Statistical Domination

[hogesar 9,672]

2 minutes ago, Paint Me Yellow said:

Got to love a bit of spreadsheet football after a $h!t-shower of a game. Can't wait for the slide-show to follow.

It'll only have one slide with "But...but...2nd" written on it.

 

It sounds like discussion forums aren't really for you fella.

[Paint Me Yellow 200]

3 minutes ago, hogesar said:

It sounds like discussion forums aren't really for you fella.

You're right. I prefer the actual football games, not what Fifa 23 tells me should've happened.

[hogesar 9,672]

11 minutes ago, Paint Me Yellow said:

You're right. I prefer the actual football games, not what Fifa 23 tells me should've happened.

Begs the question why you even bothered looking at this thread then. 

[king canary 7,606]

1 hour ago, Paint Me Yellow said:

You're right. I prefer the actual football games, not what Fifa 23 tells me should've happened.

These posts are always fun because it lets me know the person both doesn't understand the stats and has also never played the videogames they mention.

[TeemuVanBasten 3,327]

On 14/08/2022 at 14:15, hogesar said:

Posted elsewhere but thought it was worthy of separate discussion.. 

Statistically we've dominated. Not just that, our XG has been higher than our opponents in the two games we've dominated. Why is that important?

One, it shows we are creating better chances than our opponents. This is either because our chances are more guilt edge, where the probability of scoring is considerably higher, or we are restricting the opposition defensively to very low probability or speculative attempts. More likely, a combination of the two.

Two, xG is the one stat that regularly, over a season, tends to match results and table positioning. If you consistently have a higher xG you end up at the right end of the table 99 times out of 100.

Finally, xG is based upon tens of thousands of shots from different locations and is based upon the likelihood of an average player scoring that goal. So if you have a better striker than the rest of the division, you can exceed xG. Although its rare, the only player in world football to regularly out perform their xG is Messi, which actually suggests most players have similar finishing ability and the differences aren't as huge as fans make out. I digress.

Edit to show how we compare to the rest of the division overall.

Edit: Adding Huddersfield match

xG cannot measure whether a shot was the best option as it can't determine the positioning of other players, it's a load of junk. 

 

[Midlands Yellow 3,984]

3 minutes ago, TeemuVanBasten said:

xG cannot measure whether a shot was the best option as it can't determine the positioning of other players, it's a load of junk. 

 

Simon Lappin rates it. 

[TeemuVanBasten 3,327]

6 minutes ago, Midlands Yellow said:

Simon Lappin rates it. 

Is this a game where I name a player who doesn't rate it, then you name another who does, and we continue until somebody wins 367-366?

[Capt. Pants 4,234]

Novice question... how is it our overall xG is 2.07 putting us top of the league when our best rating is 1.82 against Blackpool?

[Midlands Yellow 3,984]

24 minutes ago, TeemuVanBasten said:

Is this a game where I name a player who doesn't rate it, then you name another who does, and we continue until somebody wins 367-366?

Yes!! 

[SwearyCanary 1,179]

9 hours ago, wcorkcanary said:

Someone messed up spectacularly with Haalands XG!! To allow  @Feedthewolf to get 25/1 on him scoring 3 hattricks in the whole season then . 😇

Just to clarify, xG is NOT a prediction of what goals a team will score, but a statistical measure of how many goals they SHOULD have potentially scored based on the circumstances of the chance. Haaland’s xG can only be calculated after a match has already been played. You can use longer term xG stats to get an idea of a teams form over certain games or over a certain period, but not longer than a season assuming their formation, personnel and other such variables don’t actually change much. Plus it doesn’t account for how well the opposition are set up to affect the xG by their defensive formation etc. 

But I did note the cursory emoji get out clause in your post 😝

[SwearyCanary 1,179]

48 minutes ago, TeemuVanBasten said:

xG cannot measure whether a shot was the best option as it can't determine the positioning of other players, it's a load of junk. 

 

Pretty sure it does account for them, but yes you’re right, it won’t tell you whether the shot was the best option, only the odds I’d the chosen option being successful relative to other similar circumstances 

[hogesar 9,672]

59 minutes ago, TeemuVanBasten said:

xG cannot measure whether a shot was the best option as it can't determine the positioning of other players, it's a load of junk. 

 

That it doesn't tell you something different to what it does tell you doesn't make it junk. It just means like every stat it has limitations, it just happens to be one of the least limiting.

[wcorkcanary 4,334]

23 minutes ago, SwearyCanary said:

Just to clarify, xG is NOT a prediction of what goals a team will score, but a statistical measure of how many goals they SHOULD have potentially scored based on the circumstances of the chance. Haaland’s xG can only be calculated after a match has already been played. You can use longer term xG stats to get an idea of a teams form over certain games or over a certain period, but not longer than a season assuming their formation, personnel and other such variables don’t actually change much. Plus it doesn’t account for how well the opposition are set up to affect the xG by their defensive formation etc. 

But I did note the cursory emoji get out clause in your post 😝

Crikey , I didn't expect a serious response to what was, clearly, a bit of a jolly post.  My point stands though, who the hell decided on Wolfos Odds. If I'd seen it I'd have come out of betting retirement ( I quit while I was about      £3-45  ahead in 1980). 

FWIW  stats interest me a bit , but not enough to bother arguing about their relevance.   If someone thinks they are bollix , they won't change their mind.

Edited October 10, 2022 by wcorkcanary

[TeemuVanBasten 3,327]

Has anybody pointed out yet that there are multiple sources of xG data which often vary significantly and Hogesar hasn't caveated this or explained why he has chosen this particularly source? 

[TeemuVanBasten 3,327]

19 hours ago, hogesar said:

That it doesn't tell you something different to what it does tell you doesn't make it junk. It just means like every stat it has limitations, it just happens to be one of the least limiting.

Our xG stats vary significantly by provider for a start, I'd like to know which one you use and why you choose that particular source before we determine whether your data is junk. Looks like you've just taken graphs from a twitter account.

Understat? Driblab? Let's agree a common provider before we talk about scrutinising data shall we?

[hogesar 9,672]

24 minutes ago, TeemuVanBasten said:

Has anybody pointed out yet that there are multiple sources of xG data which often vary significantly and Hogesar hasn't caveated this or explained why he has chosen this particularly source? 

 

17 minutes ago, TeemuVanBasten said:

Our xG stats vary significantly by provider for a start, I'd like to know which one you use and why you choose that particular source before we determine whether your data is junk. Looks like you've just taken graphs from a twitter account.

Understat? Driblab? Let's agree a common provider before we talk about scrutinising data shall we?

You're not interested in xG. You posted without understanding even the most basic principles of it, then tried to attack me having got several other parts of my other posts wrong and now you've come back to a thread all about xG for...what reason?

I'll leave you to your own devices, but yes there are multiple data collection agencies / data scientists that collate data, and some of them differ. I think most people are aware of this, it may even have been mentioned earlier in the thread.

Experimental 361 is good, but often have to wait a bit longer for the results. Fotmob is a big provider but many others think their xG for certain leagues is undervalued (that's been reflected in the xG / goals scored this season).

 

[wcorkcanary 4,334]

19 minutes ago, TeemuVanBasten said:

Our xG stats vary significantly by provider for a start, I'd like to know which one you use and why you choose that particular source before we determine whether your data is junk. Looks like you've just taken graphs from a twitter account.

Understat? Driblab? Let's agree a common provider before we talk about scrutinising data shall we?

Tbf you should average the data from multiple reputable sources, it'll be great fun. I think.😇

[king canary 7,606]

30 minutes ago, TeemuVanBasten said:

Has anybody pointed out yet that there are multiple sources of xG data which often vary significantly and Hogesar hasn't caveated this or explained why he has chosen this particularly source? 

Badger gave some good background on the chosen source here...

On 07/10/2022 at 09:18, Badger said:

I don't think that the Hoilett pass would be included in xG because there was no attempt on goal. The data comes from experimental 361, who has a comprehensive website as well as a twitter account:

https://experimental361.com/category/divisions/championship/

S/he states that all of the data comes from publicly available sources which he then creates charts from using automated software. Most of the original data is from the site below, where you can download it in csv format. The data seems the same as data from other sites like FootyStats. 

https://www.football-data.co.uk/englandm.php

The owner of the site, Ben Mayhew, is Head of Data Analysis at the Press Association. He had previously worked for KPMG as Manager in the Insurance Risk Advisory team. He has a pretty good pedigree!

https://www.linkedin.com/in/maybewhen/?originalSubdomain=uk

 

23 minutes ago, TeemuVanBasten said:

Let's agree a common provider before we talk about scrutinising data shall we?

This might make sense if the OP you seem to be going after had posted from multiple different sources at different times in order to paint a picture but as far as I can see he's stuck to one source. So it seems a common provider isn't an issue.

It all feels like we're getting into weirdly antagonistic territory again because of a disagreement about the usefulness of stat? Doesn't seem worth it.

[hogesar 9,672]

10 minutes ago, king canary said:

Badger gave some good background on the chosen source here...

 

This might make sense if the OP you seem to be going after had posted from multiple different sources at different times in order to paint a picture but as far as I can see he's stuck to one source. So it seems a common provider isn't an issue.

It all feels like we're getting into weirdly antagonistic territory again because of a disagreement about the usefulness of stat? Doesn't seem worth it.

I'll be honest I'm a little lost on it all. If you're not interested in xG or other stats thats completely fine - there's absolutely zero need to be and you can enjoy football all the same and hold whatever opinions you like. It has previously been limited to just this thread really. I'm still learning about some elements of it myself, and what they're looking to do in the future.

Edited October 11, 2022 by hogesar

[shefcanary 2,408]

6 hours ago, TeemuVanBasten said:

Has anybody pointed out yet that there are multiple sources of xG data which often vary significantly and Hogesar hasn't caveated this or explained why he has chosen this particularly source? 

Consistency of data is more important than source, but sticking with one reliable source should remove some of the subjectivity inherent in the stats.

[littleyellowbirdie 2,550]

6 hours ago, TeemuVanBasten said:

Has anybody pointed out yet that there are multiple sources of xG data which often vary significantly and Hogesar hasn't caveated this or explained why he has chosen this particularly source? 

There are, and the meta-analyses of the predictions against the actual outcomes will help inform the refinement of the models, which will probably converge.

This is the way with all scientific and technological development. You see more variation in the infancy, and more convergence as the knowledge develops. 

[nutty nigel 7,535]

On 09/10/2022 at 22:26, nutty nigel said:

I don't understand it enough to answer questions.

I looked at the Reading WBA game and 9/4 for a home win 5th v 22nd looks huge.

Is that xg influence? Because if so, and xg is nonsense, we should fill our boots!

Another game on Bet 365 where xg seems to have affected the price is at Luton on Saturday.

Luton (10th) 11/10 v QPR (3rd) 5/2

Draw is 12/5.

I don't fully or even partly understand xg but if it's not 'a thing' we might soon be seeing bookies on a bike 🚲

[nutty nigel 7,535]

Both games comfortably went in favour of the bookies and xg. 

Followers of the league table are on the bikes tonight🚲

Edited October 15, 2022 by nutty nigel

[Gibbo 131]

On the subject of statistical domination... I see Ipswich had 33 shots today to Lincoln's 3, and still lost 1-0 😄 

[hogesar 9,672]

I didn't see the game, so I can't really comment on performance via 'eyes test' but the xG and timeline, as well as the Top Players probably tells enough of a story.

Watford won a pen (which accounts for 0.76 xG I think) and created other good chances in a first half where we (apparently) created next to nothing. Looks like Sargent scored from a very low-probability chance - i.e we didn't create a clear chance for him. Good to keep Sargent on form.

Concerningly none of our players created a really good direct chance - it appears Hanley had a decent opportunity so I'm assuming a set piece.

I quite like Dowell as a player, have no idea how he actually played but to be honest, on the xG and xA player lists, you'll often find Aaron Ramsey cropping up. He creates chances. I know some don't seem to like him because he's a loan player (I think that's why?) but he's generally been one of our more creative options.

The graph shows we only really had one good chance to score. 

Looking to Luton for comparison, as they're next, they created good chances vs QPR according to xG, Carlton Morris creating their best chances!

 

[TIL 1010 4,746]

On 10/10/2022 at 19:11, Midlands Yellow said:

Simon Lappin rates it. 

Simon Lappin also reckoned that if people are bored watching Norwich this season they would get bored watching Man City.

Quite a comparison that one.

Edited October 17, 2022 by TIL 1010

[aBee 151]

1 hour ago, TIL 1010 said:

Simon Lappin also reckoned that if people are bored watching Norwich this season they would get bored watching Man City.

Quite a comparison that one.

Funnily I have a friend who was in a group that shared a bunch of Man City STs for many years. They gave them up a few years ago because they found watching the team largely boring with the only atmosphere being in the really big games and games against middling/lower teams so dully inevitable that even their goals barely got celebrated. I went to the Etihad with him in the home end last season v Brentford and it was largely silent where we were bar muttering about why they hadn’t scored yet and not much more than polite applause when they did from two errors. 
 

They’re tremendously skilful, organised in Pep’s style and almost metronomically consistent but I think they lack the joy you get from players doing unlikely and surprising things. Haaland is very much in their image because he mainly just plays perfectly over and over again rather than producing individually memorable or audacious goals. 

[hogesar 9,672]

21 hours ago, aBee said:

Funnily I have a friend who was in a group that shared a bunch of Man City STs for many years. They gave them up a few years ago because they found watching the team largely boring with the only atmosphere being in the really big games and games against middling/lower teams so dully inevitable that even their goals barely got celebrated. I went to the Etihad with him in the home end last season v Brentford and it was largely silent where we were bar muttering about why they hadn’t scored yet and not much more than polite applause when they did from two errors. 
 

They’re tremendously skilful, organised in Pep’s style and almost metronomically consistent but I think they lack the joy you get from players doing unlikely and surprising things. Haaland is very much in their image because he mainly just plays perfectly over and over again rather than producing individually memorable or audacious goals. 

They have so much of the ball too, that it never feels like there is huge intensity in most Man City games. When you consider the most exciting parts of football are generally counter-attacks where there's been a turnover of possession, it checks out I guess.

[aBee 151]

3 hours ago, hogesar said:

They have so much of the ball too, that it never feels like there is huge intensity in most Man City games. When you consider the most exciting parts of football are generally counter-attacks where there's been a turnover of possession, it checks out I guess.

It’s possibly also why they haven’t won the CL. They haven’t an extra gear because they’re almost constantly in top. Much like that time when Michael Schumacher did most of Monaco stuck in fifth- incredibly skilled driving to complete the race and not get passed but no excitement watching him do it. 

[hogesar 9,672]

So, our worst (xG) performance since the first game of the season.

Neither team created much, although worth noting Luton had a couple dangerous balls across our box that just needed a tap in, but no Luton player really got close. xG will not attribute that because there was no shot, but we could equally say we defended the area's pretty well.

Worth noting Luton's goal came from one of the 'best' chances in the match. Our xG is higher from having more shots rather than better quality chances as the graph shows.

Think it's also the first time we "haven't deserved to score a goal" according to xG since Cardiff first game, although that's being incredibly simplistic.

I don't disagree with the likelihood metric that suggests a draw would have been fair but of course, we never done enough during the match and then the sending off makes things even more difficult. 

Not a good performance in any metric other than defensively we were generally pretty good - Luton have had a pretty solid xG all season, they've generally created good chances in games this season.

On a sidenote, Stoke will be wondering how on earth they didn't win their game last night! One of the biggest disparities xG has seen this season: