2007-08 NHL Goal Differentials

[Doctor No]

This is a little statistic I calculate each season - it's basically a linear transformation of save percentage (so if you don't like save percentage, you won't like this either, and I'm not going to argue with you about it here).


http://hfboards.com/showthread.php?t=253883 and http://hfboards.com/showthread.php?t=381137 link to the past two years of this statistic; the first link also explains a bit more about what the hell I'm doing. One big change in the calculations this year - when taking the league average save percentage, I now remove the individual goaltender's contribution.

The statistics presented:
GD (Goal Differential) - how many goals better (or worse) a netminder is than a league-average NHL goaltender facing the same number of shots.

GARG (Goals Above Replacement Goaltender) - how many goals better (or worse) a netminder is than a replacement-level NHL goaltender facing the same number of shots. I define "replacement level" to be a save percentage 0.015 below league average (this is a bit of a SWAG). Replacement level goaltenders are plentiful and easily available.

SNW/SNL (Support-Neutral Wins and Losses) - if the goaltender had received "average" goal support, how many wins and losses would he have.

I sort this table by GARG, because there's a definite value in playing at a "league average" level for a decent period of time.

Enjoy! Some interesting results here.

[FONT="Courier New"]
[B]GOALTENDER                  GD   GARG   SNWL[/B]
Vokoun Tomas             21.78  54.97  37-30
Brodeur Martin           22.49  53.83  43-34
Thomas Tim               21.83  47.80  30-23
Luongo Roberto           16.86  47.30  40-33
Bryzgalov Ilya           19.66  47.09  33-26
Biron Martin             16.92  44.90  33-26
Giguere Jean-Sebastien   20.47  43.09  34-24
Backstrom Niklas         17.43  41.87  30-24
Huet Cristobal           16.57  38.75  30-22
Ellis Dan                17.51  34.72  21-15
Lehtonen Kari            11.00  34.40  24-20
Leclaire Pascal          13.58  34.26  26-21
Lundqvist Henrik          5.81  33.16  37-34
Price Carey              13.75  32.98  22-17
Conklin Ty               14.26  29.45  18-13
Nabokov Evgeni            0.77  27.80  38-37
Legace Manny              2.82  27.54  31-29
Gerber Martin             2.17  26.46  26-26
Garon Mathieu             5.60  25.99  24-21
Miller Ryan              -5.98  25.58  35-38
Kiprusoff Miikka         -6.73  24.71  36-39
Fleury Marc-Andre        10.74  24.38  18-13
Turco Marty               0.22  23.37  30-29
Theodore Jose             1.25  21.75  26-26
Anderson Craig           13.72  21.75  10-05
LaBarbera Jason           0.88  20.99  21-21
Khabibulin Nikolai       -0.79  20.04  24-25
Osgood Chris              4.76  19.40  21-19
Hiller Jonas             10.61  19.28  11-07
Toskala Vesa             -9.49  17.87  30-34
Ward Cam                -10.34  17.71  32-35
Dubielewicz Wade          6.03  15.44  11-08
Ersberg Erik              8.13  14.91  09-05
DiPietro Rick           -12.65  14.03  28-33
Harding Josh             -0.86  11.71  14-14
Auld Alex                -1.83  10.58  15-15
Niittymaki Antero        -1.86   9.22  11-12
Roloson Dwayne           -9.75   8.31  17-20
Tellqvist Mikael         -0.85   8.26  09-10
Budaj Peter              -4.91   7.83  14-16
Johnson Brent            -0.57   6.93  08-09
Hasek Dominik            -6.38   6.45  18-22
Valiquette Stephen        1.91   6.35  06-05
Sabourin Dany            -2.86   6.08  09-11
Halak Jaroslav            3.73   5.99  03-01
Smith Mike               -6.98   5.74  16-18
Mason Chris             -14.12   5.05  20-26
Boucher Brian             2.36   3.91  03-02
Crawford Corey            2.18   3.86  02-01
Howard Jimmy              1.64   3.06  01-01
Ramo Karri               -6.63   2.85  09-12
Lalime Patrick          -10.24   2.28  13-17
Stephan Tobias            1.64   2.24  01-00
Joseph Curtis            -0.55   2.16  02-03
Pavelec Ondrej           -0.74   2.11  03-03
Elliott Brian             1.64   2.07  01-00
MacDonald Joey            0.63   1.73  01-01
Norrena Fredrik         -11.35   1.49  15-20
Sanford Curtis           -3.57   1.12  04-04
Weiman Tyler              0.91   1.06  00-00
Rinne Pekka               0.73   0.85  00-00
LaCosta Dan               0.45   0.53  00-00
Aebischer David           0.00   0.49  00-01
McElhinney Curtis        -0.37   0.40  01-01
Leighton Michael         -0.82   0.20  01-01
Beckford-Tseu Chris      -0.18  -0.05  00-00
Weekes Kevin             -2.47  -0.07  02-03
MacIntyre Drew           -1.00  -0.67  00-01
Rask Tuukka              -2.01  -0.69  02-02
Taylor Danny             -1.09  -0.94  00-00
Greiss Thomas            -2.46  -1.71  01-01
Cloutier Dan             -5.57  -1.87  03-04
Patzold Dimitri          -2.18  -1.88  00-00
Keetley Matt             -2.00  -2.00  00-00
Quick Jonathan           -3.37  -2.44  01-02
Hedberg Johan           -18.02  -2.63  13-19
Kolzig Olaf             -24.17  -2.82  22-30
Aubin Jean-Sebastien     -9.06  -3.29  05-07
Clemmensen Scott         -4.37  -3.44  00-02
Emery Ray               -15.47  -3.47  12-17
Bernier Jonathan         -5.29  -3.52  01-03
Schwarz Marek            -4.46  -4.20  00-01
Thibault Jocelyn         -8.58  -5.37  03-06
Fernandez Manny          -7.38  -5.95  01-03
Holmqvist Johan         -23.57  -6.11  17-26
Denis Marc               -9.94  -6.96  02-04
Grahame John            -14.56  -8.20  04-09
Toivonen Hannu          -17.71  -9.22  07-14
Raycroft Andrew         -16.87  -9.23  06-10
[/FONT]

Click to read more...

 

[Doctor No]

One interesting thing to note is how much "flatter" the distribution is compared to past seasons - no one really stands out in a positive fashion.

Another interesting thing: Tomas Vokoun.

 

[IME]

hehehe...raycroft is last

 

[Metallian*]

depends what stat you look at, holmqvist is last for GD

 

[Big McLargehuge]

Another interesting thing: Tomas Vokoun.

Not surprising one bit to me.


I remember one particular game against the Penguins where he was the only Panther doing anything all game and allowed 4 goals and showed his frustration and got called a cancer for showing it...

Not many players in the league can take abuse like Vokoun faced with a smile on their face.

 

[AfroThunder396]

Great insight as always Doctor No. I really like the average goal support calculation as it puts everyone on an even playing field.

Some interesting observations:
-Florida is a playoff team if Vokoun receives average goal support
-Luongo has 40 wins if he receives average goal support
-Nabokov is only one game above .500% with average goal support
-The difference between Kolzig and Huet is astounding
-Brodeur has the highest goal differential and best win/loss differential with average goal support
-DiPietro is surprisingly low
-Lundqvist, Turco, Miller, and Osgood had relatively average seasons

 

[Doctor No]

Thank you!

Yes, I was quite surprised to see where Nabokov ended up in the totals - I haven't watched as many San Jose games as I usually do this season (usually I'm playing goal at the same time), and in the back of my head I had it figured that he must be doing pretty well.

I also didn't realize just how far Kolzig had fallen this year. He's always been one of my favourites.

Dubielewicz really shone in limited time - his GARG/60 is particularly impressive.

Manny Fernandez and Marek Schwarz both had to put in quite an effort to get where they did in limited playing time.

 

[haakon84]

I always thought Vokoun was underrated

 

[LAX attack*]

Beachball Dan sneaks in, despite playing less than 10 games this season

 

[Doctor No]

Beachball Dan sneaks in, despite playing less than 10 games this season

Even though I set the criteria for "replacement level" a few years ago, it was good to have Cloutier to help validate the hypothesis.

 

[Isles4ever11]

Great insight as always Doctor No. I really like the average goal support calculation as it puts everyone on an even playing field.

Some interesting observations:
-Florida is a playoff team if Vokoun receives average goal support
-Luongo has 40 wins if he receives average goal support
-Nabokov is only one game above .500% with average goal support
-The difference between Kolzig and Huet is astounding
-Brodeur has the highest goal differential and best win/loss differential with average goal support
-DiPietro is surprisingly low
-Lundqvist, Turco, Miller, and Osgood had relatively average seasons

Yeah, he is, but I think that has a lot to do with the Islanders defense, and the fact he played hurt for almost 3 months that essentially killed his stats for the year.

Truth is he had a very bad year for his skill level. Doesn't mean that he's not a top 10 or even top 7 goalie though.

Also, great work Doctor No.

 

[Sniper28]

go timmy thomas.

 

[Doctor No]

Yeah, he is, but I think that has a lot to do with the Islanders defense, and the fact he played hurt for almost 3 months that essentially killed his stats for the year.

Truth is he had a very bad year for his skill level. Doesn't mean that he's not a top 10 or even top 7 goalie though.

Also, great work Doctor No.

Thanks!

Good points on DiPietro - his numbers this year weren't indicative of his true skill level (my opinion, of course). Give him a full summer to get healthy and he'll bounce back.

(DiPietro's historic GARG totals are 55.36 (2006-07), 24.29 (2005-06), 19.27 (2004-05)).

 

[El_Loco_Avs]

interesting stuff!

 

[Trepanated]

Kind of odd that Fleury and Conklin end up with the exact same SNWL.

This is good stuff. The way it's presented, it does tend to favor the workhorse goalies. I guess that's what you were going for. The SNWL puts things in the context of opportunity, but would be hard to sort by. I'd like to see an additional column, perhaps something like SNWL winning percentage, to make it clear how much each goalie did with the opportunities afforded to them.

 

[haakon84]

One interesting thing to note is how much "flatter" the distribution is compared to past seasons - no one really stands out in a positive fashion.

Another interesting thing: Tomas Vokoun.

Maybe the goal differential is lower overall than previous years because their have been fewer goals scored league-wide.

 

[AfroThunder396]

Dr. No, could you explain how exactly you determine a goalies record with "average goal support"?

 

[Doctor No]

Dr. No, could you explain how exactly you determine a goalies record with "average goal support"?

I apologize - I missed this post back when it was posted, and only noticed it when I linked to it just now.

Here are the basics - first, I ignore ties. That makes things a hell of a lot simpler and doesn't really seem to bother anyone.

I'll introduce Bill James' Pythagorean Theorem - in baseball, if you know how many runs scored and runs allowed a team has in a season, you can effectively predict their record using the formula Winning Percentage = Runs^2 / (Runs^2 + RunsAllowed^2). The formula works pretty well for hockey - it's not perfect, but then again, we're starting from save percentage, and that's already not perfect, so there we are.

So the hockey formula becomes Winning Percentage = GF^2 / (GF^2 + GA^2).

Goals Against is easy to find - that's just the Goals Against that a goaltender allows.

In a support-neutral context, Goals For is basically asking - if this goaltender faced "league average" goaltending in similar situations, how many goals would his team score? The simple answer is that you take the number of shots the goaltender faced and multiply it by (one minus the league-average save percentage).

Once you have GF and GA, you can get Support-Neutral Winning Percentage using the formula above.

Then you can take the SNW% and apply it to how many decisions the goaltender had over the season.

Here's an example, using Martin Brodeur's 2007-08 campaign. Brodeur had a record of 44-27-6, allowing 168 goals on 2089 shots. League-wide save percentage was 0.90881 (I remove Brodeur's shots and saves in making this calculation because Brodeur doesn't play against himself, and he shouldn't get penalized for playing well).

Goals Against = 168.

Goals For = (1-0.90881) * 2089 = 190.5 (Note that I would ordinarily carry as many decimal places as I could, and then round at the end)

SNW% = (190.5^2) / (168^2 + 190.5^2) = 0.5625.

Brodeur had 77 decisions, so his support-neutral wins would be (0.5625 * 77) = 43.3 (rounded to 43). And his support-neutral losses would be 77-43=34.

So Brodeur's 2007-08 support-neutral won-loss record would be 43-34.

 

[Douggy]

So it's pretty clear just glancing at the charts that Andrew Raycroft has been the worst goalie in the league for the past three years.

Good work Doctor No. I love seeing hardcore statistical analysis applied to hockey!!

 

[Doctor No]

Always my pleasure! It suits my dual nature as a math nerd and a hockey nerd.

 

[Flannelman]

this is pretty great, I hadn't looked at it in the past - but surely will in the future - if you plan to continue to do this.

 

[Douggy]

I think this thread deserves a bump. How about crunching the numbers on some other years?

 

[RocheBag]

So hypothetically, Tim Thomas allowed 47.8 less goals than some average Goalie they signed to replace him would have?

This is really interesting stuff, great work.

 

[Gobias Industries]

So is Dan Ellis overhyped?..

My brain hurts from this list, half of it runs completely contrary to my expectations...

 

[Doctor No]

I think this thread deserves a bump. How about crunching the numbers on some other years?

I missed this when you first posted it - mea culpa!

I've posted 2005-06 and 2006-07 on here in the past. And now for the good news - I'm refitting my website so that I can put up all of these things on the individual biography pages. It's going to make it a lot easier for people to find them (and comment/critique them, which in turn leads to a better work product).

It should be done by the end of this month.