[Albion] Have we been lucky? (xG stat geekery)

[Bozza]

I've been getting more interested in the xG (eXpected Goals) stat that I've never really noticed until this season - maybe another feature of Premier League life.

Expected Goals is the number of goals a team (or teams) would expect to score in a match. This is determined by assigning a value to shots on goal, the number of shots, shot location, the in-game situation and the proximity of opposition defenders.

So, how are we doing on xG and xGA (eXpected Goals Against) after 7 games?

Well, on xG we are the worst performing side in the Premier League. I guess that simply means we are creating the least chances, or least decent chances.



On xGA we are approximately where we are in the actual league table thus far, and with a few teams closely grouped, it would be possible to move up quite quickly.



And how would this convert to the league itself? We'd be bottom of the pile, and have Palace laughing at us.



So, have we been lucky to earn the results and points that we have? My feeling is 'no'. So maybe xG is just a load of shit. Thoughts?

(Data sourced from >>> https://understat.com/league/EPL)

 

[Bry Nylon]

xG is just a load of shit.

That, I feel, is a fair summary.

 

[AmexRuislip]

Thoughts

Respect for all the info, but it looks like somebody from the Matrix has created it

 

[Bozza]

Respect for all the info, but it looks like somebody from the Matrix has created it

I like the theory.

For example: penalties. Between 2011/12 and 2015/16 there were 443 penalties in the Premier League and of those 347 were scored - meaning that on average 78.3% of penalties resulted in a goal. A penalty is therefore assigned an expected goal value of 0.783.

We need more penalties, BTW.

 

[AmexRuislip]

I like the theory.

For example: penalties. Between 2011/12 and 2015/16 there were 443 penalties in the Premier League and of those 347 were scored - meaning that on average 78.3% of penalties resulted in a goal. A penalty is therefore assigned an expected goal value of 0.783.

We need more penalties, BTW.

I think I'm either being given a lesson in stats or being told to shut my big trap
Penalties would be a bonus

 

[GreersElbow]

I think it’s quite a load of shit, far too many variables at play to determine what an expected value is. How much weight is given to the selected factors etc

 

[Giraffe]

I remember seeing an article about just this somewhere and the conclusion was that it is likely to be used more and more by teams assessing individual players ability as well as opposition. The view of this article was also that gamblers, such as Tony Bloom, put an awful lot of weight on such stats.

 

[Giraffe]

Interestingly the idea is not hat far removed from what Graham Taylor used to form the tactics that took Watford from the fourth division to second in the first division.

 

[Johnny RoastBeef]

I remember seeing an article about just this somewhere and the conclusion was that it is likely to be used more and more by teams assessing individual players ability as well as opposition. The view of this article was also that gamblers, such as Tony Bloom, put an awful lot of weight on such stats.

Looking at last seasons premier league player xG stats, Josh King at Bournemouth scored highly. Looks good on paper and from what I saw in the cup match, he would definitely be a player I would be happy for us to target.

As for the xG ranking teams, there doesn't appear to be a clear pattern to validate its efficacy.

 

[Iggle Piggle]

It's easy to dismiss but most professional gamblers use this (those that follow football anyway) as thier main basis to beat the market. They make a livng from it so there is lots in it. That said, week after week Reading outperformed the stats last year and they never reverted to mean like the were expected to. West Brom score loads more from set pieces than other teams (and Stoke before) and thier are countless outliers over time.

The key is understanding if this is a problem. I'm pretty sure given Blooms penchant for stats that Hypia wasn't sacked sooner (alright, he resigned) as the numbers said we were unlucky and due to pick up. In this case, it's the opposite. The stats might say we are in trouble, my eyes tell me this not the case.

 

[Not Andy Naylor]

I'd expect us to score more goals if our midfield players had a few more shots.

 

[sydney]

That, I feel, is a fair summary.

Aye.......who thinks this shit up ffs ... seems to be an avalanche of obscure , pointless and usually unwanted stats issuing from the traps of most commentators these days.

 

[Jimmy Grimble]

I took a fair bit of comfort in these stats last year as I'm sure it consistently suggested we would go up, so I do find seeing these stats a little worrying.

 

[dingodan]

Unexpected goals I guess would be difficult chances, so I guess it just means we have scored some quality goals taking chances which were not easy.

Palace have had easier chances but haven't been able to score them.

Permission to carry on laughing at palance, granted.

 

[FatSuperman]

xG and xGA doesn't take into account the quality of the players. So if a particular shot has a 15% chance of going in on average, a high quality player would beat that 15% more regularly than an average one.
The
So we may have better finishers than we believe, or we have been lucky. Same with chances against, a top notch defence will outperform the xGA.

 

[Machiavelli]

I like the theory.

For example: penalties. Between 2011/12 and 2015/16 there were 443 penalties in the Premier League and of those 347 were scored - meaning that on average 78.3% of penalties resulted in a goal. A penalty is therefore assigned an expected goal value of 0.783.

We need more penalties, BTW.

The example of penalties might just provide part of the answer to how we're faring this season. Over the last two (and a bit, not that that bit matters, as your last point indicates) seasons, I'd estimate that our penalty ratio is higher than that average. Not by much, perhaps by about 10% -- I'm not basing this on firm data, but I reckon we've scored 14 or 15 penalties, with Hemed missing 2 in that time (Wolves, plus that late one away last season when we were 2-0 down).
Now, if you generalise that 10% above the average, that might help explain how we seem to be outperforming expectations.

 

[Machiavelli]

xG and xGA doesn't take into account the quality of the players. So if a particular shot has a 15% chance of going in on average, a high quality player would beat that 15% more regularly than an average one.
The
So we may have better finishers than we believe, or we have been lucky. Same with chances against, a top notch defence will outperform the xGA.

This might be another way of saying what I was attempting to do in the post above.

 

[Stat Brother]

https://audioboom.com/posts/6332897-xg-ogs-and-cliff-richard-s-greatest-hits

Assuming that's the feed for AC Jimbo's podcast of 25.09 Opta's Duncan Alexander does a deep dive into xG ratings and how they work.

 

[Jim in the West]

I'd expect us to score more goals if our midfield players had a few more shots.

This is one of my frustrations at the mo - we seem incredibly shot-shy. I'm assuming people like Stephens are following orders - he seems to get in loads of good shooting positions, but doesn't shoot!

 

[Triggaaar]

I've been getting more interested in the xG (eXpected Goals) stat that I've never really noticed until this season

So you managed to ignore all our links to Experimental 361 last season?
https://experimental361.com/2017/04/29/e-ratings-update-championship-29-apr-2017/