Hockey analytics: Does size really matter in the NHL?

[Sabres Fan in NS]

Like everything else in this life ... the size is not critical ... how you use it is.

That's what she said anyway.

[SwampD]

18 minutes ago, The Ghost of Doohickie said:

Like everything else in this life ... the size is not critical ... how you use it is.

That's what she said anyway.

She’s lying to you.

[triumph_communes]

2 hours ago, SwampD said:

Thanks Mr Wizard. I think the disconnect is that I’ve only ever heard the term selection bias, as it relates to the NHL, being that teams are more likely to draft older players.

 

Ah yes, that’s a different context altogether though it does loop back in. Does the NHL choose the develop older players or taller players and that’s the root cause?

Statisticians have looked at many things and the one that stands out the most is date of birth.  They made correlations over many different sports over the decades and found links that the older a player is for any age bracket (Jan or July depending on locale), the more likely they are to go pro. Ultimately they’ve tied it all back to the more physically mature players get more coaching attention.  The amount of coaching attention one gets throughout their career is the strongest indicator for success than anything else, and coaches will usually give more attention to those more physically mature.

This is the argument one always should have about players needing key minutes and they won’t develop while suffocating for 8 minutes a night.

So if the league selects (drafts) larger players then they’re going to have a better chance at development and it was a self fulfilling prophecy, not an indication on size itself.

 

Again the answer here I believe has changed with time as the barriers/privileges for success for smaller players has gone away over the years and the results speak for themselves. A relevant case study I think is with goalies. The league absolutely drafted tall goalies only, our very own GM calling goalie drafting voodoo magic so he left it to raw attributes and luck (and boy have we suffered from this).  Despite this trend, large goalies have turned out to bust more often than not and a lot of teams have been burned by this approach the last handful of years. Size and fundamentals can get goalies pretty far, but to compete in the NHL it can fall short wildly quick. The best goalies have the ability to track the puck and where it’s going next without seeing it, and they don’t need to be 6’7” to block that. A few more inches of height and Dell would not have made that save in the final seconds last night, he was off the mark by a few feet.  
 

So does size matter?  Yes. Does simply drafting someone because they’re bigger mean success?  No. Are there small towns full of 5’9” Martin St Louis’ out there grumpy that they didn’t get drafted and their height was the reason?  Nah, he’s an outlier and they always will exist. Look at Gretzky or Lemiux numbers sometime versus any other player of their respective time period and those numbers shouldn’t exist in the eyes of statistics.  There’s too much there in the mythical ‘hockey sense’ to throw off any model ever created which is proof alone that your factor of choice should not be enough to overcome a well informed gut feeling.  This isn’t baseball, where nothing actually matters other than being able to hit the ball or not. This is why it was able to be moneyballed. Anything else can be coached up to passable. Silly sport, but I digress.

2 hours ago, SwampD said:

Edited November 22, 2021 by triumph_communes

[Marvin]

1 hour ago, triumph_communes said:

Ah yes, that’s a different context altogether though it does loop back in. Does the NHL choose the develop older players or taller players and that’s the root cause?

Statisticians have looked at many things and the one that stands out the most is date of birth.  They made correlations over many different sports over the decades and found links that the older a player is for any age bracket (Jan or July depending on locale), the more likely they are to go pro. Ultimately they’ve tied it all back to the more physically mature players get more coaching attention.  The amount of coaching attention one gets throughout their career is the strongest indicator for success than anything else, and coaches will usually give more attention to those more physically mature.

This is the argument one always should have about players needing key minutes and they won’t develop while suffocating for 8 minutes a night.

So if the league selects (drafts) larger players then they’re going to have a better chance at development and it was a self fulfilling prophecy, not an indication on size itself.

 

Again the answer here I believe has changed with time as the barriers/privileges for success for smaller players has gone away over the years and the results speak for themselves. A relevant case study I think is with goalies. The league absolutely drafted tall goalies only, our very own GM calling goalie drafting voodoo magic so he left it to raw attributes and luck (and boy have we suffered from this).  Despite this trend, large goalies have turned out to bust more often than not and a lot of teams have been burned by this approach the last handful of years. Size and fundamentals can get goalies pretty far, but to compete in the NHL it can fall short wildly quick. The best goalies have the ability to track the puck and where it’s going next without seeing it, and they don’t need to be 6’7” to block that. A few more inches of height and Dell would not have made that save in the final seconds last night, he was off the mark by a few feet.  
 

So does size matter?  Yes. Does simply drafting someone because they’re bigger mean success?  No. Are there small towns full of 5’9” Martin St Louis’ out there grumpy that they didn’t get drafted and their height was the reason?  Nah, he’s an outlier and they always will exist. Look at Gretzky or Lemiux numbers sometime versus any other player of their respective time period and those numbers shouldn’t exist in the eyes of statistics.  There’s too much there in the mythical ‘hockey sense’ to throw off any model ever created which is proof alone that your factor of choice should not be enough to overcome a well informed gut feeling.  This isn’t baseball, where nothing actually matters other than being able to hit the ball or not. This is why it was able to be moneyballed. Anything else can be coached up to passable. Silly sport, but I digress.

Italics: There are people pushing for the age groups to be 6 months rather than a year apart in juniors because of exactly this reason.  The idea is to more equalise size at that level to get more useful coaching for the vast majority.

Bolded: It is the job of people in jobs like operations research and industrial statistics to quantify that kind of information.  Indeed, there are analyses in managerial statistics which try to quantify when one person "lifts" her/his team mates, has business sense, etc. -- my uncle bought me a book on that for last Christmas.  (He is pissed off that I am not on a managerial track and don't have an MBA.)  There need to be a lot of models tested and vetted before this becomes a standard analysis.  I am very curious about how one would quantify hockey sense, grit, and "lifting" one's teammates.

Edit -- And in case you think I'm nuts: https://link.springer.com/book/10.1007/978-3-319-63026-7

Edited November 22, 2021 by Marvin, Sabres Fan
added book reference

[Taro T]

14 minutes ago, Marvin, Sabres Fan said:

Italics: There are people pushing for the age groups to be 6 months rather than a year apart in juniors because of exactly this reason.  The idea is to more equalise size at that level to get more useful coaching for the vast majority.

Bolded: It is the job of people in jobs like operations research and industrial statistics to quantify that kind of information.  Indeed, there are analyses in managerial statistics which try to quantify when one person "lifts" her/his team mates, has business sense, etc. -- my uncle bought me a book on that for last Christmas.  (He is pissed off that I am not on a managerial track and don't have an MBA.)  There need to be a lot of models tested and vetted before this becomes a standard analysis.  I am very curious about how one would quantify hockey sense, grit, and "lifting" one's teammates.

Edit -- And in case you think I'm nuts: https://link.springer.com/book/10.1007/978-3-319-63026-7

To the bolded, that's a really interesting idea.  Don't see how it could be made to work except for in a few very unique places.  There simply aren't enough coaches nor available ice to increase the # of teams in most areas.  And in the few like Ra-cha-cha where collectively across several organizations they could make it happen w/out appreciably increasing the # of teams, the adults running things don't play nice w/ each other.  (Which is why there are so many different organizations in the 1st place.)

[triumph_communes]

5 hours ago, Marvin, Sabres Fan said:

Italics: There are people pushing for the age groups to be 6 months rather than a year apart in juniors because of exactly this reason.  The idea is to more equalise size at that level to get more useful coaching for the vast majority.

Bolded: It is the job of people in jobs like operations research and industrial statistics to quantify that kind of information.  Indeed, there are analyses in managerial statistics which try to quantify when one person "lifts" her/his team mates, has business sense, etc. -- my uncle bought me a book on that for last Christmas.  (He is pissed off that I am not on a managerial track and don't have an MBA.)  There need to be a lot of models tested and vetted before this becomes a standard analysis.  I am very curious about how one would quantify hockey sense, grit, and "lifting" one's teammates.

Edit -- And in case you think I'm nuts: https://link.springer.com/book/10.1007/978-3-319-63026-7

Puberty varies by years kid to kid so there is still going to be a preference they can’t control. Reducing the range will help minimize some, but hockey players who go to the NHL almost all skip a bunch of leagues anyway so this wouldn’t affect the top. Leaves us joe smhoes worrying about our kids going nowhere and that’s wasted effort imo 

Yes organizations are trying to do that work, I’m intimately aware. The only thing worse than not using statistics is improperly using statistics to make decisions. Corporates and upper management are notorious at doing this - it takes a lot of work that the type of people attracted to management don’t have the attention span for and if you take shortcuts you do more harm than good. 
 

The new analytics the NHL is collecting such as player and puck position at any given time has the ability to develop algorithms to start measuring the “it” factors. Makes it much easier to come up with metrics tracking all defensemen to tell whether they always shoot the puck when they get it like Miller, continue the cycle d to d and hope someone else makes the move like most our replacement level defenseman, or break down other teams with crazy passes like Dahlin does regularly, as well as tracking success rates and taking advantage of massive sample sizes.  I’m sure there’s at least a few people out there pioneering that kind of work and making up some funny KPIs

Edited November 22, 2021 by triumph_communes