Wednesday, October 01, 2008


Curtis McE: the right man at the right time

Could it be:
Bertuzzi - Langkow - Iginla
Bourque - Lombardi - Cammalleri
Glencross - Boyd - Moss
Primeau - Conroy - Roy
(Nystrom, Prust)

Maybe! Boyd looked awfully comfortable in the pivot last night; that line was terrific, and the numbers don't merely bear that out, they put a flashing neon sign around it:

This in barely 8 minutes of EV ice time. Translation: when they were on the ice, they had the puck in the Oilers end the entire time.

McElhinney is highlighted only because his totals are the team totals, but I tell you, I'm thrilled that he got the shutout and looked sound. As I have ranted about a number of times (starting well before there were any concerns about Kipper), Keenan & Co. should be planning on giving #34 no more than 65 starts. Also: as the season rolls on and the ability to add salary increases geometrically, the last thing I want the Flames using cap space on is a freaking veteran backup G. Here's a question I'd love to see one of my peeps in the paid media ask Darryl Sutter:

"In both the 05/06 and the 07/08 seasons you started with a rookie backup. In both seasons he was used very sparingly, and then just past mid-season you brought in a veteran, citing the tight playoff race, ,a demanding schedule, and the benefit of experience.

We already know that the schedule in February and March will be demanding [starting Feb.2, CGY plays 8 games (5 road) in 13 nights, and starting Feb.27 CGY plays 9 games (7 road) in 16 nights]. And it's virtually certain that your team will be fighting to either stay in, or get into, a playoff spot at that stage. So: should we expect a veteran goalie to be brought into McElhinney's place at mid-season?"

Yeah, that's too long and windy, but there must be some way to get at the gist. And still on goaltending, here's something from Cosh at the end of Monday's comment thread that I'd like to address:
Don't imagine this would make the most scientific poll here at this site, but if we did:

Who is most likely to devastate his own team's fans with his crappiness?
* Mathieu Garon
* Miikka Kiprusoff

The answer is obviously Garon because nobody in Calgary appears to give a crap that Kipper ranked 30th in save percentage last year and is staring down the cold barrel of yet another 75-game workload. Whichever two goalies you inserted into such a question, the goalie who was better last year is going to be the answer 9 times out of ten, so what exactly does it prove?

The answer is Garon, and I think what it illustrates is this: although the Oilers have a lot of arrows pointing in the right direction, this idea that "things can only get better" is not correct, and it's not just the OT/SO record that figures to slide. See Why You Shouldn't Bet on Mathieu Garon ("...the stats suggest that Garon is no Huet... I'd expect Garon to drop to about the .905 range in 2008-09."). And DMFB's comment -- that most Oiler fans wouldn't be surprised to see Garon's performance drop off -- is wrong, I think. The very fact that goaltending has been a (relatively) lightly-discussed matter this offseason suggests to me not only that most Oilfans are assuming that Garon will pick up where he left off, but also that he'll be doing it for 65-75 games and mitigating concerns about the reupholstered sofa. But we'll certainly see, won't we.

**Lastly, here's a question that's been in the back of my mind for a while: In his 3 NHL seasons, Dion Phaneuf has been assessed 0 major penalties (excl. fighting), has been investigated w.r.t. supplementary discipline on 0 occasions, and the number of injuries he has inflicted that caused a player to miss more than the remainder of a period is 0. So: Considering how reviled the guy is by opposing fans now, how bad would it be if he actually hurt someone? 'Cause right now his reputation as a cheap-shot artist exceeds his resume by a pretty substantial -- one might say infinite -- multiple.


He doesn't have the balls to be a real cheapshot artist. He's more of a "Take a cheap shot that won't injure and then cower behind the ref" kind of guy.

You're right that most Oilers fans aren't too worried about Garon, Matt. I hear very, very few people ever mention this as a major concern. It's certainly not one of mine.

As you say, we shall see.

My bigger concern has been my ongoing gripe that Lowe made a huge error in not signing up GlenX.

I know many folks don't think GlenX was up to being on one of the top three lines, but that's not how I saw it. He's entering his prime years, he did a helluva job on the fourth line, why not give him a shot at a better job? He earned it.

I've noticed that too, Matt. Many Phaneufophobes consider his physicality coupled with a yapping demeanor as being some sort of sign of a cheap-shot artist. He's involved in significantly more on-ice happenings than other players due to the very nature of his role as an offensive d-man / power hitter. That he revels in being the bad guy doesn't do anything to help endear him to the universal hockey fan. He comes across as a jerk, therefore it is assumed that he actually acts like a jerk, regardless of the stats to prove this.

The thing many critics fail to acknowledge is that he doesn't need to take cheap shots in order to be successful. It might help to place Phaneuf into the context of past star d-men. For example, he's big and strong enough that he doesn't have to throw his elbows in order to win a battle (unlike a young Chris Chelios), yet he's not so big that his elbows are consistantly at head-level (unlike Chris Pronger). Moreover, he's a tremendous skater (unlike Scott Stevens in twilight) with a very low centre of gravity, meaning that he can get into position for a clean hit (like Ray Bourque) where less talented individuals would be forced to use their sticks or knees (Ulf Samuelsson comes to mind).

Most important to this topic, his detractors mistake Dion hitting-to-hurt for hitting-to-injure, which is a significant difference. He understands the boundaries and he toes the line as good as any other physical player in the league, and that pisses people off.

Ultimately, the complaints against Phaneuf generally fall under two headings: that he's ugly, and that he's an asshole. The former is nothing short of petty; the only question with regards to the latter is, Since when was it a bad thing for a hockey player to be an asshole?

I just want to make sure that the fairness, attentiveness, and collective wisdom of NHL officiating is being used as our yardstick here. Just want to make sure that's happening.


I'm not worried about Garon because I'm expecting him to drop to .905 or so and I'm fine with .905 or so at under a million dollars a year. Ditch Roloson in the minors if necessary and bring in a Thibault-type for pennies, use the savings to get that veteran defenseman we all want. Where the hell's the flaw in this plan?

LB, the point about Garon is fair enough. I suppose the flaw in the plan, such as it is, is that with a worse Garon (or a Thibault type) the Oilers allow more goals and presumably earn fewer points in the standings. Maybe acquiring a good vet D would cancel that out; I'm not so sure.

As for your first point: the fact that he's never injured anyone is undisputed, AFAIK. But if you know of an incident or more where he ought to have been punished more harshly than he was (or wasn't), I'm all eyes.

Rob has it right. "Phaneuf will facewash a guy on our 4th line and then decline to fight him, waaaahhh." So he's a dink. You don't have to like him, but any rep for cheapshotting would seem to come from the imaginations of fans pissed that he's pwned their team so often.

I can definately see disliking Phaneuf, hell, I wouldn't want to play against him, but the hate is insanely disproportionate to the actual damage he's causing.

He's very much controlled chaos in his effort to piss people off.

Yes, the Oilers would allow more goals for each shot taken with The Real Mathieu Garon, but if they allowed fewer shots, then I think they'd come out okay while ending up with a better all-round team.

Of course, my management experience is "former second-round playoff participant in an Eastside Hockey Manager league", so take my opinion for what it's worth. :P

And I don't think Phaneuf is a cheapshot artist in the way that, say, Pronger is. I think he's a cheap player, but there's more to that than going knee-on-knee and boarding guys. There's the Avery aspect as well as the Marchment aspect. My point was merely that NHL officials probably aren't a barometer you want to use.

(Injuries are a better barometer, although since they're so random you'd be foolish to rely on them exclusively.)

Since when was it a bad thing for a hockey player to be an asshole?

When he's not on your team. And that, in my opinion, sums up the Phaneuf issue completely.

Re Garon's Definitely Going To Get A Lot Worse, But Kipper Is Kipper No Matter What His Stats Are:

Is there actual evidence that there is any point in breaking overall save percentages into smaller, noisier samples? Or is this just cargo-cult sabermetrics (it works in baseball, so...)?

There's certainly many hundreds of team-seasons worth of evidence that PP shots go in at a higher percentage than EV shots, is that what you mean? Saying "Mathieu Garon won't have a .915 PKSV% next season" is somewhat like looking a guy who has a .955% overall SV% 12 games into the season and saying "He won't keep that up". It's outside the range of what our experience tells us that even elite performance can cause over the long-haul.

I know Vic has stated before that most of the swing in individual SV% year-to-year comes from the PK, and that EVSV% is remarkably consistent, but I'd have to dig around for it.

Since we know that PP shots go in at a higher rate, breaking out PKSV% is not the same as, I dunno, breaking out SV% in the last 10 minutes of the 2nd period. At least that's my take on it, and I think that addresses your question. Am I missing something?

Not to question the random guy on the internet and his stats, but last season Garon was at a .913SV% for the full year. Whereas, his situational stats, supposedly, measure out to:

2007-08: .922 at ES, .919 on the PK, .913 on the PP

Can someone explain how he has a .922 ES and .919 PK SV% but a .913 total SV%? Or are those numbers wrong and we're basing our entire, statistical belief on Garon's ability normalizing to a career average on busted numbers?

(Not to say I disagree, but considering we're relying on a 19 year old and a patchwork defense with bad shoulders, I don't have time to wonder what happens if our goaltenders fall below average)

Mr. Cosh, if you ever happen to start a sports site, I pray that you'll name it "Cargo-Cult Sabermetrics".

Quain: the Special Teams Goaltending numbers (and thus BiaF's citation of them) are off.

Basically, what they have marked as Saves is in fact Shots Against. This is why (apart from Cosh's general point) I don't like to squint at them too hard.

Oh, and in case no one noticed, tonight Kipper allowed 6 goals on 23 shots; 3 of them were fucking awful; and Ray Ferraro commented, "Miikka Kiprusoff has been miserable tonight". Which, if anything, was kind. I blame Mike Keenan. Should be a fun season!

Kipper was bad, but their defense did fuck all (including Celine) to help.

The score probably should have been 6-3 or 6-4 if the Flames buried their chances, but the Nuck didn't even have their first line (the sisters) so 6 goals against is an embarrassment.

Matt: I'm not even sure if that explanation is right. The team stats say we faced 366 PK shots against... the totals for Garon/Rolo in the SV column eclipse 500.

Other fun things from the NHL Stats site? Garon and Roloson ranked 60th and 68th in SH SV%.

Quain, "TS" means "Times Shorthanded" (or Tethered Swimming), not Total Shots.

Hey, you know, Roy Halladay had a hell of a 2008 season even by his own standards. Let's comb through his record and see if we can find a reason to expect some regression.

Hey, did you notice he was 5-1 against the Yankees? Well, hell, you'd be crazy to expect even Walter Johnson to go 5-1 against the New York frickin' Yankees every year. All our experience tells us it's just not realistic.

Archimedes said he could move the earth if you gave him a place to stand. Most people think he was referring to the power of the lever, but maybe he was talking about what you can accomplish given a free choice of statistical endpoints.

Colby Cosh: "But Kipper Is Kipper No Matter What His Stats Are"

Sounds like it just furthers Matt's "The answer is obviously Garon because nobody in Calgary appears to give a crap that Kipper ranked 30th in save percentage last year and is staring down the cold barrel of yet another 75-game workload."

As an Oil fan, I have remained quiet about the goaltending situation because it scares me. I am not fully confident in either goalies. I expect a few months of wait-and-see by MacT and then a decision to be made - Garon/Roli are a definitive #1 or a trade will be made.

Cosh, that is a fine (if grouchy and disingenuous) restatement of your original argument, but I'm still not clear what you're objecting to. What evidence are you looking for?

In 06/07, two goalies had >.900 PKSV%. Their overall save percentages, between that season and this past one, dropped 27 and 9 points. Is that the same as parsing this fellow Halladay's record for little slices that aren't likely to be repeated? I don't think it is, but again, I don't know what it is you want to see.

As long as we're talking about Halladay, the obvious answer is that the Blue Jays scored 43 runs in those six games he pitched against the Yankees; Cosh probably could have gone at least 4-2 himself with 7.2 runs/game worth of run support. That's 1159 runs over the course of a season, better than the modern day record of 1067 set by the 1931 Yankees.

Of course, in Halladay's other starts, the Jays gave him 3.96 R/G, which is about the run scoring rate of the Giants and Padres this year, two teams that have their pitchers bat, play in pitcher's parks and suck. And Halladay still went 15-10 in those starts.

So I don't know that Cosh' example works, particularly where we're not talking about W-L records.

Matt's got an interesting way of looking at it there, the idea of just looking at the guys at the extremes of PPSV% as being likely to regress. It works for 2006-07 too - Miller and Lundqvist were both at .899 on the PP in 2005-06 and .914 and .922 overall, respectively. The following year, when save percentages went up league wide, they sagged to .878/.917 and .846/.911 respectively. The difference between their overall numbers is worse than it appears - both likely faced more 5x3 in 2005-06 as a percentage of all PP shots. Miller's ratio of ES/PP shots went from 3.41 in 2005-06 to 4.16 in 2006-07 and Lundqvist went from 2.52 to 3.31. You would expect their save percentages to rise with that number, as ES shots are more likely to be saved.

Oh, and this Oilers fan is concerned about the goaltending. I think Calgary's got a bigger problem, in that Kipper seems to no longer be the same guy on the ice and he's certainly different off the ice (he now has a contract that pays him huge money forever) but the Oilers goaltending is one of many, many, many things that seems to me has to come up aces for them to make the playoffs.

Mirtle has Edmonton sixth, Calgary ninth. Tons of respect for the guy but that strikes me as beyond insane.

In 06/07, two goalies had >.900 PKSV%. Their overall save percentages, between that season and this past one, dropped 27 and 9 points. Is that the same as parsing this fellow Halladay's record for little slices that aren't likely to be repeated?

No, because what you're doing there (in a half-assed way using a couple samples) is checking a hypothesis, defining it by a fixed criterion, and looking for empirical evidence that it has some meaning, at which point it might be suitable to apply it going forward. This is a good summary of what would be a correct procedure. What Other Internet Guy did was to say "Garon? Really? I bet you can find a hole in his record somewhere if you don't have to define an endpoint in advance." And lo, a hole there was.

The key to the Halladay example is that it is literally true that he's probably not going win 5 of 6 decisions against the Yankees next year. (He probably won't even get 6 decisions against them.) It is Halladay's entire record that still provides the best possible estimate of what Halladay will do over all of next season.

To put it another way, you've now provided a figment of evidence that for goalies whose PK numbers are suspiciously high, the PK numbers drop next season, instead of the ES numbers rising even higher. But do we actually know that's how it usually works?

Mirtle has Edmonton sixth, Calgary ninth. Tons of respect for the guy but that strikes me as beyond insane.

No word of a lie - I have yet to see a single preseason prediction that has Edmonton out of the playoffs. That's both encouraging and worrisome at the same time.

Here's a scenario to ponder: the Oilers are media darlings this season, yet have a season where the bounces don't go their way, finish lower in the standings than last year (despite probably being a better squad personnel-wise than the end of 2007-08). Next fall, the media picks them to finish out again, sighting only marginal additions of some veteran 3rd line center and a supposed "depth" defenseman. They then go onto "surprise" and win the division.

I could totally see something like that playing out. However, I think they'll be good enough to make the playoffs this season.

Cosh, I think that's probably unfair to Other Internet Guy; he may well have written at length (at his site or elsewhere) that very a high PKSV% is highly influenced by luck and has not been shown as repeatable whatsoever.

Slightly larger sample: I looked at the top 5 guys in PKSV% in 05/06 and 06/07; 8 of 10 saw their overall SV% drop the following season. Not rigourous -- as Tyler reminds us, there was a lot of 5-on-3s in the first post-lockout season -- however:

I really don't have a problem separating out the PKSV% for goalies. I think it makes as much sense as looking at PP and EV offense separately for both teams and players. Characterizing PK time as a smaller, noisier sample is unwise, I think: it's *different*, due to inherent structural factors that only change incrementally (i.e. they're more stable than Yankees Rule!).

As "The Other Internet Guy", I think I should probably clarify what I did. I was researching situational save percentages, noticed that PK save percentages were substantially more variable than ES save percentages, and looked to see who overachieved last year and is likely to regress this year. Mathieu Garon and Dan Ellis both fit the profile, but Garon especially because his history is average even-strength save percentages and mediocre penalty kill save percentages. In 2007-08, he had a typical even-strength year and an outstanding performance on the penalty kill. That's characteristic of a guy who got lucky.

We don't know for sure, he could be improving, but if so he is only improving on the penalty kill. By the way, the numbers are indeed messed up for 2007-08. Garon's actual ES/PK split was .916/.911. For comparison's sake, he was .923/.859 in 2006-07 and .916/.823 in 2005-06.

Colby Cosh, I think Tyler Dellow and Vic Ferrari have discussed PK save percentage variance before, I haven't done a rigorous study to prove it but I don't see why it is controversial. Do you think Alexei Kovalev is going to score as many power play points this season as he did last year? Are you betting Patrick Sharp is going to score 7 shorthanded goals again? I agree with Matt that the penalty kill is a substantially different game than 5 on 5, and as a smaller sample size it is also subject to more randomness/luck.

By the way, the numbers are indeed messed up for 2007-08.

That's for sure. In their special teams goalie data, they double counted the shots that went into the net, so that all Sv% in every situation are overstated. It seems to be a systemic error that applied to every goalie I tested.

i.e. instead of
[Sv% = Sv/SoG],
the equation effectively is
[Sv% = SoG/SoG + GA].

No word of a lie - I have yet to see a single preseason prediction that has Edmonton out of the playoffs. That's both encouraging and worrisome at the same time.

I don't see how it can be encouraging. It's the flipside of the same thing that had them (IMO) underrating the Oilers in 2006-07. They're focusing on the loudest, most readily noticeable parts of the past 12 months which, in the Oilers case, is that they went on a tear down the stretch and then added Cole and Lubo, without asking questions about how likely the shit that happened last summer was. In 2006-07, the Oilers likely would have blown away the media projections (such as the notorious THN twelth place projection) but for the Smyth trade and subsequent record losing streak. I don't put a lot of stock in them although, like I say, Mirtle is a smart guy.

BTW - Bruce and Matt are right. The NHL has fucked up the goalie stats on their site for 2007-08. It's every single one of them, as far as I can tell.

Mr. Contrarian goaltender: How much of Garon's improvement in penalty-kill numbers is due to luck, and how much is due to a new team which has traditionally been very good on the PK?

Are Roloson's (and before him, Conklon and Markannen's) numbers unusually good as well?

I'll field that one. Conkannen got bombed behind the Oilers penalty killers in 2005-06. Roli did well once he came over and again in 06-07.

Markannen got bombed in 06-07 too.

Yeah, Roloson did well in both Edmonton and Minnesota, but Conklin and Markkanen did poorly throughout their Edmonton careers (Conklin 266 shots at .835, Markkanen 652 shots at .851).

How much of Garon's improvement in penalty-kill numbers is due to luck, and how much is due to a new team which has traditionally been very good on the PK?

It's difficult to say, but definitely moving from Los Angeles to Edmonton would have an effect. We can guess based on a few assumptions, though. Let's say the Kings are 10% worse than average at penalty kill defending and the Oilers are 10% better than average. That means Garon's adjusted save percentage would be about .855 in Los Angeles, and he would be expected to come in around .870 in Edmonton. That would mean it would be about 60% luck, based on Garon's 2007-08 numbers and assuming he wasn't getting any better. Now there is a good chance that he did improve at least a bit, but it appears that luck was a good part of it.

While we focus on Garon's likely regression to his personal mean we're not focusing on his impact to the Oilers. If he plays 65-75 games at his career numbers (plus a bit better on the PK) that reduces his value compared to 2007-2008 Garon, but he absolutely kills Roloson from last year.

Garon being a clear number one starter, at weaker numbers, probably gives us a better GA this year than last. There is a large gulf that allows Garon to be worse than last season and still see us have better overall goaltending. Which is good, because he's not saving that many PK shots again.

Of course, if he hits 65 GS I'd be surprised.

I checked the numbers, and since the lockout all Los Angeles goalies other than Garon are at .847 on the penalty kill, and all Edmonton goalies other than Garon are at .870, so becoming an Oiler would definitely have helped Garon.

Interestingly the even-strength numbers for those two groups were Los Angeles .907, Edmonton .908.

I guess that's a bit more support for the position that 5 on 5 and 4 on 5 are two different games.

Even if EDM's better on the PK than LA, how much of that improved PK will show in SV%?

Has EDM been better on the PK because they simply allow less shots against, or have they been better because they allow a similar shot total but just, on average, a poorer quality shot.

(法新社a倫敦二B十WE四日電) 「情色二零零七」情趣產品大產自二十三日起在倫敦的成人網站肯辛頓奧林匹亞展覽館舉行,倫敦人擺脫對成人網站性的保守態度踴躍參觀,許多穿皮衣與塑膠緊身衣的好色之徒擠進這項世界規a片模最大的成人生活A片下載展,估計三天展期可吸引八萬多好奇民眾參觀。








Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?