Canuck Penalty Kill: Reasons for Improvement

[JamesB]

We all know that the Canuck PK struggled during the losing skid in November and improved later. But I am not sure we all know how much. I just checked the numbers.

 

Over the period Nov. 6 through December 4 (when the Canucks lost 13 of 15 games), the PK was giving up 13.41 goals against per 60 minutes (GA60)  -- dead last in the NHL.

 

Over the period Dec. 5 to the present the Canucks have given up 4.25 GA60 -- good for second best in the NHL. Even in the last 3 games when the Canucks struggled overall, the PK was very good, giving up only 1 PK goal over those three games.

 

The question is why? What can cause such a big turnaround. One obvious answer is injuries. Another obvious answer is that the most important player on the PK (the goalie -- mostly Marky in this case) got better. Both those things are important. And luck is involved. The samples are small enough that a few bounces here and there can make a big difference in the numbers.

 

However, there is another factor that I think is interesting and overlooked. That is player deployment. There are big differences in player PK performance. I think it is easier to assess PK performance than just about anything else because the objective is very clear -- just stop the other team from scoring. And the usual problem of dealing with quality of opposition (QO) and quality of teammates (QT) is much reduced as all the opposing forwards are good and all the PKers are good defensively (in theory) so there is much less variance in QO and QT for PK assessment than for 5-on-5 play.

 

Here are the GA60 numbers for all Canucks who have played more than 30 PK minutes this season.

 

 

There are some remarkable differences here. Stecher (who gets relatively little PK time -- more on that later), Sutter and Erikson have GA60s of about 2. MDZ and Roussel are over 12 -- about 6 times worse!

 

One problem with GA60 is that, over half a season, it is a fairly small sample. I think it is useful to also look at  scoring chances against per 60 (SCA60), taken from Natural StatTrick. Here are the results.

 

 

 

This makes sense. There is less variation overall. As the sample size increases for GA60, the graph would probably come to look more like  for SCA60, although it would probably always retain at least a bit more variance, as GA60 captures some things that SCA60 does not. For assessing what is going on I think that both these numbers (GA60 and SCA60) should be considered. 

 

Looking at SCA60 does not change the story all that much. Eriksson, Stecher, and Sutter all look good -- as does Ben Hutton. At the bottom end things change a bit, with Schaller looking very bad.

 

There are some mitigating factors. Even though QO has less variance for the PK than for 5-on-5 there is still some. In particular, Tanev and Edler are up against the first unit PP most of the time, so their numbers are going to look a bit worse than they "should". Also, Stecher almost never got D-zone starts. Most of his starts were "on-the-fly" or in one of the other zones, so that helps his numbers.

 

Considering those factors makes Sutter look even better, however, as he gets more than his share of D-zone PK starts and more than his share of time against the first unit PP.

 

Okay, so what about deployment and injuries.

 

It is clear from the data that getting Sutter and Edler back from injury helped the PK a lot. Beagle helped but not as much.

 

During the November losing streak, the forwards with the most PK ice-time were Motte and Granlund. The D's with the most were Tanev and MDZ.  Replacing MDZ's time with Edler was huge and replacing Granlund, Motte, and Roussel with Sutter and Beagle also helped a lot.

 

But deployment is also interesting. One thing that Green has done recently is to increase the PK time of Eriksson. Eriksson did not get much PK time early but now he is one of three forwards (along with Sutter and Beagle) carrying the load. In fact, the PK has become more specialized in that those 3 guys get most of the (forwards') time.

 

I can remember some of the Canuck blogs and journalists making the point that Eriksson was under-utilized on the PK before Green made the change. I can also understand Green wanting to get a large enough sample size before making changes in deployment.

 

The only remaining anomaly in deployment is Stecher, whose numbers appear very good and who gets relatively little PK time. On the numbers, it appears he should get some of Guddy's time. In "theory" Guddy should be good given his reach and strength in front of the net. And his numbers are ok. But I wonder if we might see Stecher get more time down the road, just as he eventually got moved to the second D-pairing at 5-on-5.

 

I am not sure if anyone besides me is interested in all this but, if so, I would be interested to read any comments.

 

 

 

 

 

 

 

 

 

[BlastPast]

Interesting stuff OP.  Shows how potentially underrated aspects of some players game can be , and subsequently their overall contribution.

[AK_19]

So based on this, our first PK unit should be:

 

Sutter Eriksson

Hutton Stecher

 

And our second:

Beagle Granlund

Edler Tanev

 

Makes sense to me.

[Duodenum]

43 minutes ago, AK_19 said:

So based on this, our first PK unit should be:

 

Sutter Eriksson

Hutton Stecher

 

And our second:

Beagle Granlund

Edler Tanev

 

Makes sense to me.

Not quite. Edler-Tanev get the first unit and start the pk in the defensive zone most times so their numbers will look worse than they actually are. 

[Duodenum]

Good post OP, I had been wanting Eriksson and Stecher to be getting more pk time as both have been great in that regard. Eriksson's time on ice has gone up, which is nice to see. Would like to see how Stecher fares with increased ice time shorthanded. 

 

Another factor for the better penalty killing is, of course, goaltending as well. Markstrom has been a lot better since that losing streak. 

[Glug Datt]

nicely laid out thought process.. a lot can be inferred with a handful of stats. 

 

I wonder how much is personnel, and how much is coaching..

 

as they say, special teams win games

[Noble 6]

Nice analysis with evidence to back it up, but you forgot one thing. Advanced stats don't matter and therefore you're argument is invalid. 

 

[JamesB]

3 hours ago, AK_19 said:

So based on this, our first PK unit should be:

 

Sutter Eriksson

Hutton Stecher

 

And our second:

Beagle Granlund

Edler Tanev

 

Makes sense to me.

 

2 hours ago, Duodenum said:

Not quite. Edler-Tanev get the first unit and start the pk in the defensive zone most times so their numbers will look worse than they actually are. 

Yes, I would probably go with Edler-Tanev for the first unit PK. Considering that they are consistently out against the 1st unit PP their numbers are actually pretty good -- but more so for Edler than Tanev.  Edler is having a very good year this year.

 

I would like to get a look at an Edler-Stecher pairing. I would not expect changes right now though on the grounds that, if it ain't broke don't fix it.

 

I think AK's forward pairings are right.

 

One thing I did not mention in the OP is that of course one of forwards has to be a center to take face-offs. Both Sutter and Beagle are good on the dot, which is another reason their PK ice-time is high. Granny has the advantage that he can take face-offs if the center gets waved out.

 

There is also a case for putting Beagle and Eriksson together as they have developed good defensive chemistry playing on the "4th line" together.

 

1 hour ago, Glug Datt said:

nicely laid out thought process.. a lot can be inferred with a handful of stats. 

 

I wonder how much is personnel, and how much is coaching..

 

as they say, special teams win games

This is another point I forgot to mention in the OP. The improvement in the PK has been a big reason for turnaround in overall performance. We sometimes forget how important the PK is.

 

I have been of critical of Sutter over the years and never liked the deal that brought him to Vancouver. But I have to give him credit for his work on the PK. Similarly with Eriksson, we all agree he has a terrible contract but if he were being paid 2 or 3 million a year as a defensive specialist who can chip in with secondary scoring, we would be happy with him.

 

Of course none of that matters if a team isn't getting primary scoring. But with EP on board (and Boeser and Horvat playing well) we are getting pretty good primary scoring and the forward group is looking much better than it did last year.

[PrinceGeorgeGoon]

Ian Clark.  

[spook007]

53 minutes ago, JamesB said:

 

Yes, I would probably go with Edler-Tanev for the first unit PK. Considering that they are consistently out against the 1st unit PP their numbers are actually pretty good -- but more so for Edler than Tanev.  Edler is having a very good year this year.

 

I would like to get a look at an Edler-Stecher pairing. I would not expect changes right now though on the grounds that, if it ain't broke don't fix it.

 

I think AK's forward pairings are right.

 

One thing I did not mention in the OP is that of course one of forwards has to be a center to take face-offs. Both Sutter and Beagle are good on the dot, which is another reason their PK ice-time is high. Granny has the advantage that he can take face-offs if the center gets waved out.

 

There is also a case for putting Beagle and Eriksson together as they have developed good defensive chemistry playing on the "4th line" together.

 

This is another point I forgot to mention in the OP. The improvement in the PK has been a big reason for turnaround in overall performance. We sometimes forget how important the PK is.

 

I have been of critical of Sutter over the years and never liked the deal that brought him to Vancouver. But I have to give him credit for his work on the PK. Similarly with Eriksson, we all agree he has a terrible contract but if he were being paid 2 or 3 million a year as a defensive specialist who can chip in with secondary scoring, we would be happy with him.

 

Of course none of that matters if a team isn't getting primary scoring. But with EP on board (and Boeser and Horvat playing well) we are getting pretty good primary scoring and the forward group is looking much better than it did last year.

Nice thread James. Interesting stuff...

 

Just a couple of comments... Don't know if it is quite as straight forward as this.

The Edler - Tanev pairing, when they have both been in the line up, always faces the toughest opposition, and as such their numbers are probably 'stronger' than it shows...

Not sure if Stecher would be able to deal with big first line forwards as well as Tanev, but only by trying him on the unit, will we know.

This however is unlikely to be tested in the near future, with the PK doing so well... Maybe if Tanev goes down again at some point, it would be worth a shot.

 

Glad you highlight Eriksson, as I also think his work rate on the PK has gone fairly unnoticed (and if not unnoticed then unappreciated) on these boards. I don't know, how much of a specialist PK'er he was at Boston prior to coming here, but it's almost as if he has had to reinvent himself here. With his price tag folks expected a stronger scoring spree, and rightly so, but if he haven't put the puck in net quite as often as hoped for, he certainly has been preventing it.

 

And yes Markys exceptional play has been a huge help to all the statistics.

[therodigy]

7 hours ago, spook007 said:

Nice thread James. Interesting stuff...

 

Just a couple of comments... Don't know if it is quite as straight forward as this.

The Edler - Tanev pairing, when they have both been in the line up, always faces the toughest opposition, and as such their numbers are probably 'stronger' than it shows...

Not sure if Stecher would be able to deal with big first line forwards as well as Tanev, but only by trying him on the unit, will we know.

This however is unlikely to be tested in the near future, with the PK doing so well... Maybe if Tanev goes down again at some point, it would be worth a shot.

 

Glad you highlight Eriksson, as I also think his work rate on the PK has gone fairly unnoticed (and if not unnoticed then unappreciated) on these boards. I don't know, how much of a specialist PK'er he was at Boston prior to coming here, but it's almost as if he has had to reinvent himself here. With his price tag folks expected a stronger scoring spree, and rightly so, but if he haven't put the puck in net quite as often as hoped for, he certainly has been preventing it.

 

And yes Markys exceptional play has been a huge help to all the statistics.

Eriksson has been very underated for us, and unfortunately, most of the criticism stems from his contract. Fact his, he's been huge for us on the PK, is our most defensively responsible winger, and he's on pace to beat his season high since signing with us (the last point doesn't say much as the most points he's accumulated in Vancouver is 24pts in 65GP, though it's worth mentioning that he's consistently produced at a 3rd line level since signing in in 2016).

[cyoung]

9 hours ago, PrinceGeorgeGoon said:

Ian Clark.  

He's been our best pker

[Fanuck]

Canucks are the second most penalized team in the entire league  (only about 4 seconds/gm away from being the most penalized team per gm in the NHL).  This combined with being in the lower half of the league in gf/gp and top half of ga/gp is a BAD combo.  

 

Talking about penalty killing deployment is great,  and I really appreciate James' analysis - but the fact remains this team shouldn't be taking this many bad penalties per/gm.  We can't kill them effectively well and we don't score enough to make up for taking bad penalties. 

 

I've liked the addition of Roussel overall but him leading the league in pim/gm AND outright pim is hurting us imo.  He's not like a Kane or Rat-boy who also take a LOT of bad penalties but make up the difference by scoring regularly. 

 

Less bad penalties overall > better PK, for this version of the Canucks anyway. 

[Rob_Zepp]

22 hours ago, JamesB said:

We all know that the Canuck PK struggled during the losing skid in November and improved later. But I am not sure we all know how much. I just checked the numbers.

 

Over the period Nov. 6 through December 4 (when the Canucks lost 13 of 15 games), the PK was giving up 13.41 goals against per 60 minutes (GA60)  -- dead last in the NHL.

 

Over the period Dec. 5 to the present the Canucks have given up 4.25 GA60 -- good for second best in the NHL. Even in the last 3 games when the Canucks struggled overall, the PK was very good, giving up only 1 PK goal over those three games.

 

The question is why? What can cause such a big turnaround. One obvious answer is injuries. Another obvious answer is that the most important player on the PK (the goalie -- mostly Marky in this case) got better. Both those things are important. And luck is involved. The samples are small enough that a few bounces here and there can make a big difference in the numbers.

 

However, there is another factor that I think is interesting and overlooked. That is player deployment. There are big differences in player PK performance. I think it is easier to assess PK performance than just about anything else because the objective is very clear -- just stop the other team from scoring. And the usual problem of dealing with quality of opposition (QO) and quality of teammates (QT) is much reduced as all the opposing forwards are good and all the PKers are good defensively (in theory) so there is much less variance in QO and QT for PK assessment than for 5-on-5 play.

 

Here are the GA60 numbers for all Canucks who have played more than 30 PK minutes this season.

 

 

There are some remarkable differences here. Stecher (who gets relatively little PK time -- more on that later), Sutter and Erikson have GA60s of about 2. MDZ and Roussel are over 12 -- about 6 times worse!

 

One problem with GA60 is that, over half a season, it is a fairly small sample. I think it is useful to also look at  scoring chances against per 60 (SCA60), taken from Natural StatTrick. Here are the results.

 

 

 

This makes sense. There is less variation overall. As the sample size increases for GA60, the graph would probably come to look more like  for SCA60, although it would probably always retain at least a bit more variance, as GA60 captures some things that SCA60 does not. For assessing what is going on I think that both these numbers (GA60 and SCA60) should be considered. 

 

Looking at SCA60 does not change the story all that much. Eriksson, Stecher, and Sutter all look good -- as does Ben Hutton. At the bottom end things change a bit, with Schaller looking very bad.

 

There are some mitigating factors. Even though QO has less variance for the PK than for 5-on-5 there is still some. In particular, Tanev and Edler are up against the first unit PP most of the time, so their numbers are going to look a bit worse than they "should". Also, Stecher almost never got D-zone starts. Most of his starts were "on-the-fly" or in one of the other zones, so that helps his numbers.

 

Considering those factors makes Sutter look even better, however, as he gets more than his share of D-zone PK starts and more than his share of time against the first unit PP.

 

Okay, so what about deployment and injuries.

 

It is clear from the data that getting Sutter and Edler back from injury helped the PK a lot. Beagle helped but not as much.

 

During the November losing streak, the forwards with the most PK ice-time were Motte and Granlund. The D's with the most were Tanev and MDZ.  Replacing MDZ's time with Edler was huge and replacing Granlund, Motte, and Roussel with Sutter and Beagle also helped a lot.

 

But deployment is also interesting. One thing that Green has done recently is to increase the PK time of Eriksson. Eriksson did not get much PK time early but now he is one of three forwards (along with Sutter and Beagle) carrying the load. In fact, the PK has become more specialized in that those 3 guys get most of the (forwards') time.

 

I can remember some of the Canuck blogs and journalists making the point that Eriksson was under-utilized on the PK before Green made the change. I can also understand Green wanting to get a large enough sample size before making changes in deployment.

 

The only remaining anomaly in deployment is Stecher, whose numbers appear very good and who gets relatively little PK time. On the numbers, it appears he should get some of Guddy's time. In "theory" Guddy should be good given his reach and strength in front of the net. And his numbers are ok. But I wonder if we might see Stecher get more time down the road, just as he eventually got moved to the second D-pairing at 5-on-5.

 

I am not sure if anyone besides me is interested in all this but, if so, I would be interested to read any comments.

 

 

 

 

 

 

 

 

 

Wow.   Great post.    Lemme guess, slow day at work?

[oldnews]

Sutter

Beagle

Edler

Tanev.

 

direct correlation to their top 4 penalty killers health. 

[Kragar]

I love watching Stecher check. Reminds me I'd my little terrier cross we used to have.  Maybe it's because he is closer to the puck (short joke, sorry), but it is amazing how good he is at fishing the puck out of the corners, regardless of the size of the competition.  Feisty, just keeps trying, and often gets the job done.

[Toyotasfan]

I’m ok with our PK, got any suggestions for the PP ?

[Duodenum]

On 1/28/2019 at 5:34 PM, Kragar said:

I love watching Stecher check. Reminds me I'd my little terrier cross we used to have.  Maybe it's because he is closer to the puck (short joke, sorry), but it is amazing how good he is at fishing the puck out of the corners, regardless of the size of the competition.  Feisty, just keeps trying, and often gets the job done.

Him and Tanev are two of the best at taking the puck away from the opposition forward and making the smart play or smart pass to get the puck out of the zone and going the other way. Edler is really good at using his size to do the same. Feel like Hutton could learn from Edler in that regard.