Thursday, January 01, 2009

"That's killed us all year long. It killed us again tonight."

Waking up today hangover free could only have been some sort of sign - a message from the champagne gods, perhaps, that I must seize this miraculous day and do something positively spectacular with it.

But then it hit me, all at once sort of, as I did my daily sit-up in bed and aimed myself mostly successfully towards the bathroom to do what had to be done.

I imagine that the players and personnel responsible for the Oilers penalty kill felt much the same way this morning, with the added bonus of having felt that way last night as well while I was thirstily warning 2009 what will happen to its supply of booze if it's not careful.

Craig MacTavish's quotes in the dailies today basically echo what any warm blooded Oiler fan with a working set of eyes or ears or access to anything Dennis has written since roughly October has known for some time now: the Oilers can't kill a fucking penalty to save their life. Edmonton's needs list no matter who you ask right now, and summed up pretty accurately by Lowetide today consists very obviously of players who can kill penalties and who can win a faceoff. Should be obvious, no?

Today I set out to try to get a read on which of those two qualities is the most pressing. That is, would getting a player or two who can win a fucking faceoff generally push towards a better pk? After all, this is the year that every penalty kill starts in the pk team's defensive zone - surely this must have some measurable effect on how many pucks end up in the back of the net? And finally, if Edmonton sucks at A and also sucks at B, could there be a correlation there?

Well, if you look at the image (click to enlarge), the answer seems to be a resounding no. I don't claim to be one of the Oilogospheres heavy hitters but I think that this particular case is relatively straightforward. When you look at the overall lack of positive correlation, especially in 2008/2009 now that the rules have changed, you can't conclude that a good faceoff percentage = a good penalty kill. This might be common sense but based on what I'm reading around the internet today I think people are overestimating the impact of faceoffs on Edmonton's pk train wreck this season. What I'm trying to say is that if one conclusion can be drawn it's that there are too many elements to a penalty kill to suggest Edmonton should be looking at adding a faceoff specialist alone - faceoff % should take a back seat to players who can actually kill the damn thing off. After all, even a successful faceoff win and dump only shaves 10-15 seconds off the penalty anyway - you still have to kill the whole damn thing off regardless of how it begins.


Blogger JLikens said...

Interesting post.

A short while back, I tried to see if team faceoff percentage was correlated with team performance (shot differential, goal differential, penalty differential). The correlations were disappointingly low or non-existent.

I think that part of the reason for this is restriction of range. Almost all teams in the league fall between 46-54% in faceoff percentage, and therefore each team wins approximately half of all faceoffs.

If teams were sufficiently varied in faceoff percentage, certain relationships might emerge.

But winning an extra 3-5 faceoffs per 100 draws is going to confer a very small advantage that will be difficult to detect with statistical methods.

1/01/2009 2:08 pm  
Blogger Showerhead said...

Good post Jlikens; I think you've summed up the reason for lack of correlation perfectly.

The problem with following the NHL of course is that the one or few goals per season that come off the draw are anecdotally blown out of proportion. Sure it happens and sure it's not always luck but if you have a bad team you have to take an incredibly micro view to think you can fix everything by fixing your faceoff %. It's almost up there with bad penalties being harder to kill than those that prevent a goal :)

1/01/2009 4:05 pm  
Blogger HBomb said...

With regards to the faceoff "range restriction", and I am not claiming to be a stats genius here by any means, but couldn't there be some sort of numerical "trick" done here to make this clear up?

The start might be to express every team's faceoff percentage as a ratio relative to league average (i.e. 50%). That might not be enough to do it though. Make the lowest of these ratios zero, the highest one, and assign numbers between proportionally to each ratio? Does anyone think what I'm saying here makes sense and might lead to a stronger correlation? Or does it not matter?

1/01/2009 4:15 pm  
Blogger mc79hockey said...

Showerhead: Great stuff and it seems to jibe with my sense of things. Everyone remembers the games like the Nashville game earlier this year where there were a clump of goals scored off faceoffs and starts looking for those so that they can say "Ah ha! Another goal off a lost faceoff! Oilers need that faceoff man." What they don't forget are those with less obvious roots, such as the pure barrage of rubber faced by the goalies.

@HBomb: I don't think that it matters. The Oilers were, last I looked, just getting blasted in terms of shots on the PK. Over 60 an hour. Would winning more faceoffs help? Sure. Is it enough of an issue that it makes an appreciable difference? I doubt it.

1/02/2009 7:50 am  
Blogger HBomb said...

mc79: Since Corsi includes blocked shots, I'm wondering if there's a way to look at the team's short-handed corsi number as a whole (per hour of PK time) and how it's trended the past four years, and run that up against the teams PK performace in terms of goals allowed per sixty.

My theory, based on the "flamingo" move we've seen this year (i.e. unwillingness to block shots) is that the PK Corsi number is going to be roughly the same, but more shots are getting on net and, as such, other teams are having more success scoring PP goals. Of course, one would have to look at save percentage as well and see if there's been a drop-off in that area.....

1/02/2009 9:59 am  
Blogger HBomb said...

Short version of what I just said above: I don't think it's the faceoffs, I think it's the shot blocking, or lack thereof, that's making the PK look like crap.

1/02/2009 10:00 am  
Blogger MikeP said...

JLikens, do you have your study online somewhere? I remembered somebody doing something like what you describe, but couldn't find it again.

Of course, the reason I was trying to find it was to look at something like what Showerhead did, except from the other side.

1/02/2009 11:07 am  
Blogger YKOil said...

Honestly HBomb, imo anyways, it is both.

Combine O-zone drops with fewer faceoffs won and the result is more pressure faster with fewer chances to eat up time with down-ice pressure via a decent clear.


Add to that a reduced ability to disrupt an attack plan (fewer blocked shots) and... well...

welcome to the Oilers 2008-09 PK.

In terms of face-offs, to me, the areas requiring study would be:

1. review face-off winning % with PK ability and look for the guys who can:

a. do both
b. do one and not the other
c. more specifically, look at the guys who are good PK guys but not good face-off/PK guys

The point of 'c' is this - much like Cole can't play LW are there guys out there who have a greatly reduced effectiveness on the PK if they also have to fill the face-off function? Are we looking at a narrower skill set and the deficiency that results when said skill set is lacking?

2. correlate the o-zone start to overall time spent in the offensive zone by the PP team vs. prior year results. Also look at results.

Great thread.

1/02/2009 12:09 pm  
Blogger Vic Ferrari said...

Great stuff, Showerhead, you should post more.

I`ve never really looked at faceoffs before. I tend to think of them as just another puck battle, and there are a whack of those in every penalty kill.

To my mind the difference with faceoffs in hockey is that the play stops, every fan takes a breath and has a think, the commentators blather on about the guys on the ice ... it becomes baseball for a moment.

1/02/2009 12:25 pm  
Blogger Vic Ferrari said...


If you want to make faceoffs more important I think you need to take a micro view. So if you looked at games where one team won 70% + of the faceoffs at evens, then their corsi+/- would be much better than expected.

If you looked at games where one team won 70% + of the faceoffs on the PK, then their PK GA and shots against would be much better than expected.

The problem is, as JLikens says, that there is just no real sustain in whopping faceoff numbers. In the long run most teams just aren`t all that much better than the rest.

Oates said the same thing in different words as an Oiler (this when he was embarrassed by MacTavish crediting him for the Oilers improved faceoff performance) ... more or less "a good year for faceoffs is 55%, that's still just a D+".

Look at the tree and faceoffs are the bomb. Start tracking back goals to where it went wrong and you`ll be lead to that, methinks. Look at the forest, as showerhead has done here, and there is not much in it.

1/02/2009 12:33 pm  
Blogger Vic Ferrari said...

That`s a terrific take, YK. Good to see you talking at us again lately.

There is a ripple effect with PK personnel based on their ability to take faceoffs for sure. I mean if Cogliano was good on the dot then a whole bunch of things change for the Oilers PK ice times no doubt.

1/02/2009 12:40 pm  
Blogger andy grabia said...

All of this begs the question: who is out there that would help, who would we want, and how many of them are available?

1/02/2009 2:02 pm  
Blogger Showerhead said...

A lot of good discussion - I'm glad I found the time to post again.

The biggest issue I still have a problem with is the idea that there is a single keystone element to the PK that we can fill and see immediate results. Edmonton is not winning faceoffs, their forwards are all over the goddamned place, shots are not being blocked, passes are not being disrupted, and pressure is not being applied at the "right" times. The one area I will give credit is that once the puck is on the wall, Edmonton's D-men are doing an average job IMO of winning the battles.

Earlier in the season I remember posting to the effect that the PK didn't worry me - one game in particular, I remember thinking that there was no way Edmonton could give up that many odd man rushes consistently. I think I was right about odd man rushes but dead wrong about not needing to worry.

Back to the lack of a keystone issue, I don't think any of the things Edmonton needs to work on are quick fixes ;). How did MacT's team get this bad at the X's and O's of penalty killing?

MC79/HBomb - re: shots, I have been wondering about that as well. If X is total shots that get through, Y is total shots blocked, and Z is total shot attempts, how has each changed since past years? I'd wager that Y has gone down, Z has gone up, and X has therefore gone up but is any one of Y or Z more to blame?

1/02/2009 4:36 pm  
Blogger HBomb said...

showerhead: I'd bet Z is damn near constant when you look at the rate per sixty. Might not just be an Oiler phenomenon either.

On the PK, it's 5-on-4. Even if that's Atlanta vs. Detroit, an extra guy on the ice is a big frigging advantage. Teams are going to get their shot opportunities (i.e. it's unlikely a team kills many penalties by ragging the puck and preventing entry into their zone - teams will get their shots even against Draper/Maltby/Lidstrom/Rafalski). The good PK teams will have certain things going for them:

- Good goaltending (i.e. superior PK SV%)
- The ability to block shots
- The ability to force missed shots

The first one, IMO, is the "luck" element in essence. The second and third are a reflection of the team's PK system and how it's functioning (and the players within it, by extension). Teams on the PP will get their shot opportunities, but the teams that are good when down a man will be the ones with the abilities to turn shots that hit the net into blocks and misses. Corsi, as such, shouldn't fluctuate wildly from season to season.

1/02/2009 5:24 pm  
Blogger Showerhead said...

HBomb - I don't mean compared to other teams or to 5 on 5 - I mean Edmonton's pk this year compared to last or other years. I'm not sure if I made that clear or if I properly understand your post.

1/02/2009 5:31 pm  
Blogger Bruce said...

Interesting to track the Oilers performance relative to league. In the past four seasons Oilers have finished 1st, 7th, 3rd, and 4th in blocked shots; currently they stand 17th. Surely that represents a change in approach?

That said, 17th is still middle of the pack, or at least a lot closer to it than 27 th in faceoff % or 29th in PK %.

1/02/2009 5:31 pm  
Blogger HBomb said...

HBomb - I don't mean compared to other teams or to 5 on 5 - I mean Edmonton's pk this year compared to last or other years. I'm not sure if I made that clear or if I properly understand your post.

Showerhead: what I'm saying is that I think this extends beyond the Oilers year to year.

- Other teams probably have similar trends in their "PK Corsi" being pretty constant year to year.

- I also think there's probably not a tremendous spread in the PK Corsi from team to team in any given year. However, the good PK teams are going to increase the number of missed shots and blocked shots, not the overall number of attempted shots.

Make some sense?

1/02/2009 6:46 pm  
Blogger Showerhead said...

OK but do you have the evidence to support that or is it still in theory stage? I would suspect that good pk teams clear the puck enough, just once more per kill would do it, to skew the amount of shots thrown at the net because of zone time alone. This doesn't get into puck recovery off rebounds or missed shots which should also have an effect, albeit a lesser one, from my point of view. Basically what I'm saying is that IMO there should be no surprises if the total number of shots directed at the net IS greater this year than last.

Of course I don't have these numbers to support my claim so I could just be blowing smoke :)

1/02/2009 8:12 pm  
Blogger HBomb said...

Nope, just a theory at this point. Seems like the nights the Oilers have against a good PK, they`re still getting set up, but whatever the other team is doing (be it pressuring the point heavily or whatever), it forces shot attempts that are either easily blocked or miss the net.

But you`re right - even one more clear per PK, that`s a big deal.

1/02/2009 8:34 pm  

Post a Comment

<< Home