Smart recruitment...

[alcoblade]

There's no real 'model' that only finds players that only financially benefits its owner.

United got scammed and instead of accepting it and learning, have retreated straight back into their comfort zone

We bought players from plenty of other places too, not just the owner. The model was just shit.

I've no idea what we've reverted to, and suspect you don't either.

But there's a tendancy for people to talk about AI as a simple modern step forward, and refusing to take that step as being simply backward.

United will use data like all other clubs will, it's just about extent. But the successful models that people dream of aren't a simple step forward, they cost an absolute fortune in data acquisition, and are a pipe dream for most clubs / owners.

 

[Ron_Justice]

I'm probably being nieve here, but when 'international' scouting discussions surface I always have the same thought:

As an example, let's use France. Set up a scouting office in Paris. Employ 3 full time scouts on €50K a year, plus a couple of admin staff and office rental costs and rates - €250/300K a year? Now roll that out across the capitals of Spain, Belgium, Germany and the Netherlands for example.

€1.25/1.5M a year, you have your fingers in loads of pies.

It's probably not that simple though.

 

[Carlton Blade]

We bought players from plenty of other places too, not just the owner. The model was just shit.

I've no idea what we've reverted to, and suspect you don't either.

But there's a tendancy for people to talk about AI as a simple modern step forward, and refusing to take that step as being simply backward.

United will use data like all other clubs will, it's just about extent. But the successful models that people dream of aren't a simple step forward, they cost an absolute fortune in data acquisition, and are a pipe dream for most clubs / owners.

Every 'AI' player was from Bulgaria, other than Jeff who was apparently a vanity project from Russo.

United under Wilder sign the same player type from the same leagues. That's not using data to inform your decisions. I've given examples of how other similar sized and smaller clubs use data - targeted plans, offering options to do further work on.

Where have I ever said that a data model would fix anything on it's own? Using it as part of your toolkit, informing where to put your scouting resources, understanding player traits and player development from certain leagues is all key. But United had one foolhardy, quick fix contract with a conman and gave up

 

[alcoblade]

Every 'AI' player was from Bulgaria, other than Jeff who was apparently a vanity project from Russo.

United under Wilder sign the same player type from the same leagues. That's not using data to inform your decisions. I've given examples of how other similar sized and smaller clubs use data - targeted plans, offering options to do further work on.

Where have I ever said that a data model would fix anything on it's own? Using it as part of your toolkit, informing where to put your scouting resources, understanding player traits and player development from certain leagues is all key. But United had one foolhardy, quick fix contract with a conman and gave up

So you don't think the model was used to assess players like Tom Cannon, for example?

You know to what extent ai was used to make all our summer signings?

You know we've just given up?

Where do you get your insider information from?

As for "Where have I ever said that a data model would fix anything on it's own?" - stop it with the straw men. I didn't say you said that, or anything like it.

 

[10p]

Of all the ones to miss!!

Is that irony? You miss the one who is often missing, even when he’s on the pitch.

 

[10p]

Perhaps the first 2 questions to ask before a signing:
1. Is he fit or capable of getting fit.
2. Is this yet another last minute signing (I.e. have all the players we initially identified signed elsewhere and we are left selecting from the poorer players that no one else wanted).

 

[Carlton Blade]

So you don't think the model was used to assess players like Tom Cannon, for example?

You know to what extent ai was used to make all our summer signings?

You know we've just given up?

Where do you get your insider information from?

As for "Where have I ever said that a data model would fix anything on it's own?" - stop it with the straw men. I didn't say you said that, or anything like it.

All underlying data showed Cannon as a bottom 10% CF against his peers. So no, not even alittle bit. Wilder wanted him in the summer, previous owner pulled the plug.

Fair, but "But the successful models that people dream of aren't a simple step forward, they cost an absolute fortune in data acquisition, and are a pipe dream for most clubs / owners" is just not true. Asking a data company for access to specific information in set parameters is literally however clibthat doesn't hold proprietary data works. It's not a pipe dreamat all. Talented amateurs can do it

 

[alcoblade]

All underlying data showed Cannon as a bottom 10% CF against his peers. So no, not even alittle bit. Wilder wanted him in the summer, previous owner pulled the plug.

Fair, but "But the successful models that people dream of aren't a simple step forward, they cost an absolute fortune in data acquisition, and are a pipe dream for most clubs / owners" is just not true. Asking a data company for access to specific information in set parameters is literally however clibthat doesn't hold proprietary data works. It's not a pipe dreamat all. Talented amateurs can do it

Regarding your first point (Cannon), you're simply wrong.

Regarding the second point - do you honestly believe that we aren't paying for, and using still / data already?

 

[Nick Jansky]

Every 'AI' player was from Bulgaria, other than Jeff who was apparently a vanity project from Russo.

United under Wilder sign the same player type from the same leagues. That's not using data to inform your decisions. I've given examples of how other similar sized and smaller clubs use data - targeted plans, offering options to do further work on.

Where have I ever said that a data model would fix anything on it's own? Using it as part of your toolkit, informing where to put your scouting resources, understanding player traits and player development from certain leagues is all key. But United had one foolhardy, quick fix contract with a conman and gave up

Contrast our AI recruitment with Hearts in the less than a year that Tony Bloom has been fully on board. Claudio Braga, a Portuguese player who was playing in the Norwegian 2nd Division not so long ago. He would be great for the left hand side of our attack. These players are out there, the smart clubs find them.

 

[Carlton Blade]

Regarding your first point (Cannon), you're simply wrong.

Regarding the second point - do you honestly believe that we aren't paying for, and using still / data already?

How can we be paying for data? It's a pipe dream . United use data on opposition scouting pre match and to benefit their current players through performance analysis

Cannon was one of the worst players in open play in the Championship by every available data metric, as has been confirmed by his performances. If United used data to sign him, they are beyond stupid or the data they used was 'goals in a game v Portsmouth'.

He started at a big club, dropped down league, had a reasonable goal record in a small sample. Just like Brewster

 

[alcoblade]

How can we be paying for data? It's a pipe dream . United use data on opposition scouting pre match and to benefit their current players through performance analysis

Cannon was one of the worst players in the open play in the Championship by every available data metric, as has been confirmed by his performances. If United used ddata toa sign him, they are beyond stupid or the data they used was 'goals in a game v Portsmouth'.

He started at a big club, dropped down league, had a reasonable goal record in a small sample. Just like Brewster

I think you're being daft now. They wouldn't use ai to analyse the most basic of data that you and other internet bulletin board sage's can spot - everyone can do that. It's used for things that aren't obvious to the average human using basic level data.

It wasn't used to identify Cannon obviously. But it was used to help value the player which drove us to ridiculous valuations.

Of course, it failed miserably.

 

[Carlton Blade]

I think you're being daft now. They wouldn't use ai to analyse the most basic of data that you and other internet bulletin board sage's can spot - everyone can do that. It's used for things that aren't obvious to the average human using basic level data.

It wasn't used to identify Cannon obviously. But it was used to help value the player which drove us to ridiculous valuations.

Of course, it failed miserably.

There is no data available, in any sphere, and trust me I have both looked and asked those who have better access that identified Cannon as even an average Championship player.

United fucked up and used 'data' as the scapegoat. You cannot spend your whole career talking about leaders, character, 'lads' and every other intangible as being paramount and then hide behind 'data' and expect to be taken seriously. He's young, born in England, started at a big club, was very good at academy level, scored a few goals in a small sample. Apart from the fact Hoyland was at Everton with him to offer feedback, it's the exact Brewster situation.

United looked 'traits', leaning heavily on personality over athleticism. It's served Wilder well but it has limitations

 

[alcoblade]

There is no data available, in any sphere, and trust me I have both looked and asked those who have better access that identified Cannon as even an average Championship player.

Okay, if you're sure

 

[Yogi]

I think Wilder was pretty transparent regards what he liked about Cannon from a data point of view when he signed his time at Stoke showed he was good playing on the shoulder of the defence (something Wilder repeated almost constantly about him when he signed), he’s a clean finisher something wilder has valued in all his strikers outside Mcburnie and he won a decent amount of aerial duels something Wilder has looked for consistently in his number 9’s.

Added onto that as Carlton has said he fit the profile of what Wilder looks for in a players character came through at a top academy dropped down to try and get regular football and we had personal contacts Wilder trusted who could tell us what type of person he was. That ticks all the boxes really and that will be why we pursued him as hard as we did.

That said you can get players of a similar profile but without the theoretical upside for much less than we paid.

The issue in paying a high fee for a toolsy prospect is that in this era United have consistently been bad at developing strikers within the first team environment. We’ve done very very well with finished articles like Mcburnie, Bamford, Sharp, Didzy and BBD for a spell. We’ve consistently failed to develop toolsy prospects either home grown or purchased in. Osula, Jebbo, Brewster, Cannon. Let’s hope One bucks that trend but our track record is against him.

The outlier in that is Ndiaye but I’d argue it was clear he was a truly exceptional talent as soon as he hit the first team and much of the technical development to make him that player had been done in the academy set up.

 

[AdamSufc1997]

We’ll never gamble on players abroad with the structure and management in place now . It’ll always be British players only. They fucked up using AI do will go back to the safe place again. Things won’t change here unless we gamble more

 

[BangorBlade]

We’ll never gamble on players abroad with the structure and management in place now . It’ll always be British players only. They fucked up using AI do will go back to the safe place again. Things won’t change here unless we gamble more

Don't you think it's about time SUFC set up some sort of scouting structure abroad, look at the foreigners the likes of Brighton, Bournemouth & Brentford have playing for them all 3 aren't massive clubs but have some outstanding foreigners talent.

 

[MayorOJC]

We’ll never gamble on players abroad with the structure and management in place now . It’ll always be British players only. They fucked up using AI do will go back to the safe place again. Things won’t change here unless we gamble more

The Brewster and Cannon signings have done much more damage than the AI flops, so I hope that isn't the lesson learned.

 

[Houston_Blade]

Training Ground Guru | Lee Dykes: The seven stages of Brentford's recruitment

Training Ground Guru | This is THE portal for staff working in – or aspiring to work in – professional football.

trainingground.guru

Interesting that the German scout for Brentford when this article was written was Simon Tracey! Bladey Blade link and looks outside the British Isles, gerrimin

Also, can you imagine Wilder’s face if a DoF came in and started telling him there were 16 positions on a football pitch?! He’d set Ollie Arblaster on him.

 

[AWetFlannel]

Computers don't decide who to sign, they don't authorise payments, they don't investigate persinalities etc

Used properly, data will help you identify a group of players who might solve your specific issues. After that, it's people doing the work.

What United have done, is blamed ''AI' and used it as a excuse to refuse to do anything other than 'go with their gut'

Agree with what you say about data being used to aid recruitment, it's obviously not the whole process.

But what evidence is there that United are abandoning everything other than gut instinct? Wilder would talk about consulting data long before last season.

 

[CityInPeace]

He also played for their sister Club Lorient.

At least one other club in their model, I forget how big it is.

Hibs were in and around it, Auckland City possibly.

Edit..just checked.

 

[MayorOJC]

Regarding your first point (Cannon), you're simply wrong.

Regarding the second point - do you honestly believe that we aren't paying for, and using still / data already?

Got to back up Carlton on the Cannon thing. His numbers were (and still are) shocking on a number of fronts. It was very clear that he's just a poacher that offers very little during the game other than the occassional decent finish at a good penalty record. Not a chance was that decision made based on underlying data.

 

[alcoblade]

Got to back up Carlton on the Cannon thing. His numbers were (and still are) shocking on a number of fronts. It was very clear that he's just a poacher that offers very little during the game other than the occassional decent finish at a good penalty record. Not a chance was that decision made based on underlying data.

He wasn't discovered using ai, but the excitement and value around him was inflated by it.

And as I said, if ai drew conclusions from the data that were obvious (in the way you describe), there'd be no point in it.

 

[alcoblade]

Agree with what you say about data being used to aid recruitment, it's obviously not the whole process.

But what evidence is there that United are abandoning everything other than gut instinct? Wilder would talk about consulting data long before last season.

Haven't you read it on here? Wilder now decides over a pint and a packet of crisps with his mates.

 

[AWetFlannel]

Got to back up Carlton on the Cannon thing. His numbers were (and still are) shocking on a number of fronts. It was very clear that he's just a poacher that offers very little during the game other than the occassional decent finish at a good penalty record. Not a chance was that decision made based on underlying data.

I assume his numbers were pretty good for goals.

We all agree that it was a terrible signing, and that United haven't got an elite transfer strategy. But using data is far more complicated than is being made out by many.

"Are the numbers good" is too simplistic, "do the numbers give us what we need specifically for our style of play, given our current crop in the same position etc." is more like it.

 

[jt64]

I'm probably being nieve here, but when 'international' scouting discussions surface I always have the same thought:

As an example, let's use France. Set up a scouting office in Paris. Employ 3 full time scouts on €50K a year, plus a couple of admin staff and office rental costs and rates - €250/300K a year? Now roll that out across the capitals of Spain, Belgium, Germany and the Netherlands for example.

€1.25/1.5M a year, you have your fingers in loads of pies.

It's probably not that simple though.

What on earth would having a physical location(s) across the continent actually do? Other than haemorrhage money? It's 2026, any game of any note is streamed. Christ, I can see the goals of my local village team ten divisions below United within hours of the game finishing. It doesn't need people all over the place, it just needs people that know what they are looking for and have the ability to sift through lots of data to find it. Once you've done that, THEN if you, really, really need the luddite approach of watching an actual game in person (where there's an enormous chance you get the wrong impression due to sample size), you do it.

 

[MayorOJC]

He wasn't discovered using ai, but the excitement and value around him was inflated by it.

And as I said, if ai drew conclusions from the data that were obvious (in the way you describe), there's be no point in it.

I'm not sure how the data analysis underlying any AI model would increase excitement or value around Tom Cannon. When it's used properly, it's to uncover attributes that exist, but aren't being reflected in traditional statistics. Basic example would be a player than creates a lot of high value chances, but has a low number of assists.

Cannon's underlying stats were shockingly poor when it came to basically anything other than goals scored. If anything, data analysis would flag him as one to avoid.

 

[MayorOJC]

I assume his numbers were pretty good for goals.

We all agree that it was a terrible signing, and that United haven't got an elite transfer strategy. But using data is far more complicated than is being made out by many.

"Are the numbers good" is too simplistic, "do the numbers give us what we need specifically for our style of play, given our current crop in the same position etc." is more like it.

I agree and I think we're saying the same thing.

 

[alcoblade]

I'm not sure how the data analysis underlying any AI model would increase excitement or value around Tom Cannon. When it's used properly, it's to uncover attributes that exist, but aren't being reflected in traditional statistics. Basic example would be a player than creates a lot of high value chances, but has a low number of assists.

Cannon's underlying stats were shockingly poor when it came to basically anything other than goals scored. If anything, data analysis would flag him as one to avoid.

Well there's a whole raft of examples of where our ai model was shit. Perhaps this was just another one.

Anyway, I'm not shooting from the hip here. This came from very close to the horses mouth.

 

[MayorOJC]

Well there's a whole raft of examples of where our ai model was shit. Perhaps this was just another one.

Anyway, I'm not shooting from the hip here. This came from very close to the horses mouth.

Very true. Any model that spits out Cáceres as a £1m player could be outputting any shite.

I'm 99% sure this is a case of Wilder just picking the British / Irish player highest in the Championship top scorers chart though, which is how he usually picks his CF targets when he has money (see McBurnie, Brewster, Robinson, Watkins, etc.) As said earlier, he wanted Cannon in summer 2024.

 

[alcoblade]

Very true. Any model that spits out Cáceres as a £1m player could be outputting any shite.

I'm 99% sure this is a case of Wilder just picking the British / Irish player highest in the Championship top scorers chart though, which is how he usually picks his CF targets when he has money (see McBurnie, Brewster, Robinson, Watkins, etc.) As said earlier, he wanted Cannon in summer 2024.

You're right on the last point. As I said earlier, it's the value the club were prepared to pay that was inflated.