Breaking Down: The Avalanche Penalty Kill

The Avs finished the season with the NHL’s 25th ranked penalty kill and like the offensive special teams component, the power play, this was a bit misleading. In the middle third of the season, better known as the collapse, they were bar none the worst PK in the league. Other than that they were fairly solid right around 10th overall. Oddly enough, about the time that the power play was embarking on a 9-game drought in late January, the penalty kill went on a similar bender of ineffectiveness giving up 9 goals in 7 games. The straw that broke the camel’s back was a 3 goal nightmare against the Maple Leafs on February 12th. After that, changes were made and the unit got back on track. For the rest of the season and through the playoffs they were good to very good so hats off to the staff for fixing half of the special teams.

Penalty kill success is highly system related in the long term, in the short term it’s about avoiding mistakes Unlike the power play there’s no way to make up for poor strategy with talent. The Avs had a top 5 PK in 2017-18, regressed a little this year and regressed a lot in the middle third of the season. I’d like to know why and I bet you would too.

I’m going to use some similar type metrics as with last week’s power play article, complete with RainbowVomitCharts© and some system demo pictures, which should tell us what we need to know.

Efficiency

The Avs PK gave up 58 goals in 272 chances for a 78.7% efficiency rating, good for 25th in the NHL and broken down as such:

First 3rd (27 games): 19 goals against in 99 chances (80.8%, 11th in NHL)
Second 3rd (27 games): 26 goals against in 90 chances (71.1%, 31st in NHL)
Third 3rd (28 games): 13 goals against in 83 chances (84.3%, 9th in NHL)

– 13 times they gave up more than one PK goal in a game. Twice they gave up 3 in a game (1/8 vs WIN and 2/12 vs TOR)

– 38 games they blanked the opponent, including the final 6 games of the season which was their best stretch

– There were 2 stretches where they gave up goals in 5 straight games, from Oct 27th – Nov 9th (7 goals) and Jan 23rd – Feb 9th (6 goals)

– The Avs blanked 5 teams – Flyers, Sabres, Hurricanes, Senators & Knights

– 9 shorties ended up tied with San Jose for 9th in the league. Soderberg and Compher had 3 apiece and Nieto, Dries and Kamenev each had one. Nieto ended up with the only 2 short-handed assists.

Even taking out the hideous middle third of the season they didn’t live up to last year’s numbers. In fact, they allowed only 45 PK goals against all last year, which was reached in just the first 54 games of this season. The promising part was that the final third was their best and the last 6 games they gave up nothing. In the playoffs they fared much better than they did against Nashville in 2018 so there was growth.

Shot Metrics

Unlike the power play the Avs PK shot metrics lined up well with their overall efficiency. They were 24th in unblocked shot attempts against per hour, 23rd in expected goals against per hour and 24th in goals against per hour. As much as everyone points to the goalies being a major drag during the mid-season collapse they ended up 12th in the league at .870, so that’s a positive outlier if anything.

Full Season

I’m only looking at the top 9 penalty killers by TOI, who did most of the work this season. Mark Barberio played a lot when he was in the lineup but that wasn’t very often. Sheldon Dries played a fair amount but not enough to make a difference in this discussion. Ryan Graves and Vlad Kamenev might have made a difference if they had more opportunity but they didn’t.

Let’s start with RainbowVomitChart© #1:

Legend: Dark blue is good, dark red is bad, lighter is less so in each case.

What we’re examining here are unblocked shot attempts per hour (FA/60), expected goals per hour (xGA/60), goals per hour (GA/60) then a couple of derivative stats. First of those is FA/xGA, which is how many unblocked shots the opponent attempted to equal one expected goal. The second is goals per expected goal or fancy shooting percentage. I also used Micah Blake McCurdy’s threat metric as a with-or-without-you, which you can find here for each player.

Matt Nieto and Patrik Nemeth were the best and most consistent penalty killers last year. Carl Soderberg had a really rough first 27 games but after that was on the same level as the other two. Gabriel Bourque wasn’t all that great and definitely wasn’t consistent but he did have a positive effect on threat level. The bad news here is that all but Nieto are gone.

Matt Calvert had a fairly consistent negative effect on the PK and his worst shot metrics came at the end of the year when his PK TOI went up, which isn’t very promising.

Nikita Zadorov had really good shot metrics but horrifying goal metrics. This is something we see with Z a lot, a lot of overall good play then a few mistakes shatter everything. Avs fans better hope he cleans that up with Ian Cole out in the first few months of the season or this could get ugly.

Cole & EJ were very disappointing as a top PK pair. Cole was brought in to be a PK specialist and for the most part he was ineffective. He wasn’t very effective in Columbus last season either so at least in respect to killing penalties this is another case of misidentifying talent. Their numbers are similar, poor, and they often played together so we’re left to wonder whether one or both guys are the issue. If Cole was the one dragging everyone down then problem solved at least until December.

Last we come to the newly signed JT Compher. He had the highest shot volume and quality against rates of all the regulars but was saved by the lowest goals per expected goal percentage. I don’t consider that luck but it is somewhat suspicious. Here is his WOWY threat level viz from Micah:

The big purple blob to the right of Grubi/Varly means trouble. He’s far from the lone culprit but that’s exactly what a boatload of royal road passes looks like. On defense “+” is bad, “-” is good and the Avs were +25% with and -15% without him on the ice. This was only his 2nd NHL season so we can give a little benefit of the doubt but this can’t happen next season. He will be on the 2nd PK pair and possibly the top one so he needs to become a stopper.

First 3rd

The penalty kill was fairly effective overall to begin the season, thanks to solid goaltending from Semyon Varlamov and shot suppression from the second unit. Here are the numbers:

Ian Cole played a lot but wasn’t great at suppressing volume and especially quality, which is a theme that would repeat itself through most of the season. The top forward pair of Soderberg & Nieto had the same issue but they managed to correct it as the year went on. Big Z didn’t play much but goals happened a lot when he did despite good quality suppression. Overall this was Compher’s most promising stretch of the season.

The Avs ended this stretch 11th in the league at 80.8%, a fair job but a downgrade from the year before. UFA signings Cole & Calvert were learning the system at this point. Cole struggled yet played more than any other D while Calvert was eased into PK work and did ok other than lower than average goals per expected goal.

Second 3rd

Most aspects of the Avalanche took a nosedive starting around December 2nd and the PK was no exception. Thanks to some injuries the personnel started to shuffle a bit and either by choice or necessity the coaching staff began playing a more conservative style that was very harmful. Avert your eyes, here are this segment’s numbers:

Cole and EJ’s TOI as a percentage went down significantly along with a decrease for Zadorov. Nemeth’s went up a small amount. This was when we got a bit of a carousel with Alt, Lindholm, Barberio and eventually Ryan Graves taking some PK shifts. Soderberg and Compher saw large decreases in TOI, Nieto and Bourque did as well to a lesser degree.

In the chart above take a look at JT Compher vs Gabe Bourque, who both had hideous goal against rates. Bourque was the Avs best forward at suppressing quality, Compher the worst. Bourque’s on-ice goals per expected goal was twice Compher’s. This just oozes team inconsistency and poor strategy.

Statistically, the interesting part of this period was that unblocked shot attempts and expected goal rates went down 20% yet goals per expected goal skyrocketed by nearly 90%. Most people decided that both goalies all of a sudden started playing terribly at the same time and in the exact same way, which isn’t all that logical. I’ve taken a shot at this from several different angles and while the goalies weren’t without blame, the play in front of them was very questionable. Using just the numbers there isn’t a lot to latch on to so for me it comes down to two things, systems and mistakes. Coach Bednar as much as said so when they finally tweaked the system in mid-February and the fact the PK was instantly much more effective backs that up.

Colorado finished this part of the season dead last in the league with a 71.1% efficiency rate which was partly to blame for dropping below .500 almost losing a playoff spot.

Third 3rd

After getting destroyed by Toronto’s power play on February 12th, the staff had seen enough and finally did something about the poor PK strategy. The shot suppression and quality suppression looked good on paper but it was killing the team on the scoreboard. Here’s what it looked like down the stretch:

EJ and Nemeth became the #1 PK pair and Matt Calvert’s share of PK TOI skyrocketed despite being the least effective player on paper. Just as suddenly as both goalies had forgotten how to play, they remembered and fancy shooting percentage was 60% better. JT Compher’s TOI share went up massively as well and Carl/Nieto’s numbers were incredibly strong. The staff gave Zadorov more time even though his goal against rate and goals per expected goal were alarming.

Despite the worst suppression rates of the season, the PK gave up far fewer goals than any other stretch and that’s the point. Their efficiency was also at it’s peak, culminating with no goals in 6 games leading into a solid playoff run.

Systems

Like with the power play, whole books have been written on PK strategy so instead of explaining here I’ll show a little progression through the Avs season.

This was an early season game against the Jets. It’s a triangle plus one with the one being Kamenev putting pressure on the puck carrier at the top of the circle. This is a little too aggressive. We can see Cole at LD in no man’s land, too far from the goal to help out with the net-front guy and too far from the puck carrier to do anything but screen the goalie, which he is an ace at. This ends up pulling EJ further over the royal road making it more difficult to react to a seam pass. Bourque is also in no man’s land and heading the wrong way to cut off a seam pass. This style worked fine but made them highly susceptible to cross ice passes so they decided to chill it a bit.

This was the abomination vs the Leafs on 2/12 where they let in 3 goals on the PK in 3 minutes. This is your basic collapsed box and sacrifices puck pressure for the ability to block passes and/or shots. It plays into exactly how the Leafs PP likes to attack because they are an aggressive probing team that never shoots from the point. The issue here is that when the PK isn’t blocking shots it’s not doing anything productive so they get trapped in the zone and worn out quickly. A probing team can break up the box by passing the puck inside, throwing it back out quickly then taking a shot while chaos ensues. (Note our guy Kadri at the top of the slot, he would eventually break low and score here. Bad news for our PK but great news for next year’s PP.)

This is a game in early March vs our good friends from San Jose. Even though it’s a different angle we can see the differences between this strategy and the early season one with the puck in the same place. The LD is in front but NOT screening the goalie. The RD now has the time and space to react to a seam pass. The slot defender is closer to the net and better able to cut off angles on a pass. The puck pressurer is fairly similar.

Simply put, the staff had a system that worked ok but got burned from being too aggressive so they went too conservative and it absolutely buried them. Live and learn, and they did, going back to the original philosophy with a few tweaks. Beyond getting burned on seam passes the staff might have been worried about Grubi’s side-to-side motion and wanted to suppress shots in general with the collapsed deal. They accomplished that but it ended up making it much easier for the opponent to score when they did get shots through. Looking at the two goalies’ PK stats, the collapsed system hurt Varly more than it hurt Grubi because Varly’s fine with a million shots against as long as he can see them. It’s an interesting insight into the thought process if you consider part of the long-term goal was to make Varly expendable.

Bottom line, they had a system that worked and needed to figure out how to tailor it to the new goalie. In a clumsy, roundabout and self-destructive way that’s what they ended up with by the end of the season.

Next Season

At the beginning of 2019-20 the Avs will be without Soderberg, Bourque, Nemeth and Varlamov from trades and free agency and without Cole because of injury for a while.

Forwards

Nieto and Calvert become the top PK forwards and Compher will be slotted in close. Bellemare was signed for this role even though that’s a silly thing to waste a roster spot on. So there’s your top 4 and the others can be filled out from the rest of the lineup. Kadri, Donskoi and Burakovsky didn’t kill penalties with their former teams but the Avs love square-pegging guys. Vlad Kamanev has the chops to be a decent PKer if they end up playing him. Tyson Jost could create some value for himself by playing there but I’m not sure his skating ability makes him a good fit. This is a pretty thin spot for the roster and perhaps leaves an in for guys like Kaut and especially Bowers if the need arises.

Defense

EJ is pretty much it as far as known quantities. I’d love to see Zadorov become proficient in this area and I’m sure the coaching staff would too. Connauton played PK last year in Arizona and was ok but it was the first time he had done so in the NHL with any regularity. Graves was ok last year in limited duty and Barberio used to be good back when he played regularly. If Calle Rosen can demonstrate some PK ability in pre-season it could lock up the 6th spot for him.

Goalie

Since the current system seems to have been catered to him, I’d say Grubauer should be fine. The Eagles played something similar to the Avs last year, although much better relative to their league, so Francouz ought to be fine as well.

+      +      +

Under Jared Bednar the Avs have been a defensively oriented team and issues in that area don’t concern me as much as their scoring or offensive zone problems. Despite being stubborn they tend to figure this side of the game out in due time. It’s the base of everything else they do on the ice so they put a lot of time and effort into preventing goals first and the rest of the game follows.

The lack of PK proficiency and experience on the Avs projected roster is concerning. Two years ago they had a top PK thanks in large part to Blake Comeau, Carl Soderberg and Patrik Nemeth, who are all gone now. I’m not advocating that they should have been retained but what was once a strength is at best an unknown. On the bright side, the system they have used is solid (other than the black hole they fell into during the middle third of last year) and killing penalties is one of the easier things for coaches to teach and players to learn.

Taking out that mid-season black hole the Avs were a top 10 PK which is the minimum we want to see given that they are a high-volume penalty team. That volume is an important aspect of how their special teams affect success as a whole. Team speed will create a lot of power plays, officials tend to even things out so they will also get soft calls to kill off. It’s the nature of the game and makes it important that they have special teams as a net positive.

The likely scenario this coming season is that the penalty kill goes through some struggles and inconsistency while sorting out roles early on then settles into a solid unit by Thanksgiving.