Friday, August 25, 2006

Taking The Shots

So, which Oiler players were least likely to see a shot-against when they were on the ice? ... This is the lineup, the top eight, the "they don't get shots on the Oilers net when I'm out there!" players:

M. PECA 370
R. SMYTH 376

This normalized for equal icetime, 15 hours at evens. And just shots-against rate here.

Surely sensible people everywhere are expecting the forwards to be driving the bus here, and they are. But Pronger at the top of the list ... that's an anamoly. And he's dragged Spacek (admittedly smaller sample size here) along with him. He's a helluva player, no doubt. Some fairly shitty luck with shooting and save percentages around him probably prevented him from getting another Norris. Then again, you could probably say the same for a few of the top tier D guys any year. And none of them can touch Modano, Forsberg, et al in any season by this metric.

By and large the forwards are driving the results at this, and this is the stat that the video game kids theorize has bugger all to do with them. (I maintain that the video games are programmed by geeks on the payroll of this sport's major bookies. Though I have no proof ... who else has a vested interested in making young hockey fans dumber?)

The Oilers are going to miss Pronger, no doubt. One of the few guys who really drives the results from the D position. He ain't no Bobby Orr (Orresque playoff streak aside) ... but he'll be missed.

The Hemsky thing is hard to explain. I dunno. Clearly I've been too harsh on the guy, two years in a row wih damn decent outshooting stuff, odd man rushes surrendered or no ... the chipstack is pretty respectable.

Oiler funfact: Last season Igor Ulanov had the best save% on the team behind him (nearly .960, just crazy), this season it was the worst on the squad (nearly .830). That alone explains the fall from "unsung hero!" to "Whatta asshole!". Nearly perfectly by the sums. Seriously.

All of the other Oiler D are bunched in around the 395 mark. The Oilers 4th liner types got murdered at this.



Anonymous PDO said...

I think Hemsky being that high makes sense.

I mean, it's not often you see Hemsky hemmed inside his zone. He gives the puck away at frustrating times in the wrong spot for an odd man rush, no doubt... but he's rarely hemmed inside his own zone.

If I was going to search for a couple reasons why, I'd look at the idea that he was probably on the ice for a lot more offensive zone faceoffs than defensive zone faceoffs (which is why I'm surprised that Horcoff STILL managed to be that high), and the fact that he spent most of the season with guys like Smyth, Horcoff, Peca, Moreau... coincidently all in the Oiler top 8.

He was also generally pretty good at keeping puck possession in the offensive zone once he had it.

... I haven't played an NHL game forever, care to explain the reference? I'd guess it's that a teams D is rated by their defensemen slowly and goal scorers all hit around 80+ while a guy like Mike Peca is a 70 OV?

8/25/2006 7:32 pm  
Blogger JavaGeek said...

Did you strip all the data off the shift chart images? I haven't even gotten to 2003-2004 shift charts yet. Looking at Spacek and Pronger I was shocked to see Spacek so much different than Pronger, because I have Spacek spending 83% of even strength time with Pronger, meaning even if you rescale Pronger's shots against rate to 350 you get 350*0.83+var*0.17=376 => var = 50, which doesn't make sense in this context...

I'm not sure you can conclude about who's driving the bus with this data though, due to the fact that defenseman play with all the lines their results will be a function of line 4 and lines 1-3 and as such be lower. Also some players allow better shots against than others...

Hemsky is really quite good, I would consider him one of the top defensive forwards in the league...

Did Ulanov become a worse player in the offseason? The injuries didn't help him do well. I suspect his 96% is an anomaly

8/25/2006 10:07 pm  
Blogger RiversQ said...

Interesting stuff here Vic. Somehow Hemsky's a low event machine again. How much icetime did he really have away from Smyth and/or Horcoff? This is really the burning issue at this point for a lot of these posts.

I am surprised about Pronger. I would have put him on this list somewhere, but not ahead of Pisani or Peca. Considering his healthy TOI, that's probably no small factor in the Oilers' shot prevention ability. (I would love to see his shot prevention numbers on the PK relative to the other dmen - I swear he was outstanding 4V5 for the first 20-30 games by GAA) Hopefully better goaltending and more time in the other zone will compensate.

It was great to watch Pronger play this year - he had so many subtle little plays to regain puck control in his own end. If there's a dman that can impact results with his play in his own end, he's a prime candidate IMO. Most of the others probably do it strictly by being efficient puck movers, but he can really take command of his own zone.

javageek: Spacek may have played 83% of his time with the Oilers with Pronger according to your numbers, but he played just 31 regular season games as an Oiler. They're hardly inseparable given that. The idea that the Spacek/Pronger duo was significantly different than the other Pronger pairings is not so outlandish IMO.

I would agree with you that this data doesn't indicate who's driving the bus. It seemed to me that Vic was alluding to some additional analysis related to this data...

8/25/2006 10:48 pm  
Anonymous Colby Cosh said...

Alas, we are probably quite some way from normalizing these figures for the times when Hemsky holds the puck in the other team's circle for an entire shift while the defenders look on in horror and try to keep their feet moving. (He is underrated away from the puck though.)

8/26/2006 1:11 am  
Blogger Vic Ferrari said...


Ya, the rest of the defense are bunched pretty close together in the middle, as you would expect. If they'd all played 15 hours of even strength icetime then the shots against would have been:

J. SMITH 401

Cross and Tarnstrom both wore #23 and I didn't bother to separate them.

Really within a handful of shots there. That's pretty typical really, if the Oilers depth forwards were better then the bottom pairing guys would have had the best numbers of course, Huddy does a good job of keeping them away from danger. Hopefully that problem will be rectified this season with the young forwards coming in. Pronger is the real anamoly here. Especially considering the level of opposition he often faced.

A shitty shooting percentage in front of Pronger hurt his EV+ numbers this season. And a pretty poor EVsave% behind him by league standards (.909) but better than most for the Oilers last year. Unlike 03/04, when Pronger's numbers really got stung by a poor save% behind him in STL according to the shift chart and playbyplay numbers anyways (.907 when he was on the ice, about 20 points higher when he wasn't). Shit happens.

Still, in 03/04 STL did much better at outshooting the opposition when he was on the ice at 5on5, in spite of playing very difficult opposition for the most part. Which seems to be more of a constant.

Modano is the mysterious one though. Just tragic results in 03/04 mostly down to a nightmarish .869 EVsave% behind him. He was a good bet to bounce back this season, and obviously he did. Absurdly good 5on5 numbers this year, and a decent PP year for Modo even by his own high standards.

And Smyth this year, which was sort of where I started, I have to think that at least some of the horrific .863 save% behind him was his fault. I mean I don't think I've seen one line be on the ice for so many odd man rushes each way since Bure retired. But damn, that's a harsh number. Same applies to Horcoff of course, and to a large extent Hemsky as well. I mean we know they all takes risks with the puck at times, more than most players, but maybe the bleeding with goals against is largely as Lowetide suggests, they just got more than their share of bad breaks in that regard. I dunno, but personally I'm leaning that way.

8/26/2006 2:40 am  
Blogger Vic Ferrari said...


My bad on the Ulanov number, I trusted my memory there. It was in fact .938 EVsave% behind him, and only a .864 EVsave% from the opposing goalies when he was out there. For those of us that believe that's mostly just coincidence for all players ... then he was one lucky effer in 03/04.

Jarret Stoll, of all people, had the best with an EVsave% of .941 behind him that season. A pretty strange list of guys who god favoured by the hockey gods with a string save% behind them at evens in 03/04, not that extreme as the other two, but still ... Chimera, Semenov, Oates, Bergeron ??? Go figure. This season it seems to be more the known defensive types that got the good save% behind them. Probably some of it was due to their play, but probably mostly just coincidence.

You're a VAN fan, why would Ohlund had such a crappy EVsave% behind him in 03/04? Granted it was obscured by the fact that the 'Nucks happened to have a bitchin' shooting% when he was out there. I didn't see many VAN games that year, but I'll vote for "coincidence" anyways.

8/26/2006 3:13 am  
Blogger JavaGeek said...

First off save percentage isn't all luck, a player can effect shot quality, and as such should reduce actual save percentage in the process. However, a player from year should not "change" their skill level (with the exception of injuries and other fitness issues). So differences from year to year can often be looked at as error (it makes more sense than the alternatives). One could also look at line mates...

You're a VAN fan, why would Ohlund had such a crappy EVsave% behind him in 03/04? Granted it was obscured by the fact that the 'Nucks happened to have a bitchin' shooting% when he was out there. I didn't see many VAN games that year, but I'll vote for "coincidence" anyways.
Actually Ohlund is likely a big reason for the Canucks struggles defensively, he spends approxemately 25 minutes on the ice per game with not much more than average defensive skills. I have him down for 1.1 shot quality against, meaning he allows 10% more difficult shots on average => he should have a .898 save percentage in 2005-2006.

8/26/2006 8:47 am  
Blogger Vic Ferrari said...


We seem to disagree on almost everything here. And though I am usually loathe to get into a debate about players and teams that are NOT the Oilers ... I can't leave this alone.

To my eye Ohlund is far and away the best defenseman on the Nucks, at everything except the PP. Big strong, wins battles, can log a lot of minutes, and an underrated first pass. And Crawford clearly thinks so too, because he plays the guy in a shitload of critical situations. He has for years.

I just picked two VAN home games from near the end of the season. Game#s 1158 and 1167. That's the Flames and Ducks respectively.

Game 1158: According to the shiftcharts, which is admittedly not razor accurate, Ohlund was on the ice for 76% of Iggy's 5on5 icetime. Jovo for 18% of it (and at that probably as CGY PP's expired or the first shift after, I dunno).

On the matching with Ohlund ... that's just about as good as you can do in a game with any number of penalties . It's a hard match.

Game 1167: According to the shiftcharts, Ohlund was on the ice for 77% of Teemu's 5on5 icetime. Jovo only 21% of it.


BTW: Naslund was 30% and 23% by this measure.

IMO, as good as Ohlund is, he just isn't going to be able to rack up gaudy counting numbers, or even EV+ and EV- numbers when he's carrying that kind of responsibility. Those forwards are result getters, good players and they'll get their chances eventually.

Just going by this ... I'd bet that 2nd period 5on5 icetime drops a fair bit in road games for Jovo and Naslund. And increases for Ohlund and the Sedins too, maybe Linden as well. Just a guess mind, I'm not inside Crawford's head. Is that what you saw?

I think Nonis kept the right guy of Jovo and Ohlund, and with $ considered it's a no brainer methinks. So long as Nonis has some other point option for the PP that doesn't suck.

And I suspect that whoever was playing in VAN's bottom D pairing barely got a glimpse of the leagues star players, especially on home ice.

8/26/2006 11:14 am  
Anonymous YKOil said...

One has to think that playing Ohlund like one would play Ponger (the match-ups, the TOI, etc) would HAVE to hurt Ohlund's stats. Mainly because Ohlund isn't Pronger.

To my way of thinking only Lidstrom, Pronger 7 Neidermayer can play those kind of minutes and match-ups and get superior results while a second group of Chara, Redden, Aucoin can do okay. After that it's 'do your best' time out there. Put Ohlund in the #3 spot on a strong team and he would dominate imo.

8/26/2006 11:56 am  
Blogger JavaGeek said...

You can look over the course of the entire season:

Percentage of even strength time:
Ohlund: 36.3%, Iginla: 31.7%
Expected time together: 1876 seconds
Actual: 3675 seconds.

Percentage of even strength time:
Jovanovski: 39.8%, Iginla: 31.2%
Expected time together: 1458
Actual: 885 seconds.

Over the course of 8 reg-season games vs. Calgary. [Jovanovski missed a lot of games, which is why Iginla's ice-time does not agree.]

I cannot deny the fact that Ohlund as played vs. "tougher" opposition, but I can question whether it's a good choice to make. I'll do another strength of opposition that looks at teams lines.

YKOil: well said.

I'm not saying Jovanovski is worth $7 mil...

8/26/2006 12:47 pm  
Anonymous lowetide said...

Hmm. I'd have to see a mountain of information before agreeing that Ohlund is second choice behind Jovo. Seriously. Ohlund is steady and strong, Jovo is at the high end of the chaos defensemen who give new meaning to the phrase high event.

Can my eyes be deceiving me?

8/26/2006 12:57 pm  
Blogger mudcrutch79 said...

I'd agree with you LT.

Chris - I like what you've done there comparing expected times together given what I assume would be random TOI distribution and actual. Very cool.

8/26/2006 4:54 pm  
Anonymous Anonymous said...

Ohlund always seemed much steadier to me, and LT put it perfectly, Jovo seems to be on the high event of high event defencemen.

You sure notice his flash and dash, but whenver Ohlund is injured their PK always seemed to suffer, I'd like to see the statisics of PK time between Ohlund and Jovo.

8/26/2006 4:58 pm  
Anonymous PDO said...

I have to agree that Ohlund >>>> Jovo.

Just curious; when we're mentioning ES SV%... shouldn't that be a bit of a red herring on strength of opposition?

I mean, stronger opposition should theoretically be the Demitra's, Forsbergs, Jagr's, etc... correct? So wouldn't it only make sense that they'd be more likely to score than a Moraeu or Yelle?

We're calling it bad luck that Pronger had a shit ES SV% behind him in 03-04... personally, I'd be surprised he if didn't, given the type of players he played against.

8/26/2006 5:27 pm  
Blogger JavaGeek said...

I'm avoiding delving too deeply into the discussion because it's by far too complicated Ohlund is a very different defenseman than Jovanovski. However, most GM's would pay more for a Jovanovski than Ohlund...

You sure notice his flash and dash, but whenver Ohlund is injured their PK always seemed to suffer, I'd like to see the statisics of PK time between Ohlund and Jovo.
Ohlund missed 4 games in 2005-2006 (none in 03-04) and the team went 4-28 or 85.7% compared to the team's score of 81.8% (note the differences are well within expected error). But it doesn't appear the PK suffered...

What about 02-03, he missed 23 games...
It was 78% (78 - 100) vs 84.1%. (Salo replaced Ohlund and Baron + Salo was horrible on the PK). The PK appears to at least have suffered a bit. This may be the result of choosing a poor alternative to Ohlund...
When Jovanovski was gone there was no change in PK% (84.4% vs 84.1%) Same for 03-04.
Jovanovski was still statistically the best PK guy in 2002-2003 for defenseman. It's likely Being paired with Salo likely hurt Ohlund's PK statistics in 2005-2006...

Just curious; when we're mentioning ES SV%... shouldn't that be a bit of a red herring on strength of opposition?
Unless of course you give the puck away in the slot... (it wont matter who it is, it has a good chance of going in). Or give up a break-away or odd man rush. But the opposition likely has at least some effect on shooting percentage against.

I'm redoing my strength of opposition theories (fix a few bugs), so I'll see how Ohlund and Jovanovski fair after that...

8/26/2006 9:30 pm  
Anonymous MikeP said...

One thing that comes to mind with Hemsky is (and keep in mind I didn't get to watch much RS play and I'm going mostly from memory here) - it seems to me that while he doesn't give the puck away or miss checks or the like as often as he's decried for doing, when it does happen it seems that it's often a good chance going the other way. Any way to adjust for shot quality? I'd be willing to bet that Pronger would lead the pack even after adjusting for that too.

Or, failing a quality adjustment, can you check for shots resulting in rebounds? That would be the next best thing, I suppose.

8/28/2006 6:35 pm  
Anonymous oilerdiehard said...

Vic said:

The Hemsky thing is hard to explain. I dunno. Clearly I've been too harsh on the guy, two years in a row wih damn decent outshooting stuff, odd man rushes surrendered or no ... the chipstack is pretty respectable.

I agree with the comment above (by a different poster) that Hemsky is underrated in his own zone (not to mention he comes back harder on the back check than most take the time to notice). When he does give away the puck it is mainly semi-deep to deep in the offensive zone which allows plenty of time for forwards and D alike to adjust and defend the situation. As opposed to a loopy no look back pass by Torres at our blue line or just outside our blue, little to no time to react and a quality chance is often at hand.

I noticed especially in the 2nd half Hemmer was getting very good at making the safe play out of the zone and even doing a respectable job fighting for pucks along the boards and coming away with the puck (I would not mind if he applied this a bit harder in the O zone though).

8/30/2006 9:20 pm  

Post a Comment

<< Home