B Team Thread 22/23 Season

[Worthy4England]

The pay walled one, I can't get to. The other two have a long way to go to convince me of the contention being made. The iospress one is using whoscored ratings...

What is wrong with whoscored?

Nothing, particularly, great for forum and pub debates. But their ratings aren't based on showing or measuring peak performance, yet all the polynomials, bivariate and all the rest of the pseudo babble is based just on those ratings and an age, when the researcher can't even explain how the rating is reached....feels like crap in, pseudo-scientific crap out!

[BWFC_Insane]

The pay walled one, I can't get to. The other two have a long way to go to convince me of the contention being made. The iospress one is using whoscored ratings...

What is wrong with whoscored?

Nothing, particularly, great for forum and pub debates. But their ratings aren't based on showing or measuring peak performance, yet all the polynomials, bivariate and all the rest of the pseudo babble is based just on those ratings and an age, when the researcher can't even explain how the rating is reached....feels like crap in, pseudo-scientific crap out!

Whoscored use opta data and their own algorithm to construct their ratings. Given there is no direct measure of performance I’d think using opta data and some statistical reconstruction across 200 data points isn’t a bad way to build a statistical model to analyse.

The other study uses opta data directly and builds into their own model rather than using whoscoreds.

Not sure I really see the issue. If you take exception to a statistical measure of ‘performance’ then fine but I guess all these studies are worthless in that case.

[Worthy4England]

Well, of course it could be definitive, but then I suspect we might have to change some tunes..that Pub player Santos was our second highest rated last season. Morley was 5th...just to pick a couple out, almost at random The top 5 this season are JDC, Dion, Demps, Vic and Iredale...so don't be telling us they're shit!

[Dave Sutton's barnet]

FWIW (which may be nothing) – IMO WhoScored's very useful for quantitative analysis, but I don't trust it for qualitative. In other words - useful when it's counting things, but its 'player ratings' are questionable.

[BWFC_Insane]

Well, of course it could be definitive, but then I suspect we might have to change some tunes..that Pub player Santos was our second highest rated last season. Morley was 5th...just to pick a couple out, almost at random The top 5 this season are JDC, Dion, Demps, Vic and Iredale...so don't be telling us they're shit!

I mean the top 5 seem about right to me….

But there is a difference between a statistical measure of performance based on actions performed in a game and a subjective one.

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

However for the purposes of these sorts of analyses based on large data samples you have to use some sort of proxy as a performance measure and so long as you use one consistently, and one based on the opta data I doubt it matters too much since you are comparing the same performance variables across ages. I guess the only way in which you might argue is if you say the opta data in totality is not a measure of performance at all and at which point yes I’d agree you have an issue. And for what it’s worth I don’t think you can fully assess performance using data but I suspect for such studies it’s a pretty good proxy.

[Dave Sutton's barnet]

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

This is a very good point. For instance, it always felt to me like Radhi Jaidi was good for a goal a game; sometimes it'd be for us, but more often it was one for them from his defensive error.

Guess it all depends on the weighting but then that's different per position. As you note, a goalkeeper making one howler per game is far worse than a striker doing the same.

Am aware, BTW, that I sometimes quote xG and that's also a qual measure. But I'm much happier with it (and I trust Opta) than WhoScored's algorithm; the site feels to me a bit too much of an attempt to turn football into a management sim. I guess that's the lingua franca these days – see also JDC talking about dressing-room banter over ratings on the EA Sports FC 24 game (formerly FIFA) – but I don't trust it... just a preference.

[Worthy4England]

FWIW (which may be nothing) – IMO WhoScored's very useful for quantitative analysis, but I don't trust it for qualitative. In other words - useful when it's counting things, but its 'player ratings' are questionable.

That's where I'm at with it. It's not collecting data to prove "peak performance"

[Dave Sutton's barnet]

FWIW (which may be nothing) – IMO WhoScored's very useful for quantitative analysis, but I don't trust it for qualitative. In other words - useful when it's counting things, but its 'player ratings' are questionable.

That's where I'm at with it. It's not collecting data to prove "peak performance"

Must admit I haven't read the links BWFCi provided so I'm just generically chirping on the sidelines

As I do

[Worthy4England]

Well, of course it could be definitive, but then I suspect we might have to change some tunes..that Pub player Santos was our second highest rated last season. Morley was 5th...just to pick a couple out, almost at random The top 5 this season are JDC, Dion, Demps, Vic and Iredale...so don't be telling us they're shit!

I mean the top 5 seem about right to me….

But there is a difference between a statistical measure of performance based on actions performed in a game and a subjective one.

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

However for the purposes of these sorts of analyses based on large data samples you have to use some sort of proxy as a performance measure and so long as you use one consistently, and one based on the opta data I doubt it matters too much since you are comparing the same performance variables across ages. I guess the only way in which you might argue is if you say the opta data in totality is not a measure of performance at all and at which point yes I’d agree you have an issue. And for what it’s worth I don’t think you can fully assess performance using data but I suspect for such studies it’s a pretty good proxy.

If the peak age was part of my £2m purchasing decision, I wouldn't be relying on either of those analysis. That is all.

[Prufrock]

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

This is a very good point. For instance, it always felt to me like Radhi Jaidi was good for a goal a game; sometimes it'd be for us, but more often it was one for them from his defensive error.

Guess it all depends on the weighting but then that's different per position. As you note, a goalkeeper making one howler per game is far worse than a striker doing the same.

Am aware, BTW, that I sometimes quote xG and that's also a qual measure. But I'm much happier with it (and I trust Opta) than WhoScored's algorithm; the site feels to me a bit too much of an attempt to turn football into a management sim. I guess that's the lingua franca these days – see also JDC talking about dressing-room banter over ratings on the EA Sports FC 24 game (formerly FIFA) – but I don't trust it... just a preference.

I agree with all that but it supports Worthy's point rather than Insano's [he chirps].

If your subjective analysis is more useful than your objective analysis then it shows your objective analysis isn't very good (wrong, or more likely here the wrong proxy/an incomplete picture).

So it doesn't really take your anywhere if you need the subjective to to fill in because that's where people are going to disagree.

[BWFC_Insane]

Well, of course it could be definitive, but then I suspect we might have to change some tunes..that Pub player Santos was our second highest rated last season. Morley was 5th...just to pick a couple out, almost at random The top 5 this season are JDC, Dion, Demps, Vic and Iredale...so don't be telling us they're shit!

I mean the top 5 seem about right to me….

But there is a difference between a statistical measure of performance based on actions performed in a game and a subjective one.

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

However for the purposes of these sorts of analyses based on large data samples you have to use some sort of proxy as a performance measure and so long as you use one consistently, and one based on the opta data I doubt it matters too much since you are comparing the same performance variables across ages. I guess the only way in which you might argue is if you say the opta data in totality is not a measure of performance at all and at which point yes I’d agree you have an issue. And for what it’s worth I don’t think you can fully assess performance using data but I suspect for such studies it’s a pretty good proxy.

If the peak age was part of my £2m purchasing decision, I wouldn't be relying on either of those analysis. That is all.

No but the point I’m making is that this notion that players now play till they are 38 meaning they don’t need to breakthrough till they are 25 or whatever is wrong.

There are a number of studies that show player level ultimately is determined much much earlier in most cases and whilst there are outliers you aren’t likely to see players in their twenties who haven’t been judged good enough to play league football suddenly breakthrough.

Statistically you need to be intercepting players much much earlier in the development process.

[BWFC_Insane]

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

This is a very good point. For instance, it always felt to me like Radhi Jaidi was good for a goal a game; sometimes it'd be for us, but more often it was one for them from his defensive error.

Guess it all depends on the weighting but then that's different per position. As you note, a goalkeeper making one howler per game is far worse than a striker doing the same.

Am aware, BTW, that I sometimes quote xG and that's also a qual measure. But I'm much happier with it (and I trust Opta) than WhoScored's algorithm; the site feels to me a bit too much of an attempt to turn football into a management sim. I guess that's the lingua franca these days – see also JDC talking about dressing-room banter over ratings on the EA Sports FC 24 game (formerly FIFA) – but I don't trust it... just a preference.

I agree with all that but it supports Worthy's point rather than Insano's [he chirps].

If your subjective analysis is more useful than your objective analysis then it shows your objective analysis isn't very good (wrong, or more likely here the wrong proxy/an incomplete picture).

So it doesn't really take your anywhere if you need the subjective to to fill in because that's where people are going to disagree.

Not really because you are comparing a lot of data points objectively across a large data pool.

There will of course be outliers where their subjective performance way exceeds the stats measured by opta of course. But that’s simply a statistical model, you will have outliers but that doesn’t invalidate it.

Unless as I say you believe that objective measures of performance are invalid and we all know modern football disagrees with that.

[Abdoulaye's Twin]

Most players don't breakthrough after 25 because 99.9%* of players don't get the chance because they've long been discarded before getting to 25.

* I made that number up but am confident it isn't a million miles away...

[Prufrock]

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

This is a very good point. For instance, it always felt to me like Radhi Jaidi was good for a goal a game; sometimes it'd be for us, but more often it was one for them from his defensive error.

Guess it all depends on the weighting but then that's different per position. As you note, a goalkeeper making one howler per game is far worse than a striker doing the same.

Am aware, BTW, that I sometimes quote xG and that's also a qual measure. But I'm much happier with it (and I trust Opta) than WhoScored's algorithm; the site feels to me a bit too much of an attempt to turn football into a management sim. I guess that's the lingua franca these days – see also JDC talking about dressing-room banter over ratings on the EA Sports FC 24 game (formerly FIFA) – but I don't trust it... just a preference.

I agree with all that but it supports Worthy's point rather than Insano's [he chirps].

If your subjective analysis is more useful than your objective analysis then it shows your objective analysis isn't very good (wrong, or more likely here the wrong proxy/an incomplete picture).

So it doesn't really take your anywhere if you need the subjective to to fill in because that's where people are going to disagree.

Not really because you are comparing a lot of data points objectively across a large data pool.

There will of course be outliers where their subjective performance way exceeds the stats measured by opta of course. But that’s simply a statistical model, you will have outliers but that doesn’t invalidate it.

Unless as I say you believe that objective measures of performance are invalid and we all know modern football disagrees with that.

It's not that they are invalid, they are just only a partial picture.

They're used as filters or ranges. You don't have a model that you put a load of objective stats into that coughs out a reading that ranks a player.

So somebody who is clearly outstanding from the pack on a number of particular metrics, that's useful and persuasive. But you can't bung them in and say oh well 23rd out of 500 is clearly better than 38th out of 500 because look we did the objective data.

[Dave Sutton's barnet]

If I were a teenage player (or more likely his father) I'm not sure I'd be sprinting towards the club that eviscerated its academy and downgraded to a Category Three. These categories are decided precisely by how much money is spent on them (coaches etc). At that level, we are waaaaaaay behind others.

The B team will help pick up some of the later/older strays, but it will be a while before we're attracting the brightest and best teenagers. Unless we are absolutely sure of the potential of someone mysteriously discarded by the four/five/six bigger clubs within an hour's drive.

We've got a lot of ground to make up. But we have to start somewhere.

[Worthy4England]

For example a keeper could make a string of superb saves every game but also every game make a howler, statistically they might look good based on their overall performance but subjectively you might take a different view.

This is a very good point. For instance, it always felt to me like Radhi Jaidi was good for a goal a game; sometimes it'd be for us, but more often it was one for them from his defensive error.

Guess it all depends on the weighting but then that's different per position. As you note, a goalkeeper making one howler per game is far worse than a striker doing the same.

Am aware, BTW, that I sometimes quote xG and that's also a qual measure. But I'm much happier with it (and I trust Opta) than WhoScored's algorithm; the site feels to me a bit too much of an attempt to turn football into a management sim. I guess that's the lingua franca these days – see also JDC talking about dressing-room banter over ratings on the EA Sports FC 24 game (formerly FIFA) – but I don't trust it... just a preference.

I agree with all that but it supports Worthy's point rather than Insano's [he chirps].

If your subjective analysis is more useful than your objective analysis then it shows your objective analysis isn't very good (wrong, or more likely here the wrong proxy/an incomplete picture).

So it doesn't really take your anywhere if you need the subjective to to fill in because that's where people are going to disagree.

Not really because you are comparing a lot of data points objectively across a large data pool.

There will of course be outliers where their subjective performance way exceeds the stats measured by opta of course. But that’s simply a statistical model, you will have outliers but that doesn’t invalidate it.

Unless as I say you believe that objective measures of performance are invalid and we all know modern football disagrees with that.

It's not that they are invalid, they are just only a partial picture.

They're used as filters or ranges. You don't have a model that you put a load of objective stats into that coughs out a reading that ranks a player.

So somebody who is clearly outstanding from the pack on a number of particular metrics, that's useful and persuasive. But you can't bung them in and say oh well 23rd out of 500 is clearly better than 38th out of 500 because look we did the objective data.

Eggsackerley. And if the stats aren't measuring "peak performance" just some data points that may or may not be related to the data set for peak performance, then you need to disaggregate to only those that do. Any extrapolation or analysis is flawed before it's started. Compound pseudo-science as it were...

[Dave Sutton's barnet]

Further to my earlier post, I wondered which teams have which category academies (this season). Couldn't find a definitive list but a bit of digging turned up these names.

CATEGORY ONE - mostley Prem or Prem-adjacent, except Reading.
Arsenal
Aston Villa
Blackburn Rovers
Brighton
Chelsea
Crystal Palace
Derby County
Everton
Fulham
Leeds United
Leicester City
Liverpool
Man City
Man Utd
Middlesbrough
Newcastle United
Norwich
Nott'm Forest
Reading
Southampton
Stoke City
Sunderland
Spurs
West Brom
West Ham United
Wolves

CATEGORY TWO - largely Champo, but note Burnley, Crewe, Fleetwood (!) and Wigan.
Barnsley
Birmingham
Bournemouth
Bristol City
Burnley
Cardiff
Charlton
Colchester
Coventry
Crewe
Fleetwood
Hull
Ipswich
Millwall
Peterborough
QPR
Sheff U
Sheff W
Swansea
Watford
Wigan

CATEGORY THREE - frankly, largely fourth-tier and former fiscal failures
Accrington
AFC Wimbledon
Blackpool
Bolton
Derby
Doncaster
Mansfield
Morecambe
Preston
Salford City
Stockport
Sutton United
Walsall

Interesting to note that Preston are in that same group as us. In Nov 2022, PNE chairman Craig Hemmings stated that the club's Category 3 Academy costs £1.2million per year to run and that moving to Category 2 would increase costs to around £2million per year. DoF Peter Ridsdale added that Category 2 status requires an indoor football facility which would cost between £4.5million to £5million to build. Didn't we have one of those, or did it blow down in a storm?

Another thing I discovered: only two clubs that have attained Cat One status since its 2012 introduction have since downgraded. That'll be Swansea and us.

Oh and: Brentford, having abandoned academies for the B-team model, last season joined back in again at Category Four level (yep, there's worse than us). Huddersfield, another B-team model, downgraded in 2017 from Cat Two to Four but this summer applied for Cat Three. And let's not forget that they already have one of the country's most productive acdemies.... https://trainingground.guru/articles/hu ... y-rankings

[Mar]

I do actually like the fact that we've got a b team, regardless of whether it's successful. It's something that pops up on my twitter feed that's semi interesting at least.

I think it'd be nice for the club to turn it into broadcastable events although I suspect the general interest levels won't warrant that, but it's a tried and tested model and other clubs are doing it so why not us.

The b team approach always seemed a strange one. I can see the thinking in lieu of spending a tonne of money on u21 resources at a time when we didn't have money. Those cat A facilities don't come cheap and when you're losing players like Holding for a fraction of what he's worth it's clearly not worth it.

We're in a tough spot catchment wise. Loads of academies nearby so it's not like we're going to get blessed with an abundance of talent that's not already been touted elsewhere. Were not first choice on anyone's list but we could be given the right strategy.

I'm fairly confident the b team barely getting a sniff during the FLT matches isn't likely to help. Hopefully next match gives a more visible pathway to the first team as ideally we could do with a marker that says if you're good enough you'll break through ala Thomason style.

[BWFC_Insane]

There is a big gap between Cat A academy and the model we have.

My issue is the age and profile of players we have in the B team. It needs to be much younger I think if we are to really turn some into genuine first team prospects.

[Prufrock]

Geography clearly a big part. Three of the big outliers there are us, Burnley and Preston. Wigan to an extent buck the trend, but then it's easier to have an expensive academy when you're not paying for anything.

I can see Bwfc-I's point to a degree, but for me it's not just age, it's age and profile. The lads we've signed are not just late teens early 20s, but also haven't come from the biggest academies. They're getting released lower down at an age where they haven't shown much.

But:

- I think at the moment we're basically doing a second string on the cheap. You need some sort for injuries and the odd gamble.
- the current lads are basically just fodder I think. You hope one might come good but you need the infrastructure there for down the line. Matheson is a much better example of what we should be trying for. If he comes in and contributes this season then he probably justifies a lot of the spend just on his own, and you can't sign him if you haven't done the last couple of years.
- so my main point is it's not something you're going to see many results from in 2 or 3 years.
- final point is that the B team doesn't replace anything from the academy, and we're largely in line with other L1 clubs. Particularly given our catchment area you can see why disproportionately focusing on the academy isn't the way to go in a way it is for say Southampton.
- fully agree with Insano's point that we should be after the 14 and 15 year olds released from the big local avenue academies but...I assume we are doing that too! If you sign them it's going to take 4-5 years until you'd start to see them coming through and 4-5 years ago we were... well I don't need to remind anyone of the basket case we were. Even three years ago signing a full team from scratch in L2 with a promising but unknown manager we wouldn't have been that attractive. You'd hope we've jumped massively up the queue recently and Bradley and Traff are surely the poster boys for our academy Prospectus. But it'll be years before that pays off.