- Thread starter
- #21
Nope.
Got something to say or just want fewer pesky ads? Join us... 😊
[Albion] Have we been lucky? (xG stat geekery)
- Thread starter Bozza
- Start date
Nope.
No, because the stats that Bozza posted are just based on this season, and we haven't scored a penalty this season. Actually over the last couple of seasons we've performed close to Experimental361's E Ratings, which is based on the expected goals method.
Nope you didn't ignore them? Or nope you didn't see them? Because all those lovely charts I posted last year were based on the expected goals method.
- Thread starter
- #25
Nope, I didn't ignore them. As I said, I'm taking more interest this season, probably because of xG popping up at the end of MotD highlights for each game.
That is hardly new though is it?
2, 2, 7 then 2 goals again last season means he's not in the team for his goals.
That said if he was, he wouldn't be in this team at all, as he would have been sold on a couple of years ago, for considering more than the 50p increments Burnley were offering.
I think I'm either being given a lesson in stats or being told to shut my big trap
Penalties would be a bonus
I'm unsure I am going to open my trap as find it all confusing and I have more questions than answers. Definitely could do with a few more penalties - at least 3 per game.
AmexRuislip
Trainee Spy 🕵️♂️
I'm unsure I am going to open my trap as find it all confusing and I have more questions than answers. Definitely could do with a few more penalties - at least 3 per game.
your very brave
brightn'ove
cringe
I think all its good for is judging how well a team is actually playing, beyond just looking at their results. As said, it doesn't take into account the quality of the players. At the moment palace could get a chance that has an expected goal score (?) of 99.9% and would still miss. On the other hand, hemed's header against West Brom probably had a very low xG score, but he still finished it.
With enough data, and a data scientist who knows what they're doing working with that data, it is entirely possible to model the variables and for that model to (generally) produce reasonably accurate predictions.
So the question then becomes: do we have enough relevant data? And is the data scientist who's using that data know what they're doing?
As far as relevant data: for the most part, there's plenty enough historic data to be able to get reasonably close to a pretty accurate model. Where the modelling for *this season* should be taken with a pinch of salt, however, is that we're not far enough into the season to have enough data on any 'new' variables that this season introduces. An earlier post touched on those: the variables that need to be considered include the player(s) involved in trying to score the goal, as well as those involved in trying to prevent it. Even more, it depends on what sort of form those players are in.
For our stats, I'd suggest the modelling is yet to have received enough data to be able to accurately model how good our attack is at finishing, or how good our defence is at blocking. I think the time to start really keeping an eye on these stats is from half season onwards (once we've played all teams once).
Edit: for this season so far, there's the massive underperformance of Crystal Palace in attack to consider as well. Looking at the Prem table as a whole, the modelling has predicted 9.85 goals more than have actually been scored. Crystal Palace accounts for 7.14 of that difference on their own. Similarly, looking at those teams who have already played against Palace you can see that they all (with the exception of Liverpool, who leaked against Watford, City, Arsenal early season) have outperformed the Goals Against metric by a good margin.
Last edited:
I'm pretty sure this is the case, tbh.
I've been watching them come up on MotD and trying to rationalize them, but there are so many games where a team has won 3-1 and it says their 'expected goals' are 0.3 or something, that it just seems to be statistical nonsense.
Would be worth seeing the goals in games like that. You do occasionally get those freak games where one side scores goals that never should have gone in, largely due to defensive howlers or massive deflections from what we're otherwise low percentage shots.
E.g. Stockers having goals go in off his back after deflections from the woodwork.
- Thread starter
- #33
The xG for our game with West Brom was:
WBA: 1.03
BHA: 0.44
That 0.44 seems very low to me and must mean that our chances were deemed not very good chances at all...
Gross #1: low because of the number of defenders between him and the goal?
Gross #2: low because of the distance from goal and covering defenders?
Hemed #3: low because covering defender, position of the header?
I'm struggling to think that the mean for those three chances (not to mention any others we had) is only 0.44 of a goal.
- Thread starter
- #34
xG map for Albion v Newcastle:
[tweet]912118715835125760[/tweet]
[tweet]912118715835125760[/tweet]
- Thread starter
- #35
Hmmmm - this is a different total for the WBA game, although still has WBA as having created the better chances.
[tweet]906552464631091200[/tweet]
[tweet]906552464631091200[/tweet]
- Thread starter
- #36
We created enough to score a goal against Watford (if only!)
[tweet]901477838553055232[/tweet]
[tweet]901477838553055232[/tweet]
- Thread starter
- #37
Dolph Ins
Well-known member
Probably about where I hoped we would be at the beginning of the season. I enjoy looking at the stats but they often can't be matched to the result. It just seems footie can't be broken down into a mathematical formula. Maybe that's what makes it so interesting.
Taybha
Whalewhine
I love all stats and they really are easy to digest concerning football , if you take the time to look at the below mathematical solution to saturday's game v Man Utd the result is clearly visible .
Dolph Ins
Well-known member
So 2-2 then.