A Baffling Expected Goal Measurement

Filthy Dangles

Registered User*
Oct 23, 2014
28,356
39,706
That makes more sense.

The thing is, 8% is about average for a shot on-goal. xG is measuring the likelihood of a shot attempt going in.

Half of attempts don't get to net. Teams miss the net 1,500-2,000 times a year.

That's not relevant though because that puck is getting on net. There's no traffic in front of the shooter or anyone bothering the shooter to to contest or disrupt the shot in anyway. That's the point, XG don't take this into account and accurately reflect the realtiy an dlikelyhood of the puck going in the back of the net.

If we can't agree that specific shot or "Shot attempt" doesn't go in way more than 8/100 times i don't know what else we have to talk about really
 

Machinehead

GoAwayKakko
Jan 21, 2011
141,276
109,813
NYC
That's not relevant though because that puck is getting on net. There's no traffic in front of the shooter or anyone bothering the shooter to to contest or disrupt the shot in anyway. That's the point, XG don't take this into account and accurately reflect the realtiy an dlikelyhood of the puck going in the back of the net.
I think it's a fair criticism that xG doesn't take into account space between the shooter and the defender.

That being said, the Lightning led the league last year in 5v5 goals and 5v5 shooting percentage and only 5% of all their attempts went in. That's the best in the league. 8% is still quite elevated. There are a lot of variables. Saying any particular moment in time has an 8/100 chance of resulting in a goal still seems to me like a pretty good scoring chance.
 
  • Like
Reactions: Filthy Dangles

aufheben

#Norris4Fox
Jan 31, 2013
53,444
27,140
New Jersey
Idk, looks in the ballpark.

upload_2021-2-5_23-21-39.png


I visit MoneyPuck daily and I have to say I think they do a good-enough job. I follow their goal bots on Twitter, so every time the Rangers score I’m notified of the goal-scorer, game score, shot type, shot location, shot distance, shot angle, assists, 5v5/etc., time of goal, win % change. If that’s not cool I don’t know what is.
 

biotk

Registered User
Jan 3, 2017
7,091
5,520
Buffalo
I think it's safe to say just throw out all xG metrics (assuming they all use the same data source as MoneyPuck).

As far as I know they all use the same source data from NHL.com, which I would assume is simply tracked by one person for each game. I have tracked a few games before just to see how accurate the information was and found the data to be significantly flawed, to put it mildly. Although, I don't know why we would expect the publicly available to be that accurate (why would the NHL bother to put a lot of resources into it? and other sites just have computer algorithms crunch those numbers) and is a reason why teams pay to collect far more accurate data.

I have posted this clip before which is from one of the games I tracked (Buffalo at Colorado, February 26th 2020, 5v5 a little more than 7 minutes into the third period - the clip, of course, is from a replay which shows what happened better). and have asked people what they see happening here. Everyone agrees that it is two high danger scoring chances. On NHL.com it indicates that nothing at all happened here, and all of the advanced stats sites that I checked also indicated that nothing at all happened here (not even a corsi - zero xGF). I saw ridiculous stuff every game I tracked (although this was the worst).
 
Last edited:

Henkka

Registered User
Jan 31, 2004
31,078
12,078
Tampere, Finland
As far as I know they all use the same source data from NHL.com, which I would assume is simply tracked by one person for each game. I have tracked a few games before just to see how accurate the information was and found the data to be significantly flawed, to put it mildly. Although, I don't know why we would expect the publicly available to be that accurate (why would the NHL bother to put a lot of resources into it? and other sites just have computer algorithms crunch those numbers) and is a reason why teams pay to collect far more accurate data.

I have posted this clip before which is from one of the games I tracked (Buffalo at Colorado, February 26th 2020, 5v5 a little more than 7 minutes into the third period - the clip, of course, is from a replay which shows what happened better). and have asked people what they see happening here. Everyone agrees that it is two high danger scoring chances. On NHL.com it indicates that nothing at all happened here, and all of the advanced stats sites that I checked also indicated that nothing at all happened here (not even a corsi - zero xGF). I saw ridiculous stuff every game I tracked (although this was the worst).

I can nothing but confirm this. Seen also some totally ridiculous things.

Most funny was the 1.05 xG for an empty-netter attempt. More than 1 goal.

C'mon.
 
  • Like
Reactions: Filthy Dangles

Ad

Upcoming events

Ad

Ad

-->