Statistical Bias in the Hits Category

[Mike]

If you haven't checked out our series of articles (http://www.leftwinglock.com/content.php?1-the-front-page&page=6) on an effect called the @home bias, you should give it a read over the holiday break (or this morning instead of doing your job). After a Twitter conversation with @ngreenberg (https://twitter.com/#!/ngreenberg) and @SWhyno (https://twitter.com/#!/SWhyno) last night, we decided to update our data for the Hits category.

What you see below is fairly self-explanatory. The second to the last column is the important one as it tells you the hit differential per game for 2011-2012: meaning, how many more hits/game does a team record at home compared to on the road. The results are sorted in such a way that the teams at the top exhibit a large, positive @home bias (more hits at home) and the teams at the bottom exhibit a large, negative @home bias (more hits on the road).

http://www.leftwinglock.com/images/blogposts/hitbiasdec21_2.png

Many of the teams that exhibited a particular bias last season are revealing a similar bias so far this season. You can see last year's bias in the last column in the table above (LY Hit Diff). There is one major exception here: the Washington Capitals. Last season, their @home bias was less than 2.0 (they recorded somewhere between 1-2 extra hits/game during home games). This season (as of Dec 21, 2011), they are recording 12.8 extra hits/game at home! Makes you wonder if the Capitals got a new scorekeeper.

[DaLeafers]

I've never had to play w/ this stat category and I hope I never do as I think it's a very .... for a lack of a better word, opinionated stat category.

What exactly defines a hit? If I'm not looking at somebody but make contact and run them over, is that a hit? If I'm Nathan Gerbe and I try to 'hit' Zdeno Chara, there isn't going to be any noticeable affect (other than me flying backwards) but does that count as a hit?

Unlike Goals, Assists, +/-, PIMs, etc., hits are not as simple and clear cut to 'score' as hits all the time.

If the puck crosses the goal line, it's a goal.
If you touched the puck before your teammate scores, it's an assist.
If you're on the ice, even if you're TKO'd, and your team scores @ reg strength, it's a +1 for you.


I understand Hits arguably gives DEFENSE more value but eh...

[flubber]

FOW and Blocks are also somewhat opinionated stats-- Heck I could even argue that PIMS are based on opinion since one hit gets called a cross-check when a near identical hit goes uncalled.

I agree its not as precise as goals and assists but if pools are only about scoring, frankly I find that lacks depth or complexity.

I was aware of this scorer bias effect all along. Thanks for enumerating it but I don't think anyone can be surprised by this

[flubber]

Oh and not ALL of the effect is necessarily scorers bias-- Many teams have very different rates of scoring or powerplay success home or away so it is feasible that some teams "play differently" at home versus away

It was not in this context but I heard the pundits on sports radio talking about some teams being willing to play a more grinding style on the road as they did not feel pressured to be more entertaining.

Don't get me wrong-- A 'hit" is subjective so there WILL be scorers bias. But thats not the whole story here

[Mike]

Thanks for enumerating it but I don't think anyone can be surprised by this

I'm very surprised by it. I knew it existed based on earlier studies and our own series we did this Summer. What shocks me are the Washington numbers. Their hit totals during road games are down 46% from last year, so clearly they are playing less physical. How does their @home bias go from 1.6 to 12.8?

[flubber]

I'm very surprised by it. I knew it existed based on earlier studies and our own series we did this Summer. What shocks me is the Washington numbers. Their hit totals during road games are down 46% from last year, so clearly they are playing less physical. How does their @home bias go from 1.6 to 12.8?

Sorry -- I wasn't surprised by the general concept of scorers bias on hits-- Thats what I was generally trying to say

I cannot account for the Washington year over year change other than as you say, a scorer change . I a cknowledge the subjectivity of the stat even while I quite like it

[mbr311]

This category drives me insane. How can Hanzal have 12 hits last night and 0 hits tonight???

It's not like SOG where maybe you're just missing or someone on the ice is blocking it...if you're a physical player then how does it vary like that? 12 hits to 0? I see this happen with Callahan as well quite frequently. Right now I'm praying I can get 3 more BLOCKS to win that category but I don't think it's gonna happen. I think these are also stats that are never looked back at (like goals and assists) and can be "corrected" if you willp