| + Visit Cardiff FC for Latest News, Transfer Gossip, Fixtures and Match Results |
I know many are very sceptical about xg ratings and I must say, after watching all forty six of our games this season, I'm struggling to see how we can be as high as we are in this table. Millwall and QPR are a lot lower than their actual finishing positions, but I'd say City are the side with the biggest difference between their finishing and their position in the xg table.
https://twitter.com/EFLStats/status/...942336/photo/1
Does that mean we have been creating chances but not putting them away? (Not sure how it works). I’d tend to go for that, if it is the case.Originally Posted by the other bob wilson
I also think we’ve been quite unlucky under Morison on more than a few occasions. We’ve really missed Kieffer Moore this season, even when he played he wasn’t really here. I think if we’d had Hugill (or a similar replacement) from the start we’d be much higher up the league.
Hugill scored 4 goals in 18 appearances for us. He wasn't exactly prolific in front of goal.
But still ten times better than our other options
Over a season that would equate to about 11 goals, plus the other work he does. Would certainly have added a good few points .
Xg is the biggest load of bollox in football
Seconded. It's utter nonsense.
xG: 'Cardiff would be a lot higher in the table if they had scored more goals, conceded less, won more matches and lost fewer.'
Yeah, we know. But they didn't. Hence they finished 18th.
I think it mostly reflects the amount of time we spent chasing games, teams tend to generate more xG if they're trying to get back into a game
Just an excuse for managers and fans to clutch at straws when a result goes against them.
I very rarely see Xg brought into the conversation after a positive result.
Not too much to say about this, but barely even interesting if you do a comparison of all data in the table with the league table itself.
We conceded 13 more goals than we were expected to. More than any team in the division. Equates to how shit we've been at the back. Scoring 8 less than we were expecting to isn't the highest - West Brom were expected to score 23 more goals than they did.
I concur with others, I think xG is nonsense.
That makes as much sense as anything to explain it. I’m not convinced by xg, but I did find the difference between our positions in both tables fascinating when most sides position in the xg table was so similar to their positions in the league table.
Just been looking at each club's top scorers. Apart from us, only Barnsley failed to get anyone into double figures. Flint and Colwill were top for us with 6. (Barnsley's top scorer had 9 - so almost double figures).
That has to change next season.
XG itself isn't 'total bolox'. However, you have to be careful when interpreting what it means. This example of use IS total bolox as you can't use is to project what would have been over an entire season across all teams and all games. The average punter, and sky sports folk have no idea how to use it but trust me there are some incredibly useful applications for it (and other stats) used by team statisticians for performance management and scouting etc.
Yes, it has its uses although it isn't the magic riddle that solves football. I see people thinking that 55% possession means the team deserved to win, which is a much worse way to interpret a statistic.
If you think of xG as just how many shots did the teams have, and how good were those shots (six yards out vs 30 yard speculative) then that's it. They doomed themselves from the off by calling it 'expected goals'.
XG is just wank*ng material for statto’s
That's not really what XG is. It's all about the quality of the chances created rather than the end product. It's a complex formulae and is often misunderstood. Source: I've done statistical analysis for a living for 20 years
You're probably already aware of this anyway, but one of Colwill's goals came in the FA Cup, so City's top Championship scorers this season were:
6 - Aden Flint
5 - Rubin Colwill
5 - Kieffer Moore
4 - Jordan Hugill
That's pretty grim.
Correct me if I'm wrong (and I may well be), but doesn't xG differ from site to site? Aren't the figures open to interpretation and based on opinion rather than fact (ie: one person's view of what is a clear opportunity by comparion with another's)?
As far as I'm aware, there's no central, definitive xG source. Is that right?
I'm pretty sure there a standard algorithm provided by Opta.
There's an "interesting" article about it here https://theanalyst.com/eu/2021/07/wh...cted-goals-xg/
Maybe I didn't explain myself clearly but that's what I meant. A shot from the six yard box is higher quality than a 30 yarder from a centre back. They might both be on target but the closer one is more likely to go in, and xG 'grades' that
I know a few guys who us xg when it comes to making their weekend bets, and I can see how that might help. I can also see that your average pundit might make use of the additional information. But I'm more interested in the Expected Points column, where we came up short by 16 points.
We’ve undoubtedly had a crap season but thinking back to our championship winning season, our top scorers only got 8. Personally I’d be more than happy with every member of the squad scoring 5 or 6.
yes there are different people calculating their own versions of xG. some of which take into account different things.
the most basic are just looking at the position on the pitch the shot was taken from and comparing it with how many shots from that position in their database went in.
other ones also take into account various other things, like proximity of defenders, position of goalkeeper, where the pass has come from, body position of the player striking the ball etc.
it usually depends on what data is available for the League you are interested in, the lower down the pyramid the less data.
all of these are usually pulled directly from the video feeds by AI rather than humans manually adding the data, which can be a source of error as well.
that said even the most basic version of the stat can be very useful, and the most complex even more so. and there are a range of even more abstract stats like xT or xgChain that have different uses too, often more practical.
lots of the xG content we get in the public domain is often weak, and presented by people with a limited understanding of stats i.e. individual shot xG calculations for a basic style xG might look interesting, but the error bars on them due to the other factors mean that they're not actually very helpful.
looking at rolling xG for like 10 games or more helps to average out some of the problems with the data. "better" xG would possibly need less games (i.e. less shots) to minimise those errors.
one of the big problems with it is how affected by game state it is. we know the first goal in a game changes the way teams set up dramatically.
if we looked at xG teams were creating only while the scores were level (as I've seen some people do) then I think we'd be fairly near the bottom.
that's the most commonly seen one, but there are others
Posting Permissions
Reply With Quote