A Defense Independent Rating of NHL Goaltenders

by Matt Evans SNC on Nov 16, 2011 8:39 PM EST

There was a recent paper written that offers forth a new way of analyzing goalie's performance. The paper was presented at the Sloan Sports Analytic Conference, which is a big deal in Academia. The paper was written by Michael E. Schuckers and has a few interesting points I'll bring up for those who don't want to read it. If you are a nerd like me and get into the math, its laid out in the paper, otherwise I'll summarize the paper below. Also, here's the link to the paper

The new stat tries to take away how the defense effects the goalie ratings (GAA and SV%). For example, some people may argue that Martin Brodeur is not as good as some other great goalies because he had a stellar defense in front of him. This paper attempts to prove this gray area by measuring where shots come from and the probability that each goalie would stop that shot. There were 74,300 shots in the 2009-2010 season; the paper uses every shot on the season against the probability each goalie would stop that shot from there. He does this by spatially mapping out where each shot in the 2009 to 2010 season came from and then analyzes which goalies would stop the shot and then rates them with some fancy math.

Interesting enough the the worst goalie by this rating was Tim Thomas, the best was Ryan Miller. I would say that each of these goalies are in the top tier of the league, not at opposite ends. Martin Brodeur landed at #29 out of 40.

For those of you who want the specific results, go to the paper and read section 4 (Application and Results). Its easy to read and very interesting where this new rating puts different goalies.

All FanPosts and FanShots are the respective work of the author and not representative of the writers or other users of In Lou We Trust.

Tweet Comment 7 comments | 0 recs |

Do you like this story?

Comments

2010-2011 DIGR Stats

After a little looking I found last year’s DIGR stats so you can compare across seasons.

DIGR 2010-11

by Matt Evans SNC on Nov 16, 2011 9:49 PM EST reply actions

Very good work.

by Mihali4 on Nov 17, 2011 6:19 AM EST reply actions

There’s been a bunch of articles counter-discussing the DIGR article. Some good discussion ended up in this link. Basically, there are two problems:

a) NHL shot location data is god awful. Even beyond the MSG guy that thinks everything happens in the slot. There’s another link out there which I can’t find now where Gabe of Behind the Net showed that they can’t even properly locate the faceoff circle.

b) There’s no evidence that this predicts future goaltending behavior any better than the current best measure available (even strength save percentage).

Blueshirt Banter - Where Rangers' Fans Matter
Tracking the Rangers - Numbers don't lie. They just don't agree with you.
Twitter: RangerSmurf

by George E. Ays on Nov 17, 2011 9:18 AM EST reply actions

Anybody know where the faceoff dot is?

In Lou We Trust: SBN Blog of the New Jersey Devils

by Matthew Ventolo on Nov 17, 2011 2:03 PM EST up reply actions

This also gives no consideration to the shooter. Ilya Kovalchuk is going to make more of his shots from any given location than Cam Janssen. That’s a fact.

Come Say Hello! Section 231 - Row 2 - Seat 20

by Alamoth on Nov 17, 2011 1:07 PM EST reply actions

Yes, that's a fact

But only because you chose the right player. Kovalchuk’s the only guy in the league that has been shown observed skill in scoring more often from a given location than others.

http://www.arcticicehockey.com/2010/7/29/1593561/observed-distribution-of-shooting

Blueshirt Banter - Where Rangers' Fans Matter
Tracking the Rangers - Numbers don't lie. They just don't agree with you.
Twitter: RangerSmurf

by George E. Ays on Nov 17, 2011 2:55 PM EST up reply actions 1 recs