from Amazon
Wednesday, May 19, 2010
How much talent is there with NHL goalies?
Thanks to Tom Awad for the data, here’s what I did.
I took all goalies that faced at least 1200 shots over the last three seasons in 5-on-5 play. I calculated each goalie’s save percentage (Vokoum was at .933), compared it to the league average of .921, and figured his z-score of 2.82. That is, his performance was 2.82 standard deviations from the mean.
That by itself tells us nothing, as we expect someone to lead at some level. What interests us is to take the standard deviation of all the z-scores. If there was no such thing as goalie talent, we should get back 1.00. If there is goalie talent, it’ll be higher than 1.00. The more talent, the higher the number.
In the data I have, the standard deviation of the z-score is 1.25. So, yes, there is NHL goalie talent. This is based on 2650 shots faced.
What can we do with this? We will follow the exact same methodology as I used for measuring HR talent for pitchers.
With 2650 shots faced, and a .921 save percentage, the standard deviation of the binomial is .0052. We put all this into our trusty equation:
1.25^2 = variance(true)/(.0052^2) + 1
variance(true) = .0039^2
And there we have it. The spread in goalie skill per 5-on-5 shot is one standard deviation equal to .0039 goals per shot.
Now, let’s figure out our regression equation.
1/(1-r^2) = 1.25^2
And so, r=.60 when n=2650
We also have a general equation that says:
r = n / (x+n)
And in this case:
.60 = 2650 / (x+2650)
That makes x = 1790
Therefore, our correlation equation is:
r = Shots / (Shots + 1790)
Our regression equation is 1-r. And so:
regression rate = 1790 / (Shots + 1790)
If you have 1790 shots, you regress the SV% 50% toward the mean. If you have 3580 shots, you regress 33% toward the mean. Simple enough?
Vokoum had a .933 save percentage on 4652 shots. This means we regress his save percentage 28% toward the league mean of .921. Therefore, our best estimate of his talent level, limit to this data, is .929.
This is the list of all the goalies, ranked by their talent level, and how many goals they are worth per 29 even-strength shots.
talent goals Goalie
0.929 0.23 TOMAS VOKOUN
0.929 0.23 TIMOTHY THOMAS
0.928 0.19 JONAS HILLER
0.927 0.17 ROBERTO LUONGO
0.926 0.15 CRAIG ANDERSON
0.926 0.12 JAROSLAV HALAK
0.925 0.11 MARTIN BRODEUR
0.925 0.10 PEKKA RINNE
0.925 0.10 TY CONKLIN
0.925 0.10 HENRIK LUNDQVIST
0.924 0.08 RYAN MILLER
0.924 0.08 ILJA BRYZGALOV
0.924 0.06 JEAN-SEBASTIEN GIGUERE
0.923 0.05 KARI LEHTONEN
0.923 0.05 EVGENI NABOKOV
0.923 0.05 JAMES HOWARD
0.923 0.04 NIKOLAI KHABIBULIN
0.923 0.04 CAREY PRICE
0.922 0.02 NIKLAS BACKSTROM
0.922 0.02 CAM WARD
0.921 0.00 DAN ELLIS
0.921 0.00 MARC-ANDRE FLEURY
0.921 0.00 SCOTT CLEMMENSEN
0.921 (0.00) JOSH HARDING
0.921 (0.00) MARTIN BIRON
0.921 (0.01) ANTERO NIITTYMAKI
0.921 (0.02) JONATHAN QUICK
0.920 (0.03) CHRIS MASON
0.920 (0.03) MIIKKA KIPRUSOFF
0.920 (0.04) JOSE THEODORE
0.920 (0.05) STEVE MASON
0.920 (0.05) ALEX AULD
0.919 (0.06) RICK DIPIETRO
0.919 (0.07) MANNY LEGACE
0.919 (0.07) MARTIN GERBER
0.919 (0.07) JASON LABARBERA
0.919 (0.07) DWAYNE ROLOSON
0.918 (0.09) MARTY TURCO
0.918 (0.09) MIKE SMITH
0.918 (0.10) PETER BUDAJ
0.918 (0.10) CRISTOBAL HUET
0.918 (0.10) JOEY MACDONALD
0.917 (0.14) ONDREJ PAVELEC
0.917 (0.14) JOHAN HEDBERG
0.916 (0.15) BRIAN ELLIOTT
0.916 (0.16) MATHIEU GARON
0.916 (0.17) PATRICK LALIME
0.915 (0.19) CHRIS OSGOOD
0.914 (0.20) PASCAL LECLAIRE
0.914 (0.21) VESA TOSKALA
0.914 (0.23) ANDREW RAYCROFT