And Its Smith Posted April 3, 2021 Report Share Posted April 3, 2021 Everyone: one of the worst misses of all time xG: 25% probability Link to comment Share on other sites More sharing options...
steveybadger Posted April 3, 2021 Report Share Posted April 3, 2021 39 minutes ago, And Its Smith said: Everyone: one of the worst misses of all time xG: 25% probability Struggling to think of a worse one. With other famous misses usually the ball is at least bouncing when it’s shanked over, or played at pace or bobbles etc to give a small degree of ‘difficulty’. That was pretty much gently rolled to him with nobody there! Link to comment Share on other sites More sharing options...
Major Isewater Posted April 3, 2021 Report Share Posted April 3, 2021 Miss Yesterday Link to comment Share on other sites More sharing options...
pride of the west Posted April 3, 2021 Report Share Posted April 3, 2021 Xg of 25%? Does that mean it's a 1 in 4 chance of scoring? Link to comment Share on other sites More sharing options...
Davefevs Posted April 3, 2021 Report Share Posted April 3, 2021 58 minutes ago, pride of the west said: Xg of 25%? Does that mean it's a 1 in 4 chance of scoring? Wyscout 0.28. When will @And Its Smith ever understand xG and use it to frame a debate properly? https://www.thesignificantgame.com/portfolio/do-naive-xg-models-underestimate-expected-goals-for-top-teams/ Link to comment Share on other sites More sharing options...
And Its Smith Posted April 3, 2021 Author Report Share Posted April 3, 2021 59 minutes ago, pride of the west said: Xg of 25%? Does that mean it's a 1 in 4 chance of scoring? Yes it does. xG is dead! Link to comment Share on other sites More sharing options...
Lanterne Rouge Posted April 3, 2021 Report Share Posted April 3, 2021 1 minute ago, JonDolman said: How can this be a 1 in 4 chance? Thought it was based on where players are positioned and where ball is. That has to be a 99 out of 100 chance at least! What is the xg for a penalty? That chance has to be better than one as there`s no keeper there. Link to comment Share on other sites More sharing options...
Davefevs Posted April 3, 2021 Report Share Posted April 3, 2021 1 hour ago, JonDolman said: To be honest I don't know anything about XG. I'm wondering if where opponents are positioned is even relevent to the XG number? You proved that with the first post. Joking 1 hour ago, Lanterne Rouge said: What is the xg for a penalty? That chance has to be better than one as there`s no keeper there. 0.76 (so a shade better than 3 in 4). XG models do vary, company to company. Generally they take position of shot as a basic....then advanced by position of keeper, defenders, height of ball, trajectory of shot, player taking shot, whether foot or head, etc, etc. I don’t think it’s unreasonable to say that if there was a big enough sample of Joao’s exact scenario, that it would score higher than 0.28. But based on shots from 15 yards with the keeper to be beaten (despite his poor position), etc, I suspect conversion was much lower in the sample dataset of similar-ish chances. Fletcher’s header was 0.29 yesterday. 7 yards out, nobody marking him....but the fact it basically ricocheted to him means in reality it was a much harder chance. It is far from perfect as a model, but the problem is that people use it too readily to justify things, e.g. results. Now that we know a penalty is 0.76 (irrespective of taker or keeper), to say the taker “should score” is wrong. Much better to say he should score 3 out of 4. Link to comment Share on other sites More sharing options...
BanburyRed Posted April 3, 2021 Report Share Posted April 3, 2021 No chance he'd have done that for us. He'd be stuck out on the wing. Link to comment Share on other sites More sharing options...
pongo88 Posted April 3, 2021 Report Share Posted April 3, 2021 A bad miss? You can not serious. Has everyone forgotten Baz? Link to comment Share on other sites More sharing options...
Lrrr Posted April 3, 2021 Report Share Posted April 3, 2021 2 hours ago, JonDolman said: How can this be a 1 in 4 chance? Thought it was based on where players are positioned and where ball is. That has to be a 99 out of 100 chance at least! You really think he's scoring the next 99 of those if it was done over? No miss hits by the player, the keeper never makes a save in 99, the player never kicks the ball into the other Reading player? Link to comment Share on other sites More sharing options...
Lrrr Posted April 3, 2021 Report Share Posted April 3, 2021 2 minutes ago, JonDolman said: The shot was hit too far in the way of the goalkeeper to miss the target, yet he still could not get a touch. It was out of the goalkeepers reach when he had the shot. That's an easy pass into an open goal. For him to not the other Reading player he'd have to almost kick it sideways. I think that's an incredible miss. And yet he did, he's not scoring 99 in a row after Link to comment Share on other sites More sharing options...
Davefevs Posted April 3, 2021 Report Share Posted April 3, 2021 37 minutes ago, JonDolman said: I'd back him to pass a ball into an open net 99 times in a row. It was a freak miss. The fact that there was a keeper, changes your “pass a ball into an open net” dynamic massively. Real time reactions, etc changes that dynamic further. Dont get me wrong I still think it’s a better scoring chance that the xG quoted of 0.28. But you have to take into account it’s done on a sample of similar situations, not exactly the same situation. Some of those will have different factors, perhaps a keeper well positioned, not left stranded by a suicidal, no look back pass. But that 1000 chance archive of similar shots produced 280 goals. The way chances get bracketed into the same sample is part of the issue here. But you have to bracket chances together to get a decent sample size. Otherwise there’s probably never been a Joao chance exactly the same ever. Link to comment Share on other sites More sharing options...
Edge1981 Posted April 3, 2021 Report Share Posted April 3, 2021 My favourite miss was this beauty from Ronny Rosenthal Link to comment Share on other sites More sharing options...
reddoh Posted April 3, 2021 Report Share Posted April 3, 2021 5 hours ago, And Its Smith said: Yes it does. xG is dead! XG is dead long live XH. Link to comment Share on other sites More sharing options...
IAmNick Posted April 3, 2021 Report Share Posted April 3, 2021 XG is a statistical analysis tool, and as such doesn't do well for incredible outliers like this instance because there is almost no sample size at all for it to reference. If this situation was happening regularly it'd be more accurate, obviously. People complaining about it just don't understand how it works once again it seems. Link to comment Share on other sites More sharing options...
Davefevs Posted April 4, 2021 Report Share Posted April 4, 2021 7 hours ago, JonDolman said: Yep which means every XG chance number is not accurate, as every chance is unique. But the other stats we have always seen like shots on target, possession, fouls etc I assume are 100% accurate. Yep, if you want to look at it like that. But of course you’ll get plenty of people that will say “we had more shots / SOT than our opponent, we should’ve won”. Which is in fact bollocks. 3 tame shots from 25 yards straight at the keeper or 1 point blank volley from 3 yards that crashes against the bar and counts as off-target. All xG is trying to do is give a better granularity to a chance, to allow comparisons over a period of games, sometimes one game (which I don’t like). I think all xG does is prove that for a striker in a one-on-one that “he should score 9/10” is wrong. Its being mis-used, and therefore it’s spread into football parlance and further mis-used. Very useful in comparing a striker with another over a season, but.....don’t use it alone, bunch it with other things. And the danger in using averages, e.g. xG per 90 is that you miss the distribution. Is Pkayer X’s xG per 90 of 0.5 made up of 3 games of 0.00 and 1 game of 2.00? Or is it 4 games all around the 0.50 mark? Proper analysts in football will be looking way deeper than a number. Link to comment Share on other sites More sharing options...
Red-Robbo Posted April 4, 2021 Report Share Posted April 4, 2021 Whatever the metric says, it's got to be harder to miss the goal there than to score. Link to comment Share on other sites More sharing options...
Warwickshire Red Posted April 4, 2021 Report Share Posted April 4, 2021 the Iwelumo miss was pretty bad. At least he scored for City on loan under GJ iirc. Link to comment Share on other sites More sharing options...
Davefevs Posted April 4, 2021 Report Share Posted April 4, 2021 1 hour ago, JonDolman said: I don't really pay much attention to shot stats other than if we've had no shots on target! a lot then I guess it's one of those things some managers would find useful and some wouldn't. All personal preference. Most will use all manner of stats put together by the analyst teams. But like all these things, you need to understand how they’re collated, how they are interpreted, what they are telling you, etc. That’s the skill. I can easily throw up the following: which basically showed for 15 minutes of Friday’s game we allowed Stoke to pass it around freely with no tackle, foul, etc....whereas for the other 75 mins we didn’t. Why did that happen. It’s not all the Bakinson for Bell sub, but you can start to draw conclusions. In that little spell of 15 minutes, Steven Fletcher got 2 x 35 yard passes into his feet from the defence, with no pressure from Mariappa, who allowed him an easy touch, and subsequent layoff. So not just Bakinson’s fault but Mariappa too. Or was it Lansbury’s fault for not screening that passing lane? Link to comment Share on other sites More sharing options...
Davefevs Posted April 4, 2021 Report Share Posted April 4, 2021 5 minutes ago, JonDolman said: Yeah Bakinson can't really press that well and not particularly great movement either when we have the ball. Suddenly direct balls were less effective chasing them or trying to win 2nd balls. I thought Wells was not very good but then he might have been better had Bell not got injured or say Watkins came on instead of Bakinson. Lansbury I don't think is the best defensively. I don't think he's ever played this deep before. I'd like to see Vyner come in as a more natural defender with more mobility too. I thought Wells and Bell were looking encouraging in that opening 15....unselfish running (Semenyo too) to put Stoke under a lot of pressure....even if we didn’t back that up. Service was poor from wide, crossing especially. I thought Massengo would come into the 10, and Bakinson alongside Lansbury, but as you say could’ve used Watkins instead....wouldn’t have minded seeing Wells more as the 10, and Watkins as the 9 in fact. Vyner - definitely. Link to comment Share on other sites More sharing options...
italian dave Posted April 4, 2021 Report Share Posted April 4, 2021 On 03/04/2021 at 17:55, JonDolman said: How can this be a 1 in 4 chance? Thought it was based on where players are positioned and where ball is. That has to be a 99 out of 100 chance at least! Simple. It could have gone in the goal, left of the goal, right of the goal, over the goal. 1 in 4. That’s right, eh @Davefevs Link to comment Share on other sites More sharing options...
Lrrr Posted April 4, 2021 Report Share Posted April 4, 2021 5 hours ago, Davefevs said: I thought Wells and Bell were looking encouraging in that opening 15....unselfish running (Semenyo too) to put Stoke under a lot of pressure....even if we didn’t back that up. The balls in behind putting the back 3 under pressure dried up a while after Bell went off, Bakinson was never going to look to do the same, Wells and Antoine can only do so much when you take away the other runner Link to comment Share on other sites More sharing options...
Lrrr Posted April 4, 2021 Report Share Posted April 4, 2021 Just seen on motd 2 Che Adams put a 1v1 with the keeper wide of the post from 15 yards which was completely under control, how many shots out 100 should be on target.... is that a 1 in a 100 off target as well? Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.