HFBoards An idea for assessing goalies
 Register FAQ Members List Search Today's Posts Mark Forums Read
 Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
 By The Numbers Hockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

# An idea for assessing goalies

 09-03-2012, 09:48 PM #1 Zombie Mike Murphy Registered User     Join Date: Mar 2011 Posts: 737 vCash: 500 An idea for assessing goalies So, I had an idea come to me today, that I thought I'd bounce off the mob. This is purely a thought exercise as it would take _serious_ resources to compute this for even a subset of a single season. The idea is fairly simple - instead of looking at a goalie's SV%, a number which while easy to compute, is also poorly isolated, you look at how he handles specific shots vs the other goalies in the league. So, each shot taken that the goalie played will go into a bucket - ideally, a single bucket should represent a specific type of shot - one timer, slapper, deflection, whatever - taken from a small region of the ice - 1 yard squares maybe. The idea is to neutralize for the differing shot selections faced by goalies behind good and bad defenses. Does this approach have any merit? Has a study along these lines ever been tried?
 09-03-2012, 10:41 PM #2 metalfoot Karlsson!     Join Date: Dec 2007 Location: Manitoba, Canada Country: Posts: 1,568 vCash: 500 Seems like an interesting idea, but how will you generate the necessary data for analysis?
 09-03-2012, 11:04 PM #3 Zombie Mike Murphy Registered User     Join Date: Mar 2011 Posts: 737 vCash: 500 Yea, that's the rub. if there was some way to get the X/Y location and shot types from the NHL.com replays that might be a start.
09-04-2012, 03:59 AM
#4
Registered User

Join Date: Nov 2007
Posts: 11,922
vCash: 500
NHL Teams

Quote:
 Originally Posted by Troll Ward So, I had an idea come to me today, that I thought I'd bounce off the mob. This is purely a thought exercise as it would take _serious_ resources to compute this for even a subset of a single season. The idea is fairly simple - instead of looking at a goalie's SV%, a number which while easy to compute, is also poorly isolated, you look at how he handles specific shots vs the other goalies in the league. So, each shot taken that the goalie played will go into a bucket - ideally, a single bucket should represent a specific type of shot - one timer, slapper, deflection, whatever - taken from a small region of the ice - 1 yard squares maybe. The idea is to neutralize for the differing shot selections faced by goalies behind good and bad defenses. Does this approach have any merit? Has a study along these lines ever been tried?
NHL teams use a similar approach from game films etc. larger than 1' squares.

Journal de Montreal does an approximation with its game story of the Canadiens the last few years.

09-05-2012, 10:03 AM
#5
seventieslord
Moderator

Join Date: Mar 2006
Location: Regina, SK
Country:
Posts: 25,757
vCash: 500
Quote:
 Originally Posted by Troll Ward So, I had an idea come to me today, that I thought I'd bounce off the mob. This is purely a thought exercise as it would take _serious_ resources to compute this for even a subset of a single season. The idea is fairly simple - instead of looking at a goalie's SV%, a number which while easy to compute, is also poorly isolated, you look at how he handles specific shots vs the other goalies in the league. So, each shot taken that the goalie played will go into a bucket - ideally, a single bucket should represent a specific type of shot - one timer, slapper, deflection, whatever - taken from a small region of the ice - 1 yard squares maybe. The idea is to neutralize for the differing shot selections faced by goalies behind good and bad defenses. Does this approach have any merit? Has a study along these lines ever been tried?
This sounds an awful lot like shot quality adjusted save percentage.

09-05-2012, 02:55 PM
#6
TurdFerguson
Registered User

Join Date: Jul 2012
Posts: 1,024
vCash: 500
Quote:
 Originally Posted by seventieslord This sounds an awful lot like shot quality adjusted save percentage.
Though it can be seen that way as an overall measure, I think its real merit is in knowing a goalie's strengths and weaknesses.

 09-05-2012, 03:41 PM #7 bluesfan94 #BackesforSelke     Join Date: Jan 2008 Location: St. Louis Country: Posts: 11,996 vCash: 500 Wouldn't you also need to know the context of the shot, such as who was shooting, whether or not there was a screen, off a rebound, etc?
 09-05-2012, 07:40 PM #8 Zombie Mike Murphy Registered User     Join Date: Mar 2011 Posts: 737 vCash: 500 That would fall under type of shot. Who's shooting is probably going too far - don't want to be overly specific. Ideally, every starting goalie will face at least 20-30 shots from each "bucket".
 09-17-2012, 09:03 PM #9 Jason MacIsaac MARS Officer     Join Date: Jan 2004 Location: Halifax, NS Country: Posts: 16,658 vCash: 500 Shotkeepers are far too inconsistent for this to work. ES SV% is what I look at.
09-18-2012, 10:38 AM
#10
Global Moderator

Join Date: Aug 2006
Location: Brooklyn
Country:
Posts: 42,610
vCash: 500
Quote:
 Originally Posted by Jason MacIsaac Shotkeepers are far too inconsistent for this to work. ES SV% is what I look at.
Wouldn't the inconsistency of shopkeepers also affect save percentage?

09-18-2012, 11:31 AM
#11
Jason MacIsaac
MARS Officer

Join Date: Jan 2004
Location: Halifax, NS
Country:
Posts: 16,658
vCash: 500
Quote:
 Originally Posted by TheDevilMadeMe Wouldn't the inconsistency of shopkeepers also affect save percentage?
Not as much seeing that shots will be counted as one large group instead of position and type oriented.

Forum Jump