Stats 24/25 Season

Is there some kind of anomaly taking place this season? If xG worked as you'd expect then you'd think it would deliver an approximate balance of 0 if you added up the xG total for all the teams.

Not exact I know but as a rough figure it should be approaching zero since that's the principle behind it. Some teams would overperform and some underperform and, with some deviation from the average, the cumulative total would be dragged towards a neutralish figure.

Yet look at the table for this season so far:

View attachment 16786

Only three teams are overperforming and the total is cumulative is 73.67 fewer goals than expected - or an average about 3.7 goals per club under the expected total.

Seems a bit odd. Either something is out with the model or there's something strange going on. Any thoughts @Matt Stevenson as our resident expert?

Or is it like this every season? Doesn't make sense to me if that's the case.

I used AI to do the maths on the image so might be off but it's 761 (rounded up) total xG so it's off by about 10%
 
The model is a bit more complex and will consider a lot of the goal is blocked by players, the ball needs to travel through the gap between the two defenders and then beat the 'keeper who is positioned (IMO) pretty well.
View attachment 16785

The model will only consider the league average and not the individual player. This means if JK is particularly proficient at these types of opportunities you would expect him to overperform his xG. Also scoring twice from similar situations doesn't make him less likely to score from there in future, if anything it could be evidence that he's quite good at those kinds of shots.

You are correct, also that shots from there can be used defensively from a corner but you would expect a much more congested box which I would expect the model to manage.

It's something I've been thinking about with us underperforming generally against xG. I wonder if the style of football means we're getting players in situations where we can (and do) take shots but those situations are frequently not where the player taking them is strongest.

I don't know how relevant it is to the stats but Dango literally dragged their defender to the left to create the gap for that shot. Probably helped.
 
Dango's header was a 1 in 4 chance? Never. The height he leapt, and the way he hung in the air, were remarkable.
 
AFCB had 33 touches in the opposition’s penalty area with Ouattara having 13. N Forest had 34 touches in our penalty area.

View attachment 16760

AFCB created 9 chances (3 big). N Forest created 12 chances, 1 big. Semenyo created 3 chances.

View attachment 16761

We had 16 attempts (4 big) noting that chances can be provided after saves, deflections or a defensive error and not created. We hit the target108 times. N Forest had 18 attempts, one big and hit the target 4 times. Ouattara and Semenyo had 4 attempts each. The big chances are reported as falling to Ouattara (3) and Semenyo.

View attachment 16762
We hit the target108 times.

Impressive.
 
Is there some kind of anomaly taking place this season? If xG worked as you'd expect then you'd think it would deliver an approximate balance of 0 if you added up the xG total for all the teams.

Not exact I know but as a rough figure it should be approaching zero since that's the principle behind it. Some teams would overperform and some underperform and, with some deviation from the average, the cumulative total would be dragged towards a neutralish figure.

Yet look at the table for this season so far:

View attachment 16786

Only three teams are overperforming and the total is cumulative is 73.67 fewer goals than expected - or an average about 3.7 goals per club under the expected total.

Seems a bit odd. Either something is out with the model or there's something strange going on. Any thoughts @Matt Stevenson as our resident expert?

Or is it like this every season? Doesn't make sense to me if that's the case.

Unfortunately, I've nothing important to add. I think they are likely to recalibrate the models over time if they find that the models aren't fitting well. But it could just be a one-off season where people keep hitting the post etc, and that if adjusted there could be a big over-performance next season.
 
xG is based on past performance.
As they say in investing, past performance does not necessarily indicate future results!
Players and styles of play all change every season so it makes sense there will be a bit of a gap between expectation and reality.
 
This Opta page for AFCB appears to have been updated after the Sunday matches:
1737989875648.png
Bournemouth now ranked 19 globally by Opta, the 6th highest among EPL teams. Their Predicted Final Points total for AFCB is up again, and I count about a 64% chance on that histogram of finishing 6th or higher.
Elsewhere on the Opta site, the latest season simulation places both Forest and us as likely to finish 5th, 6th, or 7th.
This next one is amazing:
1737990297181.png
Those two rightmost columns are probabilities for a Champions League and Europa League spot, respectively. They are not tabulating figures for the Conference League.
It appears to me that the top two League spots are mostly settled, and that a 5, 6, or 7 team mini-league has formed for the remaining European spots, and AFCB are in the thick of that.
 
Is there some kind of anomaly taking place this season? If xG worked as you'd expect then you'd think it would deliver an approximate balance of 0 if you added up the xG total for all the teams.

Not exact I know but as a rough figure it should be approaching zero since that's the principle behind it. Some teams would overperform and some underperform and, with some deviation from the average, the cumulative total would be dragged towards a neutralish figure.

Yet look at the table for this season so far:

View attachment 16786

Only three teams are overperforming and the total is cumulative is 73.67 fewer goals than expected - or an average about 3.7 goals per club under the expected total.

Seems a bit odd. Either something is out with the model or there's something strange going on. Any thoughts @Matt Stevenson as our resident expert?

Or is it like this every season? Doesn't make sense to me if that's the case.

It's down to the same point I've made a few times. Take us as the example, we will take on low xG shots that other clubs simply will not take.

So for example, if we shot on average 10 times more often than other clubs, because we don't mind taking a 0.1% shot, then by the end of the game our expected goals would be 1, despite never creating a real big chance on goal.

Other clubs will clock up lower xG chances over a number of games and so overall as you get deeper and deeper into the season if you add everything up over 38 games it becomes less and less relevant as a statistic.


On a game-by-game basis it can give more of an accurate feel on where the game could have been based on "big chances", for example if the game had finished 3-1 (the xG score) on the face of it no one would have argued that.
 
Sorry, can't help myself from posting this again.

We're now 17th.
A year or two ago, we'd have been happy with being 17th in the PL and out of the relegation zone.
But this is globally. Yowsers.
 
The model underlying this points tracker did not expect much from the Liverpool match, so AFCB remain about 9 points ahead of this 51-point pace:
1738509891886.png
Across the first 5 matches of Block IV, AFCB have gathered 10 points against an expected 6.3 according to Opta's match predictions, and an expected 5.7 according to this model. The team does seem to have ascended to another level, and that may happen again this season as players continue to return from injuries.
 
THe 1.6 xG against Liverpool was once again very near our season average against decent defensive sides:
1738510901491.png
The xGA of 2.5 is on the high side, but does include the Liverpool penalty, and by itself that figure is not a statistical outlier:
1738511088791.png
I will have another look at Opta's Bournemouth page once all the weekend PL matches have concluded and they have completed another season simulation run. I expect to see us still north of 60 predicted points and likely to finish in a European spot.
 
Here is the xG from the Liverpool game. The game started with chances, a minor one for Semenyo, but two big chances (both around 1 in 6) for Liverpool from MacAllister and Diaz, both of which were blocked. AFCb were attacking but the final ball wasn’t there, Semenyo almost created a goal from nothing tricking Liverpool defenders before firing against the post with Allison beaten (rated 1 in 40) before Liverpool took the lead through a disputed penalty (there’s been enough talk about this on the forum) which Salah scored (79%). Liverpool. Liverpool appeared to have a very good chance directly after (1 in 2) when Diaz headed at Kepa, although he was offside, but as play continued the xG countered. Liverpool had a few more presentable chances, from Trent Alexander Arnold (1 in 10) and Szoboszlai (1 in 12), but Brooks had a goal disallowed as Kerkez was slightly offside in the build-up. The xG at the break, was AFCB 0.18 Liverpool 1.88.

AFCB started the second half well, with two good chances, for with a reasonable chance, with Ouattara header (1 in 9) and a Kluivert shot (1 in 7). Two minutes later Semenyo had an excellent chance (1in 2) but couldn’t lift it over the outrushing keeper. Liverpool weren’t offering much and then AFCB had another great chance, a long shot from Tavernier hit the inside of the post, rebounding back to Kluivert, who with nearly an open goal shot wide (1 in 2). This sparked Liverpool into life and following a quick break Salah expertly curled one around Kerkez into the far corner (1 in 20). There were only minor chances after this, bar a late attempt from Jones who headed wide from close range (1 in 3). The final xG was AFCB 1.58 Liverpool 2.52, with understat having a similar margin, at 1.65 vs 2.72.

xG Timeline.png
 
There were no changes to the team who won at Newcastle and home to Forest, which was unsurprising given the good results and the lack of alternatives. An interesting point is that Christie (10) sat in front of Zabarnyi (27) and Adams (12) did the same with Huijsen (2) having two banks of 2, with the full backs, Cook (4) and Kerkez (3) as advanced as the centre midfielders. Ouattara (11) was again dropping deep with the furthest forward being Kluivert (19) and Semenyo (24). In the 67th minute, Tavernier (16) returning from injury replaced Brooks (7) and in the 80th minute (at 2-0 down), Jebbison (21) replaced Christie), presumably playing as centre forward with Ouattara (11) moving to the wing.

Av Pos.jpg
 
Semenyo had most touches with 64, he had most touches in the opposition half (60) and in the final third with 47.


Touches.png

Zabarnyi attempted most passes with 56. Semenyo made most passes in the opposition’s half (28) and in the final third with 21. We averaged 79% pass completion with only Kluivert of the starting 11, at over 85% although many were over 80%.

All passes.pngPass Opp Half.png
 
AFCB players tried to take on a man 27 times being successful 12 times (Kluivert was successful 4 times out of 7).

Take Ons.png

AFCB players were tackled 21 times with Semenyo being tackled 8 times.

Tackled.png


AFCB attempted 17 crosses, being successful 4 times (twice from Ouattara), although Kerkez was none from 5.

Crossing.png
 
AFCB had 29 touches in the opposition’s penalty area with Semenyo having 15. Liverpool had 34 touches in our penalty area.

Pen Area Touches.png

AFCB created 8 chances (2 big). Liverpool created 15 chances, 3 big. Christie, Ouattara and Kluivert created 2 chances each.

Chances Created.png

We had 14 attempts (3 big) noting that chances can be provided after saves, deflections or a defensive error and not created. We hit the target 4 times. Liverpool had 19 attempts, 4 big and hit the target 7 times. Semenyo had 6 attempts each. The big chances are reported as falling to Ouattara, Kluivert and Semenyo.

Attempts.png
 

;