This is the final installment of Killing Time, a project where I tracked every single penalty kill by the New Jersey Devils in the 2014-15 season to determine how much time did they keep their opposition out of their end, who made the exits out of New Jersey's end and how, and who got targeted on the entries. This is officially part 5, though there have been five previous posts in this series. I would recommend reading each of them if you want the full breakdown of how I did this project and what the findings were, era by era.
- Part One: The Introduction of Killing Time
- Part Two: The Bryce Salvador Era - also known as: The Awful First 15 Games of Penalty Killing for the Devils
- Part Three: The Peter DeBoer Era - also known as: The Best 21 Games of Penalty Killing for the Devils
- Interlude - Did the Penalty Killing Tactics Change Between Head Coaches? A Picture Post
- Part Four: The Multi-Coach Era - also known as: The 46 Games Ending Up Between the First Two Eras
In this post, the title says it all: this is the summary of the whole season of penalty kills and it will include plenty of conclusions. I learned quite a lot about penalty killing, tracking events, and how the Devils performed when shorthanded. There's a lot in this post. Let's get right to it with the summary of results.
The Summary of 268 Penalty Kills
Across all 82 games, the Devils had 268 shorthanded situations. That's approximately an average of 3.26 per game. The Devils only one game where they were not shorthanded, so one could count on the PK to play about six minutes per night. They finished the season 21st in success rate, killing 80.6% of their penalties. While the bottom third of the league isn't necessarily a good place to be in, it is an accomplishment given how the first fifteen games went for the killing penalties. Besides, the league median was 81% so the Devils weren't too far from what one would consider to be decent. A success for the league is simply getting through the shorthanded situation without giving up a goal. For this project, I looked at how much time they took off the clock. I can now tell you how much that was over the 82-game season.
|Era||# of GP||# of PK's||Total PK Time||Killed Time Total||% Killed||# PPGA||# Games > % Killed||# Games > 50% Killed|
Surprising me, the amount of time killed was similar throughout all three eras. Yes, the DeBoer Era - the 21 games where the PK was hot - was much better at killing time than the abysmal Salvador Era; but the gain was just that. The larger set of the MultiCoach Era was just a bit lower from a percentage standpoint. Overall, the Devils' PK kept their opponents out of their end of the rink for 39.54% of their penalties. Of these 268 situations, twenty ended up having no exits at all, whether it was a quick score by the opposition, an abbreviated power play, or the Devils getting pinned for two minutes but without a goal against. For the remaining 248, the Devils were wasting just under 40% of the clock for all situations.
Ahead of the project, I would've figured that a penalty kill that managed to waste half of the penalty, a minute out of your common two-minute minor, would be a sign of success. It still would be a sign of success, just a very positive and uncommon sign. It didn't happen regardless of how hot the PK was at killing penalties. Knowing what the overall percentage time killed for the whole season was roughly 40%, that would be a more appropriate mark for what should be considered as a good amount of time killed. Given that the shorthanded team is literally down at least one man, making sure the opposition spends about 48 seconds out of their end or more may be seen as a positive.
That said, the fact that three eras - a bad run, a great run, and a run somewhere in between - weren't that far off in terms of how much time they were able to keep the puck out of their end as a percentage is very telling. It suggests that taking time off the clock isn't necessarily a barometer of whether the PK really is good or not. Sure, it's absolutely worth doing. But over a larger set of games, it . The reality is that it was rare for the Devils to consistently kill the same amount of time within a game. Often there would be the case there would be one shorthanded situation situation where the Devils' PK was just on-point and took a lot of time off the clock; but later in the game, it wasn't so sharp and struggled to get exits and so not a lot of time was taken off the clock. Over the whole game, it may have been a fine number but it evens the ups-and-downs. I believe that's where 40% comes from.
The Summary of Exit & Entry Types
Let's move on to the types of zone exits from New Jersey's end and zone entries made into the Devils' third of the rink. Over the whole season, I counted 1,102 exit and entry types. Not all exits and entries logged were actual exits or entries. I also counted attempted exits that were denied as well as goals against, no exits at all, and no entries at all. Here's the full breakdown by type.
The most common exit type made by the Devils were icings. Moves that sent the puck beyond the neutral zone, often behind the goal line and/or collected by the goalie. The runner-up was no exit at all; but the number of denied exit attempts. Unforced outs came in third; those were the exits caused by an errant pass, a shot missing that careened out of the zone or out of play, or some other instance like the end of a period. Clearances were clears that were picked up in the neutral zone; enough to force the opposition to get back onside, but not enough distance for them to go far. Exits by carrying the puck and forcing the puck out by a Devils' body or stick were close but ended up fifth and sixth, respectively. Goals against are a type of exit - they ended the penalty early - and those finished seventh. Exits by passing the puck out of the zone were quite uncommon, though not as uncommon as no exit at all. There were two situations where the Devils didn't concede a goal, but did not get the puck beyond their own blueline. Needless to say, those were tense minutes at the time.
I was not totally surprised that icings led the way. One of the rule changes for a penalty kill is that there is no whistle for icing the puck. The shorthanded team is allowed to do it. Clearly, they take advantage of that change. The Devils sure did as the vast majority of their exits were pucks sent across all three lines. Should someone want to ensure power plays are more dangerous and potentially increase scoring in general, perhaps someone should propose icings on penalty kills. I think that would be really unfair for the team down a man or two, but it would emphasize that those calls can really be damaging. Speaking of the opposition, here's how they entered the zone.
With the opposition team being up a man, their preferred way of gaining the zone was to carry the puck in. Again, this shouldn't be a big surprise. Someone should be open among the five when it's a five-on-four. What was interesting was how teams did. It was common to see a breakout essentially consist of one or two passes, where someone's at the blueline and they carry it in after a pass. Sometimes that pass would be just inside the zone, making it a pass-in entry. But most would do it before the blue. Of course, there were plenty of standard "one man has it and he's going to charge forward because he's good" carry-ins.
Dump-ins and pass-ins were neck and neck in terms of occurrence, which is a bit surprising since a pass-in was often close to being a carry-in given how most teams do their breakouts on the power play. Dump-ins were rarely done regularly - first game against the Flyers' excepted - and often done as a result of just struggling to breakout or lack of an option. For the most part, the teams that could carry it in did, even if they did a soft-dump after entry. The second most kind of entry was none at all; a combination of ends of penalties (with or without goals) or ends of periods. Errors by the Devils that led to an entry and faceoffs moved to New Jersey's end did happen, but not nearly so often in an 82-game season.
My main takeaway from this is that the expectation for the Devils' penalty kill is to worry about the how the puck is carried into the zone. Based on what I saw in the interlude, they did expect it and play in response to that. It didn't always lead to a quick exit, but that's going to happen when the other team has an extra man to play with.
Who Made the Exits?
Rather than hit you with a lot of charts for each exit type, I'm going to give you one giant summary of every single Devil that had an exit this season. All exits were counted by me for this project. The shorthanded ice time (SHTOI) nearly all came from War on Ice. I manually counted Steve Bernier's. Yes, he had an exit, so he's here.
|Devil||Who Made Exits?||Exits||Exit/Min||I||C||X||FO||P||Denied||SHTOI|
Your leader for the 2014-15 Devils penalty kill in total exits, icings, clears, and denied exits is the man who played more shorthanded minutes than anyone else on the squad: Andy Greene. He even beat out Nobody, the "one" responsible for those unforced exits. Greene had a bit of everything as he just played a ton of minutes across all three eras I defined from last season. #6 was a constant presence on the first pairing for penalty kills. If he wasn't there, either he was really tired from the prior shift, the game's no longer competitive, or he's in the box serving the penalty. Other Devils have higher rates of exits per minutes, but nearly all of them are a function of either position (it's harder for defensemen since they're usually deeper in the zone), time, or both. Among defensemen who've had more than a token appearance, only Jon Merrill had a higher rate of exits per minute than Greene. Basically, Greene was leaned on heavily for PK's, followed by Adam Larsson from a usage standpoint.
In retrospect, perhaps Merrill should've received more time in shorthanded situaitons than some of the other defenders at the time such as Peter Harrold, Mark Fraser (here as 32-D since Mike Sislo wore #32 too), Damon Severson, and Bryce Salvador (who received a ton of time in his 15 games, only three defenders had more ice time than him on the PK all season). While Merrill had more shorthanded ice time overall, his usage wasn't consistent and was often limited to being the third defenseman on kills at best when he was used. It would be beneficial for the coming season if he could prove to be someone who can handle kills, making more exits than most other defenders.
Among the forwards, Patrik Elias reigned among the number of exits made and his rates of exits per minute. While other heavily used forwards like Travis Zajac, Stephen Gionta, Jacob Josefson, and Adam Henrique were no slouches at all, this was an area where Elias was particularly effective. Given how he was declining as a player at even strength last season, this provided some solace. The rates from forwards look stronger in part of the fact that they didn't play as much. The Devils have maintained in recent years a rotation of forward sets. Given their style, a box that shifts to and from a wedge-plus-one, they are able to plug in various forwards to do a decent job. It's not particularly complex, especially when job one is to get the puck out when they get it. Which is why guys like Tim Sestito and Mike Sislo can take a shift or two and not look entirely out of place. Some were better than others, but for the most part, Elias, Mike Cammalleri, and the rest of the more common forwards were able to make things happen if or when they got the puck.
Who Got Targeted on the Entries?
Among defensemen only, I tracked which side the opposition power play targeted on their entries. While upon entry, the defensemen would skew to one side or the other at times, I picked which side they were meant to cover. Hence, I made this similar chart, breaking down who got targeted and by what type of entry.
|Devil||Who Got Targeted?||Targeted||Entries/Min||X||P||D||E||SHTOI|
Since Greene played the most, he got targeted the most among all types. This was most apparent at the beginning of the season where Greene would not only play a lot but he'd switch sides from left to the right depending on his partner. While opposition power plays seemingly tended to run their breakout regardless of who it's against, the rates here are telling. While Greene had the most targets in gross, Merrill, Mark Fraser, and Seth Helgeson (who's 25+33 because he wore both numbers last season) were much higher. It seems Merrill and Fraser were specifically targeted, as was Helgeson in his brief appearances on penalty kills. That may speak to why Merrill didn't get so much time as well as why Fraser and Helgeson appeared to struggle at times - especially when the entries came in fast. Arguing against that would be Salvador's numbers. He wasn't targeted all that much considering his lack of speed and declining skill. Maybe opposition power plays figured going up against a heavily used #6 would be the better play. I am leaning towards my theory that power plays did how they drew it up as opposed to adjusting to pick on a player.
How Much Time Did an Exit Take Off the Clock for New Jersey?
Before I jump into what I learned from this whole experience, I've saved my most interesting summary for last. How much time did a type of exit take off the clock on average for the Devils? During my tracking, I recalled Steve Cangelosi and Ken Daneyko briefly touching on that subject before a defensive zone faceoff. Cangelosi suggested winning that draw and icing the puck would take twenty seconds off the clock. Daneyko didn't really argue it. While I don't recall the game, that event did happen and it took about fifteen seconds. In general, that just above average for an icing exit within this past season:
|Time Killed by Exit Type in Seconds During '14-'15 New Jersey Devils PK End of Period/End of Penalties Removed|
|Type of Exit||Icings||Clears||Force Outs||Carry Outs||Unforced Outs||Pass Outs|
|# of Exits||384||105||63||68||116||13|
|Time Killed in Min.||92.42||19.58||13.67||18.23||23.68||3.82|
On average, any zone exit took at least eleven seconds off the clock. How much more than that depended on the type of exit. The most effective by average were exits where the Devils passed the puck out. It's almost better than a carry, which also took plenty of time off the clock, as it was direct pass to someone outside of the zone. It's enough to lead someone on for an attack. Whether they actually took a shot or not is immaterial, but the option to do so was often there. Icing the puck - which was the most common type of zone exit - finished about third in terms of average time between exit and entry. Not bad, but not as good as the others. So why didn't the Devils attempt more carries or passes out of the zone?
The answer becomes obvious when watching a PK: the option is rarely there to carry the puck out or pass it to an open player beyond the blueline. The carries and passing exits often happened as a result of the puck being put into a space where there was no one in front of the carrier. That could be a deflection, a block, a rebound, an errant pass by the opposition, and so forth. If the opportunity was there and the player saw he had enough time and space to do it, then he'll take it. They weren't going to take someone one-on-one, risking taking themselves out of the play should they lose it or have it stolen. So the opportunities to do that were rarely there. Since shorthanded teams have the ability to ice the puck with no whistle, the safer and often more available play is to collect it and fire it down the rink. That turns out to be a better play time-wise than just getting into the neutral zone, hoping the opposition or the game clock makes a zone exit, or even forcing the puck out by deflection or a block.
More importantly, icing the puck carries less variation. While all of these exits had a moment or two where the opposition almost-immediately entered the zone after the exit, icing the puck had a smaller standard deviation than the other types of exits. A carry out or a pass out of the zone may take more time off the clock on average, but it's varied as to how much. It can either be a lot or even a lot less than the average. Granted, that there were many more icings than other exits helps nail down the standard deviation. But the common sign is a clearance down the length of the rink, the opposition collects it, and they run their breakout. That alone is good for at least 10-11 seconds and the breakout itself may extend that a bit. The other types can have quicker turnarounds just based off distance alone. That said, to really go beyond the average time in seconds killed for an exit, more needs to happen than just the exit itself.
What Did You Learn From the Killing Time Project?
I've noticed quite a few things while tracking 268 penalty kills. It's hard not to. While I've watched a lot of hockey, there's a lot you can pick up when you focus on a particular set of events or game situations. Here's what I picked up from Killing Time:
Faceoffs can be quite important at the start of a special teams situation. In the larger hockey analytics community, faceoffs have been determined to have been not that important with respect to possession. However, winning that first faceoff can go a long way for the power play to immediately get set up and attack or get a clearance to truly start that power play at the end of the rink. Often times, a won New Jersey faceoff is followed by Larsson or Greene taking the puck, maybe taking a stride or two, before attempting an exit. While an icing will rarely take twenty seconds from exit to entry, taking off 12-17 seconds is a better beginning than scrambling into formation after the opposition has the puck. For the opposite case, many of those zero-exit situations or times where the PK got pinned back started with a simple faceoff win.
Teams will run their plays on special teams. One of the consistencies I've noticed from watching 29 other power plays operate is that they have their own methods and philosophies for breakouts. For example, Washington's breakouts made it such that every icing by the Devils took about 16-17 seconds off the clock. That's because there will be a carry out from their end, a pass into the neutral zone, a turnaround and pass back to someone else at their blueline, and that player will charge in. That's the most exotic I've seen and the back pass makes it take a little more time than most other breakouts. But it got them their zone entry and often led to their terrifying power play to get set up. I'm confident in what I stated among the targeted Devils, that teams will run their breakout plays regardless of the situation. It's why there weren't many quick passes up ice after a clearance or an icing (the risk helps that too is an icing of their own, which is a big one for two minute situation as it helps the defense). It's why when the power play is ending, teams rarely just dump it in or force something quick just to get something out of it. Teams practice their special teams to run a certain way and they will do that. This applies to the Devils PK as well. Their formations on entries and in their own end didn't change much throughout the season despite personnel and head coaching changes.
A lot can happen in a second. I can understand if you look at this and think something like "OK, so this type of exit takes off two more seconds on average than this one, big whoop. What can happen in a second or two?" The answer is: a lot. It often took about two to four seconds for the iced puck to actually reach the end of the rink. A lot of denied exits had the actual denial just in front of the blueline; those took a second between attempt and denial. The varying breakouts by teams made some exits take more time off than others, but typically the range was 12-17 seconds. And when I recorded the time of exit and entry, I often had to pause and rewind and pause again to make sure what I saw was correct: that a puck or a player can move quite a bit within a second. An entry can happen within it as well as an exit. So being able to consistently take a second or two more off the clock on a PK does carry value since making exits happen isn't necessarily easy - especially when the opposition is set up and maintaining puck possession. Likewise, decisions with the puck have to be quick because within a second, an open player can easily be covered. Hence, the prevalence towards icing the puck or deciding real fast to move it out as opposed to sitting in your own end with the puck - which happened only one time that I can recall.
If you want to extend time between exits and entries - disrupt the breakout. Ideally, the best thing a penalty kill can do is not just get an exit but to score a goal as well. However, if the goal is really to kill the clock, then it's superior to get an exit and then mess with the other team's breakout. A bad pass by them, a denial at the blueline, a shot on net that forces a faceoff in their end, or even a forecheck to delay their advances all takes up precious seconds that they could've used to attack. All of those exits that lasted twenty seconds or more before their entry had something like this. An effective forecheck and/or some plays in the neutral zone to essentially have the other team attempt their breakout twice or three times will take off more clock than just skating it up ice and attacking. The reward for the shot is supreme, but if the plan is to take off time, then the focus has to be on disrupting the breakout.
Your goalie really does have to be your best penalty killer. In a project where I didn't track shots against - not that I have to, sites like War on Ice or even NHL.com do that to a degree - it seems weird I have a conclusion about the goalie. Well, I've shown that the Devils kept the opposition out of their end just under 40% of the time while shorthanded. That means the opposition was in their end just over 60% of the time. That's plenty of time time to spend getting set up, looking for an attack, and attempting quite a few shots. Even if the Devils can push that to a 50-50 split, that's still about a minute for the PP to operate. For the penalty kill to ultimately be a success, the goalie's got to make the stops when they inevitably come. Cory Schneider did an admirable job last year in those situations with an 89.2% save percentage. I have my doubts whether he can keep that up. Should the PK take a step back in success rate, I would look to that first before the play of the skaters.
It is superior to have two capable defenders than one. The night-and-day difference between the Salvador Era and the DeBoer Era prior to his firing was the lack of Bryce Salvador. Salvador was notably slow, his positioning wasn't on point, and if he was beat, then he was beat. After those fifteen games, Adam Larsson became Andy Greene's regular partner. While neither are particularly quick, both were more than capable at chasing down loose pucks, getting into positions faster, moving within their own zone more effectively, and making quick decisions on the puck. That led to more exits as well as a more effective penalty kill. Go back and look at the numbers. While Salvador played a lot, he didn't attempt or make a lot of zone exits. Greene carried that load. With Larsson, it wasn't always on Greene as #5 quickly proved he can do the same. Having capable defenders who can make exits as well as defend in formation is crucial to any penalty kill.
When you track anything, you need to define what you're doing and check whether it meets your goal. I did Killing Time with the intention of determining how much time the Devils actually killed on a penalty kill. I met that goal; I have a hard number from last year and plenty of conclusions from the entire project. However, there's plenty of questions that arise from this that I can't really answer. I kept my exit types simple; so I can't tell you if an iced puck to the corner or an iced puck received by the goalie ahead of the goal line had a difference in time killed. I can tell you what type of entry was made, but not where on the ice it was - not even with simplified signs. I can't tell you much of anything of what happened after the exit or after the entry beyond what I recall. I can't tell you whether the Devils stack up well or whether 39-40% of time killed is about normal for a PK. Of course, the purpose of this project wasn't to provide those answers. They may be worth exploring should I do this kind of endeavor again.
Most of all, I can't tell you whether this is really a way of determining if the penalty kill is doing well or not. Again, the worst run of last season killed about 38.5% of all shorthanded ice time, the best run killed 39.9%. Seconds matter, but is that significant of an improvement? This goes back to how we define a successful PK. The main goal is to not get scored on, but the modification in the rules plus the tactics have more to do with taking time off the clock as opposed to goal prevention. So this doesn't really satisfy my original hope that this could lead to a new way of judging PK. It was very instructive and I would say it would lead to some more actionable discussions. Such as, "We take an average of X seconds off with every icing, what can we do to extend X?" Or, "We know disrupting the other team's breakout will help kill off the clock, so how and where are they making the entries that we can try to prevent? And how aggressive should we take to go after that breakout?" I think this project provided that a lot of useful information, but the larger question of whether a PK is doing well or not remains just that to me: a question.
For 2015-16, the Devils need to find a capable second defensive pairing for the PK. You want a takeaway that applies for next season? Check this out. The 2014-15 Devils leaned heavily on Greene and, when he was in the lineup, Larsson. I would expect them to play a lot of minutes again if only because they're the most experienced in shorthanded play among the other defensemen. However, should they get hurt or have a bad run of games, their PK will suffer greatly without one and/or the other. Not that Harrold, Fraser, or Salvador were great or even good on the PK, but they got the minutes over Merrill and Severson. I would pencil those two as being the next ones up behind Greene-Larsson. With Merrill, there's some signs to be confident, due to his solid exit per minute rate. Severson doesn't have that, but he was also a rookie last year. I'm hopeful he can be better. In a way, they pretty much have to be unless John Hynes and his staff wants to find out whether Eric Gelinas, John Moore, or David Schlemko can kill penalties. While Moore and Schlemko are new to the organization, I'll need to see something before I believe they really can. Should Merrill-Severson or other defensive combination - just in case Moore could kill penalties - function on the PK, that will make the unit as a whole stronger as it'll be less reliant on Greene-Larsson playing heavy minutes nearly all of the time. The forwards may have a similar issue now that their without Zubrus and Elias is a year older, but they are deep enough to absorb those setbacks. The defense beyond Greene-Larsson remains a question mark for the penalty kill, if not at even strength as well.
I hope this post and this series taught you as much about the Devils' penalty kill from last season and perhaps penalty kills in general as it did for me. If you have the time, something like NHL Gamecenter, and a burning question regarding the game itself, then I suggest considering doing some tracking of your own. Just watching a lot of video will allow you to not only answer that burning question, but learn much more about the game. What did you think of the overall results? What did you think about my conclusions from this project? What did you identify that I may have missed? What did you learn from all of these posts? Please leave your answers and other thoughts about this post and the Killing Time series in the comments. As always, thank you for reading.