Everton 2-1 Arsenal - Saturday 19th December, 2020

[RandoEFC+]

Yeah I think those percentages are based off comparable xG between the teams in those games.

It's just silly, all he needs to know is that if the other team scores more than your team you had/have 0% chance of winning so that's what he needs to sort out.

[LFCMadLad]

34 minutes ago, Dr. Gonzo said:

Using stats blindly without context is stupid, and it appears that's exactly what he's doing. It looks like he's looking at Everton's xG and Arsenal's xG and thinking "wow statistically we should have won that" but Everton have a low xG for a few reasons - the Arsenal own goal doesn't give anything for xG. DCL's header, by virtue of being a header, gives low xG (you get a low xG stat for all headers unless they're right on the goal line). Meanwhile Arsenal got a penalty, which counts as 1 xG.

It demonstrates a fundamentally shit understanding of the statistic he's looking at. It's also very strange because you think if you're thinking really hard about the xG stat... you'd know how headers count towards that particular statistic. But he's also a manager that's said if you cross enough you'll get goals because "it is statistical probability" - so you'd think he'd be aware of how headers are viewed for expected goals? But also in that statement it doesn't seem to account for the quality of crosses or whether they're crossing into a box where there's a high probability of an Arsenal player getting on the end of it.

Statistics are important to any football manager. But you've got to at least understand the concepts behind what those statistics mean if you want to get useful information out of them. Looking at xG and just taking that as the statistical probability of whether you should have won the match is just... waaaaay too simplistic of a way to view a side's odds on whether they should have won. It's just a fundamentally flawed approach to viewing the statistic - there's flaws in what sorts of shots are considered "expected goals" imo (see the DCL header/how headers count towards xG).

The whole second half of that match demonstrates just why exactly xG isn't the best stat to judge whether a side should have won. Everton were 2-1 up against a side that struggles to threaten a defense - how much impetus does that give Everton to open up against a side that struggles to break down sides that haven't opened up? Not much, right? So the xG stat doesn't account for the momentum of the match at all.

A robot would read stats better. Arteta's more like Eric Cartman in his Awesome-O 3000 suit, hoping he can convince enough people he is what he isn't. But instead of trying to convince us he's a robot, he's trying to convince us he's a football manager that belongs at a club like Arsenal. Ironically, what he's done here is make it abundantly clear he's a fraud.

...and he definitely looks like a puppet from thunderbirds. Type it into Google images and see for yourself! 

[Dr. Gonzo]

5 minutes ago, RandoEFC said:

Yeah I think those percentages are based off comparable xG between the teams in those games.

It's just silly, all he needs to know is that if the other team scores more than your team you had/have 0% chance of winning so that's what he needs to sort out.

I've seen a lot of people on the internet go crazy about xG and churn out some "statistical analysis" that I think doesn't even come close to telling the whole picture about certain matches they're analysing. xG by itself is just one metric that tells just a bit about how a match went. Some of them freely admit they've not seen any footage of the matches in question... but then they'll only breakdown matches with xG. These articles by hobbyist stattos isn't useful analysis for anyone, really, unless all you care about is xG.

But everyone should be aware that xG is just one metric, and while it does provide some useful information, it is far from a full picture and if used out of context (like... for instance, with this match we're in the thread of talking about) you can end up using statistics to paint a totally inaccurate portrayal of the match a la Arteta.

I'm just surprised he's come out and using xG stats in this manner that demonstrates he doesn't know how to best look at the stat. I just find it hard to believe a top flight manager would demonstrate such a simplistic understanding of football stats, like he's read one of these shite xG blogs from some hobbyist online and really taken it to heart.

I want to believe he has a better understanding of stats and that he's just come out and said this to take some pressure off him, because I think people are (rightly) thinking he should be sacked. But it also has me wondering... maybe he's just more out of his depth than I could have ever thought?

[RandoEFC+]

10 minutes ago, Dr. Gonzo said:

I've seen a lot of people on the internet go crazy about xG and churn out some "statistical analysis" that I think doesn't even come close to telling the whole picture about certain matches they're analysing. xG by itself is just one metric that tells just a bit about how a match went. Some of them freely admit they've not seen any footage of the matches in question... but then they'll only breakdown matches with xG. These articles by hobbyist stattos isn't useful analysis for anyone, really, unless all you care about is xG.

But everyone should be aware that xG is just one metric, and while it does provide some useful information, it is far from a full picture and if used out of context (like... for instance, with this match we're in the thread of talking about) you can end up using statistics to paint a totally inaccurate portrayal of the match a la Arteta.

I'm just surprised he's come out and using xG stats in this manner that demonstrates he doesn't know how to best look at the stat. I just find it hard to believe a top flight manager would demonstrate such a simplistic understanding of football stats, like he's read one of these shite xG blogs from some hobbyist online and really taken it to heart.

I want to believe he has a better understanding of stats and that he's just come out and said this to take some pressure off him, because I think people are (rightly) thinking he should be sacked. But it also has me wondering... maybe he's just more out of his depth than I could have ever thought?

I find xG really interesting. It's more informative than shots and shots on target when it comes to representing how much each team should have scored.

If Arteta has requested someone model the percentage chance a team had of winning a game after the game had ended though, that's a massive waste of resources. If it's something that Arsenal do and have done since before he took over then it's less his fault but still a massive waste of resources and he's listening enough to the feedback to quote the numbers in a press conference.

If you were being sympathetic then maybe you could say, alright if a manager wants to stay calm and believe in the tactics and approach because it gave them statistically a less than 10% chance of losing those games then fine, the guys that do this statistical modelling are way smarter and go into more detail than you or I can imagine. But it doesn't hold water here. Arsenal have had more than one or two losses that you can put down to statistical anomalies. The model is failing to take into account some important stuff, like their inability to create chances and how prone some of their individual players are to mistakes that give away goals.

I think sports science has a massive role in the game and that the elite managers do make a lot of use of it. Unfortunately though, you'll never be able to win a football match with a calculator or a spreadsheet even if they improve the model. Confidence, man management, decision making, are massive ingredients too and Arteta and Arsenal are lacking those to different extents.

I'm all for having faith in sports science and statistics but coming out with that stuff in a press conference just comes across a bit unhinged as if he's in denial about how poorly they're doing. It's the sort of thing you look back on when a manager gets sacked and recognise it as the point that pretty much everyone knew it was over.

[Dr. Gonzo]

1 hour ago, RandoEFC said:

I find xG really interesting. It's more informative than shots and shots on target when it comes to representing how much each team should have scored.

If Arteta has requested someone model the percentage chance a team had of winning a game after the game had ended though, that's a massive waste of resources. If it's something that Arsenal do and have done since before he took over then it's less his fault but still a massive waste of resources and he's listening enough to the feedback to quote the numbers in a press conference.

If you were being sympathetic then maybe you could say, alright if a manager wants to stay calm and believe in the tactics and approach because it gave them statistically a less than 10% chance of losing those games then fine, the guys that do this statistical modelling are way smarter and go into more detail than you or I can imagine. But it doesn't hold water here. Arsenal have had more than one or two losses that you can put down to statistical anomalies. The model is failing to take into account some important stuff, like their inability to create chances and how prone some of their individual players are to mistakes that give away goals.

I think sports science has a massive role in the game and that the elite managers do make a lot of use of it. Unfortunately though, you'll never be able to win a football match with a calculator or a spreadsheet even if they improve the model. Confidence, man management, decision making, are massive ingredients too and Arteta and Arsenal are lacking those to different extents.

I'm all for having faith in sports science and statistics but coming out with that stuff in a press conference just comes across a bit unhinged as if he's in denial about how poorly they're doing. It's the sort of thing you look back on when a manager gets sacked and recognise it as the point that pretty much everyone knew it was over.

Fully agree with all of this.

xG is a useful statistic - I think it can be more informative than shots and shot on target... it also sometimes isn't. It depends on the match, tbh. Because of the way some chances count towards a side's xG (for instance, headers)... there's ways you can glean information from statistics.

But statistics alone aren't going to improve Arsenal. And pointing at these things in a really misleading way is, imo, very irritating tbh because it encourages people being sloppy or misreading statistics - or making simplistic arguments that hold weight when you just look at the stats... but don't hold weight once you look at other stats that weren't mentioned. I know of a guy who does/did this with Liverpool... and he's fairly popular (or at least he was when we weren't as good as we are now) but he's done some pretty questionable statistical cherrypicking to always have us looking like we were statistically fantastic. Even if we were shite by many other metrics... including the league table

I think he's got to be under serious pressure to be coming out with this stuff. I almost feel bad for him, because I think he bit off way more than he could chew in taking this job and it's not really his fault he was offered the job without really being ready for it. But if I'm honest, I didn't understand him being linked with either of the jobs he was linked with considering his complete lack of experience. Both clubs were too big for a complete novice manager - tbh I think each top flight club would probably have had a more reasonable alternative than the assistant manager at a world class side.

It's easy to say now with hindsight... for sure. But in my defense, I think I said this pretty early on when he was first appointed and now that it's easy to say I'll be damned if I don't repeat myself.

[Devil-Dick Willie]

XG doesn't take certain intangible factors into account though. 

Example. Timo Werner is in 1-1 with a goalkeeper. He hits it straight at the keeper. 
XG will say this was an expected goal. However, we know it's not, as it was Werner. 

Shit metric. 

[6666]

8 hours ago, Dr. Gonzo said:

lol Arteta making up statistics now - what the fuck is he talking about?

To be fair, Arteta actually gave a lot more detail.

 

[Danny]

On 19/12/2020 at 19:58, RandoEFC said:

This is a better stat than the Mina one. 

This is a stretch, some of those were at home no? Feel like it can only be a proper “grand slam” of a city if it’s away wins.

And yeah, you’re just having a laugh, and yeah maybe I’m being pedantic but if I’m not going to be pedantic then who will be?!?!?!?!