Players value to their team at ES

[Jacob8hockey*]

Hey guys its my first time ever trying to do something like this and it may have already been done before but here is my data and I was wondering if I could get anyone's opinions on it.

Basically my data finds out who each teams best players at ES when it comes to offense and defense. Both are weighted equally so if a player is amazing offensively and poor defensively the difference it weighted out.

Here is how it is calculated:

Goals for that the player is on the ice for(ES)/Goals for whole team (ES) (Data 1)
Goals against that player on the ice (ES)/Goals against whole team (ES) (Data 2)

Data 1/Data 2 = Data 3

Data 3 x Player's ES TOI.

Heres an example for Kessel and Clarkson


69 GF 61 GA (Kessel) 181 GF 198 GA (Team)
19 GF 31 GA (Clarkson)

69/181 = 0.38 [How many teams total goals for ES they were on the ice for]
61/198 = 0.31 [How many teams total goals against ES they were on the ice for]
0.38/0.31 = 1.24 [Was it a positive outcome (1+) or a negative outcome (0.99-)]
1.24 x 17.30 (ES TOI) = 21.41 [Multiply by TOI to represent how much of the ice time was theirs to compare between players]

19/181 = .10
31/198 = .16
.10/.16 = 0.67
0.67 x 14.01 = 9.39


Here are the results for two teams so far.

:

Forwards:
Kessel
Bozak
JVR
Bodie
Kadri
Raymond
Holland
Lupul
Kulemin
Clarkson
Bolland
Ashton
Smith
McClement
D'Amigo
Orr
Smithson

Defense:
Gardiner
Phaneuf
Gunnarsson
Ranger
Rielly
Franson

:

Forward:
Kopitar
Gaborik
Toffoli
Williams
King
Carter
Stoll
Brown
Clifford
Pearson
Vey
Carcillo
Richards
Nolan
Lewis
Frattin
Fraser

Defense:
Doughty
Mitchell
Martinez
Muzzin
Voynov
Regehr
Greene


Anyone feel like the data came up with good results? I felt like the Leaf list was pretty accurate. Big flaw would be the that not every player played every game which will throw off the numbers.

Click to read more...

 

[IPreferPi]

Here are the results for the Yotes (min 10 GP played)



Forwards:
Erat
Vermette
Doan
Vrbata
Ribeiro
Boedker
Klinkhammer
Hanzal
Moss
Bissonette
McMillan
Korpikoski
Kennedy
Chipchura
Halpern
Szwarz

Defensemen:
Michalek
OEL
Murphy
Yandle
Schlemko
Morris
Stone
Summers
Klesla
Rundblad

 

[Jacob8hockey*]

Some time tomorrow I'll update this with more teams.

Someone told me it was very similar to the GF% which it is but also gives different results. Seems to be very accurate so far in terms of quality of players. Some are surprises.

 

[Episkey]

I did the Sharks (min 10GP)



Forwards:
Joe Pavelski
Brent Burns
Marty Havlat
Tomas Hertl
Logan Couture
Joe Thornton
Patrick Marleau
Tommy Wingels
James Sheppard
Matt Nieto
Tyler Kennedy
Bracken Kearns
Andrew Desjardins
Eriah Hayes
Mike Brown
John McCarthy
Adam Burish
Freddie Hamilton

Defensemen:
Marc-Edouard Vlasic
Justin Braun
Brad Stuart
Jason Demers
Matt Irwin
Dan Boyle
Scott Hannan

 

[Episkey]

And I did the Hawks just for fun (min 10GP)



Forwards
Jonathan Toews
Marian Hossa
Patrick Kane
Patrick Sharp
Peter Regin
Brandon Saad
Brandon Pirri
Andrew Shaw
Kris Versteeg
Jeremy Morin
Ben Smith
Marcus Kruger
Brandon Bollig
Michael Handzus
Bryan Bickell
Joakim Nordstrom

Defensemen
Duncan Keith
Brent Seabrook
Johnny Oduya
Niklas Hjalmarsson
Nick Leddy
Michal Rozsival
Sheldon Brookbank

 

[Jacob8hockey*]

Heres an update version which includes QOC which was requested. Changes the results a bit. Only have 3 teams completely done because I noticed an error with the TOI that I had. NOT TO COMPARE CROSS TEAMS ONLY WITHIN (for now)

 

[silkyjohnson50]

I like the idea. I'd love to see how players rank across the NHL. Like top 25 or so per each season. And if I'm not mistaken, this could be taken back quite a ways as hockey reference has these numbers right? Obviously QoC wouldn't go back nearly as far. I usually use behindthenet and I know they only go back to 07-08.

 

[Jacob8hockey*]

I like the idea. I'd love to see how players rank across the NHL. Like top 25 or so per each season. And if I'm not mistaken, this could be taken back quite a ways as hockey reference has these numbers right? Obviously QoC wouldn't go back nearly as far. I usually use behindthenet and I know they only go back to 07-08.

Yes it could be taken back as many years as the data is available. I used extraskater and nhl.com for almost it all.

I think rating players across team would be difficult without some other factor because someone like Bozak is tied with Kessel on the Leafs but would also be first place on every other team that I posted above.

I guess it is possible that Bozak's offense and defense at 5 on 5 is enough to make him on par in those standards but I just don't see it like that.

I think it could be a great tool at comparing players within their own team.

 

[Hivemind]

This is just a weighted +/- metric. Even when you incorporate QoC, it's still not going to compensate fully for player utilization. Beyond that, any goal-based metric is going to be awfully noisy given the relatively rare occurance of goals.

 

[Jacob8hockey*]

This is just a weighted +/- metric. Even when you incorporate QoC, it's still not going to compensate fully for player utilization. Beyond that, any goal-based metric is going to be awfully noisy given the relatively rare occurance of goals.

My first goal was to try and do something that proves +/- can be improved on. My original charts did have a +/- column as a comparison. My intentions were to show if +/- is accurate and how it can be improved on which is what I tried to do here. For example just because a player is a negative doesn't mean he should be rated lower because he may face more competition and receive more ice time which is what I weighed in here.

Hence why Kessel is at the top for the Leafs.

His offense (+) heavily outweighs his defense (-) which in turns makes him still the most valuable player to the Leafs.

For another player, lets say Bodie for example would have his defense (+) and his offense (-) to show how he is a good piece to the Leafs and could probably play a bigger role because of his defense outweighing his offense.


*NOTE - I do understand how +/- works even though I made it sound completely different in my post.

 

[NHLPaul]

This is just a weighted +/- metric. Even when you incorporate QoC, it's still not going to compensate fully for player utilization. Beyond that, any goal-based metric is going to be awfully noisy given the relatively rare occurance of goals.

I was thinking the same thing. Good job and thank you for all the work you put into this thread but I don't think this helps analyze player's performances too much. It's a slightly improved +/-

 

[Ohashi_Jouzu*]

The results certainly seem to "rank" the players on a given team at ES better than simple +/-, so I kind of like where this is going. Lists for the teams so far look pretty representative of last year's performance(s). Thought Bergeron was a bit far down Boston's list until I noticed that it's not actually fully sorted by any particular column.

 

[BroadwayJay*]

I was thinking the same thing. Good job and thank you for all the work you put into this thread but I don't think this helps analyze player's performances too much. It's a slightly improved +/-

I thought the same thing.

Still seems like you're better off with Corsi...

 

[Czech Your Math]

My first goal was to try and do something that proves +/- can be improved on. My original charts did have a +/- column as a comparison. My intentions were to show if +/- is accurate and how it can be improved on which is what I tried to do here. For example just because a player is a negative doesn't mean he should be rated lower because he may face more competition and receive more ice time which is what I weighed in here.

Hence why Kessel is at the top for the Leafs.

His offense (+) heavily outweighs his defense (-) which in turns makes him still the most valuable player to the Leafs.

For another player, lets say Bodie for example would have his defense (+) and his offense (-) to show how he is a good piece to the Leafs and could probably play a bigger role because of his defense outweighing his offense.

*NOTE - I do understand how +/- works even though I made it sound completely different in my post.

I like your efforts here. Not sure multiplying by TOI is the best.

Would recommend you peruse this thread before going too much further, and at least become familiar with the concept of adjusted plus-minus, which essentially uses same ratio you have discovered and multiplies it by the number of ES scoring events while the player was on the ice:

Adjusted Plus Minus