• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

Totally Unofficial 2019/2020 Premier League Thread (NWT)

Is there a tracker of the xG for each club over the 38 match season? And xGA (expected goals against)?

Then let's compare that to actual goals scored/allowed. If those numbers are off too then I'll rest my case. If they are close then I'd be willing to admit that there might be something there, just not on a match by match basis.
 
deacvision7 said:
Is there a tracker of the xG for each club over the 38 match season? And xGA (expected goals against)?

Then let's compare that to actual goals scored/allowed. If those numbers are off too then I'll rest my case. If they are close then I'd be willing to admit that there might be something there, just not on a match by match basis.
Click to expand...

Sure, IDK exactly how you want the numbers formatted but fbref.com has xG numbers by club and by match.

yrzmyfh.png


Their model is off by about a half a goal a game for xG and a goal per game for npxg (non-penalty expected goals) across the season for all 38 clubs.

Correlation coefficient of .942, pretty damn good.
 
It's not meant to be all that useful on a match by match basis, just tallies chances based on shots taken from those positions (and lots of other tracked data). I will not die on the hill of xG being a perfect stat, just sometimes a useful comparison beside a scoreline and the eye test. It's fairly predictive of table position and individual player success, but the best players in the world do regularly outpace the model for multiple seasons before adjustments get made.
 
Here's Man United across the season

uIPOj8b.png


Great correlation in sum for Goals Scored and xG (32 vs 31.6) but on a match by match basis not the best. And expected goals against is quite a bit off (thought not awful).
 
TownieDeac said:
It's not meant to be all that useful on a match by match basis, just tallies chances based on shots taken from those positions (and lots of other tracked data). I will not die on the hill of xG being a perfect stat, just sometimes a useful comparison beside a scoreline and the eye test. It's fairly predictive of table position and individual player success, but the best players in the world do regularly outpace the model for multiple seasons before adjustments get made.
Click to expand...

xG is very useful for knowing when a good run of form has substance or whether it’s some good fortune. Just last season Arsenal went 22 games unbeaten but we were outperforming xG and xGA both by substantial margins. We came crashing down to earth shortly thereafter. It’s definitely a good trend statistic, but you are correct it’s not a good game by game stat.
 
TownieDeac said:
Correlation coefficient of .942, pretty damn good.
Click to expand...

TownieDeac said:
Here's Man United across the season

uIPOj8b.png


Great correlation in sum for Goals Scored and xG (32 vs 31.6) but on a match by match basis not the best. And expected goals against is quite a bit off (thought not awful).
Click to expand...


OK, so I'll bend and say it is useful over multiple trials/samples (matches) but it kinda is bunk when it comes to a singular match. The SD looks to be through the roof judging from just MUFC's xG and real goal numbers in the matches displayed.
 
deacvision7 said:
OK, so I'll bend and say it is useful over multiple trials/samples (matches) but it kinda is bunk when it comes to a singular match. The SD looks to be through the roof judging from just MUFC's xG and real goal numbers in the matches displayed.
Click to expand...

In single games you'd actually expect something with a .95 r to occasionally be eye-opening.

Spectacular goals are hard to model for. There's a reason a stunner from 35 yards only goes in a few times a season, so when you have one of those that's a 0.05xG chance, and 1 goal scored, there's basically a 1 on your differential.

I think it's actually useful on a game-by-game basis more from a shot location chart perspective when coaching up strikers or people on the end of set pieces. Look at the big chances and say if we create more opportunities towards the center of the box near the penalty spot, at head or at feet, we're likely to score more goals than if we shoot from wider or further.
 
TownieDeac said:
In single games you'd actually expect something with a .95 r to occasionally be eye-opening.

Spectacular goals are hard to model for. There's a reason a stunner from 35 yards only goes in a few times a season, so when you have one of those that's a 0.05xG chance, and 1 goal scored, there's basically a 1 on your differential.

I think it's actually useful on a game-by-game basis more from a shot location chart perspective when coaching up strikers or people on the end of set pieces. Look at the big chances and say if we create more opportunities towards the center of the box near the penalty spot, at head or at feet, we're likely to score more goals than if we shoot from wider or further.
Click to expand...

I mean, that's fucking obvious though, isn't it? I coached that for a long damn time without this xG stat at my disposal. And many coaches before I was born were saying the same thing, I'm sure. OF COURSE a chance from 10 yards out in front of goal is better than a shot from 22 yards out near the corner of the penalty area.

And not many goals are scored from corners hit long into the box. It's actually better to play it short and get the defense moving around before playing a ball into the area.
 
xG just seems like a made-up stat to try to seem smart when really it isn't saying much at all. Seems to treat all 20 field players on the pitch the same in terms of finishing ability when that is clearly not the case, which is why the StdDev is so grotesquely high.

Not everything in sports can/needs to be boiled down to some mathematical equation. And I love math.


K, I'm done on this one for now. I'm sure we will revisit in September or so. :thumbsup:
 
deacvision7 said:
I mean, that's fucking obvious though, isn't it? I coached that for a long damn time without this xG stat at my disposal. And many coaches before I was born were saying the same thing, I'm sure. OF COURSE a chance from 10 yards out in front of goal is better than a shot from 22 yards out near the corner of the penalty area.

And not many goals are scored from corners hit long into the box. It's actually better to play it short and get the defense moving around before playing a ball into the area.
Click to expand...

One would think it's obvious, but players still take a ton of low quality shots! It's still a colossal problem even at the highest levels of world football and a massive waste of good possession when you've worked your way into decent scoring position. It's good for player evals to look at shot charts to see who's taking high quality shots. More than shot position too are other factors around shot quality.

Take Mason Greenwood for example, he was massively outperforming his xG and he's coming back down to earth a tiny bit. But he also has one stat in particular, which is shots taken on the ground, which is actually predictive of elite striking abilities. Think about players who score the most goals on the ground in the league (Kane, Sterling, Mane) and they're all consistent 20 goal scorers, another easily coachable piece for developing young talent.
 
BillBrasky said:
Barca have been a disaster since the restart. Nope.
Click to expand...

I don't think that has much to do with Messi and more with giant club issues for the last couple of years. He is still playing out of his mind and is still the best player in the world.
 
You must log in or register to reply here.
Back
Top