This is what I've been screaming in my head for so long now. A good example being hoping a lad can replicate his stats when he's moved across Europe away from family to live out of a suitcase in a place where he doesn't speak the lingo.
You're right of course. We could just go out and sign a 20 goal striker or two like Birmingham did in the summer. It only cost them the best part of £20m. Whilst I get that some people are really anti data, you can't deny that for a club with a smaller budget such as ours, it's one of the ways that we're going to uncover a diamond in the rough. It's not going to work every time, but as someone says above, if you find someone who underperforming their xg, and put better players around them, they might take chances for you that they didn't in another season. It shouldn't be the only measure used for recruitment, but used correctly, and it can become invaluable.
I don't have the numbers to hand, but I'm fairly certain from memory Carlton Morris had quite a high xg in his loan spell at MK Dons prior to signing for us, even though he'd only scored 3 times.
No it's the difference between stats and reality.. it's what you get when you rely on computers instead of someone 'with a good eye' for a player. ( Not anti computer but it's not the 'go too' for everything) This also explains why we sign the same kind of sideways then back midfielders.. because the stats say they're good.. They can't compete, run and challenge for 90 odd minutes but they're tippy-tappy techy stuff is decent.. This really does explain the reason for our on the field decline. Those signing players are not sports people but statisticians.. For all they're good intentions and maybe work ethic and endeavour.. They are just the wrong people doing the job..
I'm not particularly interested in expected goals. It's actual goals I'm more interested in. And he's actually scored 12 goals in 67 appearances.
But that doesn't tell you whether he's a good striker who has been unlucky, or a striker who gets in good positions but can't hit a cow's arse with a banjo. Kayode Odejayi would have had great xg stats because he was strong and good at running off defenders onto through balls. What xg wouldn't have accounted for is the fact that when it got to actually finishing the 1 on 1 he was like a spider on acid.
I agree with that, and as I say, it's not the only measure that should be used. Proper old fashioned scouting would then be helpful to determine whether their poor goal return is down to luck or application. I just don't think xg should be dismissed as a load of *****, because it isn't. A club with a smaller budget has to find ways to identify players who might improve their squad as they can't simply throw tons of cash about, was my point. Some are going to work, some aren't. Fundamentally, Watters has been a let down. Though it's better that we've paid a nominal fee for him who hasn't done the business, than say spent a couple of million on a striker who then goes on to score 12 goals in 70 games.
I haven't looked into Xg much, but from what I can gather it's a judgement call isn't it? Like the scores given to a gymnast by the judges after the floor exercise, or the rounds awarded to a boxer. It's not hard data like the timing of a 100m race or the number of passes completed. It may be an algorithm making that judgement rather than a human, but same difference, just, when it's this nuanced, less accurate. That throws up red flags for me. I don't know what could give a striker a high Xg if he's only scored 3 goals. Maybe he converted all his chances, in which case I'd be questioning where he's standing. Or it may indicate he was getting lots of chances but missing them, in which case I'd question his shooting. Neither seems all that appealing. However, if you went to actually watch Carlton Morris and didn't think he was a good player, there's summat up with you.
XG measures the quality of the chance, rather than the ability of the player. The ability is measured whether or not a players exceeded his Expected Goals. The easiest way to explain is via penalties. The XG of a penalty is around 0.8, meaning that 8 times out of 10, a professional footballer should score a penalty. If Max Watters takes 10 penalties and only scores 3, we shouldn't be thinking "great, Watters has an XG of 8", we should be thinking "christ Watters is crap, he's only scored 3 times from an XG of 8" The higher the XG, the bigger the chance. Darren Barnard against Huddersfield was probably a 0.01 XG, but he scored cos he was class. Odejayi probably had a 0.85 XG against Cardiff - but he missed cos he's a Spider on Acid (which did make me chuckle). A high XG doesn't indicate ability. It's whether or not you're performing at the expected level based on your chances.
So, from your explanation, Xg on its own is not a metric you would ever use. For example, a ball trickling along the goal line would, you would think, have an Xg of 0.99, as you'd expect a goal in all but the most freakish of circumstances. But if I was presented with 100 of these chances and missed them all, I'd still have an Xg of 99. It should only be used in conjunction with goals scored. And even then, I could miss a load of chances and still have a few goals to my name if it's ricocheted in off my arse a few times. As for the Carlton Morris example above, only a few goals but a high Xg suggests a lot of missed chances doesn't it? And that's before we tackle the issue of who or what is assigning a score to the entirely unique chances a player is presented with. Does a penalty on a blustery day, on a pitch with a tendancy to give way underfoot, away from home in front of the opposition fans score the same as one at home in August when their keeper has been replaced by an outfield player? And that's just from a static kick 12 yards out, imagine the nuances in open play.
Absolutely yes. XG is based on the collection of data over thousands of matches over however many years they've been doing it. It needs to be used in conjunction with the actual number of goals scored. If Jay generates an XG of 20 across the season, but only scores 5 goals - it means you're a crap finisher. But if Harry Kane generates a XG of 20 across the season and scores 30 - it means he's a very good finisher. It's why you should also never look at it in small sample sizes either, as you correctly say. Chris Wood this season is hugely over performing his XG and has been on a real streak - whereas if you look at his career XG, it's probably higher than his actual goals You can also use it to judge whether a keeper is any good too - because you can use it in reverse. If a team has an XG for example of 1.3 during 90 minutes but score 8 goals, you can ask questions of your keeper.
So the formula for both a player or a team should be goals scored divided by Xg. If that is above 1 then you've got a clinical team/very proficient striker. Below 1 and the closer it gets to zero, the fewer chances are converted. It's not the whole story. A striker may score highly with such a formula but fail to get himself in the right position often enough to get many chances in the first place. And someone like Sam Winnall, who did miss chances, so may not score very highly, was that good at creating space for himself and had that much confidence in himself that even when he missed a couple he'd go on to score a couple. I can see how, if used correctly, it would be useful.
This is the absolute key of it I think. If we can find a 19 year old striker in the National League who consistently over performs XG and who looks like he could have all the attributes to make it, then it's a no brainer. The trouble is, everyone is doing the same these days. That's my reading anyway.
