Metrics And Minnesota: A Love Story

Apparently this is the topic du jour in the Wild Blogosphere: So consider this one that causes the hats to fly for the hat trick:

Vexing Wild Driving Eggheads Crazy - Hitting The Post
Comparing The 2011-12 Minnesota Wild To Past Teams - Hockey Wilderness

Vexed indeed- for the Advanced Stats disciples have long railed against Minnesota's success this year: that Mike Yeo's system belies the issues this team has. What is that issue exactly? Well, our Corsi and Fenwick is terrible.

Our....what? Ya, because our Fenwick is terrible, so is Minnesota.



In case your late to the party, Corsi and Fenwick are "measurements of possession", which has some truth to it. Basically, they serve as a measurement of pucks directed at nets in either direction; shots on goal, shots missed, and blocked shots; Fenwick eschews block shots for its number. So, boiled down to its very wicked essence, if you "generate" more shots at your opponent's net than what is generated at yours, you're a good team- because to shoot the puck, you need possession of the puck. And because our Fenwick/Corsi isn't very good, it means we "struggle to generate shots." Which just isn't true- every single Minnesota Wild fan whose watched a game on the television or in person has had the urge to yell "SHOOT THE FRACKING PUCK" because someone (and there are serial offenders here- I'm looking at you Mikko Koivu. And you Jared Spurgeon. And you Dany Heatley. And you Devin Setoguchi) passed up a prime scoring chance instead choosing to defer to a teammate, and inevitably we end up with nothing. Its not that we "struggle to generate shots", we "struggle to just shoot the goddamned puck instead of passing it off." Hey, if improving our shot total, hence improving our Fenwick, would get the Metrics guys off our backs, we could just lob every puck at the net...or the direction of the net.

Ya, about shots...and save percentage.

"Shot" has a rather ambiguous quality to it; its basically a puck sent in the direction of the goal- no specifics in terms of velocity, height, or direction; that means a shot from the corner counts the same as a shot from the high slot, which also counts the same as a shot from the blueline; a shot is a shot, despite the quality of the shot in the slot being of much higher quality than that of the sharp-angled shot from the corner (in the business we call it a "low-percentage shot") and also from the point. However, there is no distinction here, and despite the systemic philosophy of forcing shooters to the outside, a shot is a shot is a shot, and therefore a combined save% of .936 (three goaltenders) is "unsustainable"- despite that Minnesota forces shooters (not always, but in general) to take shots that any NHL-level goaltender (yes that means you Steve Mason, but not you Vesa Toskala) SHOULD stop.

Where's the context?

It's getting there- friend of FRB Corey Sznajder is apart of an ambitious project where scoring chances in NHL games are being tracked and recorded; he uses use as a general reference, along with a few other exceptions. Now, this being said- I could understand the notion of unsustainable goaltending if Backstrom/Harding/Hackett/Denis Lemieux were stopping 94% of PRIME scoring chances- which during the Todd Richards years would have been near-miraculous- but is it really that far fetched that this collective is stopping the pucks they are supposed to?

Alas things change- shot totals may go up, and the goaltending may, yes, "regress to the mean." But will it matter- talking with Timo Seppa of Hockey Prospectus about these very issues- he intimated that Fenwick/Corsi is valuable early in the season based off the limited sample size, but the deeper the season gets that goal-related statistics may carry more value.

But do we really need numbers to tell us what we already see for ourselves?

2 comments:

  1. The way the scientific method is *supposed* to work, is that when the model fails to reflect reality, it is the model that is supposed to give way and be tweaked for another try.

    Too often, the faux-scientist will try to insist that reality is to blame.

    ReplyDelete
  2. Btw, here's another theory on Wild-hating by statisticians. If they tear them down now while they are #1, then when the Wild end up at #4 in the conference, they can say "See, I told you so!". . . having successfully through all the arm-waving over the Wild's unworthiness to be #1 distracted from the fact that they actually picked the Wild to be #13.

    ReplyDelete