Penalty Kill faceoffs matter, but not enough to stress over

USATSI_9156547
Photo Credit: Dan Hamilton/USA TODAY SPORTS

Young Frederik Gauthier steps into the defensive zone faceoff circle against Ryan Kesler. Zach Hyman just responded to an admittedly rough situation by punching a guy, which the referees didn’t like, putting the Leafs on a crucial penalty kill. Give up a goal, and the team is in huge trouble.

Kesler assembles his group the way he wants them. He knows that the four players Toronto have placed on the ice; Gauthier, Connor Brown, Roman Polak, and Matt Hunwick, aren’t mobile enough to conclusively win a race if the play doesn’t work out. He also knows that Gauthier doesn’t have a ton of experience against him at the dot and that his handedness is just right for this set play.

The puck drops and Kesler wins the faceoff clean. Nobody knows what’s going on. Cam Fowler skates into a now-open one-timer and blasts the puck past Frederik Andersen. 3-2 Ducks.


“That’s why you need a faceoff specialist on special teams,” bellowed the broadcasters, the analysts, and the fans (especially the broadcasters; I’m pretty sure faceoffs are at least a quarter of all Sportsnet talking points). “Games are won and lost at the dot.” Heck, if you want a direct quote, Leafs coach Mike Babcock opened his breakdown of the game with “they dominated the circle against us”.

Indeed, the faceoff percentages in this one closely mirrored the final score; 3-2 Ducks and 66.7-33.3% at the dot. The game-winning goal was a direct impact. How can you argue against that?

Well, a few ways.

Do Elite Faceoff Teams Kill Penalties Better?

The first thing I noticed upon looking at the last decade of 5-on-4 and 5-on-3 data is that, well, the penalty killing team more often than not loses the draw. Only Winnipeg (50.7%) and Detroit (50.1%) are above 50% at the dot on the kill (San Jose is exactly 50). Most teams fly between 44 and 47%, the median number is Buffalo’s 46.33, and the league-wide average is 46.25%.

That isn’t terribly surprising, to be honest; the team on the powerplay has an extra body to help with the tie up, and given that many teams still follow “best players on the PP, role players on the PK”, the fact that the powerplay team has the edge over the 136,843 faceoff sample isn’t a massive shocker. Building a penalty kill around the prospect of probably winning the draw, in the long run, already seems like a losing battle. But I decided to see if winning faceoffs lines up with having a great penalty kill in the long run.

Using the same block of data, I figured out that Winnipeg, with the top faceoff percentage, is 27th of 31 teams (Jets and Thrashers separated) in actual PK%. Detroit is 12th, San Jose is 11th. Carolina is 6th at the dot but 20th overall. The Leafs, you’ll be pleased to know, rank 9th in faceoffs! They also have the worst assortment of penalty killing units of the decade, rolling at 78.6%.

Just for fun, I also looked at unblocked shot attempts, shots on goal, and what percentage of penalty kill shifts started in the defensive zone to see if perhaps those stats had stronger connections.

Team PK% FA60 SA60 DZF% FO%
St. Louis Blues 1 2 1 4 11
New York Rangers 2 4 4 7 22
Pittsburgh Penguins 3 11 13 15 29
Montréal Canadiens 4 10 8 16 10
New Jersey Devils 5 3 5 11 5
Vancouver Canucks 6 7 7 25 20
Minnesota Wild 7 8 12 29 7
Boston Bruins 8 9 10 8 4
Los Angeles Kings 9 15 9 30 8
Philadelphia Flyers 10 6 2 10 14
San Jose Sharks 11 12 11 22 3
Detroit Red Wings 12 5 6 17 2
Ottawa Senators 13 17 23 12 17
Anaheim Ducks 14 21 17 5 21
Nashville Predators 15 1 3 19 23
Washington Capitals 16 19 21 26 25
Buffalo Sabres 17 28 27 24 15
Calgary Flames 18 22 18 3 12
Chicago Blackhawks 19 14 14 2 31
Carolina Hurricanes 20 13 20 21 6
Columbus Blue Jackets 21 18 16 13 28
Tampa Bay Lightning 22 24 25 18 26
Dallas Stars 23 20 15 14 19
Colorado Avalanche 24 16 24 1 16
Florida Panthers 25 30 31 27 24
Edmonton Oilers 26 29 28 31 27
Winnipeg Jets 27 23 22 28 1
New York Islanders 28 25 26 20 30
Arizona Coyotes 29 31 29 23 13
Atlanta Thrashers 30 26 30 6 18
Toronto Maple Leafs 31 27 19 9 9

My unscientific opinion here is that there’s not a huge connection between winning draws and being good on the penalty kill. There’s a slight one, as there should be, but more seems to be gained from preventing attempts, either by blocking them or by boxing off shooters to force them to miss the net. Even starting the play in other zones doesn’t help a ton, presumably because you’re stuck with a potential odd-man rush if you don’t win the draw, which, again, you probably won’t even if you’re near the top of the league at it.

It seems like more emphasis should be put into having your skaters that are best at getting in the way of and clearing/skating out loose pucks when you start in the defensive zone, and having those who are stronger at gap control in other zones. If your centre isn’t proficient in those regards, getting him out there in hopes that he’ll have a slightly better chance than the other guy to win the draw probably won’t do you a ton of good.

Does Winning The Draw Prevent Goals?

But just for the benefit of the doubt, let’s make a crazy scenario. Let’s make faceoffs super impactful. By that, I mean, let’s pretend that all 13,453 powerplay goals scored in the past ten years were the exact same as Cam Fowler’s goal. One of 136,843 faceoffs were won, the puck went to the defenceman, and it was blasted in three seconds. Every time. We’ll even ignore zones; Fowler can blast these theoretical goals from Cloutier or even Toskala distance. Because faceoffs matter a lot in preventing goals, we’re going to create an absolute worst case scenario for our penalty killers.

In that world, how many goals is your faceoff specialist saving you every year by winning the draw?

Using the data we have at our disposal, one can expect that one in every 4.71 faceoff losses will lead to a goal against. Teams take 6.198 penalty kill faceoffs a game, meaning that we’d be looking at about one Fowler every 1.32 games.

Assuming your draw wizard takes half the penalty kill draws (3.1) in a game, he’ll win 1.9 of them. Over an 82 game season, that’s 153 draws; or 32 Fowlers prevented! Impressive, right? Well…

FO% 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45
FOW 1.9 1.8 1.8 1.8 1.7 1.7 1.7 1.6 1.6 1.6 1.6 1.5 1.5 1.5 1.4 1.4
82GP 153 150 147 145 142 140 137 135 132 130 127 125 122 119 117 114
GA- 32 32 31 31 30 30 29 29 28 28 27 26 26 25 25 24

A faceoff specialist, winning faceoffs at 14% above league average on the penalty kill, who slots in for half of your draws over an entire average season in this scenario prevents seven goals against over a player who wins at the average, and ten more than someone who wins 40% of draws.

Keep in mind that our scenario involves more than 20% of faceoff losses turning into a one-timer goal. Most of the time, that’s not what happens. The team on the powerplay sets up, maybe gets a shot attempt or two off, the killers clear it, they do it again, the goalie freezes it, another draw happens, the same thing happens, and 80% of the time, the penalty killers come out unscathed. Someone smarter than I could probably figure out the true percentage of faceoffs that turn in into powerplay goals before the opposing team gains possession, but I’d imagine that it’s in the low single digits, meaning the real word difference between 60 and 40 here is probably closer to a goal or two a season.

At that point, it probably makes more sense to have a skater who can close off a lane, get in the way, clear it, or carry it over a guy who has an extra 1-in-20 chance of affording his teammate the chance to do that. If your best natural penalty killer happens to be the guy who wins the draws, that’s great; there was a time when the aforementioned Ryan Kesler was that guy for Vancouver, though it seems Anaheim does better at preventing shots now when the super speedy Andrew Cogliano hits the ice instead. Most of the top penalty killers in the league in actually keeping the puck away from the net are players you know more for their mobility than you do for their draws (Patrice Bergeron, of course, being the exceptional man that can do both).

An alternative strategy is to simply not take penalties, which is also something that subbing out depth players with speedy, skilled players helps with. You draw more calls when you’re hard to keep up with and half the puck, and while some of Toronto’s penalty killers (Matt Martin, Leo Komarov, Matt Hunwick) have been adept at keeping the puck away from the net, their lines often contribute to initiating the penalty kills to begin with by having to chase the play. 

Where Are You Going With This?

It’s obviously a shame to see a situation like last nights happen, especially when it’s a guy like Gauthier, who has been hyped up as a prospect who will one day dominate the dot and kill penalties, who gets burned. Ultimately, what happened in that play cost the Leafs some standings point, and a bit of ego as Randy Carlyle got to feel smug for a few minutes. Toronto also gave up both of its powerplay goals while serving roughing minors, though, so that might be something to be more concerned about.

Don’t let the few examples that burn into your head of a quick conversion make you believe that the long-term impact of having an “expert” at the faceoff means much. Yes, winning the draw and getting initial possession is something that’s very nice to have, but it’s a situation that normalizes itself quickly if you don’t have the talent on the ice to maintain it, and the additional odds you gain by having a specialist are negligible in the long run. The best way to kill penalties is to focus on players who can keep the puck away from the net for not just the two seconds that get committed to draws, but the other 1:58. Or, draw more penalties than you take.

This is all before we even consider the vast majority of the game that’s being played at even strength. If your player isn’t impactful there, but they’re in your lineup just to win special teams draws, you’re probably giving up many more goals than you’re preventing. But we’ll save the “No but seriously, waive Ben Smith” argument for another day.

  • LukeDaDrifter

    That is a long article with a lot of data to come up with a minority point of view. One thing that is missing is: What would the data be like if teams didn’t care about whether a center can win his share of faceoffs? Every team have at least one top guy. On the powerplay losing the draw in the offensive zone kills 30 seconds of your power play as your team has to retrieve the puck, wait for line changes to complete then attempt a zone entry. My argument would be where would teams be if they didn’t have a couple of guys like Bozak, and Smith that could win half the draws? When you win that draw on the PK in the defensive zone no shots have to be blocked. The goalie doesn’t have to make any saves. Your PK guys get to make a quick line change. The are so many advantages of gaining procession of the puck off the draw, I can not see how anyone could possibly argue it doesn’t matter in the long run, or short run either. .Especially a guy who normally argues how important procession is.

    • Brad M.

      He never argued that it doesn’t matter.
      In the HEADLINE of the article, he even wrote, “Penalty kill faceoffs matter… but not enough to stress over.”

      I think it’s pretty obvious that traditional analysts (and some coaches) tend to over value faceoffs in general.
      Killing off 30 seconds on the PK (because of a successful draw) means very little over a large enough sample, because… if your players suck… they still have 90 more seconds to prove their suckage. And they often do, according to the historical stats.
      That’s why Jeff is saying that capable penalty killing (blocking shots, being positionally sound, covering passing lanes etc., is more important than employing a faceoff specialist).

      I think having both is ideal, but a faceoff specialist on the PK is a luxury more than a necessity.

      • Jeff Veillette

        Yeah. Basically my point is being a good player that is good at faceoffs is nice to have but if you’re only there to win faceoffs you aren’t going to help the team very much.

        • LukeDaDrifter

          It is not that what you are saying makes no sense at all. A coach must deploy the players that the team has acquired for him in the best manor he sees fit. It looks like Bozak and Smith will be out at least until the holiday break. Your vision has a good chance of coming to life on the Leafs. If Nylander establishes he is ready to play a top nine center role at 20 years old, then our top faceoff guy Bozak could be moved to the fourth line. Bozak certainly will be better than Smith when it comes to doing more to help the team on the 5 on 5’s. That should also give us the most expensive 4th line in the league. $4,200,000 to a 4th line center, $2.5 million for 4th line left winger, and $612,500 to the right winger. Other things to consider are who is going to play the defensive role on the JVR – Nylander – Marner line? How will Matthews line look with Brown taking Nylander’s spot? There certainly is a good possibility. to see Bozak on the 4th line after Christmas.You could always trade for a top line center to play the 4th line role. It will certainly cost you money (how’s your cap room) and you will have to send something decent back the other way. You would now be running a top 12, which might be the way of the future.

      • LukeDaDrifter

        I respect you opinion but in this case you are profoundly wrong. I will rebut some quotes in Jeff’s article.

        “Building a penalty kill around the prospect of probably winning the draw, in the long run, already seems like a losing battle.” (So what should teams do, just give the other guys the puck and get ready to block a shot)

        “there’s not a huge connection between winning draws and being good on the penalty kill.” (even if there is only a slight edge, of 10 goals over the season, on winning the draw, should we ignore it?”

        “it probably makes more sense to have a skater who can close off a lane, get in the way, clear it” (All of the guys on the penalty kill team must also have those qualities to make the team)

        “An alternative strategy is to simply not take penalties” ( that certainly would be simple alright, except every coach in the league knows they will take some penalties and must have a plan to defend them)

        “The best way to kill penalties is to focus on players who can keep the puck away from the net for not just the two seconds that get committed to draws,” (The draw may only take 2 seconds but it leads to 30 seconds of relief. Why isn’t best way to kill a penalty to score a short handed goal)

        Most of the game is played at even strength.”you’re probably giving up many more goals than you’re preventing.” (the opposite is true. Our top 6 players with the best goals against for 60 minutes are in this order. Matt Martin, Matt Hunwick, Roman Polak, Ben Smith, Connor Brown, Zack Hyman. All guys who are chosen for the penalty killing team who are also the best defensively at even strength.

        Jeff’s argument involves using a hypothetical “crazy scenario” ( That the scenario is “crazy ” is the truest part of the article)

  • FlareKnight

    As always Jeff is really battling the fight that faceoffs don’t matter. Which still makes no sense. With a game focused on possession, how it getting initial possession that meaningless? You win it on the PK, you can ice the puck. The PP has just wasted a lot of time going back, trying to get back in and setting up. When if they had won the faceoff…they are just setting up and trying to score immediately.

    Especially when you are trying to keep shifts short, the less time you spend chasing down the puck the better. I just don’t get it.

    Sure having guys that can only do one little thing can be questionable. The more well-rounded all players on the ice are the better. But I sure as heck want all the centers on the Leafs to master this skill. Because it does help you kill that penalty, it does help you score that PP goal, and is useful on 5 on 5. There are always faceoffs. You aren’t going to have a game where they drop the puck 3 times (only at the start of each period).

    • Jeff Veillette

      Getting the initial possession and moving the puck out makes sense in a vaccum. An individual faceoff is something you always want to win.

      But the greater point is that difference between a great and a bad faceoff person doesn’t lead to that many extra faceoffs won over the course of the year, and perhaps there’s more to gain by getting someone who can drive the puck out throughout the shift rather than those first few seconds. Teams don’t usually score on their first possession of the powerplay and while it’d be nice to make sure they don’t have it, there just isn’t anybody that’s better enough at it to make a sizeable impact on goals for/against in the long-term.

      • LukeDaDrifter

        The reason there “isn’t anybody that’s better enough at it to make a sizeable impact on goals for/against in the long-term.” is because the powerplay center is always facing-off against the other teams best faceoff guy.. One only has to think how the stats would look if teams constantly put a guy out, that loses most of his draws.

        Using the same argument that the stats show there is not really much difference doesn’t always hold water. Stats show pulling your goalie increases the chance you can score. You could use the same logic to conclude there is a better chance to score a goal if you played the whole game with the extra attacker.You know for sure you will score more goals and you would be correct. The other part of the game is the reason coaches only take the gamble in the last couple of minutes.

      • LukeDaDrifter

        Jeff this is a true story. My bother once asked me to play a game with his church league guys. I never turned down a chance to play hockey. Turns out we drove up to Beaverton and played against five guys who played for the local junior C team. They didn’t have a goalie so no one was their in net. They beat us 17 to 1. The game was also a lot rougher than one would expect in a church league. Nobody was that interested in turning the other cheek. I know I got caught crossing the blueline with my head down.

        • The China Wall

          Hey Luke, great story.

          You and your brother were on the wrong end of a real stats whooping… lol

          Your CA must have been about 99% and your CF about 1%! 🙂

          • LukeDaDrifter

            The funny thing is most of our guys weren’t convinced. We wanted to go an extra period. Hey! we had all played a lot of pond hockey and road hockey down by sixteen was nothing.

      • LukeDaDrifter

        Re- “Getting the initial possession and moving the puck out makes sense in a vacuum.”

        It also makes sense in a hockey game. Or are we talking about playing hockey in outer-space?

    • LukeDaDrifter

      My scientific calculator is working. After spending all night crunching the numbers, I have concluded Jeff really does have a good point. Having better players would be helpful.

  • The China Wall

    Ah, a new battleground for the eye-test gang and the whiz-bang analytic posse.

    I grew up in an era where the eye-test and simple stats ruled but that doesn’t mean I am so set in concrete as to be closed off to these new analytic stats.

    But here is the issue I see with the current discussion.

    In the past three games, the Leafs have lost a face-off in their D-zone while on the penalty kill and the puck has ended up in their net within 3 seconds (ANA game) and 4 seconds (SJ game). The SJ goal tied the score and sent the game to OT (which the Leafs subsequently lost) and the ANA goal won the Ducks the game.

    Seems to me this is a pattern and isn’t a pattern something that you’re supposed to look for in numbers? Isn’t a pattern an important marker?

    But hey, what do I know, after all my TI Stats Calculator stopped working years ago… lol

  • Bob Canuck

    Jeff,

    Very interesting read.

    Like you, I think much too much importance is given to faceoff wins. Yes, when on the penalty kill, it is better to win a defensive-zone draw than to lose it but the impact of a loss is not nearly as great as many suggest. For those not familiar with the work of Gabriel Desjardins, he estimated that you had to win 27 faceoffs in the offensive zone on the power play to generate an additional goal. See the work at the link below.

    http://www.arcticicehockey….

    In the 2015-2016 regular season, there were 1,429 power play goals; there were 5,953 power play faceoff wins in the offensive zone. Arizona led the NHL with with 250 wins; Edmonton was last with 160.

    If you attributed all power play goals to faceoff wins, you would need to win a little more than 4 such faceoffs to generate a goal. Using Desjardins estimate of 27 faceoff wins to produce a goal, on average only 220 (15.4%) of the power play goals were the result of faceoff wins (5,953 faceoff wins / 27). In other words, in the 2015-2016 regular season, 84.6% of power play goals were not directly attributable to faceoff wins.

    Systems and non-faceoff player skills are far more important in killing a penalty than winning a faceoff in the defensive zone. Hopefully, in the future, broadcasters will put faceoff wins into perspective.

      • Bob Canuck

        I think the arguments made by Jeff and I are consistent with the points made by Dom in his article.

        Per the article, “… When someone asks me why analytics matter this is what I think about. I think about finding an edge, an inefficiency, an improvement over the ways of traditional thinking”.

        Desjardins, Jeff and others have questioned the traditional view that faceoff wins are very important, particularly on the penalty kill. Their work has shown that faceoff wins are not nearly as impactful as many believe.

        Dom noted that a team can capture some of the extra 2% by optimizing the fourth-line and bottom-pairing personnel decisions. Dom called Smith a below-replacement level player and that Peter Holland was sitting in the press box while Smith wins some faceoffs. Looking at all situations and faceoff locations, if you apply Hollands faceoff winning percentage over the past three seasons to the number of faceoffs handled by Smith this season, Smith wins 23 more faceoffs than Holland (169 versus 146). You do not capture any part of the extra 2% with a below-replacement level player (Smith) who wins 23 more faceoffs than a more serviceable player (Holland).

        As Jeff stated in a comment, if you are only on the team because you can win faceoffs, you are not going to help the team very much. This point is consistent with the points made by Dom in his article.

        • LukeDaDrifter

          The reality is the four guys Dom identified as being problem players for the Leafs have the best GA60 ratings on the team. Dom’s proposal to make the Leafs better by playing Marincin, Corrado and Holland would in fact remove the four players that have the best goals against rates on the team and replace them with some of the worst GA/60.

          My information comes from David Johnson who Dubas and others say is one of the smartest analytic guys around if not the smartest.

          I read the article you refereed to. It concludes you will only gain about one goal extra by winning faceoffs. This simply is not true. This year already after watching the Leafs play only 31 games I have seen them score about 5 or 6 goals in aprox. 10 seconds after winning the draw on the powerplay. I have also seen other teams do the same thing to us often enough. This tells me his model is imperfect. Theoretically it must be missing some important data or not using the correct data.

          • LukeDaDrifter

            Just because I don’t necessarily agree with someone’s conclusions, it doesn’t mean I don’t find them interesting and worth while thinking about. I never rule out they could be right and I could be wrong. Usually mine and theirs are never 100% wrong anyways. When analyzing the finer points of hockey all is not black and white. There are still are lot of grey areas and a lot of exceptions to the rule.

          • Bob Canuck

            Your claim that the Leafs have scored 5 or 6 power plays within 10 seconds of winning a draw seemed awfully high to me. So I decided to check your statement. I used hockey-reference.com/boxscores and NHL.com‘s play-by-play feature and checked every power play goal that the Leafs have scored to date this season; there have been 17 Leafs power play goals so far.

            The Leafs have scored exactly 1 power play goal within 10 seconds after winning an offensive zone faceoff. It occurred in the Arizona game, the most recent game in which they scored a power play goal. Talk about recency bias!

            This is a perfect example of how unreliable memory can be and why data is so important.

        • LukeDaDrifter

          The heart of Dom’s article was a 2% improvement can be the difference between an average team and a consistent winner. Even if Dom was incorrect in the example he presented, with parity a real thing in the NHL, that 2% is something worth looking for.

  • espo

    David Steckel won a career avg of 58.3% of his faceoff draws, at which point he went back to being a useless liability while his team may or may not have won the next puck battle.

    Which also means 41.7% of the time he was just a pylon while his team didn’t have possession, slowly getting in the way of whoever was directly closest to him, either team really.

    winning a faceoff is all well and good until you can’t contribute at all afterwards