XG? Love it hate it

Status
Not open for further replies.
I like it. It's a math based stat, and I believe in math. If you play good and create chances it will pay off in the long run.
 
Seems to be one of the stats that holds up though, small chances, big chances etc.

Also it's no surprise statisticians from Opta and the like are hired by analytical football clubs.

Attacking players who have a high rate of chances per game are the ones stand out at analytical level, players in a league who didn't score much then come to another team built from analytics of chances created/big chances key passes created in the final 3rd and that attacking player who didn't score much but had multiple chances in games starts scoring frequently all of a sudden.

No surprise the likes of Astrophysicists and people who have had a background at CERN are hired by these clubs, Number crunching by the likes of people in those backgrounds are perfect for analytics.

Brentford have for 4-5 years been selling attacking players to the PL, like Andre Gray, Ollie Watkins, Benrahma, Maupay, Scott Hogan and still not missing a beat, now they have Ivan Toney.

I'm surprised a number cruncher accountant like Moshiri hasn't embraced analytics, maybe the recruitment would have been better under him if he did.
 
Brighton were xG kings last season. I can remember people saying all they needed was a striker...
You are correct and this year their actual goals are slightly exceeding their expected and they are doing better in the league.

To me these stats are a good indication of whether or not you have gotten lucky with your results. Perhaps luck is a bad word, but it can indicate how you "should" perform given the chances you have created and allowed. Still too early in the season to put too much stock in it, especially for us since we've had a favorable schedule compared to others. We should have a higher xG than teams with more difficult schedules.

Feels about right to me. I'd say we've generally had results that we deserved. Not surprising that we've scored a bit more than deserved and conceded a bit more. I think we've generally done well in keeping games under control (except for that 15 minute spell at Villa), so its also not surprising to see that our actual GA is close to our xGa.

Can't argue with last year either. The same site says we got lucky, should have finished 12th and got 8 fewer points. That to me feels about right. There were quite a few games where I felt relived that we got 3 points and maybe didn't deserve it. Basically the beginning of last season flattered us. I think that is less true this year.
 

I’m not really bothered.

it all feels very “FIFA” generation to me.

Goals, Shots on Target are your indicator of how well you should have done in a game in an attacking front.
The point of stats like xG is to tell a better story about what happened, and therefore predict with greater accuracy what will happen going forward. The idea is that not all shots, saves and shots on target are created equal. A tame effort hit directly at the keeper from 20 yards out counts the same in those stats as a ball from eight yards that the keeper makes a wonder save to keep out.

xG tries to fix this. It finds that players that get into position to score a lot of goals tend to score a lot of goals. It can't explain why some players outperform while others underperform expectations. But then again, neither can a metric like penalties scored vs. penalties taken. Whether an outperforming player is good at his job or lucky is in the eye of the beholder. As the data series gets large, whether we're talking penalties or xG, it becomes possible to say just how unlikely it is that the player is lucky rather than good.

Brighton were xG kings last season. I can remember people saying all they needed was a striker...
So far the stat suggests they're creating fewer chances, but finishing the ones they get.
 

CF45528B-1F95-498F-988C-E6A2ED72F2F5.jpeg
 
Understand that the better the XG, the better the chance, but who actually puts a number to that, and how do they work it out? Eg how is it quantified
 
Understand that the better the XG, the better the chance, but who actually puts a number to that, and how do they work it out? Eg how is it quantified
To start with it was literally where the shot was taken from (the closer to goal and more central the higher the value) and the body part (head or foot).

It was calculated by taking the coordinates of a shot (and body part used) from hundreds of thousands of logged shots and saying if a goal was scored or not. So if 100/1000 shots were scored from that location then that equals a 0.1xg shot.

That did actually work, teams that got better shots from better locations score more goals.

However that had obvious problems, there is a huge difference between a shot into an empty net from 20 yards and a shot from 12 yards with 10 defenders in the way, but the "naive" model would rate the open goal shot as harder.

So then some providers added in where the shot was placed (called post-shot xG) which took into account goalkeeper position and where the shot went.

Now some take snapshots that plot where the defenders were positioned when the shot was taken which makes it more accurate on a game by game basis.

Every club uses xG to some extent but it is normally part of a package of stats used to monitor performance. It is just new terminology for "we had the better chances" to be honest.
 
To start with it was literally where the shot was taken from (the closer to goal and more central the higher the value) and the body part (head or foot).

It was calculated by taking the coordinates of a shot (and body part used) from hundreds of thousands of logged shots and saying if a goal was scored or not. So if 100/1000 shots were scored from that location then that equals a 0.1xg shot.

That did actually work, teams that got better shots from better locations score more goals.

However that had obvious problems, there is a huge difference between a shot into an empty net from 20 yards and a shot from 12 yards with 10 defenders in the way, but the "naive" model would rate the open goal shot as harder.

So then some providers added in where the shot was placed (called post-shot xG) which took into account goalkeeper position and where the shot went.

Now some take snapshots that plot where the defenders were positioned when the shot was taken which makes it more accurate on a game by game basis.

Every club uses xG to some extent but it is normally part of a package of stats used to monitor performance. It is just new terminology for "we had the better chances" to be honest.
Wow, didn’t know that, nice one ??
 

Status
Not open for further replies.
Top