The NHL's Best: A Stat Story by @ToonsBrian - PensInitiative | Pittsburgh Penguins Blog | Rumors | News

The Latest

Post Top Ad

Tuesday, March 18, 2014

The NHL's Best: A Stat Story by @ToonsBrian

    Admittedly, I'm not always a fan of stats. Primarily because people tend to lean too heavily on them when it comes to arguing who's better than who or who's worth what contract or who should be traded or any of a number of arguments, insane or legit.

    At the same time, I'm a number freak. I've always been keen on numbers and messing with mathematics at all levels, including statistics.

    I also love "best of" lists.

    Which is why, short and sweet, I compiled a list of stats (pretty much all of them) for all 848 skaters that have touched the ice at the NHL level this season. Honestly, it created the largest spreadsheet I've ever compiled, weighing in at merely 78,864 cells worth of stats and calculations that, as of 10:00 AM Sunday, March 16th were current.


    All to see who the statistical "Best Players in the NHL" are.

    Here's what I found and how:

    First off, one of the first things that I realized was that "best" is, at its heart, a relative term and that to determine the "best" of anything, I would need to set all player stats relative to, well, the best. And I did that for each category that I tracked.....a long freakin' list of them, I assure you.

    To find a relative score for each category, save +/-, I took each player's actual stat and divided it by the maximum value. For example, Patrick Sharp's 29 goals are 64.4% of Alex Ovechkin's 45 goals. As a result, Alex Ovechkin, the league leader in that category, scored a "relative rating" of 1.000 while Sharp scored a 0.644.

    Plus/Minus ratings were handled similarly, but with negative numbers a possibility some manipulation was required. To determine a relative score, each player's actual +/- had the Minimum score added to it and that sum was then divided by the subtracting the Minimum rating from the Maximum. Therefore, Patrice Bergeron's league-leading +32 and Nail Yakupov's worst -33 created an effective range of 65. Bergeron would receive a relative rating of 1.000 and Yakupov would get a 0.000. Meanwhile, a player between the two, would receive a score based on their placement between those two extremes.

Best of the Best(eses)

    To figure out the best all around skaters, statistically speaking, this season, I simply added every relative score, save Giveaways and Penalty Minutes, which I subtracted. The results were not that surprising, to be honest, with some of the biggest names in the sport finding themselves in the Top 10.


    But.....HOLD YOUR HORSES!!!!!!! These are all centers. Clearly the 10 best players in the league can't all be centers! Right?

    So we massage the stats a little.

The Real Best of the Bestsssss

    Clearly, the center position has a decided advantage statistically based on the fact that they take the majority of faceoffs. The obvious solution is to eliminate all of the faceoff stats then. So I eliminated all the stats that seemed to favor centers. To make it fair, of course.

    That's better. I mean, sure, most of the Top 10 are still centers, but Ovi's on top now. And St. Louis, Steen and Perry make appearances as well. This is clearly the most accurate representation of who the best players in the league are.

    Unless you're a defenseman. Where's THEIR representation? By design, they can't put up the offensive numbers a forward can. Especially numbers that would even approach cherry-pickers like Ovi.

    So, we go back to the drawing board.

The Fairest Bestest of the Bestest

    This is really getting exhausting. But in the interest of getting it right, it must be done.

    This time around, I weight all defensive stats such that there was more of an emphasis on things like Giveaway/Takeaway ratio, +/- and hits.

    But I was smarter than that this time, see? I realized that that would clearly be unfair to forwards. So I did the same thing by weighting offensive stats, too.

    So, offensively, you have this:


    And defensively, you have this:


    Which, of course, are two completely dissimilare lists that have only muddied the waters further!

    So let's just average those puppies because at this point, why not. And we end up with this 2-way rating:





    There. That's fair.

    Unless.......your favorite player has been injured throughout this season. So we've got to compensate for that! I mean, Steven Stamkos' name has not appeared even once!

The Real-for-Real Best Skater in the NHL

    So I'm just going to figure out a nice multiplier that will set each player on even grounds in terms of ice time. Therefore, adjusting everybody's initial stats from the very first graph to per 60:00 TOI, we see some changes:





    And then you have to correct for number of games played, as well. So setting all stats to an 82 game equivalent:


    So there you have it! The best skater in the NHL, all things equal, is not Sidney Crosby. Or Alex Ovechkin. Or Phil Kessel or Claude Giroux. It's Montreal's Joonas Nattinen, who in 1:45 of total ice time this season was able to amass 2 faceoff losses and a blocked shot.

The NHL's Best Skater in 2013-14

    You may not agree with this "definitive" analysis, but let's face it: Stats don't lie.

2 comments:

  1. What's the point of this article? I enjoy stats as much as anyone, but there's no takeaway from this blog post.

    ReplyDelete
    Replies
    1. It's simply to point out that you can make stats say anything you want if you manipulate them enough.
      Everybody focuses so much on stats, advanced stats, every type of metric that we tend to forget that the fun of the game is why we watch to begin with. And stats do very little to enhance that experience.

      Delete

Post Top Ad