NHL to add Corsi/Fenwick to league stats

[rkhum]

If Corsi sucks, then please tell me what advanced ARE good metrics for evaluating players and teams AND where to find.

 

[Doctor No]

Corsi/Fenwick are pretty decent in conjunction with zone starts and strength of opposition.

I don't know where they are offhand, other than Rob Vollman's Hockey Abstract books. I'm sure someone will speak up.

 

[charlie1]

war-on-ice.com

Corsi doesn't suck. Some people just put too much weight on it.

 

[hatterson]

stats.hockeyanalysis.com (or their new UI at puckalytics.com) has a good set of stats, including full WOWY charts/data

behindthenet.ca also has a good set of data.

 

[rkhum]

I really, really want to use advanced stats.
It's very subjective and hard without advanced to measure how good certain positional players are (defense).

Yet, even after reading the primer, I'm still not clear.
I want to know what stats are best to evaluate how good a player is, and how good a team is as well.

Any help? That premier isn't that good.

 

[Devilsfan992]

So is SAT supposed to be CORSI/FENWICK?

 

[Goose of Reason]

So is SAT supposed to be CORSI/FENWICK?

SAT is Corsi, USAT is Fenwick.

I have one question that's confusing me looking at the stats. The SAT Rel says it's a players shots-his team's shots when he's not on the ice, so how does a player like Nic Deslauriers have a 712? I'm sure I'm missing something fairly obvious but that one sticks out for me.

EDIT: Are they just using absolute values?

 

[Ogopogo*]

If Corsi sucks, then please tell me what advanced ARE good metrics for evaluating players and teams AND where to find.

Nobody tracks what would be useful. What makes advanced stats useful are when you measure individual players in isolation - Corsi and Fenwick both do not and, as a result, provide nothing of value for individual players.

Useful stats would be real possession stats - number of touches. If a player touches the puck 50 times in a game, he controls the game a heck of a lot more than player who touches the puck 10 times. You combine that with a real individual measure of performance (when a player makes a good play, something above the norm, give him a +. goal or no goal, shot or no shot. If a player makes a bad play, give him a -.)

You take those two items and you will have something infinitely more useful to evaluate individual players than Corsi and Fenwick. Using a team stat to measure an individual is complete foolishness.

 

[Sabretooth]

SAT is Corsi, USAT is Fenwick.

I have one question that's confusing me looking at the stats. The SAT Rel says it's a players shots-his team's shots when he's not on the ice, so how does a player like Nic Deslauriers have a 712? I'm sure I'm missing something fairly obvious but that one sticks out for me.

EDIT: Are they just using absolute values?

SATrel is players SAT (not shots nor SAT-for) minus team SAT while player is not on the ice. Players SAT is "shot attempts for minus shot attempts against while player is on the ice". It is worded very badly in the rollover description of SATrel.

I came here looking for a discussion on this because SATrel doesn't seem to be a very useful stat as NHL.com is presenting it. It needs to be at least normalized to a per minute/per 20/per 60 basis.

If you assume a team like the sabres who are always negative SAT give up SAT against at a pretty consistent rate, then someone who plays 15 minutes is obviously going to have a less bad SAT than the team the rest of the 45 minutes he's not on the ice. In the case of the sabres, you have players with +719 SATrel because Nic deslauriers is only a -317 SAT, while the Sabres must have a -1036 SAT when Nic is not on the ice.

On the flip side, a team that has a strong positive SAT, a player who plays 15 minutes and has a positive SAT is obviously still going to have a less good SAT than the rest of the team in the other 45 minutes. So good possession players can have a negative SATrel. All but 2 players on the Kings have a negative SATrel for example, despite the fact the Kings are the best in the league in SAT. Kopitar has a +226 SAT, yet is a -25 SATrel. That means in the ~40mins/game that Kopitar is not on the ice, the Kings have managed to be +251. In other words, the Kings generate shot attempts nearly twice as often when Kopitar is on the ice vs. when he's off, which is really an amazingly positive stat, yet shows up as a negative when dealing with the absolute numbers not corrected for TOI.

So SATrel as presented by NHL.com is worse than a useless stat. Its incredibly misleasing. I do find it amusing that Nic Deslauriers has the best SATrel in the league, tho

 

[Ho Borvat]

Really like that the NHL is weighing things /60 minutes.

Nothing advanced about that, but things like Goals/Assists/Points per 60 adds some context to how impressive someones production may be.

 

[Epsilon]

The NHL needs to get rid of the improperly-scaled bar graphs on the Leaders page, that's just amateur-hour statistical content there:

http://www.nhl.com/stats/leaders?navid=nav-sts-league#