2008-11-26

Defence at the quarter pole


Facts, stats, and observations about the Oilers defence corps 20 games in.

Scoring **** GP, G-A-P PPP P/GP
-------------------------------

Souray ***** 19, 7-8-15(7) 0.79

Visnovsky ** 20, 4-8-12(7) 0.60
Gilbert **** 20, 2-9-11(5) 0.55
Grebeshkov * 17, 1-7-8 (3) 0.47
---
Strudwick ** 17, 0-3-3 (0) 0.18
Staios ***** 18, 1-1-2 (0) 0.11
Smid ******** 9, 0-2-2 (0) 0.22

The Oilers rely more heavily on scoring from the back end than almost any other NHL squad. Oilers currently (thru Monday's games) feature 3 of the top 25 defence scorers in the NHL (Souray 9th, Visnovsky 19th, Gilbert 25th). All 3 rank among Oilers' top 5 scorers; only Nashville and San Jose even have 2 D in their top 5, all other teams 1 or 0. Similarly, Oilers are the only team with 4 defencemen in their top 8.

Sheldon Souray leads the Oilers in goals with 7, while no forward has managed more than 5. The other 29 NHL teams have a forward as their leading goal scorer, although in Nashville there is a tie between a D (Shea Weber) and a F (Jason Arnott).

Tom Gilbert's splits:
as of Nov 1: 10 GP, 0-1-1, -6
since Nov 1: 10 GP, 2-8-10, +3


Dennis Grebeshkov's splits:
Oct 12-Oct 18: 4 GP, 0-4-4, +3
Oct 22-Nov 9: 7 GP, 1-0-1, -8
Nov 10-Nov 20: 6 GP, 0-3-3, +3

Time on ice:
Visnovsky is leaned on heavily at evens, 18:35 per game. The rest of the big four all come in right around 16 minutes, Staios 14, Strudwick and Smid 12. Visnovsky also leads the PP at a shade under 5 minutes a game, but ranks a distant 6th among penalty killers at under 1:00 per. Souray makes huge contributions to both special teams (4:45 and 3:55) and thus joins Visnovsky with 24+ minutes of total ice time. The PP has been divided 70/30 between Visnovsky/Souray and Gilbert/Grebeshkov, which is tough to argue with given both of the big money guys have earned a good reputation as a power player. It'll be interesting to see how this morphs as MacT tinkers with the pairings. It's also tough to argue with no other defenceman on the team averaging more than 0:10 PP TOI/G. Meanwhile, Staios and Souray anchor the PK, with Strudwick, Gilbert and to a lesser extent Grebs contributing appreciable time (if not results).

Special teams: On the PP so far,
Souray has been The Man. He has been on the ice for 12 of the 15 5v4 goals, with an impressive +8.21/60. Visnovsky, Grebs, and Gilbert are all between 5 and 5.5. The PP is only +2.94 when Souray is OFF the ice, whereas the other guys all see the PP improve from the low +5s to +7 or 8 when they leave the ice. On the PK, Strudwick impressed until the most recent game, and still leads the way with a modest -3.03/60. Gilbert and Souray are semi-respectable around -7, but the other guys are in minus double digits per 60 which is flat out terrible. Visnovsky trails at worse than -14 ... egads.

Corsi numbers of Oilers D (even strength):
Visnovsky +21, Souray +16, Grebeshkov +10, Gilbert -6, Smid -6, Staios -67, Strudwick -82.

ZoneShift of Oilers D (even strength):
Grebeshkov +16, Souray +12, Visnovsky +10, Gilbert +1, Smid -6, Strudwick -22, Staios -28.

From these last sets of numbers it is clear that our third pairing is getting killed out there at evens. Despite the fact Strudwick has started exactly the same number plays in the offensive and defensive zones (52 each), Oilers have been outshot a staggering 134-69 with him on the ice. Staios actually has started more shifts in the offensive zone (73-64), yet the shots against are nearly as bad at 150-93. This has begun to show up on the scoreboard in recent games.

Steve Staios' splits:
First 9 games: +5
Next 9 games: -4

Jason Strudwick's splits:
First 10 games +3:
Last 7 games: -4


On the other hand, there’s this from the same source (timeonice.com):

Sv% ON: Smid .983 (in limited minutes), Strudwick .948, Staios .947, Visnovsky .941, Souray .932, Gilbert .906, Grebeshkov .904

... which suggests that perhaps the two veterans are limiting the quality of the shots they allow even as they are getting clobbered in the territorial play. The next two on the list are also veterans who have the decided advantage of creating stuff in the other end. Meanwhile Gilbert and especially Grebeshkov have been doing well in driving the play judging from the above Corsi and ZoneShift numbers, yet both are minus players because … their goalies stink? they’ve been giving up ten bell chances? The Dice Are Rattling? Yet more grist for the Corsi/shots/shot quality debate that has been a recurring – and fascinating – theme of the Oilogosphere since I’ve been around and before. At this point I have concluded nothing more than the answer lies somewhere in the middle between quantity and quality, and neither can be trusted on their own to tell the whole story.

The Oilers' current group makes an interesting case study towards this discussion which I will continue to follow with interest.

9 comments:

Jonathan Willis said...

Good job italicizing the suggests and perhaps in reference to EV SV%, but it won't save you Bruce, it really won't.

I think Staios/Strudwick have been pretty lucky, and their EV SV% in part is a result of playing mostly crap forwards. I also think that if this continues, they're just going to get killed in the stats department, because that kind of Corsi number is just mind-boggling.

Bruce said...

they're just going to get killed in the stats department, because that kind of Corsi number is just mind-boggling

Jonathan: I agree they're "going to" get killed, starting two weeks ago. Those high high Sv% are unsustainable. I do think there's something to differing shot quality for different style players, but ~.950 is above and beyond and will surely fall somewhat toward (if not to) the mean.

As for facing crap forwards, there is nothing to the QualComp data at this point (they're all 7 in a very tight cluster between +0.03 and -0.01). I share your observation that 24-43 are playing with and against lesser players, in the manner that Greene and Smid did last year. I'd frankly much rather see Smid there this year as well.

If Staios and Struds were "steady" on the PK we could maybe shield them a little more 5v5 and use them to advantage, but as things stand, recently they've been killing us. I think any sane reading of those faceoff numbers and shots data has to conclude that. Surely the second part of their respective +/- splits listed in my post is closer to representative of the flow of play while they're out there.

it won't save you Bruce, it really won't.

On the whole shot quality v. quantity thing, I don't need saving, thank you. I consider myself very open-minded on the issue, although no doubt I am seen to have taken one side (against what I consider to be a fairly extreme and uncompromising position). I have learned lots from this discussion, and my own position is moderating as I come to more fully appreciate the merits of some of the new stats such as those collected at BehindtheNet and TimeonIce. They provide important details even as they don't tell the whole story either.

As I ponder the matter I will continue to turn over statistical rocks to see the little corsis scurrying beneath. The Oilers' group which we observe closely on an ongoing basis will make an interesting case study. As I mentioned during a recent chapter of this debate, the whole point of noticing early "tendencies" in any study is to later see if there's anything to them. There's a very large gulf between hypothesis and Theory, and I'm still very much in the realm of the former.

Jonathan Willis said...

On the whole shot quality v. quantity thing, I don't need saving, thank you.

I was kidding, Bruce. I'm one of those guys who doesn't view statistical interpretations as show-no-quarter warfare. I just liked how diplomatically you tried to put it.

slipper said...

The QualComp thing is probably more valuable as the season wears on. Somewhere on the internet VF did a decent job showing it's value by taking comparing players who ranked high in Qualcomp to the minutes they played against "known superstars".

The one thing that bothers me is how it's adjusted on a team by team basis using GF-GA/ON-OFF. So right now the Red Wings are playing stellar from the bottom of their roster up, and those who play against Zetterberg are registering "soft minutes".

@Bruce: Tell me how you are determining the quality of chances through a goaltender's save percentage. This was what I was alluding to over at LT's. As far as I can see all that can be derived from an EV SvPCT of .950 is that the goaltender is stopping 95 out of 100 hundred shots. Is there a model or method you are using to gauge shot quality that I am missing?

dstaples said...

I can tell you one thing, Visnovsky has played well overall, as his Corsi attests, but he's been on the ice for some whopper shots against, mainly because he made some horrendous neutral zone mistakes that directly resulted in goals against.

This is what Corsi and even scoring chances doesn't measure, the quality of that scoring chance, and Visnovsky, in the first 15 games, may well have directed play in a positive direction, overall, but his horrendous goal-causing errors cost points in the standings.

Was this bad luck on his part?

Was it because he has such a high range factor -- as the baseball guys put it -- and that kind of guy is prone to making more errors, since he gets to more balls. That's an interesting notion.

Anyway, I don't know why he made so many errors, but I do know he made them, far more than any other Oilers d-man.

Bruce said...

The QualComp thing is probably more valuable as the season wears on.

Slipper: Agree entirely. Last year I had a reasonable amount of confidence in both QualComp and QualTeam indices, but this early season I notice they're all over the map. Presumably confidence will rise with sample size. Nonetheless, your point about performance only within team context is bang on, and to my mind a weakness in BtN's method. Not all teams add up to 100%.

Tell me how you are determining the quality of chances through a goaltender's save percentage.

I'm not "determining" anything, it's a passive observation. As the sample size mounts, the goalie's performance on this shot or that shot should become less of a variable. Assuming (perhaps dangerously) that over time he will mostly "stop the shots he should stop", a player's Sv% ON should begin to reflect the true quality of opportunity that is happening when he is on the ice. At least, that's the hypothesis.

Matt said...

At this point I have concluded nothing more than the answer lies somewhere in the middle between quantity and quality, and neither can be trusted on their own to tell the whole story.

There will certainly be disagreement as to where in the middle, but anyone who objects to this conclusion you have drawn is nutty.

There are a few things we can be pretty confident about. No player "earns" a .950SV% behind him (or an .850%) over any reasonably long stretch. Conversely, every player has *some* effect on the sum quality of scoring chances against while he's on the ice, which over time will probably be reflected in the SV% behind him.

I would suggest that a player who is prone to giving up 5-bell chances will see that show up in the Shots/Corsi data, however (having thought it through for not very long). IMO, a lot of examples of great scoring chances against are plays that, if they had been "properly" defended, are not shots/scoring chances at all, as distinct from being a 1- or 2-bell chance.

Mishandling the puck at the opposing blueline, leading to an odd-man rush coming back. Failing to cover an open man in the high slot. Picking the wrong direction to chase somebody behind the net. Etc. The difference is between great chance/No chance, no great chance and good chance.

The odds are, if someone has high SA totals (setting aside QCOMP/QTEAM effects), they are allowing some good scoring chances, even if their SV% is impressive. How good? Like you say, the answer is certainly somewhere in the middle.

Traktor said...

"Staios actually has started more shifts in the offensive zone (73-64), yet the shots against are nearly as bad at 150-93"

I usually don't put much stock into this kinda stuff but this stat is just alarming.

Nice to see that "our bright shiny object" has been carrying the team.

PerformanceOil said...

Assuming (perhaps dangerously) that over time he will mostly "stop the shots he should stop", a player's Sv% ON should begin to reflect the true quality of opportunity that is happening when he is on the ice. At least, that's the hypothesis.

I think it is a reasonable one, thought the question is, how to test it?

Great look at the defence corps. In general though, if the forwards were getting the job done (incl. the PK), I think the defence would look a hell of a lot better overall.

Then again, maybe crappy defencemen are leading to crappy transitions chances, but I tend to disfavour this idea (purely subjective) since we know at least three of our defencemen have skill sets suited for that job (and even Souray, whose passes are often ugly, seems to be moving the puck to open skaters at the very least).

A bit off topic, but the one thing I have been struck by is how slow the Oilers have looked this season. I believe this is a reflection of the system they are playing (either in execution or formulation), since we KNOW we've got jets on this team.

What really disturbs me though, is that at the beginning of the year, execution and team-play were clearly problems. Yet, individual skill was still obvious. Now, I feel like I'm watching an AHL caliber team (and when AHL call-ups don't look a lot worse than anyone else, I'm not dissuaded from this view).

Arguments about just how good of a team (on paper) the Oilers are is beside the point. They should be good enough to compete, and so far, they haven't. And based on the limited view you've presented, it's the forwards dragging the team down, not the defence.

Hopefully an answer is found soon, because this is getting painful to watch.