Saturday, May 20, 2006

The Selanne Factor

Mudcrutch has an excellent post on the problems with the Canucks this season, focussing on the derailment of the West Coast Express line and factoring in Save%. Funny thing about that, you can do the same with any team when looking at big upsurges or failures ... and it just always seems to work. SEE Dallas Stars with Modano and Save% over the past few years ... stunningly consistent with these two things factored in, which just makes sense.

Anyhoo, Using WCE logic on the Ducks ... but with Selanne as the key factor.

Ducks at 5on5 (only as accurate as the NHL.com Event Sheets)

5on5 icetime: 3188 minutes
4on4 icetime: 278 minutes

Which is pretty much the league average split.

Selanne 5on5 minutes: 847 minutes
I've guessed here that he played 40% of the 4on4 minutes, which should be near as dammit.

So, at 5on5 with Selanne on the ice:

ANA Goals-For Per Hour: 4.2 (shockingly good, yowza!)
ANA Goals-Against per hout: 1.8 (pretty decent ... though you'd expect it to be lower given the madass scoring rate, I high-event cat is Selanne, always has been)

And, ANA at 5on5 with Selanne on the pine:
ANA Goals-For Per Hour: 1.8 (bag suckage)
ANA Goals-Against per hout: 2.2 (decent)
Factor in the great EVsave% that Jiggy and Bryz gave them ... then the fact is that they were nearly the worst team in the league, as a group of skaters, when Selanne wasn't on the ice, though WSH without Ovechkin on the ice was beyond terrible, and save them from rock bottom.

This made Selanne the league leader in the Ron-Wilson-Number category, bumping Forsberg to second place for the first time ever afaik (In fairness to Peter he faced much tougher competition and was playing injured, still ...)

All ships rise with the tide. Selanne is the tide. And it's surprising as hell, I mean who gets so much faster as they get older?

.

As an aside: If Fedorov gets his speed back in the same way (and for the sake of the game I hope he does) ... then CBJ are going to surge next season. And in stunning fashion. And mediocre Dmen and wingers, as well as mistake prone youngsters, they are going to appear much better. And the perceived IQ of coaches and GMs will skyrocket.

.

As an aside #2: Carlyle runs Selanne with the craziest shift patterns imaginable. Huge stretches of inactivity and then monster long shifts, then a series of short shifts. Clearly the game plan is designed around him, and it's nigh on impossible to match that with forwards without running a screwy bench yourself. Doing this really limits Selanne's overall 5on5 icetime too ... but it seems to work for Carlyle. In the Oilers game 81, with MacT finally abandoning the "let's develop Hemsky" mantra and inserting Dvorak up with 10/94 ... that's about as tight a match with forwards that we've seen vs Selanne this season. i.e. You have to like the Oiler's chances at home in this series. And huge props to Huddy and the off-ice coaches for getting a tremendous match of Pronger and Selanne in Game 1. That's been tough to do this season, smart guys like Babcock and Quenneville haven't done nearly as well.

13 Comments:

Anonymous Anonymous said...

you guys should make these rankings for every player in the league, kinda like what 82games.com does for basketball.
Or at least hook us readers up with all the stats you're using.

5/20/2006 1:07 pm  
Blogger mudcrutch79 said...

...then CBJ are going to surge next season. And in stunning fashion. And mediocre Dmen and wingers, as well as mistake prone youngsters, they are going to appear much better. And the perceived IQ of coaches and GMs will skyrocket.

I haven't taken a good hard look yet but I think that they're going to be my darkhorse playoff pick for next year.

You have to like the Oiler's chances at home in this series.

You have to like the Oilers chances just period in this series. Intellectually, I know that the Oil are maybe 60% to win right now. I don't see how they can lose unless the hockey gods are really against them though. It just seems like the Ducks are a team who match up incredibly favourably for the Oilers.

5/20/2006 1:34 pm  
Blogger Vic Ferrari said...

As you probably know, 61% by the bookies right now, 63% by me, but I take your point.

Then again, you had to like DET against the Oilers at every point, and OTT outplayed BUF in almost every period there and lost in 5. The bounces be the bounces.

And Selanne is still a force, easier to minimize him in the game plan than on the ice I'm sure. A lot of smart guys have tried and failed.

Fact is, the Ducks almost never win when Selanne is a minus, and rarely win when he is merely evens. Plus he's a key PP guy. They rely on this player to an extreme. And he's hard fucker to stop this year. Good on 'im. I still remember how he spoke publicly about his bitter disappointment at the Jets leaving the 'Peg (the reason he was traded and never played a game with the Coyotes btw). Good guy, none of the Oilers spoke like that when it looked like they were heading out of town, not one. At the time our local heroes cautiously played both sides of the fence, most erring on the side of "kind words for Houston" with their verbage iirc. And although I would have done the same in their shoes, my moral standard isn't the benchmark ... so for shame.

5/20/2006 1:51 pm  
Anonymous momentai said...

Yeah Vic.

At the beginning of the game, I was surprised that the Selanne/McDonald/Kunitz first line was going to be matched up against Horc. That was not something I expected out of Carlyle and I had to admit I was questioning this motive.

However, Selanne's line got "out of dodge" real quick and Pahlsson line came out in the middle of Horc's shift. Shortly nearing the end of Horc's shift, though, Selanne's line came out once again.

I remarked at how weirdly Carlyle was running his bench. I don't know if I've seen that very often.

5/20/2006 2:44 pm  
Blogger Vic Ferrari said...

Momentai:

Ya, that was strange. I think maybe it's a case of 'starting your star player at the beginning of every period so you don't piss him off', I dunno. But like you say, he's off in a hurry and then right back on in just a little while. Crazy stuff.

You can reason your way through the actions of Quenneville, Babcock, Crawford, Lemaire, etc ... they're smart guys and have things they want to do in a game, and it's easy to see within a few minutes usually, and they'll shift away into Plan B if it stops working ... but Carlyle is a bit all over the map with the forwards. I'm not sure whether that's just 'Ron Low style' erratic gut-feel stuff, or if there is a method to the madness. For reasons I can't really pin down; I tend to give Carlyle credit. Yesterday I looked through a few shift charts for the Ducks, against good bench coaches ... and you can't argue with the results. A schizophrenic quality about the pattern of Selanne's icetime though.

5/20/2006 4:12 pm  
Anonymous huffpuff said...

Once again, cool stuff, Vic!

Yeah, last night was a very impressive line matching job by Huddy considering that the last time at the Pond (2006-01-25), Selanne managed to spend half his EV time against the Ulanov/Greene pair. Ouch. Though I think all that special teams time in that game probably made it difficult to match lines. And in the end, the Giggy blow-up saved our skins.

[Of course, my words predicate on the accuracy of NHL's TOI data, and that's always a bit of a question mark....]

Looking at the past several games against ANA, there does not seem to be really any distinct matchups against Selanne as far as the forwards are concerned. The couple of defensive suspects are kept away as much as possible. And then the reliable ones pretty much split their time against him (though in that April game, Pisani & Moreau got the majority of the minutes)

As for Selanne's turn around ....if I'm not mistaken, I think the official word is that he was hampered by injuries for much of last season. IIRC, in the games that I've seen him, even with the Avs playing him on the 3rd line, he looked pretty good - skated hard & worked his butt off - but just couldn't get any breaks.

As for Fedorov, I dunno. Gun to the head, I don't know if I would bet on the same turn around. He's made the big bucks and has won all there's to win...he just doesn't seem all that motivated anymore.

A bit of an aside:

I've been hacking the NHL's TOI sheets and have generated some Excel files that allow one to look at in-game matchups. Basically, I've flattened each player's shift data into seconds-by-seconds slots (1 = on ice, 0 = off ice) and inserted them into columns. One can then use autofilter to see the desired matchups. There are some obvious errors in the data (e.g. 11 men on ice for one team at the same time) but overall it seems reasonably close enough.

Anyway, if anyone is interested, it can be downloaded from here:

http://rapidshare.de/files/20977282/toi_array_uploaded.zip.html

I've included 3 EDM vs ANA games in the regular season plus all EDM & ANA 's po games so far (except for game 3 of EDM vs SJS due to NHL totally mucking up the data for that game.

5/20/2006 7:04 pm  
Blogger speeds said...

This comment has been removed by a blog administrator.

5/20/2006 10:28 pm  
Blogger speeds said...

It's going to be pretty hard to get Selanne's line away from both the Horcoff and Peca lines, no matter how Carlyle runs the bench, and they should get Pronger out against him most of the time as well.

5/20/2006 10:57 pm  
Blogger Vic Ferrari said...

huffpuff:

Great stuff. I just summed your total icetimes for Selanne and Pronger and they match the NHL.com event sheets icetimes right to the second.

Here's the tough bit though, the PP and SH times are off, quite a bit in this case, about 1.5 minutes or so. I've had a bit of a kick at this before, looks like the NHL arbitrarily make one shift disappear for one or two players in the game. Creating the impression that there was a powerplay for one of the teams during that timeframe.

Really it doesn't matter much, if you're looking at head to head 5on5 icetime, you may or may not have one shift missing for one or both guys, just because the TOI sheets tell us it was PP time. So you show that Pronger played 75% of Selanne's 5on5 minutes, the shift chart shows 73%, neither is likely spot on but they're easily close enough for what we're doing. I mean with that we've got a pretty damn good idea of the matchup Huddy was pursuing and the extent to which he was successful (bitchin' good in this case :D )

Still, just on principle I'd like to write a script that searches for the missing shift(s) and fits it in. Just because I hate to think that these greasy effers at NHL.com have beaten me. I imagine that they have written a script that uses random numbers to choose the small error, the missing shift, that they want to impart on the TOI sheets before they publish them, so I don't think you could hit it from that side.

Maybe one day, when the mood strikes, I'll take a kick at this problem, it's like the logic puzzle from hell is this one. Better yet ... you could figure out a script first and let us know. ;)

5/21/2006 1:20 am  
Anonymous huffpuff said...

Yeah, I know...and I'm also as frustrated as you are.

I have also wondered myself if the NHL actually sabotages the data before publication. But then it's probably just mistakes by the stats keepers. Once in while, they probably forget to hit a button somewhere and find out when it's too late. In their defence, it is a pretty fast game to keep up with.

But some of the junk they've posted really makes you wonder what's going on in their IT dept. I mean, look at the garbage that they published for Game 3 of EDM vs SJS....WTF was that about????

As to searching for & correcting the errors...my script actually shows where the problem spots are. For instance, under the #On.raw column for ANA, if you filter for 2, you'll find that for the first 44 seconds of the 2nd period, the TOI sheet says that ANA only had 2 players on the ice. The shift chart shows the same thing. So there's really no alternative data source that can be used as an accurate reference. It appears that the only perfect solution is to rewatch the game & try to figure out which players were on.

My current (totally imperfect) fix is to clip that number to 4 which implies a 2 men PP for EDM. This is clearly wrong when you look at the game summary. That's one of the reasons why my situational times differ from those in the events summary. I wonder if I'd get better results if I clip all error values to 6 men instead...

I think one obvious way to alleviate this problem is to try to figure out the PP/SH/EV time regions by going thru the goals/penalties sequence in the game summary. Once this time series is determined, it can then be superimposed onto the players shift data (instead of using the diff method based on incorrect #-of-men-on-ice info). You still have no way to figure out the missing/extra shifts on an individual basis, but I think this would still be a huge improvement.

I had thought of doing this but then I really didn't feel the urge to decipher the NHL rulebook and learn the precise intricacies of all penalty situations. Besides, one gotta think that someone has already got the algorithm stashed away somewhere...

Anyway, that's enough geeky ramblings outta me for now, so I'll stop right here.

5/21/2006 6:06 am  
Blogger mudcrutch79 said...

Huffpuff:

Good post. Out of curiosity, how'd you end up posting here and following this stuff? There seem to be about six of us who've paid attention to it the past few years and I don't recall seeing your name at OF or HF.

I agree with you on the mistakes. Someone somewhere else (I can't remember who or where) was commenting on this and noted that if Smyth is on the ice and a penalty is taken and he stays out for the PP, the coding for the shift doesn't change.

I think one obvious way to alleviate this problem is to try to figure out the PP/SH/EV time regions by going thru the goals/penalties sequence in the game summary.

Something I'm thinking about doing but I've got hte smae problem-I don't want to learn the rules;).

5/21/2006 4:47 pm  
Blogger hffpff said...

This comment has been removed by a blog administrator.

5/22/2006 7:59 am  
Blogger hffpff said...

Good post. Out of curiosity, how'd you end up posting here and following this stuff? There seem to be about six of us who've paid attention to it the past few years and I don't recall seeing your name at OF or HF.

Oh...I've been lurking around the message boards & reading various hockey discussions, including that stats stuff that you guys post. Only recently have I got the time to actually do some hacking on my own. For a while I was wondering where the stats guys have all gone and then in a post I saw Vic's signature pointing to this blog. From here I've followed the links to other blogs such as yours, which is another one that I've enjoyed reading.

I agree with you on the mistakes. Someone somewhere else (I can't remember who or where) was commenting on this and noted that if Smyth is on the ice and a penalty is taken and he stays out for the PP, the coding for the shift doesn't change.

Yeah, that coding error is a bit of a careless oversight. But I think the focus of their presentation is on the shifts as individual units (i.e. how many, how long, etc). So perhaps the situational codes are just secondary info. But still, it's such an obvious issue that they've overlooked. And it should have been easily dealt with - just split the shifts into separate parts and denote each part with a sequence number...or something similar. That way they can still preserve the shifts as individual entities and make us all happy :)

Anyway, it would have been almost a nonissue - just a minor inconvenience - had all the shifts' on/off times been complete and correct. That's the biggest problem.

BTW, not sure if anyone is still interested, but I've uploaded the xls file for game 2 here:

http://rapidshare.de/files/21075519/toi_array.2005_06.P.0322.EDM.ANA.2006-05-21.zip.html

5/22/2006 8:26 am  

Post a Comment

<< Home