Assigning goal value to Toronto’s faceoff takers

USATSI_9795825
Photo Credit: Dan Hamilton/USA TODAY SPORTS

Shortly after I woke up this morning, I stumbled upon a tweet from Empirical Sports co-founder Michael Schuckers, referencing an analysis that he and a few peers had done years prior on the statistical value of a faceoff in the National Hockey League. Their finding was that you needed an approximate faceoff differential of 76.5 for your talents at the dot to be worth a goal differential of 1. They also broke it down into even strength and special teams, and into specific zones.

The final findings were summed up in this table:

C4j2cdiWYAECfnJ

This had me intrigued as to the impact that Toronto’s faceoff takers had, both in the positives and negatives. So I made a rudimentary Faceoffs Above Replacement model. By model, I dropped a few formulas into Excel, and by replacement, I mean the 50% league average rather than actual Replacement-Tier talent, but “Faceoffs Above Replacement Model” sounds academic and cool.

What I did was extremely simple: I broke Toronto’s seven players who had taken 100+ faceoffs this year into Even Strength and Special Teams results. There’s a lot of room for refinements here: Schuckers’ data might have (very slightly) varying results if it used a more recent data set, and, as I was having trouble finding aligned zone breakdowns for draws won/taken (a player getting thrown out of the draw messes with using Zone Starts), I focused just on the strengths. That means these numbers are inexact, but given that Toronto isn’t very far off from the average in zone starts at Even Strength, on the Powerplay, or on the Penalty Kill, this will still bring us to our unscientific ballpark. Just keep in mind that all numbers are approximated, and exist to give a ballpark estimate on value over an average player in the same situation.

Here’s what I found.

Tyler Bozak

Bozak is Toronto’s most frequent faceoff taker, as he has been for most of the years he’s been here. That’s because he’s good at it and mostly keeps up as a player as well; this year, he’s 56.61% at the dot and pivots a line with James Van Riemsdyk and Mitch Marner.

Compared to having a 50% draw-taker take his place, Bozak has gained the Leafs approximately 0.97 goals at even strength, 0.37 goals on special teams, and 1.34 goals total out of 862 faceoffs. For every 100 draws he’s taken, he has earned the team approximately 0.16 goals, and requires about 643 faceoffs to gain an extra goal over the 50% draw taker.

Nazem Kadri

Kadri has become Toronto’s shutdown centreman, tasked with neutralizing some of the league’s best forwards, along with Leo Komarov and a seemingly rotating winger (currently, it’s William Nylander). Mike Babcock has expressed that he’d like Kadri to improve upon his poor career performance at the dot, though, and at 45.91% this year, that’s pretty understandable. Having initial possession, after all, means less time spent having to shut those opponents down.

So far this year, Kadri has “cost” the Leafs 0.63 goals against at the dot over 771 draws or 0.08 goals against for every 100 taken. It would take Kadri approximately 1275 faceoffs to give up a goal more than a 50% faceoff man would in the same situation.

Auston Matthews

Young Auston has taken the city by storm, but he hasn’t taken the circle over just yet. This year, he’s pitching just 45.79% at the dot, doing slightly better on the powerplay than he is at even strength. The team doesn’t seem too phased, though, because faceoffs are just one form of puck battling and, well, have you seen that kid along the boards?

Even still, it’s worth noting where he stands, and where he stands is right around Kadri’s range. In 725 faceoffs, his 45.79% rate has theoretically cost the Leafs approximately 0.64 goals over a 50% faceoff man. That comes out to 0.09 goals per 100 faceoffs, or a goal every 1125 draws.

Ben Smith

This is the one that I’m sure most have been waiting for. Why? Well, if you believe the quotes, faceoffs are why Smith has his job with the Leafs to begin with. He was hyped up as the right-handed key to getting the puck out of the defensive zone as soon as possible on the penalty kill, and…

He hasn’t done super brilliantly, in theory. He’s won about 49.56 of his draws on the penalty kill (1 fewer than 50%). However, it’s worth noting that penalty killing teams tend to be the underdogs at the dot, as they have fewer bodies in tie-up situations. Still, though, his net gains, while existent, aren’t massive. over 355 draws, Smith’s performance would typically gain the Leafs 0.19 goals. That’s equivalent to 0.05 goals per 100 faceoffs, or a goal every 1900 draws. It’s no surprise that Smith, at 52.68% on the year, carries the second highest benefit of the group, but it still isn’t much of one.

The Others

I looked at three others, using the threshold of 100+ draws. 

Frederik Gauthier took 207 faceoffs during his call up here, and while he was unreal at even strength (55.3%), going 43.48% on the PK actually led to him being a slight net negative, at a value of -0.08 goals, or a theoretical goal against every 2516 draws in comparison to someone who goes 50% across the board.

Leo Komarov takes most of his draws on the penalty kill, in the event his unit needs a leftie, and he’s also a net negative. This year, he sits at -0.26 goals earned by Faceoff Wins over a 50% taker thanks to his 44.8% total, or -0.21 goals per 100 draws, or a goal against approximately every 482 draws.

William Nylander sits in a similar boat, though most of his faceoffs have come at even strength. He’s a putrid 40.57% at the circle, which probably has a lot to do with Babcock’s decision that he will play the year out as a Right Winger instead. In just 106 draws, his performance is equivalent to a cost of -0.23 goals; or a goal against every 468 faceoffs compared to the 50% mystery player.

Application


Naturally, practical deployment of players at the dot isn’t exactly cut and dry or a weighted coin toss. There are factors that will influence the potential result sometimes; a size/handedness mismatch, for example. Not to mention, a great faceoff taker will probably add a couple percentage points to the spreadsheet if he’s always facing bad players.

With this in mind, you’ll find yourself in situations where you want to give yourself a bit more of a shot at winning, and you’ll send out the best-suited player out of what you’ve got. Occasionally, you’ll even get that goal you wanted, and you’ll sit back and be happy you sent the right player out there, and that the player was capable.

But save for hot streaks, it’s rare that the results deviate significantly from the above patterns. We’re still at a point where the best specialists in the league couldn’t be reliably bet upon to win two or three draws in a row, let alone be a safe bet to win every time you need them to. 

Usually, that’s nothing to stress over, because the winning team will pass around the puck a bit and maybe attempt a shot before possession goes the other way again, and you’ll forget abut it. But when it does directly lead to a goal, it feels super, super important. It becomes the broadcast talking point, it becomes the subject of the next social media fight, it gets further cemented into the team’s list of things to work on.

Which is great, in the right context. Working with your centres for an extra few minutes during practice to get them a little better at it? Sure. Get them in the video room to watch their upcoming opponents’ habits? Definitely. Try to find optimize the matchups of the players you happen to have dressed? Absolutely. Work on how the other two to four skaters react after the draw is won or lost, to ensure maintenance or regaining of possession? That would be massive.

But when deciding who should be in the lineup, or whether a player should be acquired at the expense of another, or whether it’s worth giving a nice free agent contract to the player with the best numbers in the circle available, you have to ask yourself; is the gap between bad and good at this role enough to make them a more impactful overall player than the other option? Is it worth the asset and/or financial cost of making a replacement, or can you acquire a bigger goal impact in a more efficient way?

As it stands, we put a gigantic premium on being 5% more likely to be in a position to have a 1 in 76.5 chance of being directly involved in a goal. That’s where the issue lies. While teams should be doing whatever they can in-house to make their players better in every way (I’m all in favour of giving a Yanic Perreault or a David Steckel a truckful of non-cap money to be an official Faceoff Coach), there are probably more efficient ways to acquire a small handful of potential goals above the average than chasing faceoff specialists; especially if they’re costing you in other parts of the ice.

  • Harte of a Lion

    The one thing everyone seems to have overlooked, and I apologize if I missed it is that on many faceoffs (there was no mention of whether a draw was won cleanly or from the ensuing scramble) is that unless a faceoff is won cleanly, it is the wingers and possibly a d-man who determines who ultimately gains possession.

  • Brad M.

    I like this kind of post because the comments become littered with classic cognitive biases.

    Nobody is saying faceoffs don’t matter.
    But over a proper sample size, statistically speaking, the importance of faceoffs definitely appears to have been overstated.

    The way I look at it as this:
    People get hung up on Bozak’s 55% to Matthews’ 45%, Babcock included, but that’s basically 33 draws won to 27 in a typical game. 6 more.
    How many of those 6 can be directly linked to a shot on goal and scoring chance? Well, considering those 6 could have come from any zone, and even a d-zone loss is often blocked and cleared, I’d say 1. Despite what Hughson and Simpson tell you, most lost draws do not cause a shot against.
    And, of course, the goalie should be stopping at least 91.5% of those , leaving a goal deficit of .085 in that game, or 7 in a season.

    Those 7 goals matter, sure. Just not as much as people think. Keep in mind that the faceoff loss is not necessarily the only component that preceded those goals. Many other variables are in play between the loss at the dot and the shot entering the net.

  • FlareKnight

    Well it has been a while since Jeff’s last attempt to make a war on faceoffs.

    Starting out with the puck is really darn important and useful in every situation. Get over it.

  • Glen

    Yada Yada yada you need to have the puck to score or get the puck out of your own end. You have to have somebody to score or set up a play when you get the puck. The leafs struggles at the dot is one piece of their defensive problems. They need to get better period. If Kadri is their shutdown guy he needs to start with the puck more often.

  • tealeaves

    Faceoffs are important at the end of game when you have pulled your goalie and need to score as well as on special teams. One of the reasons Nylander is on the wing rather then playing center ahead of Smith is just that. And there is no amount of math you can produce that will convince Babcock otherwise.

    • LukeDaDrifter

      Not only Babcock. You can include every professional hockey coach in the world. As for Nylander on the wing, Marner is another example of a natural center on the wing because of his poor faceoff percentage. There was an attempt to move Kadri to the wing after coming out of junior but he was unable to adjust to the change of position.

  • Bob Canuck

    It is obvious that we are not all going to agree on how important winning a face off is or isn’t. But can we at least agree that the face-off-winning-percentage stat (“FOW%”) is not very informative?

    As the chart in the article demonstrates, goal differentials vary depending upon which zone the face off occurs and also the situation (even strength, power play, or short-handed). The FOW% stat used by broadcasters makes no distinction between where the face off occurs or the situation; it is a stat that is not very instructive.

  • LukeDaDrifter

    Using the exact same flawed logic you have used here, one could also say ” Taking shots on the net could be a waste of time”.

    The last game we played we took 44 shots and produced 1 goal. At that rate, to win that game we would have had to register 176 shots on goal to score the 4 goals we needed to win the game.

    This article needs to be filed under “Bogus Claims and False Solutions”

    • Jeff Veillette

      Gary, with all due respect, this is some next-level warping. There’s a gigantic difference between what happens in one game and what happens in a data set of multiple full seasons from the entire league. A single game is full outlier moments to patterns, but if you play that game, say, 4000-5000 times, you’re going to have an idea of what to expect on the average night.

      I can pull a one game sample and tell you that 200 foot slapshots lead to goals. Or I can combine it with the 10,000 games that have happened since and tell you that Vesa Toskala was just unlucky.

      And that’s what you need to aim for. You can’t build a team around “oh, anything can happen so we better be ready for anything”, because you will never be able to do so. Sure, there are nights you can look at the highlights and say “hey, look, three goals were scored off the draw,” but don’t you think the fact those only happen a few times a year across the league (leading to these arguments) means that they’re unpredictable?

      You don’t know when the wave comes most of the time in hockey. In full games, in single plays, whatever. Single games are ridiculously unpredictable. But the big picture isn’t, and you can prepare for the big picture.

      The big picture tells me that, given the long-term odds of a player getting a goal off a faceoff, investing too much of my time, money, and assets to find a player that’s 2-3% better at them isn’t worth the aggrevation, even if I benefit or get burned from a random luck of the draw night or two. Just like the big picture tells me that on your average night, the Leafs would’ve popped in 3-5 goals against the Sabres, so I’m not going to lose my mind due to it not working out this time.

      Coaches can coach one night at a time, players can play one night at a time, but at the end of the day, when building a team, you build for the roster that gets you through October to April-June, not through the evening.

      • LukeDaDrifter

        You may see the idea of running a successful top four format come to pass next year. The high scoring prospects we have, could very well take over that fourth line. What effect on the importance of winning faceoffs remains to be seen. As you say the extra goals would out weigh the goals generated from faceoff wins. What effect it would have on goals against would be my concern. If Babcock decides to go that way do you really want to see Bozak traded?

      • LukeDaDrifter

        Your article is assuming that the only value in winning a face-off is scoring a goal. With 40 second shifts getting possession of the puck off the draw is likely more important than simply scoring. It leads to both scoring chances and greatly reduces the other teams ability to score on you. Putting the other team constantly on the defensive eventually drains their energy. This in itself changes the momentum of the game, leading to a goal in your favour at some point down the road. Even the goalie knows that if he has a great faceoff man he will freeze the puck to have a better chance of getting it out of his zone.

        The reason faceoff percentages tend to balance out over a long period is because every NHL coach attempts to have on his team a player/players who is great on the draw. He ends up facing off against the other teams best guy. The end result is typically 50 percent. If I use weight training as an example, statistically I would find there is no advantage having all your players lift weights. The fact that every team players lift weights and become strong cancels out any advantage.

        Your conclusion that replacing Smith with a better offensively minded center could have some merit. In the end you may be able to generate more goals from that center. I think Buffalo uses that strategy by having Ryan O’Reilly take over 50% of their draws, often leaving the ice immediately after taking the draw. Even there the coach still knows that winning faceoffs is very important.

        If you are playing Pittsburgh with two minutes left in the game. You are up by one goal. The faceoff is in your end of the ice. Mike Sullivan puts out Malkin – Crosby – Honrquist – Kessel – Letang. Who are you going to send out to take the draw?

        • LukeWarmWater

          Brother Luke I recall a couple games in the last little while where the opponent just owned the face off circle in the early part of the game as they controlled the puck in the leafs zone. It set the tempo for the whole game. As you astutely point out whether on the power play or killing a penalty you lose that face off and quite often you blow 20 seconds on chasing after the puck in your own zone to reset up the power play or face the opposition having control of the puck in your zone much of the penalty kill.

          Bottom line late in a close game I want a team that is in the top five on winning face offs.

          • Bob Canuck

            The top 5 teams in FOW% (1 to 5): Anaheim; Colorado; Boston; Detroit; and Carolina.

            You really want the NHL’s worst team (Colorado), with the second worst CF% 5v5 (Score, Zone and Venue Adjusted), to protect a lead late in a close game? Really?

        • DukesRocks

          All sports have metrics that show value of a player beyond the obvious (goals. assist, runs, buckets TD etc). Some of these metrics like War in baseball and Corsi in hockey have value to further quantify a players’s offensive and defensive game. While I agree some of these advanced stats have merit, some are a reach to say the least. The bottom line is if you win faceoffs, it will drive possession, thus resulting in goals. I don’t need a percentage to tell me if I win more facesoffs, possession will be the result. The metric is a waste because faceoffs are already recorded for NHL players, thus telling everybody if player X helps drive team possession numbers.

          • LukeWarmWater

            Dukes it is out of control as in baseball they will actually discuss picking up a player as according to their deeper thinking the guy will be the reason for an extra plus 5 wins. Really, if you are that psychic become the head bookie of the largest Vegas gambling casinos. You could make a fortune being the new Nastradamus of 2017.

          • DukesRocks

            I get (save a run), when you have a fielder that robs a player of a hit or HR with a crazy catch, like our boy Pillar. But I’m with you when they try an push the envelope to the absurd. By the way Serge Ibaka, yeah baby… we needed that stretch PF.

          • The China Wall

            Dukes, I think you make the point better than I was going to make it but I will have my say anyway.

            What I don’t understand is this.

            The numbers the advanced stats crowd live and die by tend to be focused around possession. Possession for, possession against, etc.

            Yet an event (face-off) which directly determines who starts with possession after a stoppage in play, has no value? I just don’t get it…

            Having said that, I would agree that I would not advocate for the Leafs to go out and spend some assets to acquire a player who has a slightly better success rate taking face-offs, BUT I would whole-heartedly get behind the idea of exhausting whatever resources are necessary to improve the face-off proficiency of the players on the Leafs roster.

            Hell I would go down to the Mastercard Centre myself and drop pucks into the dot until the last player was exhausted and left the ice if that would help improve the Leafs ability to win face-offs.

          • Kanuunankuula

            This is from my memory of reading an article on this: The possession advantage of a faceoff win lasts for around 3-4 seconds after the draw. You don’t get a long term possession from it. Other factors come in to play and nullify the effect of the faceoff win.

            If you really think about it, it makes sense. Faceoff win just gives you initial possession. It does not help you keep it, you still have to be a good possession team to keep it.

          • DukesRocks

            China I’m with you, I don’t get it either. In reference to faceoffs, I’ve seen Nylander,Bozak, Kadri and others practice faceoff with each other on Leaf’s TV. So it’s safe to say they do practice this skill and I’m sure they might employ an expert on the matter. If they don’t have an expert to tutor the Leafs budding stars. I would suggest they do so to fast track their growth. The spotlight seems to be focused mainly on Kadri due to his age and rightfully so if he wants to be that shut down 2 way forward.

          • Kanuunankuula

            This leads to interesting dilemma. If the faceoff is so imperative to possession, should possession stats reflect that? I would argue that since the game is so fluid, that the one set piece impacts only a few seconds after it, once that’s up it’s anybodys game again.

            What I’m trying to say, is if Smith etc. faceoff specialist (or use Bozak as a example) is giving you that extra possession with his faceoffs, how come it’s not registering anywhere on the advanced stats?

          • Ben

            It would show up in CA/60, if anywhere. And there is numbers-driven evidence that CA/60 has more value than CF% in holding leads, especially as you get closer to the end of the game.

          • DukesRocks

            Good question, however since I’m not a Corsi expert I can’t anwer it, but logically winning faceoffs should help possession numbers, since you aren’t chasing the puck. Maybe Jeff could take the time to answer the question.

      • DukesRocks

        Jeff I see your argument and I also see Luke’s point. Generally decisions are based on numbers. Some numbers effect the bottom line better than others. I remember reviewing a report where I had to look at 1000 accounts to find 1 fraud account. I remember telling my boss this report is garbage due to the amount of resources (time and energy) spent on putting the data together and reviewing it. We concluded that a good report false/positive should be around 25-1, otherwise you’re trying to find a needle in a haystack. The same thing applies here to what you and Luke are saying. That at the end of the day, this faceoff data doesn’t really move the needle in making the team better.

  • G2

    It doesn’t matter what you say, the mopes on HNIC are still gonna go on and on about “he’s only 3 for 11 tonight and it’s killing them”. So thanks for all the work but…