HFBoards

Go Back   HFBoards > General Hockey Discussion > By The Numbers
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.

"Clutch" Goaltenders

Reply
 
Thread Tools
Old
07-09-2014, 12:58 PM
  #51
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Agreed on CuJo - one of the problems with this analysis is that it's hard to separate the goaltender from the team (for good or for bad).

My recollection as well is that Joseph's typically pushed weaker teams deeper into the playoffs. I do a strength of schedule metric on my site; the numbers presented below would be by how many goals a team would be favored to win, against an average opponent, on neutral ice.

With that said, here is each of Joseph's playoff series (if he wasn't a starter, I marked the series with an asterisk), his team's "strength" and the opponent's "strength":

SeriesResultJoseph TeamOpponentUpset?
1990 TorontoL0.09-0.31-
1992 ChicagoL0.110.38 
1993 ChicagoW0.030.45+
1993 TorontoL0.030.47 
1994 DallasL-0.240.14 
1995 VancouverL0.720.07-
1997 DallasW0.040.54+
1997 ColoradoL0.040.88 
1998 ColoradoW-0.090.19+
1998 DallasL-0.090.79 
1999 PhiladelphiaW0.390.41+
1999 PittsburghW0.390.20 
1999 BuffaloL0.390.51 
2000 OttawaW0.240.19 
2000 New JerseyL0.240.63 
2001 OttawaW0.330.61+
2001 New JerseyL0.331.05 
2002 NY IslandersW0.320.13 
2002 OttawaW0.320.44+
2002 CarolinaL0.320.00-
2003 AnaheimL0.750.31-
2004 Nashville*W0.67-0.06 
2004 CalgaryL0.670.42-
2008 San Jose*L0.110.42 

By this metric, Joseph won six series as the underdog and lost five series as the favorite, although there's a pretty good cluster of underdog wins there in his prime.

Doctor No is offline   Reply With Quote
Old
07-09-2014, 01:43 PM
  #52
BraveCanadian
Registered User
 
BraveCanadian's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 9,474
vCash: 500
Quote:
Originally Posted by Chalupa Batman View Post
Agreed on CuJo - one of the problems with this analysis is that it's hard to separate the goaltender from the team (for good or for bad).

My recollection as well is that Joseph's typically pushed weaker teams deeper into the playoffs. I do a strength of schedule metric on my site; the numbers presented below would be by how many goals a team would be favored to win, against an average opponent, on neutral ice.

With that said, here is each of Joseph's playoff series (if he wasn't a starter, I marked the series with an asterisk), his team's "strength" and the opponent's "strength":

SeriesResultJoseph TeamOpponentUpset?
1990 TorontoL0.09-0.31-
1992 ChicagoL0.110.38 
1993 ChicagoW0.030.45+
1993 TorontoL0.030.47 
1994 DallasL-0.240.14 
1995 VancouverL0.720.07-
1997 DallasW0.040.54+
1997 ColoradoL0.040.88 
1998 ColoradoW-0.090.19+
1998 DallasL-0.090.79 
1999 PhiladelphiaW0.390.41+
1999 PittsburghW0.390.20 
1999 BuffaloL0.390.51 
2000 OttawaW0.240.19 
2000 New JerseyL0.240.63 
2001 OttawaW0.330.61+
2001 New JerseyL0.331.05 
2002 NY IslandersW0.320.13 
2002 OttawaW0.320.44+
2002 CarolinaL0.320.00-
2003 AnaheimL0.750.31-
2004 Nashville*W0.67-0.06 
2004 CalgaryL0.670.42-
2008 San Jose*L0.110.42 

By this metric, Joseph won six series as the underdog and lost five series as the favorite, although there's a pretty good cluster of underdog wins there in his prime.
Right and even then there is no way I would pin 2002 and 2003 on him.. 1995 is one time I think he really didn't get it done followed by 1999 somewhat too. So tough to separate goaltenders from teams.

Anyways very cool to see how some of these numbers match up with perceptions in the case of Hasek etc.

BraveCanadian is offline   Reply With Quote
Old
07-10-2014, 02:58 AM
  #53
Luongofan558
Registered User
 
Join Date: Jun 2014
Posts: 140
vCash: 500
You should look at what goalies have the best even strength sv percentage in the playoffs

Luongofan558 is offline   Reply With Quote
Old
07-10-2014, 09:32 AM
  #54
blogofmike
Registered User
 
Join Date: Dec 2010
Posts: 840
vCash: 500
Would be partially incomplete if missing pre-1983, but how about Fuhr?

blogofmike is offline   Reply With Quote
Old
07-10-2014, 09:49 AM
  #55
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
I've been trying to hold off on Fuhr, since I'm this close to completing my 1980-81 and 1981-82 game logs (which will make Fuhr's record - and his tandem-mate, Andy Moog's record, complete). Here's what I have so far:

SituationGPWLActual SV%Expected SV%S+/30
All Games14590470.9000.881+0.55
Can Eliminate3319110.8910.875+0.46
Can be Eliminated15870.9050.894+0.32
Mutual Elimination6420.9060.882+0.71

(Unlike the above table, these results form do not form a partition. Mutual Elimination games are included in all three "elimination" categories)

All I really see here is that Fuhr was a great playoff goaltender. Here are his mutual elimination games (it's missing one, the last game of the Miracle on Manchester series):

GameResultScoreShotsExp SavesActual Saves
4/22/1984 vs. CalgaryW7-41412.313
4/30/1986 vs. CalgaryL2-32118.418
5/30/1987 vs. PhiladelphiaW3-12017.619
4/15/1989 at Los AngelesL3-62924.924
4/16/1991 at CalgaryW5-4 OT3026.226
5/4/1999 at PhoenixW1-0 OT3532.135

Fuhr replaced Andy Moog in the first of those above.

Doctor No is offline   Reply With Quote
Old
07-10-2014, 02:21 PM
  #56
Senators Growing
Next year again...
 
Senators Growing's Avatar
 
Join Date: Jul 2006
Location: Cumberland
Country: Canada
Posts: 8,830
vCash: 500
Can you do Tuukka Rask and, because I'm a Senators fan and thus a masochist, Patrick Lalime? Thanks!

Senators Growing is offline   Reply With Quote
Old
07-10-2014, 11:08 PM
  #57
BraveCanadian
Registered User
 
BraveCanadian's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 9,474
vCash: 500
Quote:
Originally Posted by Chalupa Batman View Post
All I really see here is that Fuhr was a great playoff goaltender.
No surprises here but it is nice to see another more objective way of looking at it.

BraveCanadian is offline   Reply With Quote
Old
07-11-2014, 12:17 PM
  #58
King Mapes
Global Moderator
HF's King
 
King Mapes's Avatar
 
Join Date: Feb 2008
Location: Halifax
Posts: 24,139
vCash: 500
Right now Quick is a playoff warrior. IIRC Turco was typically very good under pressure

Just looked it up, he had a couple good runs..

King Mapes is offline   Reply With Quote
Old
07-11-2014, 07:25 PM
  #59
Statto
Registered User
 
Join Date: May 2014
Posts: 183
vCash: 500
What a great thread and analysis. I tip my hat to you sir!

Statto is offline   Reply With Quote
Old
07-14-2014, 11:27 PM
  #60
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Here's Tuukka Rask, who is actually borderline statistically significant on the bad end of things.

SituationGPWLActual SV%Expected SV%S+/30
All Games4728190.9300.908+0.65
Can Eliminate14590.9000.908-0.26
Can be Eliminated5140.8770.903-0.78
Mutual Elimination3120.8490.903-1.62

(Unlike the above table, these results form do not form a partition. Mutual Elimination games are included in all three "elimination" categories)

Yikes. Not sure what needs to be said, other than Rask's certainly going to have some more playoff games to acquit himself. p-values above (in order) are 0.9%, 1.0%, and 0.4% (even with the sample of three games).

Here's the evidence for mutual elimination:

GameResultScoreShotsExp SavesActual Saves
5/14/2010 vs. PhiladelphiaL3-42724.523
5/13/2013 vs. TorontoW5-4 OT2825.024
5/14/2014 vs. MontrealL1-31816.415

Doctor No is offline   Reply With Quote
Old
07-14-2014, 11:32 PM
  #61
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Patrick Lalime, one of my favorites in the 1990s back to his super start with the Penguins, is somewhat comparable to Rask in this regard.

SituationGPWLActual SV%Expected SV%S+/30
All Games4121200.9260.909+0.52
Can Eliminate7340.9010.910-0.28
Can be Eliminated7340.9110.908+0.09
Mutual Elimination3030.8620.908-1.39

(Unlike the above table, these results form do not form a partition. Mutual Elimination games are included in all three "elimination" categories)

Like I said, similar to Rask without the magnitude (albeit without the chance to balance the scales in the future).

Lalime's game sevens:

GameResultScoreShotsExp SavesActual Saves
5/14/2002 at TorontoL0-32724.324
5/23/2003 vs. New JerseyL2-32724.824
4/20/2004 at TorontoL1-4119.98

Good guy Martin Prusek finished that last one.

Doctor No is offline   Reply With Quote
Old
07-15-2014, 05:38 PM
  #62
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Carey Price is the third of the trifecta being debated elsewhere, and since I've written up Quick and Rask, I thought I'd close the loop here.

SituationGPWLActual SV%Expected SV%S+/30
All Games4217210.9090.910-0.02
Can Eliminate6330.8950.913-0.52
Can be Eliminated7430.9300.907+0.67
Mutual Elimination3210.9440.912+0.96

(Unlike the above table, these results form do not form a partition. Mutual Elimination games are included in all three "elimination" categories)

Carey's overall playoff numbers aren't that great (remember from higher in the thread that playoff save percentages are typically better than average). He and the Canadiens don't seem to press the advantage, although his numbers when he can be eliminated are quite nice (although not quite statistically significant).

Price's game sevens (which all have a common theme):

GameResultScoreShotsExp SavesActual Saves
4/21/2008 vs. BostonW5-02522.925
4/27/2011 at BostonL3-4 OT3431.030
5/14/2014 at BostonW3-13027.329

Doctor No is offline   Reply With Quote
Old
07-15-2014, 06:56 PM
  #63
KEEROLE Vatanen
Always positive
 
KEEROLE Vatanen's Avatar
 
Join Date: Oct 2011
Posts: 25,112
vCash: 500
this really is a cool topic, with the advent of more statistical analysis in sports it seems like the term "clutch" is in an interesting state. people in baseball tend to believe it doesn't exist in that sport, whereas in the NBA and possibly now the NHL there is now analysis being done to show it may exist in those sports but sometimes people we think are clutch, really aren't performing that well (see kobe bryant)

KEEROLE Vatanen is offline   Reply With Quote
Old
07-15-2014, 07:11 PM
  #64
jacks
Registered User
 
jacks's Avatar
 
Join Date: Mar 2013
Posts: 5,884
vCash: 500
Quote:
Originally Posted by Chalupa Batman View Post
Okay, let's have a little bit of fun now, and play with a few goaltenders.

Patrick Roy has a pretty good reputation for clutch play.

SituationGPWLActual SV%Expected SV%S+/30
All Games247151940.9180.894+0.73
Can Eliminate5630260.9180.894+0.70
Can be Eliminated2412120.9240.897+0.83
Mutual Elimination13670.9070.897+0.31

(Unlike the above table, these results form do not form a partition. Mutual Elimination games are included in all three "elimination" categories)

A few conclusions - compared to the numbers above, Roy steps his game up in the playoffs overall. In mutual-elimination games, he doesn't look so hot; this is suggestive of small samples (although there's certainly no evidence to suggest that he's "extra clutch" in game sevens).

A look at his mutual-elimination games may be in order:

GameResultScoreShotsExp SavesActual Saves
4/29/1986 vs. HartfordW2-1 OT2521.624
4/29/1991 at BostonL1-22925.727
5/1/1992 vs. HartfordW3-2 2OT4136.739
4/29/1994 at BostonL3-53128.126
5/4/1998 vs. EdmontonL0-41715.513
6/4/1999 at DallasL1-42522.521
5/27/2000 at DallasL2-32926.226
5/9/2001 vs. Los AngelesW5-12623.325
6/9/2001 vs. New JerseyW3-12623.225
4/29/2002 vs. Los AngelesW4-02320.823
5/15/2002 vs. San JoseW1-02724.127
5/31/2002 at DetroitL0-71614.510
4/22/2003 vs. MinnesotaL2-3 OT3027.327

I'm not entirely sure what to make of this, although Roy's overall record in game sevens may be beaten to death at this point.
No one was more clutch in the playoffs than Roy including forwards and D-man.
3 Conn Smythe Trophy is actual proof.

jacks is offline   Reply With Quote
Old
07-15-2014, 07:55 PM
  #65
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Quote:
Originally Posted by jacks View Post
No one was more clutch in the playoffs than Roy including forwards and D-man.
3 Conn Smythe Trophy is actual proof.
No one's a bigger Roy fan than I am, and I agree that he's a fantastic goaltender.

With that said, did you bother to read the thread before commenting? The phrase "actual proof" is a bit insulting and suggests that you did not.

Doctor No is offline   Reply With Quote
Old
07-20-2014, 09:28 AM
  #66
TieClark
Registered User
 
Join Date: Jun 2011
Posts: 4,115
vCash: 500
Quote:
Originally Posted by Chalupa Batman View Post
Always loved Ed Belfour, even though he may have cost my Avs a Cup or two in their heyday. I gave away a bit above, but Belfour does appear to step it up when he's got a team on the ropes:

SituationGPWLActual SV%Expected SV%S+/30
All Games16188680.9200.900+0.60
Can Eliminate241850.9390.901+1.14
Can be Eliminated2110100.9280.901+0.82
Mutual Elimination6510.9210.902+0.58

(Unlike the above table, these results form do not form a partition. Mutual Elimination games are included in all three "elimination" categories)

In mutual-elimination games, Belfour plays about as well as he always does, although the results are quite good:

GameResultScoreShotsExp SavesActual Saves
4/30/1990 vs. St. LouisW8-22824.526
5/19/1995 vs. TorontoW5-22421.922
6/4/1999 vs. ColoradoW4-11917.118
5/27/2000 vs. ColoradoW3-23329.731
4/22/2003 at PhiladelphiaL1-63633.030
4/20/2004 vs. OttawaW4-13733.436

The one stinker may be bringing him down disproportionally (although it was a game seven).
I remember that one stinker... it certainly wasn't Belfour who didn't show up that night.

TieClark is offline   Reply With Quote
Old
07-29-2014, 01:35 PM
  #67
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
During lunch today, I was looking at the Joseph series record (above) and decided to compare it against the acclaimed playoff clutch hero Patrick Roy (yes, Roy's one of my favorite goalies).

In a similar format to the above, here are Roy's playoff series (if he wasn't a starter, I marked the series with an asterisk), his team's "strength" and the opponent's "strength":

SeriesResultRoy TeamOpponentUpset?
1986 BostonW0.770.34 
1986 HartfordW0.770.54 
1986 NY RangersW0.770.14 
1986 CalgaryW0.770.46 
1987 BostonW0.530.22 
1987 Quebec*W0.530.00 
1987 Philadelphia*L0.530.73 
1988 HartfordW0.61-0.18 
1988 BostonL0.610.63 
1989 HartfordW1.170.09 
1989 BostonW1.170.40 
1989 PhiladelphiaW1.170.28 
1989 CalgaryL1.171.54 
1990 BuffaloW0.560.36 
1990 BostonL0.560.58 
1991 BuffaloW0.210.01 
1991 BostonL0.210.23 
1992 HartfordW0.47-0.45 
1992 BostonL0.47-0.17-
1993 QuebecW0.570.44 
1993 BuffaloW0.570.42 
1993 NY IslandersW0.570.51 
1993 Los AngelesW0.57-0.02 
1994 BostonL0.330.37 
1996 VancouverW1.10-0.09 
1996 ChicagoW1.100.59 
1996 DetroitW1.101.47+
1996 FloridaW1.100.35 
1997 ChicagoW0.880.05 
1997 EdmontonW0.880.04 
1997 DetroitL0.880.78-
1998 EdmontonL0.19-0.09-
1999 San JoseW0.350.01 
1999 DetroitW0.350.45+
1999 DallasL0.350.82 
2000 PhoenixW0.540.06 
2000 DetroitW0.540.84+
2000 DallasL0.540.43-
2001 VancouverW1.100.02 
2001 Los AngelesW1.100.28 
2001 St. LouisW1.100.70 
2001 New JerseyW1.101.05 
2002 Los AngelesW0.550.31 
2002 San JoseW0.550.59+
2002 DetroitL0.550.95 
2003 MinnesotaL0.700.33-

By this metric, Roy won four series as the underdog, and lost five series as the favorite. Of course, Roy had more opportunities as the favorite than he did as the underdog, but it's an interesting comparison with Joseph above.

Doctor No is offline   Reply With Quote
Old
07-29-2014, 07:18 PM
  #68
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Okay, okay, okay - I'll do Brodeur and Hasek. Here's Brodeur:

SeriesResultBrodeur TeamOpponentUpset?
1994 BuffaloW0.950.70 
1994 Boston*W0.950.37 
1994 NY RangersL0.950.97 
1995 BostonW0.740.27 
1995 PittsburghW0.740.28 
1995 PhiladelphiaW0.740.42 
1995 DetroitW0.741.22+
1997 MontrealW0.54-0.44 
1997 NY RangersL0.540.34-
1998 OttawaL0.57-0.14-
1999 PittsburghL0.520.20-
2000 FloridaW0.630.20 
2000 TorontoW0.630.24 
2000 PhiladelphiaW0.630.55 
2000 DallasW0.630.43 
2001 CarolinaW1.05-0.32 
2001 TorontoW1.050.33 
2001 PittsburghW1.050.18 
2001 ColoradoL1.051.10 
2002 CarolinaL0.130.00-
2003 BostonW0.63-0.04 
2003 Tampa BayW0.63-0.08 
2003 OttawaW0.630.84+
2003 AnaheimW0.630.31 
2004 PhiladelphiaL0.450.52 
2006 NY RangersW0.110.20+
2006 CarolinaL0.110.38 
2007 Tampa BayW0.02-0.33 
2007 OttawaL0.020.67 
2008 NY RangersL-0.090.08 
2009 CarolinaL0.310.07-
2010 PhiladelphiaL0.220.23 
2012 FloridaW0.21-0.32 
2012 PhiladelphiaW0.210.34+
2012 NY RangersW0.210.38+
2012 Los AngelesL0.210.46 

Five wins as the underdog, and five losses as the favorite.

Doctor No is offline   Reply With Quote
Old
07-29-2014, 07:28 PM
  #69
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
And Hasek...

SeriesResultHasek TeamOpponentUpset?
1991 Minnesota*L0.700.06N/A
1992 Pittsburgh*L0.380.70 
1993 Boston*W0.420.62N/A
1994 New JerseyL0.700.95 
1995 PhiladelphiaL0.060.42 
1997 Ottawa*W0.20-0.13 
1998 PhiladelphiaW0.410.39 
1998 MontrealW0.410.27 
1998 WashingtonL0.410.26-
1999 OttawaW0.510.62+
1999 BostonW0.510.39 
1999 TorontoW0.510.39 
1999 DallasL0.510.82 
2000 PhiladelphiaL-0.050.55 
2001 PhiladelphiaW0.370.20 
2001 PittsburghL0.370.18-
2002 VancouverW0.950.52 
2002 St. LouisW0.950.57 
2002 ColoradoW0.950.55 
2002 CarolinaW0.950.00 
2007 CalgaryW0.820.42 
2007 San JoseW0.820.74 
2007 AnaheimL0.820.76-
2008 Nashville*W1.080.13 

First - lotta asterisks here; I didn't feel like Hasek should get credit for winning the 1997 Ottawa series or for winning the 2008 Nashville series. With that said, the totals are here, so if you have another opinion, you now have the data. Hasek's team was favored in both cases, so it really doesn't change my conclusion, which is...

Using my criteria, I see one series win for Hasek as the underdog and three upsets as the favorite.

Doctor No is offline   Reply With Quote
Old
07-29-2014, 07:34 PM
  #70
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
I do find it neat the the first "real" playoff series for both Hasek and Brodeur - two of the greatest all-time goaltenders - was against one another. And it was a tremendous series, to boot.

If anyone's looking for a great historical book premise, there you go.

Doctor No is offline   Reply With Quote
Old
07-29-2014, 07:40 PM
  #71
Caeldan
Moderator
Whippet Whisperer
 
Caeldan's Avatar
 
Join Date: Jun 2008
Country: Canada
Posts: 9,590
vCash: 50
Question - when it comes to comparing strength of schedule, is there any metric as far as comparing differences between years?

Also, while you're doing a simple upset/underdog calculation... .
I'd think there'd be some value in determining an 'equal' or 'tossup' range and looking at win/loss in those series... because if someone is significantly coming out with wins in matches that are considered to be a coin toss - on the surface they could be that 'clutch' difference maker.

For example... I'm going to arbitrarily decide that anything +/- 0.1 is considered to be a 'tossup':
In Brodeur's case... there are 6 series and he has a 2-4 record.
In Hasek's case... there are 3 series and he has a 2-1 record.
Not a lot to draw from that, but perhaps it's a matter of finding the right range?

PS - your Team column name for Hasek is mislabeled. (feel free to edit this statement out when you get around to fixing it).


Last edited by Caeldan: 07-29-2014 at 07:49 PM.
Caeldan is offline   Reply With Quote
Old
07-29-2014, 08:31 PM
  #72
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Quote:
Originally Posted by Caeldan View Post
Question - when it comes to comparing strength of schedule, is there any metric as far as comparing differences between years?

Also, while you're doing a simple upset/underdog calculation... .
I'd think there'd be some value in determining an 'equal' or 'tossup' range and looking at win/loss in those series... because if someone is significantly coming out with wins in matches that are considered to be a coin toss - on the surface they could be that 'clutch' difference maker.

For example... I'm going to arbitrarily decide that anything +/- 0.1 is considered to be a 'tossup':
In Brodeur's case... there are 6 series and he has a 2-4 record.
In Hasek's case... there are 3 series and he has a 2-1 record.
Not a lot to draw from that, but perhaps it's a matter of finding the right range?

PS - your Team column name for Hasek is mislabeled. (feel free to edit this statement out when you get around to fixing it).
Great comments (and nice catch on the header!).

Admittedly, this is a somewhat coarse exercise that I'm looking to explore further. I like the toss-up idea - it seems a bit unfair that, if a goaltender's team has strength 0.40 and loses to a team with strength 0.41, that's not an upset, but is an upset if they lose to a team with strength 0.39.

I've also toyed with the idea of summing the differences (so that an upset where the difference was 0.70 would count a lot more than an upset where the difference was 0.01).

As for your first question, I'm not sure what you mean, so I'll kind of head in that direction and see how well I do. The values are reflective of how much a team would be favored over an "average" opponent on neutral ice. To that end, the denominator is goals, and in a season with more overall goals, the best (and worst) numbers will deviate from zero by a greater degree.

To give an extreme example, if all NHL goals were suddenly worth ten points, then the strength metric values would all be 10 times their current size.

Doctor No is offline   Reply With Quote
Old
07-29-2014, 09:54 PM
  #73
Caeldan
Moderator
Whippet Whisperer
 
Caeldan's Avatar
 
Join Date: Jun 2008
Country: Canada
Posts: 9,590
vCash: 50
Quote:
Originally Posted by Chalupa Batman View Post
Great comments (and nice catch on the header!).

Admittedly, this is a somewhat coarse exercise that I'm looking to explore further. I like the toss-up idea - it seems a bit unfair that, if a goaltender's team has strength 0.40 and loses to a team with strength 0.41, that's not an upset, but is an upset if they lose to a team with strength 0.39.

I've also toyed with the idea of summing the differences (so that an upset where the difference was 0.70 would count a lot more than an upset where the difference was 0.01).

As for your first question, I'm not sure what you mean, so I'll kind of head in that direction and see how well I do. The values are reflective of how much a team would be favored over an "average" opponent on neutral ice. To that end, the denominator is goals, and in a season with more overall goals, the best (and worst) numbers will deviate from zero by a greater degree.

To give an extreme example, if all NHL goals were suddenly worth ten points, then the strength metric values would all be 10 times their current size.
I guess the short version of what I was meaning is, does a differential of 0.7 have the same meaning in 1999 as it does in 2010? Because that'd also have an impact I guess in trying to determine the value for the 'tossup' concept because you couldn't apply a static differential to every year.

Caeldan is offline   Reply With Quote
Old
07-30-2014, 11:29 AM
  #74
Doctor No
HFB Partner
 
Doctor No's Avatar
 
Join Date: Oct 2005
Posts: 4,079
vCash: 50
Ah, I see what you mean!

My impression is that from a "chance of winning" perspective, they're comparable, and I'll walk through my line of reasoning. In seasons where there is more goal scoring (like the 1980s) the better teams will have higher absolute totals. However, because there is more goal scoring, each team's variance will be higher as well.

Is the increase in standard deviation proportional to the increase in the expected value? Maybe (and maybe more, maybe less). If team abilities are normally distributed, then the standard deviation increases proportionally to the mean.

So if we assume that team abilities are normally distributed, then it doesn't matter if the raw values are scaled differently from year to year (which they would be).

One caveat would be in years where there are really bad (usually expansion) teams. The method I use to assess a schedule strength intends to wipe out the difference if one team plays a disproportionate share of their games against overly bad (or overly good teams) - for those of you who know linear algebra, it's an attempt to find the eigenvector of the schedule and results (although I do it numerically). However, it may not be a perfect representation.

Since my team strength values are just (goal differential + schedule strength), then it's possible that these get inflated disproportionally in these kind of years. I haven't tested it rigorously in either direction.

Doctor No is offline   Reply With Quote
Old
07-30-2014, 02:36 PM
  #75
ShameOnYouZidlicky
Registered User
 
ShameOnYouZidlicky's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 1,277
vCash: 500
Fantastic thread, I love seeing how these goaltenders differ in the different situations. Just a question, so you're using opposing team shot percentage as your basis for expected saves. Could you also use the goalie's season save percentage as a basis? For example, if Jonathan Quick's season save percentage was 0.95, and he faced 40 shots in a playoff game, you would expect him to make 38 saves. I wonder how this method of determining expected saves would differ from what you're using now. I wonder if it's possible to use a combination of both even.

Also, could you show us Felix Potvin?

ShameOnYouZidlicky is offline   Reply With Quote
Reply

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 01:05 AM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"
Contact Us - HFBoards - Archive - Privacy Statement - Terms of Use - Advertise - Top - AdChoices

vBulletin Copyright ©2000 - 2015, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. 2015 All Rights Reserved.