possible way to predict player worth?

[mattkaminski15]

I thought a lot about this lately. Teams want wins right? Well what if we take key parts of the game that attribute to wins (blocked shots, killed penalties, etc.) And combine their average per game "points" (points being given for each thing done to help team secure a win) and see if their points go up or down in a season. It wont be exact but I feel like it would work.


If this is already being done feel free to shun this thread.

Click to read more...

 

[McStimperi]

Blocked shots is problematic. It equals to no possession. And puck possession tends to lead to wins.

 

[insomniac]

Blocked shots is problematic. It equals to no possession. And puck possession tends to lead to wins.

Shots against also indicates a lack of possession, as does goals against. Blocked shots have a higher likelihood of the defending team recovering possession, which is important. That seems to get lost in the new-age of possession metrics - you definitely aren't going to have possession constantly, and if possession really is so important, then by extension so is recovering possession. The game doesn't end because you don't have the puck; you just need to get it back.

 

[mattkaminski15]

Shots against also indicates a lack of possession, as does goals against. Blocked shots have a higher likelihood of the defending team recovering possession, which is important. That seems to get lost in the new-age of possession metrics - you definitely aren't going to have possession constantly, and if possession really is so important, then by extension so is recovering possession. The game doesn't end because you don't have the puck; you just need to get it back.

There are players that are valued as defensive players. Players that aren't gifted to just put the puck in the net. They dont need possession to secure a win for a team. This is just a work in progress but I feel like its actually a good way to detect player worth. The teams that want scorers can have a seperate system to find their scoring 'worth'.

Let's say player X is a shut down defenseman. Doesn't want shots onhis goalie.

Let's say X gets 10 blocked shots; I'll award him 10 points towards a win.

Let's also say he had 2 penalties that he killed; I'll award him 4 points. 2 for each penalty killed and keeping the puck out of the net.

1(10)+2(2)=14 points.

He earned 14 points towards that teams win.

Find what all the other players earned and find which players have highest point numbers.

It could use some tweaking but I think it would work.

 

[Thesensation19*]

As much as I enjoy the review and discussion of advanced stats like Corsi, we cant assume that it is the end all for possession. I mean, first off Corsi assumes that all shots are the same. No discussion of quality shots. Apparently numbers say differently but I know the different when a team shoots 60x a game and its all from poor places and a team that shoots 25x and its all high quality chances.

Plus, a guy like Ovechkin has a very high Corsi. No?
No offense but I would hate for the guy to be on my team. Very low percentage of shots, very poor decision making recently. Very predictable. He has a very high Corsi rating because he uses all of his energy for sprints down the wing, shoots it quickly and thats it. No one in stats cares if there is a guy in his face with a stick on his stick. No one cares that its from 20' out and hes attacking the net on a 2 v 4 or 1 v 2. They just care that he shoots it on net. Thats a poor mans way of thinking.

Im all for the evaluation of shots for and shots against but there are other valuable stats that can also be measured. I wonder how many shots Ovechkin has blocked that never reach the net. I guess you can have Fenwick - Corsi to see that, no?

A guy blocking shots is very important to a team I feel.

 

[charlie1]

As much as I enjoy the review and discussion of advanced stats like Corsi, we cant assume that it is the end all for possession. I mean, first off Corsi assumes that all shots are the same. No discussion of quality shots. Apparently numbers say differently but I know the different when a team shoots 60x a game and its all from poor places and a team that shoots 25x and its all high quality chances.

Plus, a guy like Ovechkin has a very high Corsi. No?

No offense but I would hate for the guy to be on my team. Very low percentage of shots, very poor decision making recently. Very predictable. He has a very high Corsi rating because he uses all of his energy for sprints down the wing, shoots it quickly and thats it. No one in stats cares if there is a guy in his face with a stick on his stick. No one cares that its from 20' out and hes attacking the net on a 2 v 4 or 1 v 2. They just care that he shoots it on net. Thats a poor mans way of thinking.

Im all for the evaluation of shots for and shots against but there are other valuable stats that can also be measured. I wonder how many shots Ovechkin has blocked that never reach the net. I guess you can have Fenwick - Corsi to see that, no?

A guy blocking shots is very important to a team I feel.

No, no he doesn't.

 

[bigd]

Blocked shots is problematic. It equals to no possession. And puck possession tends to lead to wins.

If blocked shots aren't important then I'm sure that coaches wouldn't let them do it. It presents a high rate of injury. So regardless of possession, coaches and players feel blocking shots are a valuable part of the game.

 

[TheHockeyRant]

Corsi does not directly correlate to possession. As possession means "to have."
What Corsi/ Fenwick do is calculate shot differentials, which is far better than relying on plus minus. This can lead to a correlation of possession but we cannot be sure. Remember a team may have the puck in the zone for a solid 1 minute but take 0 shots (pass happy)

If you remember the World Cup they are able to track REAL POSSESSION, because whoever has the ball, clearly possesses it. In hockey there are variables. Scrums on the boards, dump-ins, 50-50 pucks after shots/ in front of the net. All of those situations can't exactly be attributed to "possession" because no one clearly has possession.

Corsi/ Fenwick should get away from this possession term and focus on being what it is ( a better plus minus shot counting statistic)

 

[charlie1]

Corsi does not directly correlate to possession.

TOA = Time On Attack (i.e., possesion).



Source: http://www.pensionplanpuppets.com/2013/9/16/4727746/leafs-attack-time-at-the-halfway-mark

One-sentence summary:

Some folks are going to be hard-headed no matter what you put in front of them, but I think my findings are simple yet illuminating; stats like Corsi and Fenwick are very effective proxies for puck possession.

 

[TheHockeyRant]

TOA = Time On Attack (i.e., possesion).



Source: http://www.pensionplanpuppets.com/2013/9/16/4727746/leafs-attack-time-at-the-halfway-mark

One-sentence summary:

Key word is "directly"
Does it often correlate, sure. But no one is tracking who holds the puck for the duration of the game. Again possession means to have not "time on attack" A team could be in the offensive zone for one solid minute and produce 0 shots. That is possession, but not Corsi boosting.

 

[Thesensation19*]

No, no he doesn't.

From what I can find, he has a 51.8% Corsi Rating. Looks to be near the top of the list in the NHL. 3rd on his team. Excellent PP Corsi as well...

I didnt even question it when I saw it. The dude leads the league in shots every year. So he would produce a very high + for Corsi.

 

[charlie1]

49.5 CF% (5v5).

485th out of 783 players.

59.5% Offensive zone-start.

Those are some very poor corsi numbers.

http://war-on-ice.com/playertable.html

 

[charlie1]

Key word is "directly"
Does it often correlate, sure. But no one is tracking who holds the puck for the duration of the game. Again possession means to have not "time on attack" A team could be in the offensive zone for one solid minute and produce 0 shots. That is possession, but not Corsi boosting.

By "does not directly correlate to" you appear to mean "is not."

 

[Patman]

I thought a lot about this lately. Teams want wins right? Well what if we take key parts of the game that attribute to wins (blocked shots, killed penalties, etc.) And combine their average per game "points" (points being given for each thing done to help team secure a win) and see if their points go up or down in a season. It wont be exact but I feel like it would work.


If this is already being done feel free to shun this thread.

Just a general reminder to the audience. Anybody can come up with a metric. Its up to the creator and the audience to decide that it has any particular worth.

----

IMO, any time you have something simple enough to explain that ties into game mechanics then you have something potentially useful.

Also, notions of "value" remain in the abstract.... and a bit cart before horse. Define value then you present yourself with the analytic challenge.

 

[mattkaminski15]

Just a general reminder to the audience. Anybody can come up with a metric. Its up to the creator an the audience to decide that it has any particular worth.

----

IMO, any time you have something simple enough to explain that ties into game mechanics then you have something potentially useful.

Also, notions of "value" remain in the abstract.... and a bit cart before horse. Define value then you present yourself with the analytic challenge.

Value as in.. Say.. Usefulness to the team. Like i said. Teams don't buy players, they buy wins. If a player can create a win their 'value' increases.