OK, that's where that's coming from. Having watched it again, it was definitely a shot on goal but it actually shouldn't be a 0.86 xG if you could truly take it in full context. He's 2 feet from goal but Pulskamp is literally right on top of him. It was going to be hard to head it around him. Looks like Beau was assigned to be the guy right in front of the Pulskamp. They jostled a little bit as the ball came in. Pulskamp tried to push him a way a little and Beau more or less held his ground, but by the time the ball is there, Pulskamp is right there like 1 foot in front of him. Jamie Watson said "This is all about the pace of the delivery!". Meh, not really. It was about the bend of the delivery. Espinoza's ball had a really nice arc to it. It bent like 3 or 4 yards from the time it entered the 6-yard box to the time it hit Leroux's head. It was far enough away as it entered the box that Pulskamp didn't think he could get it. It wasn't hit super hard, it just had a really go
Yeah, I'd still contend that the "right" xG is probably still > 0.5 on that shot since the keeper doesn't really have time to react and it's difficult to miss the target. If you had an xG model with thousands of input variables trained on an infinite number of shots perhaps you end up with something quite a bit lower than 0.86xG on that shot, but I think in practice they try to keep the regression model relatively simple (i.e. 10s of "X"'s used to predict the "Y", rather than hundreds/thousands). So much for the narrative about Espinoza's corners being poor.
I understand why xG has it that high, but I don't think it was that easy to score there, given how close Pulskamp was to him. There was really very little space to get it around him at that angle. If he went to one side or the other of Pulskamp's body it would have probably been wide due to the super narrow angle. I think I was the main one who complained about Espinoza's corners but I last complained about them a few years ago, and I've noted that they've improved a lot, and haven't complained since then. Probably, he read my posts, and realized he need to put in the work . Now if we can just get Niko to read my posts about his one-footedness. April 9, 2023:
That's why (at least to me) xG is such bullshit. It's such a subjective stat. If you had a dozen people looking at the same play I'm betting the xG would be all over the place. To me, this renders it pretty useless.
Of course it will be at times miss some critical context. But I think it's a matter of looking at the bell curve - there are always gonna be some cases that lie outside of the norm, but in most cases it's a reasonable approximation of the chance, as it's based on empirical data. So to me it's not the be all - end all, but it's not useless either.
There are certainly problems with how we use xG. For example, Beau's header was worth 0.86xG, while Josef's follow up (which he scored) was 0.56xG. But it's kind of silly to claim that the sequence generated 1.42 xG since obviously you can't really score more than 1.0 goals on a play. You would need to apply some kind of discount factor when you aggregate these xG values over the course of the game, otherwise you would be overrating the amount of actual offense generated. But as far as comparing individual shot opportunities in a vacuum, it's somewhat useful.
I think it's fine if a sequence of chances add up to > 1. It is simply a measure of a chance, irrespective of what happened prior or after. It is a snapshot of the current state at the time of the shot. But then it's all about how we use the data. With 2 games it's kind of pointless to have "MLS xG leaders" when Beau is 5th or whatever just because he had a header 2 feet from goal on one play and Josef is right behind him because he had a toe poke from 2 feet for a goal. Basically the Quakes presence in that top 10 is just a consequence of how we set up for that corner. And BTW, we've scored twice on that play already. We need to keep doing it.
Given the Quakes are using 0/3 U22 initiative slots this might be an interesting read... too bad it's behind a paywall... oh well at least they have a pic of our boy Luna Has MLS’s U22 Initiative been a success? Examining the league’s $300 million endeavor
I'm not an idiot. I understand that it takes factors like angle, distance, keeper position relative to shooter, etc. into contention, but all of those factors are also subjective. Basically, I think it's a way of pretending that objectivity can be attained by somehow rendering everything into 0s and 1s. I watch the game with Mark I eyeball - and I've been watching for decades as a player, ref, coach, and spectator and I find that not infrequently I vehemently disagree with the supposedly objective xG.
Well unfortunately I was right that Bruce has Pellegrino #2 at the CAM spot after Lopez... the question is did Bruce learn his lesson... not great that he didn't pull Pellegrino until the 74th minute, he should've been gone at halftime. MLSsoccer with the more accurate version
Regardless of how average position worked out I think the intention was for Arango to be the CAM. He was on the ball a lot more in midfield driving he attack and Pellegrino was supposed to join the attack late and make an occasional thru one. They were both playing pretty deep.
3 games is way to early but as of now Daniel is the best GK in the league... I believe he was 5th before this weekend.
Yeah, that’s what saving 5 point blank shots will do for you…unbelievable game from him. And it was all game long.
This is not exclusively Quakes-related, but an interesting article on "scanning" in soccer, where scanning is looking away from the ball to get a picture of what's happening around you. Not surprisingly, the best players tend to be high-rate scanners. But of course this is only part of the picture (if you think it's the whole picture, then you're not "scanning" in an intellectual sense ). You have to know what the "pictures" you're getting from your scans mean and how you're going to react to that information, and how well you can execute that reaction. But it seems that having those picture are a sort of prerequisiste to all of that. As Lampard (who was a very high rate scanner) says: “I think it [scanning] possibly is a talent,” he says. “But it’s a strange thing: my dad saying that I had ‘pictures’; it’s a really easy statement – just swivel your head, that’s not hard to do. And it doesn’t necessarily feel like a talent, does it? “But I think that the talent is, probably, understanding what you’re looking for and understanding what you want out of that.” There are some Virtual Reality tools out there that let you practice scanning and reacting to those scans. Some players, like Norway's Kristian Thorstvedt, use it before games to try to get sharp reading game scenarios. There is also something they call the "critical scan" which is the scan players do one to many times (or not) while the ball is coming to them. Of course you can't always tell when a player is scanning just by watching a game with the naked eye because the eye movements may not involve a noticeable head swivel but it might be interesting to see which Quakes players appear to be active scanners and which do not.
Funny, when I was coaching 12 year olds I would drill in a few principles such that every player could rattle them off when asked: 1) Go to the ball 2) Know who you are going to pass to before you get the ball etc. There's some good videos on Busquets on this (one example: )
Cool video. The VR software they mentioned is the same one mentioned in the Athletic article, "Be Your Best". I have to admit that I have not been able to fully appreciate Busquets's game because he looks generally slow and unremarkable, but I'll watch him with a little different eye now. I do think that his effectiveness is dampened a bit because of the level of physicality in MLS and, well, he's 36.
Hernan out until August (?). Josef and Niko out for March 22 game vs Charlotte. Team has 2 wins and 6 points now. How does Arena organize and run the team and hope to be within reasonable distance of last playoff spot by the time Hernan returns?
Quakes' March 29th Heritage Cup home game against Seattle will be affected by injuries. Paul Arriola just tore an ACL, and Jordan Morris and Pedro de la Vega are out for a month or so as well .
The way you'd do that so that you don't have to change formation is play Espinoza as a CAM / 3rd forward. He could still lean to the right side and cross. He just wouldn't be as wide and have as much defensive responsibility. Lima or Verhoeven as right winger. Probably Lima initially. But there are many other options. I think we should play Kaye, Harkes, and Beau at the same time. Pick one (probably Harkes) to be a little more on the attacking side. But that leaves us woefully thin at CM. Who subs in? What happens if one is injured? It will be interesting to see if Cruz Medina plays for The Town tonight. If he doesn't he'll be on the bench tomorrow night probably.
I'd like to see Espinoza come inside with the freedom to float around or interchange with Chicho...basically he gets the hernan role... which gives him license to find pockets or flank space...bring in Lima at right wingback. i don't know what's up with Kaye, maybe not completely fit yet, but in the above scenario I'd like to start Kaye with Leroux...I think they are both better than Harkes....Plus you would have Espinoza's crazy two way work rate helping in the middle too.
I think Skahan has the legs for the right wingback role too. Espinosa is perfect at wingback, but I agree with lopez out we need to move him to the 10 as a half winger. He would be a pressing demon up there.
There are a lot of ways Bruce could go with this, for the period that Hernán is out: 1) Play Niko or Skahan in Hernán's role. 2) Bring in Lima, Verhoeven, or Marie to play RW, and move Espinoza to forward. I say forward rather than CAM because I think you could treat it more like a 3-4-3, and then Espinoza is pretty much playing the role he's always played for the Quakes - a right side attacking player. 3) Play all 3 of Harkes, Beau, and Kaye, and one of them (probably Harkes) plays more of a 10 role. 4) Play Bouda or Pellegrino as a 2nd forward and Chicho plays more of a 10 role (basically what we did against MNU). Of these I think I prefer #3 and #2, and wouldn't be disappointed if we go with one of those. I prefer #3 a little more than #2 because it's only one positional change - swapping out who kinda plays the 10 role, and w/o relying on weaker players (Niko and Skahan). With #2, there are 2 changes - new starting wingback and Espinoza plays a different position. We've been training and playing with Espinoza as wingback the whole season / preseason. So it's more change. Also I think with the 3-4-3 (#2) our central midfield will be weak and subject to getting run over. With a central midfield of Harkes, Beau, and Kaye (#3), we'd be more stout in the central midfield, and could easily shift into a more defensive CM posture if the game state calls for that. Finally, of all the players that have not started, I think Kaye is the strongest player, and with #3 he's the guy that goes in.
So getting back to this, Bruce didn't wind up using any of my guesses (falling behind in his reading I guess). The closest was #2, where we bring in Lima and push Espinoza up. But I thought that would be in the context of a 3-4-3. I didn't expect Rodrigues to go out and for us to go 4 in the back. Someone described it as a 4-2-3-1, which sounds about right. -------------------------Josef-------------------------- -----Bouda---------Chicho-----------Espinoza ----------------Harkes--------Beau--------------- Ricketts-----Romney-----Wilson-----Lima--- Even though we lost I think it worked pretty well. The goals were both due to mistakes dealing with the offside trap (mostly on Ricketts, who probably played about his 5th game as a fullback in his life), and the attack generated 28 shots and 3.1 xG. That's a game you normally win. Now the question is do we have enough depth to sustain it. ----------------------------------Josef / Judd (ouch)----------- ----------Bouda / Pelle------Chicho / Niko-----Espinoza / Skahan ---------------------Harkes / Kaye--------Beau / Kaye-------------------- Ricketts / Costa----Romney / Rodrigues---Wilson / Munie----Lima / Verhoeven It's pretty thin, and we have only Kaye to spell Harkes and Beau. Really need another CM. We are pretty good on the back line, but otherwise the 2nd guys in many cases are big drop-offs (Judd for Josef, Niko for Chicho, Skahan for Espinoza), etc. Also, there's the unfortunate odd man out Paul Marie, who could fill in a number of places (inheriting TT's "Swiss army knife" mantle?). Right back, and either in Bouda's or Espinoza's place.