Match Thread Coventry City-SWFC Match Thread- Saturday 5th October (4 Viewers)

[fernandopartridge]

100%.

Think it also suits our forward players - I don't mean lumping it forward either i mean playing through the lines using Sheafs abilities to get it to our front 4 quickly and then getting at their back 4.

Just think we went back to basics and what we know Tuesday night. Hopefully see it again Sat!

I don't think we've ever not tried to play a relatively quick style even if we're having more possession, it's just having the players and confidence to execute it, it needs players willing to receive the ball in tight positions to open the game up.

 

---

[mmttww]

I don't think we've ever not tried to play a relatively quick style...

Exactly. MR has said since pre-season games that this is what's expected so that's gonna be how we train.

Players have to show some guts and make it happen. Not like it's news to them. Mostly on the midfield tbf.

 

[Perennial Lurker]

So did EMC, that was the first time I’ve seen him work back properly.
I think he’s had the facts of life explained to him about the expectations of the effort needed in the championship compared with League 1. If he continues like he did on Tuesday he will be a great asset for us.

I think he will be too , it was noticeable that the other players made a fuss of him after his hand in BTAs goal.
Just needs a scruffy goal to get up and running but he's not hiding on the pitch ,he's looking to get involved

 

[Philosoraptor]

Xg is just a quantitative measure of chances created, which is the main factor people use to discuss a teams performance anyway.

And that there is exactly the thinking that worries me,

No offense skybluecam.

 

[skybluecam]

And that there is exactly the thinking that worries me,

No offense skybluecam.

I’m not even sure what you mean

 

[Philosoraptor]

I’m not even sure what you mean

Analytically it is a very poor tool to use for team performance.

I think i can remember somewhere someone describing it as 'industry leading'.

It is nowhere near it.

Just the best of a bad bunch currently in use.

 

[skybluecam]

Analytically it is a very poor tool to use for team performance.

I think i can remember somewhere someone describing it as 'industry leading'.

It is nowhere near it.

Obviously no single number is going to be able to accurately capture how a team performed over a whole game of football.

But it is certainly very useful. I’d be interested why you think it isn’t.

 

[Philosoraptor]

Obviously no single number is going to be able to accurately capture how a team performed over a whole game of football.

But it is certainly very useful. I’d be interested why you think it isn’t.

It is certainly no Stockfish.

 

[skybluecam]

It is certainly no Stockfish.

???

Chess has a finite number of moves/states. Football is infinite.

 

[Philosoraptor]

???

Chess has a finite number of moves/states. Football is infinite.

More or Less: Behind the Stats - Are there more possible games of chess than atoms in the universe? - BBC Sounds

We investigate how the vast possibilities in chess compare to the vastness of the universe

www.bbc.co.uk

I give up!

 

[skybluecam]

More or Less: Behind the Stats - Are there more possible games of chess than atoms in the universe? - BBC Sounds

We investigate how the vast possibilities in chess compare to the vastness of the universe

www.bbc.co.uk

I give up!

You seriously don’t understand why it is impossible to mathematically analyse a football match in the same manner as a turn based board game? Lmao

There is very finite set of possible moves each turn in a game of chess. Football doesn’t even have turns.

Honestly this is one of the weirdest pseudo intellectual arguments I’ve ever heard lol

 

[Matt smith]

I agree, pretty sure he’ll stick with it anyway.

Also from what I’ve seen so far I don’t want BTA playing on the right.

Defensively you can really see the difference when he plays on the right, drifts into the Centre of the pitch too much

contributed to Swansea’s second

 

[Philosoraptor]

You seriously don’t understand why it is impossible to mathematically analyse a football match in the same manner as a turn based board game? Lmao

There is very finite set of possible moves each turn in a game of chess. Football doesn’t even have turns.

Honestly this is one of the weirdest pseudo intellectual arguments I’ve ever heard lol

Not really. I can understand that it wouldn't be a zero-sum approach. but you could take other approaches which should at least be better than xG.

 

[SlowerThanPlatt]

Get the Glicko-2 out!

 

[Philosoraptor]

Get the Glicko-2 out!

It is as good as a yardstick measurement.

 

[quinn1971]

If we want to play the same as Tuesday assante has to play up front, started the press from the front, simms doesn’t do that, feel a bit sorry for him, don’t think he’s gonna fit in to how we should be playing at home, away games maybe

 

[skybluecam]

Not really. I can understand that it wouldn't be a zero-sum approach. but you could take other approaches which should at least be better than xG.

xThreat is probably similar to what you’re suggesting but I think it’s too limited to be useful.

I suspect in the coming years AI may be able to get closer to what you want. But it’s never going to be anything like as good as stockfish is, for the reasons I mentioned.

 

[Cally Fedora]

Can’t see any scenario where Simms starts v Wednesday. They’ve rightly identified that we don’t have the back 4 to play out from the back and the ball is getting forward quicker. If you’re playing that way you need a hassler up front which brings in BTA and Bassette.

 

[ProfessorbyGrace]

I don’t wish Sheffield Wednesday any ill will, however what I will say is - and I believe I speak from a majority standpoint - I hope we smash the Alans off them on Saturday. ⚽️

‘FORWARRRRRRD!’

 

[TomRad85]

Sheffield Wednesday never win at our place.

 

[Gynnsthetonic]

Sheffield Wednesday never win at our place.

Jinxed it

 

[Gibbo]

not changing the team and being shite against Swansea will mean he is likely to change it to the xi he thinks is best to play against Wednesday

There's still a place in the modern game for that style of football . Brentford and Villa play it effectively in the Premier league.Klopps Liverpool side also

But crucially the Liverpool goalkeeper gets it away within seconds, almost instantaneously. Then they are in the opposition half.

We have been taking an eternity to get beyond the first pass

 

[shmmeee]

???

Chess has a finite number of moves/states. Football is infinite.

Football is absolutely not infinite, there’s a limited number of players in a limited space. The barrier to modelling isn’t resolution, we can accurately model the weather after all.

We can absolutely build a predictive model using some kind of tokenisation and traditional feature extraction from current video segmentation and labelling techniques. Even a relatively low resolution 3D grid would capture most of the relevant features of players and balls maybe you could get sophisticated with pose estimation and kinematic modelling of player body positions, but a simple CNN classifier to put them into a few minimal pose buckets would work too for player state: standing, jumping, falling, etc. time series gives you velocity and movement.

Once you’ve done that you can model goals and produce some kind of similarity metric which allows you to assign difficulty scores.

But I fully expect that’s what closed source xG models are doing.

TBH I wouldn’t be surprised if you could skip all of that these days and feed the video into a video transformer model similar to ChatGPT and get a direct prediction out. Next token prediction is what video generation models do, run a bunch of next frames predictions after training on all the football footage you can find and assess how many have the ball in the goal.

 

[shmmeee]

Football is absolutely not infinite, there’s a limited number of players in a limited space. The barrier to modelling isn’t resolution, we can accurately model the weather after all.

We can absolutely build a predictive model using some kind of tokenisation and traditional feature extraction from current video segmentation and labelling techniques. Even a relatively high resolution 3D grid would capture most of the relevant features of players and balls maybe you could get sophisticated with pose estimation and kinematic modelling of player body positions, but a simple CNN classifier to put them into a few minimal pose buckets would work too for player state: standing, jumping, falling, etc. time series gives you velocity and movement.

Once you’ve done that you can model goals and produce some kind of similarity metric which allows you to assign difficulty scores.

But I fully expect that’s what closed source xG models are doing.

Obviously to be perfect you want weather and wind models and all that bullshit but you’re probably looking at <1% accuracy gains for huge effort at that point.

 

[fatso]

Sheffield Wednesday never win at our place.

.

 

[skybluecam]

Football is absolutely not infinite, there’s a limited number of players in a limited space. The barrier to modelling isn’t resolution, we can accurately model the weather after all.

We can absolutely build a predictive model using some kind of tokenisation and traditional feature extraction from current video segmentation and labelling techniques. Even a relatively low resolution 3D grid would capture most of the relevant features of players and balls maybe you could get sophisticated with pose estimation and kinematic modelling of player body positions, but a simple CNN classifier to put them into a few minimal pose buckets would work too for player state: standing, jumping, falling, etc. time series gives you velocity and movement.

Once you’ve done that you can model goals and produce some kind of similarity metric which allows you to assign difficulty scores.

But I fully expect that’s what closed source xG models are doing.

TBH I wouldn’t be surprised if you could skip all of that these days and feed the video into a video transformer model similar to ChatGPT and get a direct prediction out. Next token prediction is what video generation models do, run a bunch of next frames predictions after training on all the football footage you can find and assess how many have the ball in the goal.

Football is infinite in the sense that when a player has possession there are an infinite number of things they could do with it; dribble in any direction for any distance; pass any distance, any length, with any type of height on the ball; shoot at any area of the goal with any power and curl etc. On any given move in chess you have a finite number of pieces you could move in a finite number of ways.

xG is only capturing shots that actually happened - doesn't capture anything else. That's it's limitation.

 

[Tile Hill Phil]

I think he trusted the majority of Tuesdays team to put in a performance. Think a few were picked on character

If you look at the goalkeeper and defence it’s more or less the same as last season in the game against Blackburn. I think he’s picking the players he can trust more than anything

 

[Tile Hill Phil]

Sheffield Wednesday never win at our place.

Why did you say that that’s fuckef it now

 

[Sky Blue Pete]

Football is absolutely not infinite, there’s a limited number of players in a limited space. The barrier to modelling isn’t resolution, we can accurately model the weather after all.

We can absolutely build a predictive model using some kind of tokenisation and traditional feature extraction from current video segmentation and labelling techniques. Even a relatively low resolution 3D grid would capture most of the relevant features of players and balls maybe you could get sophisticated with pose estimation and kinematic modelling of player body positions, but a simple CNN classifier to put them into a few minimal pose buckets would work too for player state: standing, jumping, falling, etc. time series gives you velocity and movement.

Once you’ve done that you can model goals and produce some kind of similarity metric which allows you to assign difficulty scores.

But I fully expect that’s what closed source xG models are doing.

TBH I wouldn’t be surprised if you could skip all of that these days and feed the video into a video transformer model similar to ChatGPT and get a direct prediction out. Next token prediction is what video generation models do, run a bunch of next frames predictions after training on all the football footage you can find and assess how many have the ball in the goal.

Was watching something or reading something and the figures of variables after about 6 moves was in the billions unless I was mishearing

 

[baldy]

Was watching something or reading something and the figures of variables after about 6 moves was in the billions unless I was mishearing

Give it a rest with all this chess moves talk ffs - I don't wanna fall asleep just yet

 

[quinn1971]

so if robins picks the same team Saturday and guessing he’ll tell them to just go out and do the same again, and we don’t,robins fault or the players ?

 

[lord_garrincha]

Give it a rest with all this chess moves talk ffs - I don't wanna fall asleep just yet

Well, if that's the end result... there are even more possible variations of chess moves than there are atoms in the observable universe.

 

[wingy]

Give it a rest with all this chess moves talk ffs - I don't wanna fall asleep just yet

Sorry I find it fascinating, well finally crack it when we've already made it with Mr at the helm and in the prem, hopefully, but not definitely!!

 

[wingy]

Well, if that's the end result... there are even more possible variations of chess moves than there are atoms in the observable universe.

It's enough to drive a machine to destruction if that is the case!

 

[Johhny Blue]

I think he’ll probably go the same team with EMC for Saka. Probably needs to make subs earlier than Tuesday though.
[/QUOTE]
Is Saturday afternoon “earlier than Tuesday”