I’d also add that when a team puts a spy on him, as the commentators seemed to suggest in the Everton game, it creates space for others.
I don't disagree with that sentiment. My point is... A) Laca needs to go and I would like to have a vet ST to lead the attack vs having to rely on youngsters in that spot B) we can only have so much turnover in one window. If we sell both Laca & PEA and then conceivably buy a new vet ST does that take away from our ability to improve another position? My comment is also contingent on CL. With the MCity suspension we have a real shot at CL via 5th place or EL trophy. If we end up in CL and feel that PEA as our vet ST can keep us top 4 then I would be ok with missing out on the capitalization of his value. If we are out of CL he likely makes the decision for us anyway.
What i take out of the turnaround, and especially having read the deep dive a couple of weeks ago, is that the performances of all our key players collapsed because Emery was dreaful So no surprise that now we have smarter setups and systems everyone is coming back up Even Xhaka is wildly better
Yes, at least until his next royal f-up... or two. But TBF he's cut back his frequency of those. p.s. Nice use of "cromulent"... sounded familiar and the context was clear, but I'd be lying if I said I didn't have to look it up to check.
This seems bad: xG map for Arsenal Everton yuuuuuuup pic.twitter.com/FTns1fp7Tv— Caley Graphics (@Caley_graphics) February 24, 2020
Given the positions from which the shots were taken, why isn't Arsenal's xG higher. Nketiah made contact in the 6 yard box and didn't really have anyone on him. Aubameyang scored from the penalty spot followed by a chance when he was essentially one on one with the keeper, who was coming off his line. I admit I don't know the assumptions behind any of the expected goal calculations, but I would have expected Arsenal's expected goal total to be higher. Especially since Aubameyang, who has finished 26% of all his shots in league play, almost certainly finishes a much higher percentage when he's not under pressure.
Headers are low value based on historical data, and Auba was under pressure from Sidibe. Not sure what Caley has for the others, but Understat had Nketiah’s goal as a .57 (which is very very high) and Auba’s first as a .34 (a little low maybe specifically for Auba in that exact scenario, but still a very high value chance). Remember all shot conversion is only about 10%, we only took 9 shots the entire match, and basically only the first two goals were above average chances.
Sidibe was chasing him, he hadn’t caught him, so it was essentially a one v one vs the keeper. Since Auba has finished all of his chances at 26 percent it probably should have been a little higher than .34. Do any of the models take into account the player taking the shot. If not, they probably should. Messi’s shots should be given a higher xG than a normal striker from the same position. Finally, I understand that headers are low value XGs, but do the models differentiate between free headers, which Aruba’s goal was, and competitive headers.
When I said he was under pressure from Sidibe, I meant on the header. And it wasn’t free; it actually deflected off Sidibe, so it couldn’t have been that easy. As I said on the throughball chance, it might be slightly low for Aubameyang, but it was still counted as very high value (more than 3 times average shot value). As far as incorporating individual player data, I don’t know if any models do, but if they did I would have questions about the purpose. We have results to tell us what happened. The goal of the models has to be to tell us what is likely to happen if any given situation arises for the average player/team/league/etc.
It may have deflected off Sidibe, but he wasn’t challenged for the header, I’d consider that a free header. However, I’m at cross purposes, I’ve argued that there should be different xG values for free and contested headers and I also suggested that it might be appropriate to consider the player taking the shot in xG models, Since a very small percentage of Aubameyang’s goals come from headers, that would probably argue against increasing the xG value of his headed goal, free header or not. I’d suggest the purpose of using individual player data is to accurately reflect the performance of the team. The fact is a player like Messi, Aubameyang, or Salah is not an average striker.
Right. The whole point is to compare your team performance versus a baseline. It doesn't matter if you 'lose' a particular game on xG, because the detail of the game matters, but obviously in the medium to long term. if we cough up 2.5xG on the regular, we will drop a lot of points because shooting overperformance won't save us.
If you regard the data as a form of decision support, then this is not an issue. e.g I know in rugby, they look at a lot of stats including dominant tackles and missed tackles and they have the exact co-ords for all of them. So say the tackle dominance is missing for a midfield player, they will then go to the tape and look at all the game situations. e.g. maybe he was getting beaten on the rush defence by a particular tactic. So we don't need to add it to the model that our player is a better than average tackler. We know this, and go into the tape to find out what went wrong (or right) This is reflected in different stats - e.g. the players xG contribution. A select few players like Messi outperform their xG+xGA But as i recall correctly, very few players regularly outperform xG and this is thought to be because the real skill in scoring is shooting positions. i.e. the secret sauce of Rene Meulensteen and van Persie
What the xG map really shows is that having gone ahead twice in the game, Arsenal gave up far too many good shooting locations - and normally you'd expect to drop points. I mean none of us would have been surprised if Everton had equalised.
Yes. IMO Mustafi's development has also really suffered from arriving at Arsenal at the peak of Wenger's decline, and then Emery being even worse. Xhaka as well. We may be about to be shown that Xhaka has what it takes to be elite under the right trainer.
van Persie was maybe the best I've ever seen at 3-5 yard movements that gave him the space to get a fast and accurate shot off.
Aubameyang is finishing 26% of his shots this season and during his career he has finished 21% of his shots, but he has never been a high volume shooter. Is his finishing rate as high as it is because he exceeds his xG, or because he generally only shoots from good positions? The latter would explain the relative low shot volume.
As i understand it, ability to get into great shooting location explains most of being an elite goal scorer - and is thus captured by xG approx 20% finishing is basically elite anything above that will revert to the mean unless your name is Messi / Ronaldo etc This is also why people who claim a striker is bobbins because "he needs 4 or 5 chances to score" are idiots Obviously with volume, for an elite striker ideally you are approaching 4 shots per 90 without sacrificing quality.