Hogging the Puck: Who's Doing It, and Who Should Be Doing It

Ben Wendorf
November 06 2012 02:36PM


Still building on the infant stages of my fantasy predictor work, I was looking across forwards with 20+ games the last five years and decided to add a metric where I took the shots attempted (technically, it would be like "Fenwick attempted"; shots plus missed shots) while a player was on the ice and determined what percentage of those shots were attempted by the player. This percentage of attempted shots (%AttSh) was my way of saying, "Okay, I don't care what line you're on, or where you start on the ice...how many of your line's shots did you take?" At the time, I was more interested in seeing if it was a stable metric for fantasy hockey prediction; if it was, I could consider it a marker of player behavior, or talent, and use it to predict how many shots a player might take. I was also interested in labeling people "puck hogs."

Full disclosure: I've played with puck hogs often in my life, and I've always been okay with setting people up (even though I have a pretty good shot). If you've ever played in an adult league, chances are you've played with a puck hog, and if you think you haven't - YOU ARE IT. So there.

Anyway, I was surprised to find that it actually was a pretty stable indicator of human behavior, and not much of an indicator of talent, making it an important adjustment for fantasy hockey but also fun to use for pointing fingers. So...who's hogging the puck?

Puck-hoggery as an inherent trait

First, I should say why I've suggested that it seems like a behavioral indicator. For one, I ran a year-to-year correlation across this forward population (n=1,472) and found that %AttSh was quite highly repeatable from one year to the next, with r = 0.74. In a picture:

Yr to Yr Corr of AttSh Percentage

Note: Excuse the non-purposeful swapping of the "%" sign up there.

Only two metrics I tested among this population had higher year-to-year correlation, and they were Attempted Shots/60 and 5v5 TOI. So maybe right now you're saying, "Well Ben, it's probably because good shooters take more of the shots." I'd respond that a.) this measure takes away the effect of being a top-line player, b.) the correlation of year-to-year shooting percentage is much lower (0.26), to the point of being not very predictable at all, and c.) here's the correlation for shooting percentage to %AttSh: -0.16. Meaning, it's nearly random, with a very, very slight lean to lower shooting percentages relating to lower %AttSh. But really too slight to suggest much of anything.

It's also important to look at %AttSh's relationship to talent a different way - does it fit player talent development and peaks? We already know that forward talent seems to develop/increase 18-24, peak at about 24-25, then slowly drop off.

Relationship of Age to AttSh Percentage

In fact, it doesn't seem to budge much at all until ages 30-31. No rise up to a peak, no substantial movement across a player's 20s, with some gradual decay in the 30s (our number of players drops down below fifty after age 35).

Well, maybe it has to do with your teammates? After testing the three Quality of Teammates metrics (+/- QoT, based on your linemates' plus-minus ratings, Corsi QoT, based on your linemates' Corsi ratings, and Corsi Rel QoT, based on your linemates' Corsi ratings controlled for team effects), I found correlations of 0.13, 0.06, and 0.05, respectively. Nothing to see there.

With high repeatability, and little connection to shooting talent, player development, or teammates, this seems much more like a behavioral activity.

Leading puck hogs

Reserving judgment for the time being, I took these forwards and looked over the last five years again. Taking all forwards with at least 3 years of 20+ GP from 2007-08 through 2011-12 (a population of 410 forwards), who was taking the higher percentage of the shots while on the ice?

Top 25

  1. Alexander Ovechkin - 37.9%
  2. David Clarkson - 36.7%
  3. Tyler Kennedy - 34.8%
  4. Shawn Thornton - 34.51%
  5. Rick Nash - 34.47%
  6. Jeff Carter - 34.3%
  7. Patric Hornqvist - 33.9%
  8. Matt Ellis - 33.5%
  9. Jed Ortmeyer - 33.43%
  10. Phil Kessel - 33.4%
  11. Cal Clutterbuck - 33.36%
  12. Evander Kane - 32.8%
  13. Ales Kotalik - 32.7%
  14. Viktor Stalberg - 32.6%
  15. Alexander Semin - 32.44%
  16. Jason Blake - 32.43%
  17. Zach Parise - 32.3%
  18. David Booth - 32%
  19. Jordin Tootoo - 31.9%
  20. Brian Boyle - 31.8%
  21. Ethan Moreau - 31.68%
  22. Mike Brown - 31.65%
  23. Marian Gaborik - 31.65%
  24. Scottie Upshall - 31.6%
  25. Sean Bergenheim - 31.52%

The average among the entire forward population was 24.4%, reflecting the percentage that our "Age vs %AttSh" seemed to revolve around.  So, the bottom guys?

Bottom 25

  1. Stephen Weiss - 18%
  2. Ville Leino - 17.94%
  3. Steven Reinprecht - 17.92%
  4. James Sheppard - 17.87%
  5. Todd White - 17.7%
  6. Brandon Prust - 17.2%
  7. Sergei Kostitsyn - 17.08%
  8. Jerred Smithson - 17.06%
  9. Tomas Holmstrom - 16.71%
  10. Alex Tanguay - 16.68%
  11. Todd Fedoruk - 16.55%
  12. Kyle Chipchura - 16.52%
  13. Saku Koivu - 16.51%
  14. Craig Conroy - 16.3%
  15. Ryan Johnson - 16.27%
  16. Matt Stajan - 16%
  17. Kyle Wellwood - 15.9%
  18. Josh Bailey - 15.8%
  19. Derek Boogaard - 15.7%
  20. Andrew Brunette - 15%
  21. Joe Thornton - 14.8%
  22. Henrik Sedin - 14.5%
  23. Tyler Bozak - 14.2%
  24. Georges Laraque - 11.4%
  25. Cal O'Reilly - 10.7%

Pretty fascinating lists, no? I mean, you have some unsurprising names up there, like Ovechkin, Nash, and Carter up top, but Clarkson actually had less variance than Ovechkin over his years! It's possible that this could be a method to better isolate 3rd and 4th liners that can drive offense in addition to providing defensive or truculent contributions.

SHOOOOOOOOOOOT

The question of who should be shooting a lot is a bit trickier; we know that just because your shooting percentage is high, it doesn't mean that you're taking a bunch of shots. So first, let's look at guys who have a higher shooting percentage than their teammates, but still have a low %AttSh. I took our 410-player sample and gave the players a rank for highest (1) to lowest (410) differential between personal shooting percentage and teammates' shooting percentage, then a separate rank for highest (1) to lowest (410) for %AttSh. I subtracted the former from the latter for an index number with the potential of ranging from -409 to +409. Higher in the positive means, based on your shooting ability compared to your teammates, maybe it would have been helpful if you shot more. Lower in the negative means that maybe the team would benefit if you shot a bit less.  So, your top 10 who could've stood to shoot a bit more:

Top 10

  1. Georges Laraque ... +406
  2. Kyle Wellwood ... +393
  3. Andrew Brunette ... +392
  4. Tyler Bozak ... +391
  5. Mathieu Perreault ... +378
  6. Jiri Hudler ... +369
  7. Rod Brind'Amour ... +362
  8. Sergei Kostitsyn ... +360
  9. Alex Tanguay ... +356
  10. Colin Wilson ... +355

Laraque is a goon, no question there, but when he did focus on playing, he could be an in-front-of-the-net garbage-goal-scorer, hence his 5v5 shooting percentage (10.1%). Keep in mind, though, that that percentage comes out of a pretty small shooting sample. As for his linemates? Perhaps they were a bit more, um, challenged (5.7%). Ew. Laraque being one of the more outspoken commentators on the steering of physical players towards goon hockey might also speak to him surmising what he could've been capable of doing offensively. I mean, he was no lithe, dynamic offensive player, but neither is Andrew Brunette.

Bottom 10

  1. Evander Kane ... -353
  2. Jason Chimera ... -353
  3. Mikael Backlund ... -353
  4. Jonathan Cheechoo ... -356
  5. Matt Ellis ... -360
  6. Jesse Winchester ... -361
  7. Mike Brown ... -362
  8. Henrik Zetterberg ... -367
  9. Patric Hornqvist ... -374
  10. Alexander Ovechkin ... -391

I think, by the end, we were all asking Cheechoo to stop shooting.

And finally, the guys who are/were taking pretty much the right amount of shots for their shooting abilities and teammate potential:

Middle 20

  1. Logan Couture ... +12
  2. Slava Kozlov ... +12
  3. David Backes ... +11
  4. Anze Kopitar ... +10
  5. Glen Metropolit ... +8
  6. Jarkko Ruutu ... +8
  7. Nick Spaling ... +8
  8. Alexei Kovalev ... +7
  9. Robert Nilsson ... +5
  10. Milan Michalek ... +4
  11. Paul Gaustad ... -6
  12. Sidney Crosby ... -7
  13. Ben Eager ... -8
  14. James Neal ... -9
  15. Taylor Pyatt ... -9
  16. Dany Heatley ... -9
  17. Martin Hanzal ... -9
  18. Todd White ... -9
  19. Joffrey Lupul ... -11
  20. Ryan Kesler ... -11

I suppose you could say there's a virtue to being a player like this; on the average, this is a list of pretty solid players, plus goons who knew they were terrible.

It's important to keep in mind that, as I'm suggesting, this could be more about on-ice behavior than most of the other factors. That said, one doesn't just flip a switch and change their behavior, particularly a behavior that's been reinforced by countless hours of practice and play. On the other hand, by reducing their own shotload and setting up teammates, some of these players might be doing their team some good.

Now if only someone could tell that to my jackass teammate...

417ea714751a4580f06253a09a79bab0
Benjamin Wendorf was co-manager of the SB Nation Winnipeg Jets blog Arctic Ice Hockey (formerly Behind the Net); he is currently in the PhD program in Africology at the University of Wisconsin at Milwaukee. You can follow his graph work on Tumblr or his tweets @BenjaminWendorf. He can also be reached at bwendorf84 AT gmail DOT com.
Avatar
#1 PRC.
November 06 2012, 03:13PM
Trash it!
0
trashes
Props
0
props

Good piece...

From a Nashville perspective, Hornqvist and Kostitsyn are interesting examples.

Is Kostitsyn's shooting ability really that much better than that of his linemates? He's got a career shooting percentage of 17.2%, but that's been buoyed by the last two seasons, where he has scored at a seemingly unsustainable rate.

Hornqvist on the other hand, may benefit from positional opportunity - he often plays a net-front role, meaning he'll get to rebounds more often than another player on the ice.

Avatar
#3 Eric T.
November 06 2012, 03:47PM
Trash it!
0
trashes
Props
0
props

I'd be interested in a scatter plot for your data at the end -- shooting percentage relative to teammates on the x-axis and percent of shot attempts on the y-axis.

I'm trying to do some thought experiments about what I'd expect that to look like...

If every player is perfectly interchangeable and efficient, obviously after a billion shots everyone lands at the same spot on that plot.

What happens to that plot when we limit players to a few hundred shots?

What happens to that plot when we give players defined roles as net-crasher or puck-handler (but keep each player efficient in his decisions about when to shoot)?

What happens to that plot when we give some players more shooting talent than others (but again, keeping them efficient)?

What happens to that plot when we introduce behavioral inefficiencies, with some players taking more or fewer shots than they should?

I'm curious to hear what people think would happen. I'm also going to fiddle with some modeling and see if I can come up with anything convincing.

Comments are closed for this article.