A look at which players generate the most quality shots (5v5 only)

krutovsdonut

eeyore
Sep 25, 2016
16,772
9,427



Pretty damn interesting, still an early take, but still really interesting seeing those players at the top
Domi, Petry and mostly Draisaitl took big steps this year.

some would say Petry is now MTL #1 over Weber and Draisaitl leading over Mcdavid on his line


uh, there are four habs in the top eight. i think pretty obviously that the stat guy in montreal is operating to a different standard than the rest of the league.
 

TomasHertlsRooster

Don’t say eye test when you mean points
May 14, 2012
33,360
25,417
Fremont, CA
According to hockey reference it does.

X goals for is defined as follows:

"▼ -- 'Expected Goals For' given where shots came from, for and against, while this player was on the ice at even strength.
It's based on where the shots are coming from, compared to the league-wide shooting percentage for that shot location."

So Alex Ovechkin is not expected to score any more goals than any random scrub player relative to average shot distance and location, meaning, xGoals for is deliberately agnostic regarding the quality and caliber of a player's shot.

First off, you’re looking at the definition of a different expected goal model from the one in the OP. Evolving Hockey’s model contains more info than just location.

Second, it doesn’t assume that all shooters are equal. Everybody who understands expected goals understands that some players will consistently out-perform their expected goal totals and others will consistently under-perform them.

Does shots on goal also assume that all shooters are equal? Is shots on goal a terrible stat for that reason?
 

ponder

Registered User
Jul 11, 2007
16,928
6,217
Vancouver
4 Habs in the top 8 basically tell you that this is a garbage/meaningless analysis. Guessing it's based on an incredibly small sample size? If it's this season only, teams have only played 3-7 games, and all stats are basically meaningless over 3-7 games.
 

Midnight Judges

HFBoards Sponsor
Sponsor
Feb 10, 2010
13,491
10,046
Second, it doesn’t assume that all shooters are equal. Everybody who understands expected goals understands that some players will consistently out-perform their expected goal totals and others will consistently under-perform them.

Right. Guys with great shots will consistently out-perform expected goals for (as calculated by hockey reference.com).

You assume people realize this. I am pretty sure they do not. The guy who created that WAR calculation certainly didn't.
 

TomasHertlsRooster

Don’t say eye test when you mean points
May 14, 2012
33,360
25,417
Fremont, CA
Right. Guys with great shots will consistently out-perform expected goals for (as calculated by hockey reference.com).

You assume people realize this. I am pretty sure they do not. The guy who created that WAR calculation certainly didn't.

What WAR calculation are you talking about?
 

HockeyDBspecialist

Habs 2019 cup champ
Jan 30, 2018
6,000
3,386
Montreal
4 Habs in the top 8 basically tell you that this is a garbage/meaningless analysis. Guessing it's based on an incredibly small sample size? If it's this season only, teams have only played 3-7 games, and all stats are basically meaningless over 3-7 games.
or that the habs take more shots from dangerous places.... like front of the net, which they did...
 

supsens

Registered User
Oct 6, 2013
6,577
2,000
Right. Guys with great shots will consistently out-perform expected goals for (as calculated by hockey reference.com).

You assume people realize this. I am pretty sure they do not. The guy who created that WAR calculation certainly didn't.

Isn't 'guys with great shots out-performing' the entire point of the stat?
 

Midnight Judges

HFBoards Sponsor
Sponsor
Feb 10, 2010
13,491
10,046
First off, you’re looking at the definition of a different expected goal model from the one in the OP. Evolving Hockey’s model contains more info than just location.

Evolving hockey considered something they refer to as "shooting talent" but they haven't gotten it right just yet, and so it's left out there as well.

Both of us wanted to include a shooting talent variable; however, the algorithm did not feel the same. This variable (in each model) was never used in any decision tree that was generated - essentially, it determined that this variable added no value for predicting goals. Of course there’s the chance our method was flawed, which resulted in a less than ideal shooting talent variable for inclusion. Even so, we’re confident it wasn’t done so incorrectly as to be completely ignored. It makes intuitive sense that shooting talent should be included in xG, at least we think it does.

RPubs - A New Expected Goal Model for Predicting Goals in the NHL
 

Midnight Judges

HFBoards Sponsor
Sponsor
Feb 10, 2010
13,491
10,046
Isn't 'guys with great shots out-performing' the entire point of the stat?

If the purpose of expected goals for is to compare to actual goals for and determine shot quality, then sure. If that were the case then actual goals for would be beside it in virtually every instance, and actually we'd be focusing on "x goals difference" or something like that as the primary output.

I am not seeing anyone do that, but now that you bring it up, maybe they should!
 

TomasHertlsRooster

Don’t say eye test when you mean points
May 14, 2012
33,360
25,417
Fremont, CA
Evolving hockey considered something they refer to as "shooting talent" but they haven't gotten it right just yet, and so it's left out there as well.



RPubs - A New Expected Goal Model for Predicting Goals in the NHL

I didn’t say they accounted for shooting talent. I said they accounted for more than just location.

If the purpose of expected goals for is to compare to actual goals for and determine shot quality, then sure. If that were the case then actual goals for would be beside it in virtually every instance, and actually we'd be focusing on "x goals difference" or something like that as the primary output.

I am not seeing anyone do that, but now that you bring it up, maybe they should!

There are plenty of people who do that.

this one:

Wins Above Replacement 2.0

He utilizes x goals for to temper actual statistics.

That guy’s WAR model literally uses actual goals - expected goals and uses that number as a shooting component that is added to a player’s value.

I really don’t mean to be disrespectful here but you have no idea what you’re talking about.
 

NHL WAR

Registered User
Sep 29, 2018
959
1,176
this one:

Wins Above Replacement 2.0

He utilizes x goals for to temper actual statistics.

How many times do we have to go down this road. x Goals isn't a big factor in the model and it is considered in addition to actual goals. If iXG and iCF were taken out so that it was all actual goals, Ovechkin's (and every other good goal scorer in the league) WAR would go down.
 

Midnight Judges

HFBoards Sponsor
Sponsor
Feb 10, 2010
13,491
10,046
How many times do we have to go down this road. x Goals isn't a big factor in the model and it is considered in addition to actual goals. If iXG and iCF were taken out so that it was all actual goals, Ovechkin's (and every other good goal scorer in the league) WAR would go down.

OK, it's small, but it still just serves to make your model less indicative.

And again, it doesn't matter if everyone's number would go down, it matters what impact it makes relative to other players.
 
Last edited:

Midnight Judges

HFBoards Sponsor
Sponsor
Feb 10, 2010
13,491
10,046
I didn’t say they accounted for shooting talent. I said they accounted for more than just location.

Sure they do, but the caliber of an individual player's shot (the thing I initially highlighted that you disagreed with) isn't reflected in there...yet.

Even the folks generating the data assert that it ought to be - which is precisely what I am saying. And that is all that I am saying. Nothing more.

I even provided you with their paragraph quoted.
 
Last edited:

BlackFrancis

Athletic Supporter Patch Partner
Dec 14, 2013
5,647
8,990
Weird, when I plug the numbers into Excel, it returns, "n = let's be reality".
 

Midnight Judges

HFBoards Sponsor
Sponsor
Feb 10, 2010
13,491
10,046
That guy’s WAR model literally uses actual goals - expected goals and uses that number as a shooting component that is added to a player’s value.

I really don’t mean to be disrespectful here but you have no idea what you’re talking about.

His model uses actual statistics, adjusts them based on circumstances, and adds an 'x goals for' component to that (which is a small factor).

There is absolutely nothing to misunderstand here.

He's using raw x goals for data as indicative of player quality and adjusting with zero accounting for a player's shot (which is already reflected in the actual statistics). In this instance, x goals for is being used to say "this guy or that guy actually shouldn't have scored as much as they did or they should have scored more than they did, and I base this assertion on the assumption that his shot (and the shots of the guys he shares the ice with) is no different than the league average."

If you think that is not what is happening in that calculation, then it is you who does not understand.
 

NHL WAR

Registered User
Sep 29, 2018
959
1,176
His model uses actual statistics, adjusts them based on circumstances, and adds an 'x goals for' component to that (which is a small factor).

There is absolutely nothing to misunderstand here.

He's using raw x goals for data as indicative of player quality and adjusting with zero accounting for a player's shot (which is already reflected in the actual statistics). In this instance, x goals for is being used to say "this guy or that guy actually shouldn't have scored as much as they did or they should have scored more than they did, and I base this assertion on the assumption that his shot (and the shots of the guys he shares the ice with) is no different than the league average."

The bold is a decent summary of that calculation.

And it is true that there are massive variations in the quality of player's shots. Despite this, shooting percentages generally regress to the mean, which is a saving grace for expected goals. For example, even Ovechkin (who has a large enough sample size of outperforming them to pay less attention to his expected goals) can't sustain a particularly noteworthy career shooting percentage. He and Alex Chiasson are both career 12.6%. A bigger factor in goals is shot volume, where Ovechkin is in a league of his own.
 

Ad

Upcoming events

Ad

Ad

-->