Jump to content
Jays Centre
  • Create Account

Recommended Posts

Posted

So, okay, I created this new metric that incorporates SH% into dCorsi called delta goal differential (dGD). The methodology is as follows:

 

1) Instead of multiplying the player’s actual CA by his actual CSv%, I multiply it with his opponent’s Corsi on-ice SH% (OppCSh%; when he wasn’t on the ice). This would be the expected SV% if the player had a “league average” goaltender behind him accounting for quality of opposition if we assume players outside the goaltenders have negligible effects on their team’s SV% (There is possibly a small effect, but I’m assuming it’s negligible right now). I'll call it "Goalie-Independent Goals Against" or "GIGA" for now. I get the goal differential by subtracting the player’s actual GF with GIGA instead of the old GA to get me "Goalie-Independent Goal Differential" or "GIGD".

 

2) Then, I calculate the expected goal differential by: multiplying the player’s exCF (based on usage) with his teammate’s on-ice Corsi SH% when he is not on the ice with them; multiply the player’s exCA with his opponent’s on-ice SH% when he is not on (like mentioned in 1 when calculating GIGA); and, calculate the expected goal differential based on this.

 

3) Finally, I calculate dGD between the player’s “actual” GD (GIGD) and his “expected” GD.

 

So, without further ado, here are the top/bottom 10 forwards/defencemen in the past 6 years using dGD (5v5 only; minimum 50 TOI):

 

Top 10 forwards (dGD):

 

1 SEDIN, HENRIK 131.4

2 CROSBY, SIDNEY 114.6

3 DATSYUK, PAVEL 96.4

4 SEDIN, DANIEL 93.3

5 TOEWS, JONATHAN 89.6

6 THORNTON, JOE 80.2

7 OVECHKIN, ALEX 75.9

8 SPEZZA, JASON 68.6

9 BACKSTROM, NICKLAS 66.2

10 RYAN, BOBBY 66.2

 

Bottom 10 forwards:

 

893 PAHLSSON, SAMUEL -39.3

894 CROMBEEN, BJ -40.0

895 NYSTROM, ERIC -40.1

896 MOREAU, ETHAN -40.5

897 BROUWER, TROY -41.4

898 MARCHANT, TODD -41.6

899 THORNTON, SHAWN -41.7

900 HANDZUS, MICHAL -48.3

901 ADAMS, CRAIG -52.1

902 MOEN, TRAVIS -63.3

 

Top 10 defencemen:

 

1 CHARA, ZDENO 88.4

2 VISNOVSKY, LUBOMIR 65.5

3 HEDMAN, VICTOR 55.2

4 LETANG, KRIS 54.4

5 ROBIDAS, STEPHANE 54.2

6 HAMHUIS, DAN 51.6

7 CARLE, MATT 48.6

8 COBURN, BRAYDON 42.4

9 GREEN, MIKE 40.8

10 GOLIGOSKI, ALEX 40.3

 

Bottom 10 defencemen:

 

475 JOVANOVSKI, ED -33.7

476 FOWLER, CAM -37.6

477 EDLER, ALEXANDER -38.5

478 ZANON, GREG -38.6

479 SBISA, LUCA -39.9

480 PHILLIPS, CHRIS -42.5

481 STUART, BRAD -46.9

482 MORRIS, DEREK -52.6

483 GIRARDI, DAN -62.3

484 JOHNSON, JACK -64.8

 

Edler was a big surprise in the bottom 10 list. I knew Girardi was overrated, but I didn't realize he was this bad. The list for defencemen make me realize that SH% may not be as important a tool compared to forwards (which, to me, makes sense; forwards tend to have greater spreads in SH% than dmen).

Posted
Amazing work, Frag.

 

Thanks! It's been a lot of work. I'm just going to re-post this here from PPP, with some modifications from the original post:

 

 

 

So, I ran regressions btwn 3-year data sets of forwards (07-10 and 10-13; minimum 2000 TOI per set). The variables I ran included: goal differential (GD), goal differential where GA = CA * Opponent on-ice Corsi SH% (GIGD for goalie-independent goal differential), delta GD (dGD), delta Corsi (dCorsi), GD20, GIGD20, and Corsi For percentage (CF%). For GD1, highest to lowest adjusted R^2 value (all significant; p < 0.01; n = 144):

 

1) GIGD: 17.9%

2) GIGD20: 17.8%

3) CF%: 14.8%

4) GD: 13.4%

5) GD20: 13.1%

6) dGD: 7.5%

7) dCorsi: 2.6%

 

Interesting to see that when you adjust the goal differential for Opponent’s on-ice Corsi SH%, it actually improves the model relative to both CF% and GD1. I guess the reason for this is that since I adjusted for goaltending behind them, GIGD better reflects what the player actually did that season than GD. Though, to be better than CF% was a bit of a surprise, but I also guess that it shows that we really shouldn’t discount SH% when evaluating forwards. Also, as I expected, dCorsi/dGD were not predictive stats of either GD or GIGD (though, dGD did slightly better than dCorsi).

 

Because of the above results, I was curious to see what happens if I ran the above variables with GIGD instead of GD (all significant; p < 0.01):

 

1) GIGD20: 30.8%

2) GIGD: 30.3%

3) CF%: 22.2%

4) GD: 19.0%

5) GD20: 18.7%

6) dGD: 13.3%

7) dCorsi: 3.0%

 

Interesting. Finally, I wanted to check the adjusted R^2 values btwn dGD (or dGD20) and GD/GIGD or GD20/GIGD20 relative to dCorsi, or CF% within each set. First, 2007-10:

 

CF%:

 

1) GD1: 40.6%

2) GD2: 46.3%

 

dCorsi:

 

1) GD1: 14.5%

2) GD2: 9.5%

 

dGD:

 

1) GD1: 49.4%

2) GD2: 58.3%

 

dGD20:

 

1) GD20 (1): 44.1%

2) GD20 (2): 51.1%

 

GD2:

 

GD1: 73.3%

 

Next, 2010-13:

 

CF%:

 

1) GD1: 34.0%

2) GD2: 45.0%

 

dCorsi:

 

1) GD1: 6.7%

2) GD2: 7.8%

 

dGD:

 

1) GD1: 44.8%

2) GD2: 62.7%

 

dGD20:

 

1) GD20 (1): 43.4%

2) GD20 (2): 61.8%

 

GD2:

 

GD1: 75.9%

 

Maybe we should adjust the GD where GA is adjusted for Opponent’s on-ice Corsi SH%? Right now, I am tired and want some rest. :P

 

I should state, this is based off of regressions using TOTAL TOI, not average TOI. Average TOI should improve the models for dGD/dCorsi slightly.

 

 

I'm going to do further tests on dGD and GIGD, but I wonder if GIGD may be a hockey version of FIP.

Posted
Wow. You may have outdone yourself on that one. Nice work!

 

Thanks!

 

So, essentially, the top 20 players in the past 6 years (excluding goaltenders) in dGD are:

 

1 SEDIN, HENRIK 131.4

2 CROSBY, SIDNEY 114.6

3 DATSYUK, PAVEL 96.4

4 SEDIN, DANIEL 93.3

5 TOEWS, JONATHAN 89.6

6 CHARA, ZDENO 88.4

7 THORNTON, JOE 80.2

8 OVECHKIN, ALEX 75.9

9 SPEZZA, JASON 68.6

10 BACKSTROM, NICKLAS 66.2

11 RYAN, BOBBY 66.2

12 HORTON, NATHAN 66.0

13 SEMIN, ALEXANDER 66.0

14 VISNOVSKY, LUBOMIR 65.5

15 ST._LOUIS, MARTIN 64.3

16 ELIAS, PATRIK 60.5

17 GETZLAF, RYAN 59.9

18 MALKIN, EVGENI 57.2

19 KUNITZ, CHRIS 55.6

20 HEDMAN, VICTOR 55.2

 

Bottom 20 are:

 

1 JOHNSON, JACK -64.8

2 MOEN, TRAVIS -63.3

3 GIRARDI, DAN -62.3

4 MORRIS, DEREK -52.6

5 ADAMS, CRAIG -52.1

6 HANDZUS, MICHAL -48.3

7 STUART, BRAD -46.9

8 PHILLIPS, CHRIS -42.5

9 THORNTON, SHAWN -41.7

10 MARCHANT, TODD -41.6

11 BROUWER, TROY -41.4

12 MOREAU, ETHAN -40.5

13 NYSTROM, ERIC -40.1

14 CROMBEEN, BJ -40.0

15 SBISA, LUCA -39.9

16 PAHLSSON, SAMUEL -39.3

17 CAMPBELL, GREGORY -39.3

18 ZANON, GREG -38.6

19 MCCLEMENT, JAY -38.5

20 EDLER, ALEXANDER -38.5

 

*Itallics = Defenceman

  • 2 weeks later...
Posted

Current free agents:

 

- Tom Gilbert

- Ian White

- Tim Thomas

- Ron Hainsey

 

Some good ones left. LOL @ Habs signing Douglas Murray instead of Gilbert or White!

Old-Timey Member
Posted

I wish the Leafs had cap room to go after Damien Brunner or Tom Gilbert or Ian White.

 

Oh, and YAYGOODNEWS! About Nonis, even! The Leafs seem to be prepared to offer Kessel around 8/70, and Kessel does want to stay in TO. All signs point to us keeping him, which makes me a happy camper. And fwiw, Nonis has spoken on Dion probably staying.

Posted
I wish the Leafs had cap room to go after Damien Brunner or Tom Gilbert or Ian White.

 

Oh, and YAYGOODNEWS! About Nonis, even! The Leafs seem to be prepared to offer Kessel around 8/70, and Kessel does want to stay in TO. All signs point to us keeping him, which makes me a happy camper. And fwiw, Nonis has spoken on Dion probably staying.

 

Oilers and Toni Rajala part ways, too. That's another one on the FA market.

Posted
At least it wasn't the Leafs. ;)

 

I like Travis Yost, though.

 

In addition, I found this ridiculously inane article about Grabovski. First, it focuses a lot on him not being a "team-player" or stupid s*** like that. Then, the author gets the team who signed Ribeiro wrong. Finally, the worst part, believe it or not, was in the author's bio:

 

I am near a PhD in Economics and I have eight years work experience as a statistician.

 

*shudders*

Old-Timey Member
Posted

Eugene Melnyk is both crazy and delusional. Borderline insane, IMO.

 

The whole Matt Cooke incident "investigation" last season was proof enough for me.

Posted
Eugene Melnyk is both crazy and delusional. Borderline insane, IMO.

 

The whole Matt Cooke incident "investigation" last season was proof enough for me.

 

roflbot.jpg

Posted
I keep hearing about what a great hockey mind the Leafs' new president is, but so far he's allowed the team to buy Grabovski out, give stupid money to David Clarkson, and extend that idiot Nonis for like 5 years.

 

AA and Dave Nonis ugh. Instead of working on getting new contracts for good young players like Kadri and Franson, they go out and blow their load on Bozak and Clarkson lol. Hey let us buyout our best possession C and bring back Tyler "possession black hole" Bozak.

 

Why waste assets on Bernier? I love how this moron pretty much flipped the bird to the guy who was the main reason they even sniffed the playoffs. What a f***ing idiot. Good luck trying to squeeze Franson and Kadri under the cap now.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
The Jays Centre Caretaker Fund
The Jays Centre Caretaker Fund

You all care about this site. The next step is caring for it. We’re asking you to caretake this site so it can remain the premier Blue Jays community on the internet.

×
×
  • Create New...