Are AI clubs too fixated on CA/PA?

[laurentius82]

I'll answer my own question: yes. The AI clubs seem to make decisions only based on players' PA and CA, and completely ignore their career history and form. For example, I have a 22-year old striker now in Vanarama National who has scored 20, 39, and 47 league goals in the previous 3 seasons in the Vanarama North. Absolutely nobody has paid interest to him, because he obviously doesn't have a very high PA at all. This feels quite unrealistic, if it happened IRL, I'm pretty sure Leicester would be knocking on my door hoping to catch the next Jamie Vardy. Moreover, it should not be possible to conclude with absolute certainty that a young player with that kind of record doesn't have any potential. 

In contrast, tens of millions will be paid by AI clubs for high-PA 18-year-olds who haven't necessarily played a single senior match. And we humans do it too, of course, but it's an entirely safe bet to take in the game because we know they'll be good if they have high PA (as judged by good scouts).

[cris182]

This is definitely something that needs improving in my opinion, I too have seen players with great goal records left in lower leagues to rot when the AI spend 20m on a brazilian who plays less than 10 games in 4 years then gets released 

[Platinum]

Yeah I've definitely noticed this too. In a previous FM I had an English 17 year old who scored 30 goals in the league , and then the next season he scored 30 goals again however his potential was only 'good premier league player'. His attribute spread obviously made him a beast. I never recieved a bid for him once which was completely unrealistic.

[Orion_]

Another thing worth fixing is AI picking players for the starting team based on position they can (theoretically play) rather than position they are good with. For example I had a striker who was very good at his role. He was banging goals like crazy but he could be also played as a left winger. So every time his national team manager and my ass man wanted to play him as a winger where he played just badly. I assume their thought process was like: Hey it's players with very high rating (a lots of goals) where he should play him? Striker? No he got not that much stars in CA. Maybe winger, our wingers don't have that good games ratings so he might be good to be played there.

This also happens when your Winger/WB/FB can play both sides but are single footed. Even if your tactic don't have inverted wingers/inverted full backs the AI will often try to put them as for example 'regular' winger on the left side if he's right footed just because he has proficiency on that position.

[shirajzl]

That has literally been a problem since the CM days.

[Jops14]

I always thought for transfers they were too fixated on rep, so you could boost a players rep, sell them for millions and then AI goes "oh his CA isnt that good we wont play him"

[Weed07]

Its definately a problem late in the game, i'm about 25 seasons into my save and alot of the big clubs play players way past their prime as its based on their high CA and reputation. For example i noticed Arsenal signed a 38 year old striker for £10.75M and played him in most of their league games getting 4 goals in 28 games. The player is still in the England squad too at 38.

Its happened in all versions of FM and is one of the reasons why the game gets too easy when you manage a big club as you can easily create teams full of wonderkids while the AI teams end up being full of older players.

[Nacaw]

Without knowing exactly how the AI work these things out (because it's not just a decision based on PA/CA, clearly), it's probably the opposite. The AI should be MORE focused on CA and PA in the decisions it makes, and especially helping players reach their PA. 

This isn't for the sake of realism, but because the AI is gimped enough compared to the average FM player, that if it started buying more players based on form, the long term squad building would be even worse than it is right now. 

Also, to all the people saying the AI should do this, how many players do you buy based on form? And follow-up, if you don't buy based on form, why should the AI?

[laurentius82]

6 hours ago, Nacaw said:

Without knowing exactly how the AI work these things out (because it's not just a decision based on PA/CA, clearly), it's probably the opposite. The AI should be MORE focused on CA and PA in the decisions it makes, and especially helping players reach their PA. 

This isn't for the sake of realism, but because the AI is gimped enough compared to the average FM player, that if it started buying more players based on form, the long term squad building would be even worse than it is right now. 

Also, to all the people saying the AI should do this, how many players do you buy based on form? And follow-up, if you don't buy based on form, why should the AI?

No you're exactly right, the problem runs much deeper than just AI not buying players based on form. We don't do it much either, precisely because even if an 18-year-old player bangs 50 goals in league 2 but has 1 star potential according to our top scouts, we KNOW with 100% certainty he will never be good enough to play in the premier league. This is symptomatic of a general problem of how the rigid CA/PA system works. To start fixing the issue, I think PA should i) be A LOT harder to evaluate for both human and AI ii) PA should be much harder to reach. Perhaps even the whole PA system should be replaced by something like an "ability to develop" attribute, removing a fixed ceiling that is very unrealistic especially when it's so often reached by very young players.   

[GreenTriangle]

On 11/05/2021 at 09:59, laurentius82 said:

I'll answer my own question: yes. The AI clubs seem to make decisions only based on players' PA and CA, and completely ignore their career history and form. For example, I have a 22-year old striker now in Vanarama National who has scored 20, 39, and 47 league goals in the previous 3 seasons in the Vanarama North. Absolutely nobody has paid interest to him, because he obviously doesn't have a very high PA at all. This feels quite unrealistic, if it happened IRL, I'm pretty sure Leicester would be knocking on my door hoping to catch the next Jamie Vardy. Moreover, it should not be possible to conclude with absolute certainty that a young player with that kind of record doesn't have any potential.

Clubs probably take into account a player's CA and/or PA ... but they certainly take into account the player's national reputation (which is hidden in the game).

About Vardy : he really had a high CA/PA when Leicester offerred him a contract. And he was not an unknown player. In terms of FM he was "a player received at youth intake" by a Championship club (Sheffield United). The strange thing is Sheffield sent him to the 7th league. Should FM simulate this sending of a young (and high PA player) to the 7th league ? On the other hand, Vardy was the best goal scorer for Halifax (in the 6th league) and the best goal scorer for Fletwood (in the 5th league). Fleetwood just promoted to League Two when Leicester spotted Vardy. In other words, his "national reputation" was not insignificant at all. Even  those few games played at Sheffield United U18 gave him a relatively high "national reputation". And the large number of goals scored and that promotion to League Two kept his reputation at that level. Your player, however, does not have a high CA/PA and, although these goals scored probably raised his reputation, he has not yet reached a high enough level of this reputation.

[BuryBlade]

50 minutes ago, GreenTriangle said:

Clubs probably take into account a player's CA and/or PA ... but they certainly take into account the player's national reputation (which is hidden in the game).

About Vardy : he really had a high CA/PA when Leicester offerred him a contract. And he was not an unknown player. In terms of FM he was "a player received at youth intake" by a Championship club (Sheffield United). The strange thing is Sheffield sent him to the 7th league. Should FM simulate this sending of a young (and high PA player) to the 7th league ? On the other hand, Vardy was the best goal scorer for Halifax (in the 6th league) and the best goal scorer for Fletwood (in the 5th league). Fleetwood just promoted to League Two when Leicester spotted Vardy. In other words, his "national reputation" was not insignificant at all. Even  those few games played at Sheffield United U18 gave him a relatively high "national reputation". And the large number of goals scored and that promotion to League Two kept his reputation at that level. Your player, however, does not have a high CA/PA and, although these goals scored probably raised his reputation, he has not yet reached a high enough level of this reputation.

Sheffield Wednesday btw not Sheffield United.

For anyone who had played OOTP baseball, would the fog of war/20-80 style of assessing players work better? 

[WelshMourinho]

I've always felt like it's less a CA/PA thing and more a reputation thing, which is probably why as other people have said you find them buying some older players later on into the save.

It's probably a balance between the two.

[SCCP1910]

On 11/05/2021 at 03:59, laurentius82 said:

I'll answer my own question: yes. The AI clubs seem to make decisions only based on players' PA and CA, and completely ignore their career history and form. For example, I have a 22-year old striker now in Vanarama National who has scored 20, 39, and 47 league goals in the previous 3 seasons in the Vanarama North. Absolutely nobody has paid interest to him, because he obviously doesn't have a very high PA at all. This feels quite unrealistic, if it happened IRL, I'm pretty sure Leicester would be knocking on my door hoping to catch the next Jamie Vardy. Moreover, it should not be possible to conclude with absolute certainty that a young player with that kind of record doesn't have any potential. 

In contrast, tens of millions will be paid by AI clubs for high-PA 18-year-olds who haven't necessarily played a single senior match. And we humans do it too, of course, but it's an entirely safe bet to take in the game because we know they'll be good if they have high PA (as judged by good scouts).

There should be a way to force us (and the AI) to recruit by stats and data instead of only going by the little stars and arbitrary numbers. I know I could restrict myself, but I'm not really good enough at the game to do that, and it'd be extremely frustrating, knowing that I'd lose out. 

[zyfon5]

9 hours ago, Nacaw said:

Without knowing exactly how the AI work these things out (because it's not just a decision based on PA/CA, clearly), it's probably the opposite. The AI should be MORE focused on CA and PA in the decisions it makes, and especially helping players reach their PA. 

This isn't for the sake of realism, but because the AI is gimped enough compared to the average FM player, that if it started buying more players based on form, the long term squad building would be even worse than it is right now. 

Also, to all the people saying the AI should do this, how many players do you buy based on form? And follow-up, if you don't buy based on form, why should the AI?

The AI will take into account on form when they are buying players. Just look at the amount of complaints that it is very difficult to sell bench players who does not have a lot of game time even if these are good players. So the AI does not only make decisions based on CA/PA. And I agree that if the AI start buying more players based on form, the situation will be even worse than it is right now. 

[Neil Brock]

Lots of things are taken into account, including players form (both recent and over a longer period), their reputation and their CA (which is determined depending on how good their scouts are) and PPA (perceived potential ability - again dependent on the scout's ability).

They'll also consider the value of the player, their wage demands, interesting in joining etc, so there's certainly more to it than just the CA/PA side. 

[laurentius82]

3 hours ago, Neil Brock said:

Lots of things are taken into account, including players form (both recent and over a longer period), their reputation and their CA (which is determined depending on how good their scouts are) and PPA (perceived potential ability - again dependent on the scout's ability).

They'll also consider the value of the player, their wage demands, interesting in joining etc, so there's certainly more to it than just the CA/PA side. 

Of course, but I don't think any amount of form will be enough to compensate for a low CA/PA?  

[laurentius82]

5 hours ago, GreenTriangle said:

About Vardy : he really had a high CA/PA when Leicester offerred him a contract. 

Jamie Vardy didn't have a PA. He had drive, determination, and other abilities and characteristics that allowed him to develop into the player he became. 

[CARRERA]

With knowing now how many parameters lead to a players evaluation for a club, i start thinking the importancy of CA/PA & Age is not valued high enough for the AI. Lets face it, what most human players look at, is the CA, PA and the Age to great success, as we know that young players with high PA will most likely become good players and players with a high CA will most likely perform well. Or am i missing something?

[laurentius82]

33 minutes ago, CARRERA said:

With knowing now how many parameters lead to a players evaluation for a club, i start thinking the importancy of CA/PA & Age is not valued high enough for the AI. Lets face it, what most human players look at, is the CA, PA and the Age to great success, as we know that young players with high PA will most likely become good players and players with a high CA will most likely perform well. Or am i missing something?

Within the game as it is, you're right that focusing on CA/PA/age makes total sense, but that is because experience has no value in FM and PA is set in stone and way too obvious (yes yes bad scouts sometimes it get it a tiny bit wrong but still). 

[CARRERA]

Just now, laurentius82 said:

PA is set in stone and way too obvious

While i generally agree with you, as long as i know PA is not fixed for players when they are younger and can still vary, independed of your scouts judgeing ability. However it somehow its the same as it is with attributes. Some might prefer them being hadden and other may like them to be visible.

[Platinum]

6 hours ago, Neil Brock said:

Lots of things are taken into account, including players form (both recent and over a longer period), their reputation and their CA (which is determined depending on how good their scouts are) and PPA (perceived potential ability - again dependent on the scout's ability).

They'll also consider the value of the player, their wage demands, interesting in joining etc, so there's certainly more to it than just the CA/PA side. 

Thanks that's good to know and I gathered that would be the case. However its not really addressing the core problem people have been talking in this thread which is that the AI often doesn't buy players whose performances exceed what you would generally think their CA would allow. Someone raised a good point asking if the human player does that and I for one do that take form massively into account. 

In fact I often wont buy a high PA player whose average rating for his current club is poor as it could indicate bad hidden attributes (assuming the player has been playing at an appropriate level for his skill).

 

[Neil Brock]

13 hours ago, francis#17 said:

Thanks that's good to know and I gathered that would be the case. However its not really addressing the core problem people have been talking in this thread which is that the AI often doesn't buy players whose performances exceed what you would generally think their CA would allow. Someone raised a good point asking if the human player does that and I for one do that take form massively into account. 

In fact I often wont buy a high PA player whose average rating for his current club is poor as it could indicate bad hidden attributes (assuming the player has been playing at an appropriate level for his skill).

 

As always in situations like this, if you have a save game example of where a player has consistently been playing very well yet is not receiving any interest from 'bigger' clubs, please create a thread in our transfers bugs forum and upload the save so we can take a look.

Thanks. 

[tiotom92]

14 hours ago, CARRERA said:

While i generally agree with you, as long as i know PA is not fixed for players when they are younger and can still vary, independed of your scouts judgeing ability. However it somehow its the same as it is with attributes. Some might prefer them being hadden and other may like them to be visible.

PA is fixed for all players.

[silentwars]

My take is that the attributes in game are too static.

For example:

Paul Mullin (Cambridge) - on the last FM20 database had an attribute finishing of 10 , he has then IRL had a fantastic season, scoring 32 goals as such on the latest version of FM21 his finishing has been increased to 15.

I had a forward at Morecambe (Liam McAlinden) who scored 25 goals from open play, yet at the end of the season his finishing attribute was 9 as it had been at the start of the game. He was let go on a free and nobody picked him up. If this form had been shown IRL his finishing attribute would almost certainly have been increased to at least 12, and on a free would of easily gotten another club based on that form.

I think attributes should be more flexible based on the form a player shows over time as this is how they are judged IRL by the FM researchers. It would also make scouting a more important aspect of the game knowing that attributes are liable to more varied changes, as you'd have to trust their opinion on if those attributes could be sustained over a period of time.

[lex311085]

yes, Patson Daka in my save who Dortmund bought, Pretty average attributes but ended up having a goalscoring rate of about 95% consistently, literally no interest in him from anyone in the 8 or so years he was there. Also a player called Ricky Jade Jones played for Rotherham in my save in the championship, scored about 80 goals in 2 seasons, literally no premier league or even top championship team interest in him at all. 

Just to add to that aswell. I created a striker in the editor and made his attributes so he'd be the best in the world however his form and goalscoring did not reflect that, rarely got into double figures in a season. Yet he has been sold to both Real Madrid and Bayern Munich at different stages in his career for 150million plus obviously just based on his CA. and regularly gets called up to International squads. He's now pushing 30 so his media description is Legendary Striker when he is far from it, whereas Patson Daka only goes down as veteran striker, even though he's won ballon d'ors and scored a crazy amount of goals. 

[GreenTriangle]

1 hour ago, lex311085 said:

having a goalscoring rate of about 95% consistently

95 % of what ?

Quote

1 hour ago, silentwars said:

had an attribute finishing of

Finishing it's just one of those important attributes for a player to score often. It is probably an attribute with a high weight, but the others (including the hidden ones) are very important. A greater variation of these attributes could make scouting more inaccurate.

Edited May 14, 2021 by GreenTriangle

[lex311085]

19 minutes ago, GreenTriangle said:

95 % of what ?

 

as in a goal a game so would score 95 goals in 100 games

[GreenTriangle]

I prefer a more used formula : goals per 90' played.

What means "literally no interest in him from anyone in the 8 or so years" ? Did you not receive any offer or did you not receive any offer at the level at which you was willing to negotiate ? Absolutely no club offered 5 or 10 times less than the amount you considered as being minimum for a possible transfer?

[Ruh Roh]

I never got a response on this bug report but it's sorta connected, even if it's just about junk data in a user database. Makes it look like the AI doesn't handle extreme situations well enough

 

[Bradley21]

I wish threads like this would have any effect on SI, but the sad fact is that this has been an issue since the CM days. It will never be improved because there is absolutely zero reason for SI to ever improve AI transfers and squad building. The people who raise this very obvious issue will buy the game every year, regardless of if SI actually change anything meaningful, so where's the benefit to them as a developer? The only effect this would have would be to drive away casual players because it would make the game more difficult, which is obviously not ideal from a business point of view. 

Well, it would also have the effect of making the game better, but let's not pretend that game developers actually care about making a good game nowadays. 

[CARRERA]

4 hours ago, Bradley21 said:

Well, it would also have the effect of making the game better, but let's not pretend that game developers actually care about making a good game nowadays. 

That’s so narrow minded. Of course any game developer is looking to create a good game, as this will increase their sales. The question to ask is though, for whom they are trying to create a good game. And are you a part of that particular group.

SI is heavily focussing on market share right know, as you can see by them trying to conquer mobile and console markets (which are future growth markets) and therefore trying to reach a very broad audience. And AI realism is proberbly not a focus feature for that audience. 

Better AI Processing will also be more demanding for people’s hardware and FM already is already that, at least if you are looking to play with many leagues loaded

[JordanMilly]

7 hours ago, Bradley21 said:

Well, it would also have the effect of making the game better, but let's not pretend that game developers actually care about making a good game nowadays. 

Let's do without that snark, thanks, not like you should need reminding given your previous warnings. If you want SI to take any notice of your input (and they do, I can literally see features that were added in FM21 that had been suggested in the suggestions forum before), insulting them isn't the best way to go.

[lex311085]

On 14/05/2021 at 18:23, GreenTriangle said:

I prefer a more used formula : goals per 90' played.

What means "literally no interest in him from anyone in the 8 or so years" ? Did you not receive any offer or did you not receive any offer at the level at which you was willing to negotiate ? Absolutely no club offered 5 or 10 times less than the amount you considered as being minimum for a possible transfer?

he wasn't playing for me, but i kept him on my shortlist as he was that brilliant that I wanted to keep tabs on him cos I was just interested. and yes literally no "wnt" icon next his name in any of the transfer windows which is weird

[rekluse]

On 13/05/2021 at 05:22, BuryBlade said:

Sheffield Wednesday btw not Sheffield United.

For anyone who had played OOTP baseball, would the fog of war/20-80 style of assessing players work better? 

I think the biggest thing FM could take from OOTP to fix this would be to simply give us, the users, settings to manipulate what the AI values in players. Almost common place for OOTP players who want a more realistic game to move the trade difficulty slider hard and move the slider to make the AI value prospects more.

If we had sliders to manipulate how much value the AI put on CA/CP, reputation, form, etc then I'm sure the community would do FAR more testing to get a better system in place than any developer could possibly have time to invest in this.

[Safe Hands]

It's a major problem. The AI squad building is ridiculous.  Right from the start of the game big clubs handed out contracts 4 years contracts to 36 year olds. Madrid always give Modrc a 3 year deal, same with City and Fernandinho. There are loads more. In real life it would only be a 1 year deal.

[wazzaflow10]

On 11/05/2021 at 05:33, cris182 said:

This is definitely something that needs improving in my opinion, I too have seen players with great goal records left in lower leagues to rot when the AI spend 20m on a brazilian who plays less than 10 games in 4 years then gets released 

Surely this happens in real life. Look at Minamino or van De Beek. it might be a bit extreme in FM but players most certainly get left rotting on the bench irl while others bang in goals in lower leagues.