looks a lot better.

Fortunately, behind the net reports two other competition and teammate metrics: Corsi Rel QoC and Corsi Rel QoT. The only difference between these stats and Corsi QoC and QoT is that they're relative to your teammates (hence the "Rel"). The formulas:

Corsi Rel QoC = On-ice Corsi QoC - Off-ice Corsi QoC

Corsi Rel QoT = On-ice Corsi QoT - Off-ice Corsi QoT

Let's say that your on-ice Corsi QoC (which is the same as your plain old Corsi QoC) is 0.56, and your off-ice Corsi QoC is -0.32. What this tells us is that you saw some stiff competition, while your teammates saw a bunch of slouches. Your Corsi Rel QoC would be 0.56 - (-0.32) = 0.88. Corsi Rel QoT is calculated the same way.

Corsi Rel QoC helps to illustrate how tough your minutes were, or in other words, how much of the heavy lifting you did for your team. A high Corsi QoC means you not only saw tough competition, but were the go-to shutdown guy for your team. Corsi Rel QoT helps to illustrate the level of help you had on the ice relative to your teammates. A high Corsi Rel QoT means that you were on the ice with not only good players, but the best on your team.

I'm now going to look at these numbers for the same 20 defensemen I looked at in my original article. And for those interested, the adjusted^ +/- stat is calculated this way: (adjusted +/-) + (Corsi Rel QoC) - (Corsi Rel QoT). The numbers:

Yup, Michalek is second in this group. Though Leopold came in 9th in adjusted^ +/-, his 0.8 pales in comparison to Michalek's 3.226. Among this group, Michalek ranks 6th in Corsi Rel QoC and 19th in Corsi Rel QoT.

Overall, the point is clear: the Penguins are getting a defenseman that put up a positive +/- on the year while literally playing the toughest minutes in the NHL. If he could survive under these conditions, imagine what he can do when he's flanked by guys like Paul Martin, Kris Letang and Brooks Orpik. Oh, and he'll also have some guys named Crosby and Malkin to pass to.

The content expressed in fanposts does not necessarily reflect the opinions of the staff here at Pensburgh.com. FanPosts are opinions expressed by fans of various teams throughout the league but may be more Pittsburgh-centric for obvious reasons.

Tweet Comment 28 comments | 0 recs |

Do you like this story?

Comments

That’s cheating, GP. lol… but stellar work, makes to excited for October all over again.

I’ve always assumed it was elementary that Michalek represented a better investment both for now and for the future. It’s like shooting fish in a barrel if we are to give the whole Leopold at 2Million against Michalek argument.

I’m personally happy that Leopold finally gotten his payday…

Is it October yet?

Just so I never have to answer this again. I'm from Singapore so whenever you see me online, I'm either sleep deprived or just waking up.

by Alighieri on Jul 10, 2010 1:56 AM EDT reply actions

Isn’t this stuff exciting? I mean, I know I’m all about adding scoring, but holy crap is the D going to be solid this year.

PensBurgh

by JustinM on Jul 10, 2010 11:22 AM EDT reply actions

I know, I really can’t wait anymore.

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 10, 2010 3:18 PM EDT up reply actions

GoPens! Nice piece. Thanks! Esp. appreciate the explanations on the stats.

Agree on Michalek > Leopold.

fwiw, in the post asking for our “Wish Lists” circa June 30th, i did include them both in this rank order:
1. Selanne, RW — w/c would have obviated the need to move Staal to 2C, btw. although i realize it was a 1% chance
2. Martin OR Hamhuis, D
3. Michalek, D
4. Malhotra, 3C
5. Leopold, D
I included Leo as a backup in case Shero couldn’t get 2 of the 3 D listed above.
He’s underrated, so long as he’s back to full health, anyway.

Question: What methodology would use to rank the Top, say, 40 defencemen in the NHL — OVERALL?

by Diomedes7 on Jul 10, 2010 1:59 PM EDT reply actions

Thanks for the props Diomedes.

Question: What methodology would use to rank the Top, say, 40 defencemen in the NHL — OVERALL?

This would be more subjective than you might think. Since defensemen do so many different things, like blocking shots, killing penalties, passing the puck, etc. it would be tough to gauge which skills are more valuable. You’d have to apply a formula that weights different skill-sets differently, and it would be tough to get everyone to agree on something like that.

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 10, 2010 3:20 PM EDT up reply actions

roger that. as soon as i wrote that, i thought, “what do you do w/ anton volchenkov?” excellent defensively, but a zero on offense—which would be hard to quantify and rank.

could you put together something where you combined the defensive aspects in one formula and the offensive aspects in another?

by Diomedes7 on Jul 10, 2010 7:06 PM EDT up reply actions

As I’ve mentioned in the past, it’s tough to actually separate the defensive aspects from offensive aspects in the game. And even if I could do that, you’d still have the same problem as before, i.e. are blocked shots more important than penalty killing? Is QCOMP more significant than QTEAM?

Don’t get me wrong, I love the way you’re thinking right here, but we already have GVT, which gives us an overall, all-encompassing ranking, and anything more sophisticated is wayyyyy beyond my capabilities.

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 10, 2010 11:35 PM EDT up reply actions

well, i don’t think GVT is reliable

re: defencemen…
for the 2009-2010 season
Z. Michalek ranks #100** in GVT

BEHIND Wade Redden, Tom Poti, Henrik Tallinder, and Mark Eaton — to name a few
so…i’m thinking there has to be a better way

**for defencemen who played 50 games or more

by Diomedes7 on Jul 11, 2010 9:52 AM EDT up reply actions

this doesn’t look right either…
for Penguins d-men
for the 2009-2010 season
w/ at least 50 games played
GVT ranks them as follows:

1. Gonchar
2. Goligoski
3. Orpik
4. Letang
5. Eaton
6. McKee
7. Leopold

by Diomedes7 on Jul 11, 2010 10:18 AM EDT up reply actions

Michalek is penalized in GVT because he doesn’t produce as much offensively as others. If you look strictly at Val D (which is the defensive part of GVT) Michalek is near the top among defensemen who played at least 40 games.

I think GVT was a huge breakthrough in the field of hockey stats, but you have to use context when looking at the numbers. Since Michalek has thus far been a defensive d-man, Val D might be better than overall GVT. Another reason he doesn’t get the measurement he deserves is that GVT doesn’t take the quality of teammates into account.

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 11, 2010 11:45 AM EDT up reply actions

Val D

if you just look at Val D…
Z Michalek is ranked 30th in the NHL — not bad, but nothing spectacular either

if you look at just Val D of the Penguins D corps (last season’s stats)…
Eaton and Goligoski are ranked 1st and 2nd on the team (?)
And McKee is ranked ahead of Letang and Gonchar (?)

I think GVT was a huge breakthrough in the field of hockey stats, but you have to use context when looking at the numbers

Perhaps. But if it is, its usefulness is not obvious. Or maybe the context isn’t obvious. I’m not sure. That’s why I was asking if you had a better metric—kind of like your adjusted +/- formula.

by Diomedes7 on Jul 11, 2010 3:51 PM EDT up reply actions

There will always be some oddities with any formula that attempts to measure overall performance. As to why our d-men were ranked where they were, you could probably gain some insight by checking out Awad’s articles that describe GVT’s calculation (here, here, and here). It’s not perfect, and no one says it is, but it’s the first overall performance metric that I know of, and when you look at the top players in GVT, they correlate pretty well with who we imagine to be the elite players in the NHL.

Perhaps. But if it is, its usefulness is not obvious.

It’s useful because it allows you to compare players from different positions using an identical metric (which is something my adjusted +/- formula can’t do). And when I look at the GVT for the Penguins this past year, it looks like it gives us a pretty good indication of who had a good year and who had a not so good year.

That’s why I was asking if you had a better metric—kind of like your adjusted +/- formula.

My metric is much less sophisticated than GVT, and I don’t really know if I’d call it an overall performance metric along the same vein as GVT. Adjusted^ +/- attempts to normalize for how tough your minutes were, and I think using Corsi Rel QoC and Corsi Rel QoT serve that purpose the best. I also only look at even strength performance and exclude other important variables. But it can be used for a similar purpose, and it’s pretty cool that you’re mentioning it in the same breath as GVT :)

Hope that helps Diomedes.

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 11, 2010 4:44 PM EDT up reply actions

i'm reviewing this stuff again. i see that it can get you in the ballpark but there are still glaring discrepancies

like the #2 OVERALL player acc. to GVT was Evgeni Nabokov
yet he couldn’t land a contract with ANY of 30 NHL teams…?

and it has Johan Hedberg > Jordan Staal
do you think Ray Shero would have traded Staal for Hedberg?
or that Hedberg had a better year than Staal?

when i see things like that…i can’t buy into the metric

by Diomedes7 on Jul 11, 2010 5:18 PM EDT up reply actions

what i liked about your Adjusted^ +/- was not that it could replace GVT, but that it MADE SENSE
ie, it made sense intuitively — the tweaks you made comport with common sense or hockey sense or whatever you want to call it
and the results did too
GVT is perhaps more “sophisticated,” but it yields results that just don’t match up with common sense

by Diomedes7 on Jul 11, 2010 5:21 PM EDT up reply actions

Nabokov’s regular season GVT was well-earned in my opinion, but he’s in the KHL right now because he had a crappy playoffs and wanted way too much money.

As to Hedberg, he was probably more valuable to the Thrashers than Staal was to the Penguins. That should explain the discrepancy in GVT. Ultimately though, I wouldn’t read too much into these instances because the vast majority of GVT rankings make sense to me.

I’m happy you dig the formula. I honestly just kinda threw it together, but it looks like it may hold some promise. Maybe I’ll try to refine it and see if it holds up when looking at the NHL overall.

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 11, 2010 11:05 PM EDT up reply actions

Well, then we can agree on your formula!, but perhaps agree to disagree on GVT…

Nabokov: No way he’s #2 overall. He wasn’t even a nominee for the Vezina—and that’s only goalies. Yet GVT has him ahead of Crosby, Ovechkin, Sedin, Brodeur, Pronger, Toews, Stamkos, etc.

Hedberg — Platooned with Pavelec and then mgmt. let him walk. Better than Staal? C’mon.

Eaton the Pens #1 defensive defenseman?
Shero wanted to SIGN a defensive defenseman, if Eaton was the #1 on the team why did he let him go?

And Goligoski as the #2 defensive defenseman (on the Pens)? = even more perplexing.

I didn’t study the list very long. These jumped out at me in the 1st few minutes, and from that point on, I had a hard time believing any of it.

by Diomedes7 on Jul 11, 2010 11:54 PM EDT up reply actions

You bring up some good points. I might have to email Tom to see what he says…

Hockey Blogger at Pensburgh.com

by GoPens! on Jul 12, 2010 12:06 AM EDT up reply actions

oh my goooooooosh, i cannot wait to see our defense. file that under things i never thought i’d say.

i like the pens, sidney crosby, jordan staal, and typing in capital letters.

by katielynn906 on Jul 10, 2010 2:40 PM EDT reply actions

I can’t wait to not see our defense, like if they do their job then we don’t even notice they are there, so we can spend more time watching Sid and Genostaalkin tearing up the other team’s D.

by PensAreYourDaddy on Jul 10, 2010 2:51 PM EDT up reply actions