Size Has No Impact On Faceoff Percentage: 2009-10 Data

 

A question that has come up a few times is whether big players tend to have an advantage when taking faceoffs. There is a certain logic to the idea that they do: after all, bigger, stronger players should be able to outmuscle their smaller counterparts in the faceoff circle.

The data, however, suggests something else entirely: that there is no advantage to being big when it comes to taking face-offs.

Advertisement - Continue Reading Below

I went back to the 2009-10 season, and took every NHL centre who had taken 100+ even-strength face-offs. Then I plotted three charts, one comparing even-strength faceoffs to height, another to weight, and a third to “size” – just the product of height and weight. I used even-strength faceoffs only to make for a fair comparison, as it’s easier to win faceoffs while on the man advantage and harder while killing penalties.

Face-offs vs. Height

Correlation: -0.02 (Scale of -1 to 1)

Face-offs vs. Weight

Advertisement - Continue Reading Below

Correlation: 0.03 (Scale of -1 to 1)

Face-offs vs. Size

Correlation: 0.01 (Scale of -1 to 1)

Conclusion

There is no noticeable advantage or disadvantage granted by size in the faceoff circle. The best faceoff men in the game last season varied from small (Scott Nichol, Kyle Wellwood, Vladimir Sobotka, Todd Marchant) to large (David Steckel, Wayne Primeau, Paul Gaustad, Vincent Lecavalier) and the worst faceoff men varied from small (Oscar Moller, Andrew Cogliano, Daymond Langkow, Toby Petersen) to large (Brian Boyle, Eric Staal, Michael Rupp, Nik Antropov).

It’s one of the few areas in the game where the playing field is relatively level.


  • Horcsky

    Funny, I guess the biggest impact on faceoffs this has is any psychological advantage gained when guys are apprehensive about their ability to beat big guys. Oh well, not the FIST time stats have shown us the light.

  • Jiri Dopita

    This isn’t too suprising to me. Faceoffs have more to do with timing. I think the advantage of size comes in when you are trying to scramble the draw and tie up the opponent.

    Maybe it’s just that I’m a small guy myself but I think in general size is overrated in hockey. being successful has more to do with decision making. It’s definatly not a liability to be big. It’s just overrated. I think small guys get passed over in the draft just because of size which is a mistake.

    • SmellOfVictory

      I would argue that size is often overrated, but that smaller players need to be better in order to be successful. You can have a 6’2, 200lb, average-skilled, average skater with average decision-making who does reasonably well as a 3rd liner, for example. You take that same set of traits and put it inside a 5’9, 160 lb body and the guy isn’t going to do so well.

      Basically, my thinking is that size can help cover for a lot of mediocrity or deficiencies in a player’s game. If they’re going to be even moderately successful and small, they have to be some combination of smart/fast/agile/skilled.

  • Jiri Dopita

    Yanic Perrault was a perennial 65% FO man and he’s about 5’10 190lbs. A good faceoff man has technique and extremely quick hands.

    btw. Ryan O’Marra seems to have found his niche finally. It seems he’s read all the scouting reports on what type of pro he was projected to be and has embraced the role. Could become a decent bottom 6 forward for Edmonton

  • Peca was a demon on the dot.

    Not a large man by any means but effective none the less.

    Why not dust the guy off,strap some scates to him and make every player spend an hour a day with the lil fellow?

    Worked for Stoll.

  • wiggs22

    Now we have it…fast hands which we know cogs does not have…

    we always talk about how fast he is and how he speeds by defenders but that his hands are too slow and he looses the puck and breaks up the play…slow hands!!

    I sir wiggs have finally uncovered the truth behind cogs and his woeful face off powers!

    ” goes back into hiding in closet untill next game”

  • Jiri Dopita

    Interesting article. Quite counter-intuitive. I agree with some of the comments about quickness being the main factor in determining face-offs wins, but you’d think that a bigger player with quick hands would have an advantage over a smaller player with quick hands owing to his superior strength. Maybe smaller players tend to focus more on their technique knowing that they can’t rely as heavily on strength. Who knows!

    Jonathan – A bit off topic, but do you have a preferred website and/or software for your scatter plots (and other graphs) and calculating correlation coefficients? I need to do some really basic stuff for work and school, but am finding that a lot of the web-based tools are far more complex than I need.

    Thanks.

  • @ Coco crisp:

    Is your indifference a product of your lack of intellect (i.e. inability to grasp charts) or your lack of intellect (i.e. inability to see why the non-relationship between size and faceoffs might be relevant on a hockey website)?

    Please consider the options carefully before answering.

  • a lg dubl dubl

    Good read JW, and merry christmas!

    Ive been sayin for awhile that a centerman doesn’t need to be a 6′ + behemoth im sure it does help in the scrums, but if he posesses the smarts and quickness to be effective whos to say he cant be a #1c, i.e Gagner. Imo id rather have the bigger players on the wings to open the ice for the “smaller” centermen. Besides if you think about it Gagner isnt really that short at 5’10, im 5’5 i think hes tall.

    Merry christmas bloggers!

  • The data from 2009-10 might suggest size has no bearing, but a glance at the top 10 from this season, as of the Christmas break, says something entirely different.

    1 Steckel 63.6 6’5″ 217

    2. Malhotra 62.8 6’2″ 220

    3 Gaustad 60.6 6’5″ 212

    4 Konopka 60.2 6’0″ 210

    5 Stoll 60.0 6’1″ 210

    6 Toews 59.6 6’2″ 210

    7 Kesler 59.0 6’2″ 202

    8 Smithson 58.1 6’3″ 206

    9 Zajac 57.8 6’3″ 200

    10 Halpern 57.4 5’11” 198

    The premise that size doesn’t matter might fit with the year you’ve selected, but it certainly doesn’t stand up regarding players at the top of the heap so far this season.

  • @ Robin Brownlee:

    11. Rich Peverley, 6′ 195

    12. Antoine Vermette, 6’1″ 199

    13. Jason Spezza, 6’3″, 216

    14. Pavel Datsyuk, 5’11”, 194

    15. Sidney Crosby, 5’11”, 200

    16. Steve Ott, 6′, 192

    17. Nate Thompson, 6′ 210

    18. Marty Reasoner, 6’1″ 205

    19. Patrice Bergeron, 6’2″, 194

    20. Mark Letestu, 5’11”, 195lbs

  • @ Robin Brownlee:

    Interestingly, here are a few names from the bottom 10:

    3. Brandon Sutter, 6’3″, 183

    4. Artem Anisimov, 6’4″, 200

    7. Eric Staal, 6’4″, 205

    9. David Backes, 6’3″, 225

    10. Patrik Berglund, 6’4″, 218

    There are more players 6’3″ or taller in the bottom 10 than in the top 10. I’d suggest a full review of the data would show the same thing in 2010-11 as it did last season.

  • OB1 Team Yakopov - F.S.T.N.F

    08/09 “small” players in the top 10:

    2. Draper 5’11 190

    8. Pavelski 5’11 195

    9. Datsyuk 5’11 197

    07/08

    1. Scott Nichol 5’8 170

    3. Draper 5’11 190

    8. Mike Sillinger 5’11 200

    9. Boyd Gordon 6 195

    06/07

    1. Yannic Perreault 5’11 190

    3. Sillinger 5’11 190

    4. Drury 5’10 195

    6. Draper 5’11 190

    7. Zigomanis 6 195

    8. Datsyuk 5’11 197

    10. Mcdonald 5’10 185

    • Average size is six-foot-one and 205 pounds.

      From James Mirtle: “NHL players have gone from an average of 5-foot-9 and 172 pounds in 1920 to 5-foot-11, 180 by 1955, 6-foot and 190 in 1980 and then, finally, up to the 6-foot-1, 205 pounds hockey players have averaged the last 15 years or so.”

      You can put “small” in quote marks, but based on the average, a lot of the guys you’ve listed are closer to being average in height, weight or both as opposed to being small. I’d suggest that over the past several seasons, average and larger than average players haved fared better than small players.

      In any case, that doesn’t change the make-up of the top 10 face-off men through the Christmas break.

      • OB1 Team Yakopov - F.S.T.N.F

        Ya, the definition of small is obviously up for debate. But if you were looking for FO guys sub 5’10 and/or 190lbs I doubt you will find many, simply because theirs probably not more then 10% – 15% of the NHL (if that) that dimension or smaller (any position)

      • SmellOfVictory

        At a glance I’d say his mean (certainly mode) is around 5’11/190. Those extra 2″ and 15 lbs make a big enough difference to categorize those guys as being smaller, I’d say.

        To give you an idea of height distribution in the average white male N.A. population, someone who is 6’4 (only 5″ taller than average) is in the 95th percentile. Quick and dirty, that would put the standard deviation in the population at around 2-2.5″.

        I’m making assumptions here, but I don’t think it’s unreasonable to say that the distribution of height among hockey players is going to be tighter than the average population (you won’t have guys clocking in at 5’3, nor at 7’2), so I’m going to assume the SD is 2″ or less. This would put the guys who are 5’11 at least one standard deviation outside the mean, which I think would warrant being considered in a different size category.

        • OB1 Team Yakopov - F.S.T.N.F

          Exactly, so if 5’11 190 isn’t small then 6’3 220 isn’t big either (assuming 6’1 205 is average)

          So using the list of this years top 10, you’d have 2 “big” guys using height and 0 “big” guys using weight.

  • Wanyes bastard child

    Excellent work Jonathan. Good stuff.

    In Edmonton, we’ve all got it in our heads that without a Joel Otto we’re sunk. But a Todd Marchant would do.

  • Horcsky

    @ Brownlee and Willis

    Any idea what the average player height and weight is this year? I’m guessing it’s close to Stoll’s size (6’1” 210lbs), but maybe a bit lighter than 210lbs.

    • SmellOfVictory

      Good point the stats are skewed as players, especially down the middle, are getting larger as teams are trying to get bigger there. These days many of the small young players that come into the league as centers are placed on the wing to adapt to the pros. Ex. Seguin and Eberle to name two.

      It’s not so much the size that matters so much as more and more centers are getting larger. Not to mention the league average in height and weight keeps going up as time goes on.

  • @ Robin Brownlee:

    I’d be surprised if there’s any significant change in the overall data from last season to this season. Very surprised.

    But as you might imagine, it takes a fair bit of work for me to go and get this data manually, so I’ll wait until the end of the season to run this year’s numbers. That will increase the number of players with 100+ EV faceoffs and make the data more accurate anyway.

  • Guys, keep tweaking this if you must. Big players dominate the top 10 in face-offs in the NHL this season through the Christmas break, which begins for me right now.

    Merry Christmas to Wanye, Bingofuel, Lowetide, Amber, Willis, Gregor and all the people who take time to read and contribute here.

    • book¡e

      The top ten are bIg relative to the size of the average guy on the street, but the top ten and just slightly above avg in terms of NHLers. They certainly do nothing to discredit the argument that JW just put up.

    • These are the attributes of the analysis presented by Jonathan that make its conclusions reliable, reputable, and accurate:

      1. A clear and unambiguous test question.
      (We actually have a 3-part test question.)
      –> How much do differences in a player’s height, weight, or size make to face-off success?
      [The question is NOT: “How heavy or tall are the people who get the best FO%?”; nor, “How can we explain this year’s FO leaders?”, nor “How does age affect FOs?, nor “What about a playoff team vs a lottery team?”.]

      2. Clearly stated assumptions, definitions, and limitations.
      (These serve to eliminate potential ambiguities and to make sure the question can be tested saliently.)
      –> height is in inches, weight is in pounds, size is in inches*pounds/100
      –> The data for the analysis is all 09/10 regular season EV FOs taken by players with 100+ such FOs.
      [That means the analysis is not about power plays or players who infrequently take draws. It also means that it is covering players most likely to be the coaches’ favorites for taking draws. That is a way of reducing the effect other variables might accidentally colour this analysis. Most importantly: when this analysis is complete we ought to be able to say something about whether size matters to players taking most of the draws, not just the best ones, nor the ones who are considered to poor to be sent on draws very often.]

      3. The sample size is statistically large enough that, if there is a correlation, it will be noticed. Conversely, uncorrelated means uncorrelated, with no ambiguity.
      –> The only way I can think of to describe adequacy of sample size is to ask that the reader imagine removing half of the dots (at random) from the scatter charts. Would the nearly horizontal slope line need to be tipped one way or the other as a result. Nope – there are no ambiguities created by not looking at enough data.
      [This is actually why Jonathan is very safe in his prediction of 2010 results — the sample size of his analysis is great enough that even if half the players are missing, or if another bunch of NHLers is analyzed, the conclusions will not change.]

      4. The analysis method remains true to the test question and the assumptions.
      –> Just plot the data on a simple graph. Voila.

      5. The conclusion is a direct answer to the test question, with no intervention on the part of the writer.
      –> “There is no noticeable advantage or disadvantage granted by size in the faceoff circle. . . It’s one of the few areas in the game where the playing field is relatively level.

      **** So with that in mind – can we all STOP worrying about how BIG Cogs and Gags are and somehow find a way to teach a very important skill that lots of players possess.

      This is going to sound stupid, but: The very best correlation that exists for future FO% success is past FO% success. Some guys just suck. Maybe they should never be allowed near the dot until they learn. But its not cuz they’re small.

      • freshpotofcoffey

        Are you a statistician?

        It is for these reasons that I strongly doubt Robin’s statements that things are different this year. The top 10 are a very small sample size. As we’ve seen in later posts, it wouldn’t be difficult to find years where the top 10 are small. Whereas I’m sure the data will suggest that there is no correlation, year after year.

        Unless we have some reason to believe that something fundamental has changed about faceoffs this year that would be creating this advantage. I’m no expert, but I daresay we don’t.

        • Hi Fresh Pot,

          I’m an Electrical Engineer. I have 5 years of university plus 20 years experience as a manager who regularly parses between good analysis, incomplete analysis, and crap. I have studied statistics in the past and I know about numbers, generally.

          The problems with Robin’s statements do not merely lie in the small sample size of 10, they are that Robin, and some others, keep moving the issue around, and keep disallowing various contrary data points by altering the question, the constraints, the method, and their original conclusions.

          This happens over and over again, regarding issues such as faceoffs, fighting, “intangibles” (Strudwick), goalies, and so on. I am positive that the mistakes are not intentional, so I tried to describe Jonathan’s analysis in layman’s terms. Hopefully we can come to understand that one reason statistical analysis is performed is that it takes the bias and other human limitations out of the issue.

          If people believe that they are being lied to with statistics, it ought to mean that they have spotted a relevant fallacy in the analysis performed. That possibility can be tested and conclusions based on the new test can be made. For example, someone can tell you that global ocean temperatures are 1 deg. C more than in 1970. And, that humans are creating 3x as much CO2 each year since 1970. Those numbers are based on statistical analysis — no one was “there” measuring “everything”. To know whether they are true, we would have to learn exactly how the numbers were arrived at, or have a deep seated trust in the people telling us the information. The conclusion that could come up next is “therefore we need to cut CO2 output in Alberta by 30% (or whatever) or we’re all going to die and the people in Zambia will be very upset so we better start shipping our money to the government by the trainload”. The lie is less likely to be in the statistics and more likely to be in the interpretation or untested conclusions.

  • SmellOfVictory

    I’m not a statistician, but a scatter plot does not seem to be the best method of analysis. It seems a regression analysis would be more suitable. That will allow you to demonstrate any corrolation between height, weight and face off %. Instead of each one independently.

  • SmellOfVictory

    @JW – I appreciate the work that you put into posts like this and find them interesting. Thanks.

    @Shingin – Along the lines of your suggestion, I would love to see someone look at whether centermen can show improvement over time and if so, how long does it take (ie: are Cogs/Fraser/Gagner etc at all likely to improve on the dot, or is what you see what you get – for eternity.)

    I know that this would be a back-breaker to compile, but in terms of the Oilers situation, I think that this would be highly relevant with regards to projected future value of these players.