4
Photo Credit: Geoff Burke / USA TODAY Sports

Evaluating the Leafs’ PK Performance in 2016/17

As you might’ve heard, the Leafs are going up against a juggernaut Capitals team in the first round. Not only is Washington excellent at driving 5v5 goal differentials, they also have one of the most dominant power plays in the NHL. Since acquiring Kevin Shattenkirk, the Capitals have generated the most 5v4 shots per 60 minutes in the NHL, and the 4th most goals. This isn’t anything new. Check out how they’ve ranked in goals and unblocked shots (“Fenwick For” or FF/60) in the previous 4 seasons:

Needless to say, the Capitals’ power play has been phenomenal since the 2012 lockout, which isn’t hard to believe when you consider that they have arguably the best passer on the planet and this guy. Their top unit has been extremely successful for many reasons. Throw Kevin Shattenkirk into the mix, along with some funky shifting around, and you’ve got a PP that can really do some damage. If the Leafs want to pull off the upset in round 1, they’ll need to slow down Washington’s elite power play. With that in mind, let’s take a look at how well the Leafs rank in the major penalty killing metrics this year:

Leafs Rank This Season
Goals Against per 60 9th
Shot Attempts Against per 60 27th
“Expected” Goals Against per 60 24th
Save Percentage above “Expected” 3rd

 

(For anyone confused by Expected Goals, it’s the number of goals a team is ‘expected’ to allow based on shot locations)

Although the Leafs have done an excellent job at preventing goals while on the penalty kill, it’s mostly because of the phenomenal 4v5 goaltending they’ve been getting from Frederik Andersen. When it comes to the shots and chances that they’re allowing, unfortunately, the Leafs have been #ActuallyBad at suppressing scoring opportunities on the penalty kill. This isn’t to say that their PK is doomed; especially considering Frederik Andersen has been one of the best 4v5 goalies since coming into the league. Objectively looking at it though, the Leafs’ 4v5 shot metrics have clearly not been ideal this season.

When it comes to evaluating the individual players on the Leafs PK, I think it’s really important to take zone starts into account, which I wrote about last week. If you don’t feel like reading 2,500 words of nerdy nonsense (and no one would blame you), I can save you some time with a quick recap here: zone starts really matter on the PK. They determine about 50% of a Forward’s shot suppression and roughly 25% for defencemen.

To make a long story short, players with tons of DZ starts on the PK (ie. Brian Boyle) have a much more difficult time suppressing shots & chances on the penalty kill. This is because they spend more time in the defensive zone against power plays that are set up in formation. On the other hand, someone like Nikita Soshnikov starts almost all of his PK shifts on the fly. This puts him in a much better position to succeed since he spends a so much of his 4v5 time in a neutral zone trap. Here’s a good way of visualizing the difference between starting a shift in the DZ vs on the fly:

DZ Start

‘On The Fly’ Start

Using some fancy math, I’ve been able to adjust for these factors so that we can put players on a level playing field when we look at their 4v5 results. From what I’ve found, 4v5 shot suppression (Corsi Against or CA/60) is typically a good barometer for how well you’re defending the blue line (denying PP zone entries), while Expected Goals Against (xGA/60) is great for determining the quality of the chances that you’re giving up. Keeping that in mind, let’s take a look at how well Toronto PKers have been suppressing shots and chances this season. We’ll start with the Defence:

Before we dive into these numbers, I just want to remind everyone that this isn’t a perfect metric. To show you what I mean, let’s talk about Jake Gardiner, who looks amazing in these charts. Although he’s put up these excellent numbers in his limited 4v5 time this season, we have to try to think about the context of his shifts. He’s only played 43 minutes of 4v5 play, which is roughly 10% of Toronto’s total 4v5 time this season. I’ve noticed that when he does get the odd PK shift, it tends to be towards the end of the penalty (ie. the last 20 or 30 seconds). This sheltered usage is logically going to inflate his shot suppression numbers since teams often won’t play one of their typical power play units at the tail end of a penalty. For example, you’ll probably notice that the Leafs put two defencemen on towards the end of their power plays instead of a 4-forward group. This is because they’re anticipating the game returning to 5v5 play and adjusting their lineup accordingly.

This isn’t to say that I think Gardiner is a poor penalty killer, far from it actually. We just need to understand that his godly numbers in these charts are probably a bit inflated given his usage (the same can be said about Connor Carrick). Even if we were to make a quality of competition adjustment though, I’d imagine that Gardiner would still be the Leafs’ best defenceman on the PK, given just how dominant his numbers are. Considering that fact that he’s been an elite 5v5 shot suppressor throughout his career, I think it’s fair to hypothesize that he would also excel at it during 4v5 play. We’ve seen evidence of him doing so in small samples over the past few seasons, so personally, I think it would be preferable to give him more PK time moving forward.

Other names that really interest me are Marincin and Polak. Although they’ve struggled with 4v5 shot suppression, they’re doing a good job at suppressing dangerous chances (since their xGA/60 is much better than their CA/60). Although Hunwick gets tons of 1st unit PK time, the data shows that Marincin is significantly better at suppressing chances on the PK (xGA/60). This was also true last year, where Marincin was Toronto’s best shot suppressing defenceman – a team which ranked 2nd in the league in 4v5 shot suppression. We all know that Marincin’s basically the opposite of Karlsson when he has the puck, but his solid defensive numbers make sense when you think of him using his 50-foot stick to deny zone entries and take away passing lanes.

For my money though, the biggest eye-opener on these charts is Rielly. His 5v5 shot suppression has been long-noted as a significant weakness in his game, and it appears that these struggles persist in 4v5 play. We know that Rielly is an exceptional skater and very effective offensive defenceman. When he has the puck he’s one of the best defencemen in hockey – his shot generation, passing, and zone exit numbers are all elite. His biggest issue has always been when he doesn’t have the puck. I understand that Babcock is probably giving him PK time to help him improve in this area, but unfortunately, he’s struggled mightily this season. Moving forward I would suggest having Gardiner take more of these minutes since it would improve the team’s 4v5 numbers and increase their chances of winning.

That was fun, let’s check out some of the forwards now:

We’ll get to Marner in a minute (he was my inspiration for writing this piece), but let’s start with Soshnikov and Brown. They rarely start their PK shifts in the defensive zone, but after we adjust for their zone starts, they come out as the two best suppressing forwards among Toronto’s main penalty killers (players who have over 50 minutes of 4v5 time this season). It’s worth pointing out that Matt Martin also looks decent on these charts in his limited PK time, however, we have to take these numbers with a grain of salt given his extremely sheltered usage (similar to with Gardiner & Carrick). I think the take-home point from this data is that the majority of the Leafs’ full-time PKers are below league average in suppressing shots and chances, which is unfortunate to say the least.

When it comes to the 1st unit PK, Komarov’s numbers are pretty underwhelming, but they’re nothing compared to Hyman’s. His results shocked me, since my eye test was convinced that Hyman was an animal on the PK. He’s potted a few shorthanded goals this year, which goes a long way towards helping his 4v5 goal differential, but it’s very disappointing to find out that the player who many fans thought was an elite penalty killer actually has very poor 4v5 numbers.

Keeping things positive, let’s talk about Ben Smith. Just kidding, we’re going to talk about all of the Leafs who have been given the faceoff role on the PK, which we know is very important to Babcock. Smith, Gauthier, and Boyle were given a lot of 4v5 DZ starts, but even after you adjust for their zone starts, these players have all been horrible at suppressing shots & chances this season (with the exception of Gauthier’s xGA/60, which may be inflated by his small sample of 4v5 minutes). After seeing these numbers, I told myself “well the zone start adjustments probably just don’t do a great job with players receiving extreme DZ usage.” Unfortunately, I was wrong in this assumption:

That, my friends, is Randy Carlyle making full use of his faceoff specialist. I can’t believe I’m actually giving Carlyle credit for out-coaching Babcock in any facet of the game (even Babcock’s gum-chewing is elite), but I have to call a spade a spade – Carlyle has done an excellent job at getting the most out of Anaheim’s talent on the PK. Despite being buried with DZ starts, Vermette has thrived in his 4v5 faceoff role. It makes sense that he would excel in this role since he’s been godly in the faceoff dot this season (62.3%). More importantly though, he’s not a liability on the penalty kill when he loses the draw. He’s able to use his foot speed to take away space, unlike players like Boyle or Gauthier who can’t take away the shooting & passing lanes as quickly.

Now I love Boyle during 5v5 play, but unfortunately, his penalty killing ability is mostly narrative driven, not evidence driven. He’s consistently been a below average shot suppressor on the PK regardless of his usage. He’s struggled on the PK in both a faceoff-role (heavy DZ starts) and a Soshnikov/Brown role where starts most of his 4v5 shifts on the fly. So now you’re probably wondering “alright Ian you’re making the argument that Boyle shouldn’t be used on the PK, so who the hell should Babcock have take faceoffs then?” I have an answer that’s going to sound crazy:

Tyler Bozak is pretty much the farthest thing from a defensive stalwart at even strength, but when Carlyle used him in the ‘take the faceoff then get the hell off the ice’ role like Babcock has with Boyle, he was actually very effective. I’ve made the joke that Bozak’s good at chasing the puck since he got so used to it under Carlyle, but he actually is effective at taking away shooting lanes on the PK. The odd time he would block a shot and get a breakaway, few players in the NHL were better at scoring on breakaways than Tyler Bozak (except this year where he seems to have lost his shootout pixy dust).

Again, I know it sounds crazy to suggest replacing a 5v5 defensive monster with a 5v5 defensive liability, but based on their 4v5 results throughout their careers, we have evidence that suggests Bozak is a more effective penalty killer. It’s also worth pointing out that Bozak ranked 6th in the NHL in faceoff percentage this season, which is an extremely overrated skill, but a useful one when we’re specifically talking about the 4v5 faceoff role.

On to my final point: Mitch Marner is good at hockey.

Like very good.

The dude’s just…so good at hockey.

When you look at the data over the last decade, the best players at suppressing shots & chances on the PK over the past decade tend to be guys who also do it very well at 5v5. Players like Datsyuk, Bergeron, Karlsson, and Subban have traits that make them extremely effective hockey players. What I think a lot of people don’t realize is that these traits can also be very useful on the penalty kill. Here’s a great example of Tavares using his skill to play some keep-away on the PK (keep in mind that puck possession is shot suppression – the other team can’t shoot if you have the puck).

The evidence indicates that players like Tavares tend to be excellent shot suppressors on the PK, while 4th line grinders like Gregory Campbell typically have a poor impact on 4v5 results. The data seems to suggest that speed and skill are the biggest drivers of success, even if a player has elite Heart per 60…

I picked that clip for a reason since it highlights everyone’s biggest fear with playing an elite player on the PK. It comes down to a tricky cost-benefit analysis, but I would personally argue that the minor risk of injury from blocking shots is worth the value that star players bring to penalty kill, especially in high leverage situations like the playoffs. I’m planning on digging deeper into this, but I’ve found that forwards don’t get injured blocking shots nearly as often as we think on the penalty kill (it tends to be defencemen in front of the net when a puck rises and catches them in the face). I completely understand the other side of the argument though – “why the hell would I willingly put one of my best assets in front of a Shea Weber slap shot?” It’s a fair question, and would you really be able to forgive yourself if you lost Marner for the year because of a mid-January PK shift? I know I wouldn’t.

To wrap things up, I think the Leafs have all the makings of a great PK moving forward, especially with the addition of Kapanen. His elite speed should make him a great penalty killer in the same vein as a Michael Grabner or Darren Helm, and I’d argue that he has more skill than those players. When it comes to filling out the rest of the PK, the evidence suggests that the Leafs can improve their 4v5 results by using Bozak in the face-off role over Boyle. The data also indicates that Toronto would be best served to give Marincin, Gardiner, and Marner more 4v5 time in the future. To compromise on the injury risk argument when it comes to using Marner on the PK, Anthony Petrielli came up with an excellent idea (which we talked about in length on my podcast). His idea was to only give your elite players 4v5 time in high leverage situations. Essentially the Leafs could minimize Marner’s risk of injury and maximize his impact on results by only bringing him out on the PK late in close games throughout the regular season, down the stretch when competing for a playoff spot, or in a playoff series against the best team in the league.

  • MartinPolak

    I’m a big Polak fan on the PK but after that Marincin 5v3 PK I’m starting to become a bigger fan of his.
    I find Hyman scoring so low on your metric a big question mark. These are the points when smart coaches like Babcock over rule #s and slot roles players because of “intangibles”. (Ok I hate to say compete but at some point we have to acknowledge that analytics doesn’t capture everything and babcock along with a handful of other NHL coaches are the best of making these kind of decisions).

    • Kanuunankuula

      I think coaches get rapped up in players “reputations” same as everyone else. Hyman’s reputation is that he is good on the PK. What’s the point of this “compete” or intangibles if they don’t show on the results, as I question always with this narrative?