AI Squad-Building: Is there actually a problem?

While I'm waiting for the final patch before starting another long-term club save, I thought I would run a ten year holiday save and see if perennial complaints about AI squad-building still hold up with this year's version. Starting with eight loaded nations (Argentina, Brazil, England, France, Germany, Netherlands, Portugal, Spain) on a large database (with no further modifications), I have found that squad-building and overall regen development is generally very good. Over ten years, teams in the top domestic leagues become noticeably stronger with average CA's jumping by as much as 15 points. Moreover, given the importance of key personality attributes in determining the rate of player development, player personalities were generally stronger on the whole as well. "Negative" personalities were nowhere to be found while Professional and Determined players were the norm.

In regards to aging players, older players in 12.1.1 do not see the rapid physical deterioration that plagued earlier versions of FM. Players are still very useful heading into their early/mid-30's with the AI given plenty of time to develop and introduce their eventual replacements. While there does appear to be a brief window between 2017-2019 where average squad age is on the old side, this quickly corrects itself with the average age for most top teams being about 25/26 by 2022.

Additionally, I did not identify any weird "flaws" among high CA players in the top leagues. For example, there were no centrebacks who can't jump and no mid-fielders with the stamina of a 70 year-old. While such aberrations may occur at the youth level, the AI appeared to be wholly competent in weeding them out before they reached the first team.

Finally, I found that "big teams" also didn't horde all the best young talent. By 2022, several potentially great players were slowly working their way up in the squads of mid-table and relegation-level teams with Fiorentina even managing to build itself into the 2021/22 European champions.

With that said, there are still a few issues that remain. First and foremost, the general quality of managerial talent declined significantly over 10 years. In England, the top teams all ended up being controlled by relatively mediocre managers as their owners evidently preferred not to take a risk on retiring players or foreign talent. However, in terms of overall game difficulty, the relative degeneration of managerial quality is more than offset by the increase in general squad quality.

Second, as the 442 creeps back into dominance, "pure" attacking and defensive midfielders find themselves played out of position, and in some cases, managers playing a 442 even buy these players despite the fact that they have no competence in the manager's preferred formation.

Third, international management becomes the playground of horrible regens with staff CA's in the 50-80 range. Even for powerhouses with loaded leagues like Argentina and Spain, national FA's are prone to giving their prestigious national management roles to literal nobodies who are barely qualified to manage in a Sunday league.

And finally, there is the issue of too many big talents coming from small, inactive nations, though this will apparently be fixed in the final patch.

However, on the whole, squad building appears to be in very good shape for FM12 with squads across the top domestic leagues growing gradually stronger as time progresses.

And for those interested, here are the starting XI for several of the teams that I looked at. Real players are listed by their name while regens are listed by current ability, personality, nationality and age.

TOP ENGLISH TEAMS

---------------------------------------------------------------------------------------

ARSENAL

Manager: Craig Levein

GK: CA 163 (Resolute/Cyprus)

LB: CA 157 (Fairly Professional/Honduras)

CB: CA 175 (Professional/Nigeria)

CB: CA 175 (Determined/Mexico)

RB: CA 162 (Fairly Determined/Croatia)

DM: CA 156 (Fairly Determined/South Africa)

CM: Jack Wilshere

CM: Aaron Ramsey

AM: CA 176 (Professional/England)

ST: CA 176 (Resolute/Turkey)

ST: CA 165 (Professional/Mexico)

CHELSEA

Manager: Gareth Southgate

GK: Thibaut Courtois

LB: CA 178 (Resolute/Netherlands/26)

CB: CA 161 (Professional/England/27)

CB: CA 169 (Professional/Italian/24)

RB: CA 190 (Model Professional/Mexico/26)

ML: CA 152 (Fairly Ambitious/Colombia/24)

CM: CA 161 (Spirited/England/26)

CM: Yann M'Vila

MR: Matias Schelotto

ST: CA 182 (Resolute/Portugal/26)

ST: CA 180 (Resolute/Argentina/26)

LIVERPOOL

Manager: Jose Mourinho

GK: Alex Smithies

LB: Jack Robinson

CB: CA 158 (Fairly Determined/Ivory Coast/27)

CB: CA 170 (Professional/Mexico/29)

RB: CA 161 (Professional/Scotland/24)

CM: CA 179 (Determined/Argentina/25)

CM: CA 177 (Fairly Determined/Germany/24)

AML: CA 175 (Resolute/Croatia/23)

AMC: CA 175 (Professional/France/24)

AMR: CA 170 (Resolute/England/29)

ST: CA 160 (Fairly Professional/Poland/22)

MANCHESTER CITY

Manager: Steve McClaren

GK: Joe Hart

LB: CA 176 (Professional/Poland/22)

CB: Danny Wilson

CB: CA 156 (Professional/Senegal/32)

RB: CA 160 (Driven/Paraguay/25)

ML: Samir Nasri (the next Ryan Giggs?!?!)

CM: CA 164 (Model Professional/Romania/24)

CM: CA 168 (Professional/Portugal/25)

MR: CA 174 (Driven/Russia/24)

ST: CA 180 (Professional/USA/24)

ST: CA 165 (Fairly Professional/Belgium/25)

MANCHESTER UNITED

Manager: Remi Garde

GK: CA 174 (Fairly Ambitious/England/29)

DL: CA 184 (Professional/Netherlands/22)

CB: Phil Jones

CB: Chris Smaling

DR: CA 188 (Fairly Professional/Liberia/27) <------------ the new Ronaldo, multiple Ballon D'Or winner

ML: CA 178 (Fairly Professional/Peru/23)

CM: Sandro

CM: CA 174 (Model Professional/Czech/24)

MR: CA 167 (Professional/Mexico/25)

ST: CA 177 (Resolute/England/26)

ST: CA 156 (Resolute/England/25)

TOTTENHAM

Manager: Alex McLeish

GK: Jason Steele

LB: CA 163 (Fairly Determined/USA/26)

CB: CA 163 (Resolute/Belgium/26)

CB: CA 160 (Model Professional/Hungary/24)

RB: CA 165 (Fairly Professional/Chile/24)

ML: Gareth Bale

CM: Jack Rodwell

CM: CA 163 (Professional/Congo DR/26)

MR: CA 174 (Model Professional/Egypt/24)

AMC: CA 173 (Fairly Professional/Zimbabwe/25)

ST: CA 175 (Resolute/USA/29)

MID-TABLE

---------------------------------------------------------------------------------------

ASTON VILLA

Manager: Ciriaco Sforza <--- best manager in the EPL in terms of staff attributes

GK: Frank Fielding

LB: CA 129 (Fairly Professional/Romania/29)

CB: CA 140 (Balanced/Chile/26)

CB: CA 150 (Determined/Ghana/21)

RB: CA 136 (Fairly Determined/Northern Ireland/25)

DM: CA 144 (Balanced/Czech/24)

ML: CA 143 (Resolute/Hungary/29)

CM: CA 154 (Fairly Ambitious/Guinea-Bissau/25)

CM: Andy King

MR: CA 161 (Determined/England/26)

ST: CA 155 (Professional/England/24)

FULHAM

Manager: Simon Grayson

GK: CA 150 (Balanced/Portugual/22)

LB: CA 137 (Balanced/France/30)

CB: CA 151 (Model Professional/Belgium/23)

CB: CA 155 (Light-Hearted/France/29)

RB: John Flanagan

ML: CA 153 (Balanced/England/23)

CM: CA 158 (Fairly Determined/England/28)

CM: CA 144 (Balanced/Argentina/28)

MR: CA 151 (Model Professional/Belgium/23)

ST: CA 155 (Fairly Professional/Croatia/23)

ST: Andy Carroll (the next Emile Heskey!)

RELEGATION

---------------------------------------------------------------------------------------

CARDIFF CITY

Manager: Alan Shearer

GK: David Marshall

LB: CA 145 (Balanced/England/26)

CB: Sebastian Coates

CB: CA 154 (Determined/Congo DR/25)

RB: CA 136 (Determined/Scotland/28)

ML: CA 141 (Fairly Determined/England/26)

CM: Aron Gunnarsson

CM: CA 149 (Balanced/Belgium/23)

MR: CA 161 (Fairly Professional/England/23)

ST: CA 157 (Fairly Professional/Ireland/26)

ST: CA 137 (Balanced/Czech/28)

WATFORD

Manager: Tal Ben-Haim

GK: Gary Woods

LB: CA 135 (Fairly Determined/England/24)

CB: CA 139 (Spirited/England/20)

CB: Andre Wisdom

CB: CA 135 (Fairly Ambitious/England/26)

RB: CA 131 (Fairly Determined/Scotland/23)

DM: CA 140 (Determined/Romania/23)

CM: CA 144 (Resolute/Switzerland/24)

CM: Alex Oxlade-Chamberlain

AM: Gylfi Sigurdsson

ST: CA 154 (Fairly Determined/Portugal/24)

RECENT EUROPEAN CHAMPIONS

---------------------------------------------------------------------------------------

BARCELONA

Manager: Andre Villas-Boas <---- now ten years in charge and the best manager in the world in terms of staff attributes

GK: David Ospina

LB: CA 171 (Fairly Determined/Spain/24)

CB: Gerard Pique

CB: CA 178 (Resolute/Portugal/28)

RB: CA 178 (Professional/Nigeria/27)

DM: CA 166 (Spirited/Spain/24)

CM: CA 167 (Driven/Brazil/24)

CM: CA 184 (Professional/Germany/23)

AML: CA 167 (Fairly Ambitious/Portugal/25)

AMR: Alexis Sanchez

ST: CA 174 (Resolute/Brazil/24)

FIORENTINA

Manager: Delio Rossi

GK: Neto

LB: CA 163 (Fairly Determined/Belgium/26)

CB: CA 158 (Fairly Ambitious/Czech/24)

CB: CA 155 (Resolute/Argentina/22)

RB: CA 155 (Fairly Determined/Netherlands/26)

CM: CA 177 (Spirited/Italy/26)

CM: CA 152 (Resolute/Netherlands/22)

CM: CA 159 (Model Professional/Denmark/21)

AM: CA 172 (Fairly Determined/Argentina/26)

ST: CA 160 (Resolute/Scotland/27)

ST: CA 181 (Resolute/Italy/23)

INTER

Manager: Gianluca Atzori

GK: Andrea Consigli

CB: CA 151 (Model Professional/Brazil/20)

CB: CA 180 (Resolute/Spain/25)

CB: CA 154 (Resolute/Switzerland/25)

ML: CA 164 (Professional/Greece/28)

CM: CA 157 (Fairly Determined/Brazil/25)

CM: CA 153 (Fairly Professional/Italy/25)

MR: CA 158 (Resolute/France/22)

ST: CA 161 (Balanced/Italy/24)

ST: CA 184 (Leader/Argentina/24)

ST: CA 179 (Resolute/Italy/24)

What are the highest CAs of players in the reserve squads of the big sides? I think that's a big problem, they sign too many good players and leave David Silva (for example) in the reserves.

I've heard very different stories with this.

I haven't bought the game this year due to this issue that has plagued the game for years. Maybe this will be the start to me changing my mInd.

Good thread, I would be interested to see how successful Man Utd have been, as in FM11 if they were not immiediately successful the debt would be out of control fairly quickly.

I don't think I have the energy to go over this in detail but buying enough players with high CA to put out a first team side with a high average CA has never been a problem for the AI. The well rounded players are a result of templates being used to create different types of players not the AI being able to evaluate players on their suitability.

In short the AI can assemble teams of high CA players but it rarely if ever builds squads that contain: enough players natural to its preferred formations, a mix of ages so it doesn't suffer a whole generation nearing retirement at the same time, Enough home grown players to field a strong team in the European competitions. It doesn't get value for its signings, It will happily buy an expensive player put him in the reserves and then extend his contract after three years.

I don't know why you use CA to judge the quality of AI squad building. We all know the AI codes dictate that it buys players purely based on CA/PA and reputation. The real problem we've been talking about is this:

Second, as the 442 creeps back into dominance, "pure" attacking and defensive midfielders find themselves played out of position, and in some cases, managers playing a 442 even buy these players despite the fact that they have no competence in the manager's preferred formation.

AI doesn't know what kind of player it's buying, and that leads to the Liverpool-Carroll, Barcelona-Zlatan, etc. problem. AI buys the wrong player for its team's style of playing. Unlike real life when it happens only once in a while and real manager stops using that player once they realized it's a mistake, in FM it happens too often and AI managers keep using him.

Any change of a SC of the 190 CA and the new Ronaldo ?

Just wanted to note that, because I play exclusively with MLS, I edit the 150 players for the draft each year. Surprisingly, the vast majority of players all seem to have 10s across the board for their "hidden" attributes (professionalism, ambition, etc.) when I use FMRTE. Outside of ambition, which can fluctuate to either extreme, these attributes all seem to be pegged at 10. I'm wondering if this is a result of the game trying to let events "mold" a player, rather than have pre-set attributes determine how a player's career plays out.

Look - No, this is a bug that will be fixed with the next patch. See here.

And having a look at what SCIAG stated would be quite interesting. I've seen quite a lot of clubs that keep buying players in a certain position, obviously can't play them often enough and after half a season a lot already want to leave due to lack of first team football and often rot away in the reserves after.

Also Inter playing 3-4-3, they didn't learn their lesson?

What are the highest CAs of players in the reserve squads of the big sides? I think that's a big problem, they sign too many good players and leave David Silva (for example) in the reserves.

Looking at the worst offenders, Chelsea has 4 good but sub-first team players (around CA 150) in the reserves, all of whom are transfer listed. The rest are players in the 110-135 range. It's a similar situation at other sugar daddy clubs with big reserve teams. The worst offense is a CA 170 wonderkid at Man City waiting to break into the first team.

But on the whole, it's not a major, systemic problem and it's certainly not found once you go outside the "big teams." I would also say this reflects a real phenomenon in two ways:

(1) Undeniably talented players like Romelu Lukaku, Tevez, dos Santos and Hargreaves just sitting in reserves.

(2) Unknown but incredible talents waiting for their big break.

Obviously, as a player, you're able to spot (2) in-game by just looking at attributes, but obviously, having a numeric representation of every player's ability is not something real managers have. Anyway, I don't see this as unrealistic given the relatively limited extent to which it occurs. Annoying? Yes, but it was also annoying of Chelsea to buy a player as exciting as Lukaku only to let him rot.

And on the whole, sugar daddy teams aren't hording all the good players and aren't preventing competitors from building great squads.

IMO, the only major issue here is that smaller teams only have about 5 or 6 players in their reserve squad.

Enough home grown players to field a strong team in the European competitions.

This seems to vary wildly depending on the team. Some teams improved, others did not. It's not clear whether there's an equilibrium point for homegrown players that the AI shoots for, but I'll do some further holidaying and see how it pans out. However, in the numbers I post below, you'll notice that many big clubs start the game without a sufficient number of HG players despite the fact that they are in the Champions League. I would also note several of these teams start with HUGE first teams whereas, after ten years, most teams normalize with a first team of about 25 players, so in some cases, the actual proportion of homegrown talent in the first team increased as enormous teams shed excess talent from the initial database.

The Ajax issue, I suspect, has more to do with Holland's weak overall youth output.

HOMEGROWN STATUS IN FIRST TEAM

Ajax

2011: 17 (11 Dutch)

2022: 6 (4 Dutch)

Arsenal

2011: 8 (4 British)

2022: 7 (6 British)

Barcelona

2011: 18 (17 Spanish)

2022: 12 (10 Spanish)

Chelsea

2011: 6 (5 British)

2022: 9 (5 British)

Liverpool

2011: 10 (10 British)

2022: 11 (9 British)

Milan

2011: 14 (13 Italian)

2022: 4 (4 Italian)

Porto

2011: 4 (3 Portuguese)

2022: 8 (5 Portuguese)

Real Madrid

2011: 12 (12 Spanish)

2022: 14 (9 Spanish)

It will happily buy an expensive player put him in the reserves and then extend his contract after three years.

As I mentioned above, I notice this happens a bit with the wealthier teams, but it's not a persistent issue that really affects the overall quality of the global talent pool. Again, I didn't notice any real player hording of the sort I saw in FM11 when teams like Man City would, for example, buy virtually every decent keeper on the planet. And to the extent that it does happen, it can be interpreted to reflect "mistakes" that do occur in real life.

Good thread, I would be interested to see how successful Man Utd have been, as in FM11 if they were not immiediately successful the debt would be out of control fairly quickly.

In this save, Chelsea dominate the next decade but United manage to win the Premier League twice and make it to 3 Champions League finals (winning once). For the most part, the Premier League Big 5 remain intact.

Good thread, I would be interested to see how successful Man Utd have been, as in FM11 if they were not immiediately successful the debt would be out of control fairly quickly.

I agree that this issue needs to be addressed. However, given the overall CA increase of squads, the big Champions League teams mostly consist of CA 160+ players who are generally capable of working very well in any system. And this is further reflected by the fact that the match engine works more on generalities than real life. For example, Ibrahimovic works brilliantly in the Barca system in Football Manager, better than Villa even, but this is because we don't really see those more subtle stylistic differences reflected in-game. Everyone is just based off the same general attributes, so everyone is generally more versatile, especially when you're talking about current ability exceeding 150. Similarly, Andy Carroll works fine with Liverpool with decent man management and Torres is utterly brilliant at Chelsea.

Mid-table and below, with the initial database, you basically just have a random mish-mash of players anyway as such teams take decent talent where they can get it (and we definitely see more transfer "mistakes" made at that level, much more than we see with the AI based on the CA system).

So on the whole, I feel the CA weighting system does its job and the squads circa 2022 are still stronger across the top domestic divisions than the squads in 2011. And that's the central point I'm trying to make. I totally agree that the AI squad-building can be made stronger, but looking at this save, I do see squads actually improving, not degenerating.

SI, obviously, can and should improve on the AI looking for players that match even more specific roles than those defined by CA weighting, but it's not a purely random system and works well enough that the squads do get better. The AI obviously isn't going to match the player in terms of overall squad-building genius, but that is more of a limitation of computer AI in general.

And regarding positional familiarity in general, based on other threads posted here recently and my own experience with FM12, I'm under the impression that effect of position familiarity has been toned down quite a bit to compensate for this.

Just wanted to note that, because I play exclusively with MLS, I edit the 150 players for the draft each year. Surprisingly, the vast majority of players all seem to have 10s across the board for their "hidden" attributes (professionalism, ambition, etc.) when I use FMRTE. Outside of ambition, which can fluctuate to either extreme, these attributes all seem to be pegged at 10. I'm wondering if this is a result of the game trying to let events "mold" a player, rather than have pre-set attributes determine how a player's career plays out.

This is an acknowledged bug. Personally, I haven't encountered it, but it will supposedly be fixed for the next patch.

You say squad building has improved, but showing 11 players with good CAs does not a squad make. How many players do teams have in their first team squads, and what are the CAs of those players? That would help paint a better picture about how the AI is working. And how many players in the whole of your game world are there with CA 160+/170+/180+?

All the above current shows us is that the AI can put together 11 players with good CA - it doesn't tell us anything about squad building.

You say squad building has improved, but showing 11 players with good CAs does not a squad make. How many players do teams have in their first team squads, and what are the CAs of those players? That would help paint a better picture about how the AI is working. And how many players in the whole of your game world are there with CA 160+/170+/180+?

All the above current shows us is that the AI can put together 11 players with good CA - it doesn't tell us anything about squad building.

The AI seems to eventually settle on a squad of about 23-27 players unless the manager has the "big squad" tendency set. Unlike the initial database, the AI tends not to amass many "useless" players far below the club's standards and I also haven't seen much of the FM11 epidemic of top level managers blowing their transfer budgets on half-a-dozen mediocre players that have no choice on seeing first team action (which was one of the symptoms of the player-hording tendency in prior versions). Unlike prior versions, the AI also appears to do a good job of fixing obvious weaknesses in their squad and not overprioritizing positions where they don't need reinforcements. Arsenal, for example, almost immediately went about addressing its fullback situation with players generally comparable to the rest of the squad. In last year's version, the AI was a little too much like Arsene Wenger. This year, it seems to be decidedly smarter than Wenger.

Anyway, looking at Tottenham as an example, their lowest CA substitute is their third-string keeper (CA 129) while everyone else in the 24-person first team is CA 140 or above. But maybe some of them are horribly flawed? Looking at them, I don't see any drastic weaknesses in positionally relevant attributes, certainly nothing more than what you can find in the Tottenham of the initial database. Moreover, the personalities of the regens are generally stronger across the board.

EDIT: And just to be clear, I'm not saying the squad-building AI is perfect in the sense of being as good as an experienced FM player. However, I do think its much improved over prior versions to an extent that makes the game consistently challenging and balanced well into a save. The "mistakes" don't create unrealistic imbalances and seem to reflect situations that you see in the initial database. At the moment, my greater concern is the development of staff (particularly in terms of AI board hiring decisions, randomly distributed tactical attributes and the prevalence of 442).

I think the AI squad building has improved since FM11 yes. Or maybe it is just their tactics that has become better (or the lack of Superb morale winning streaks), because I have a reasonably good PL team with Bradford in 2016, but it still looks to be a challenge to break into the four CL spots. This was not a problem at all in FM11, and one of the reasons for this was the incredible deterioration of AI squad quality after the first three or four seasons.

I have a large-database holiday save currently in 2046, and the big teams are still dominating, with the odd exception of some sugar-daddy teams taking up the battle with them. I don't have FMRTE, though, so I haven't really looked very deeply into the data. I just checked the transfer histories of the big clubs and there is much more activity than in FM11.

AI squad building is slighlty bizzare for me.

my 3rd season with blackpool

last season newly promoted west ham was trying to buy the basel midfield. (and got xaka)

landon donovan is at spurs

dremthe is at liverpool

dempsey at everton.

yet i was left alone to buy the likes of rodwell and gaitan.

FM12 has definitely shown improvement in terms of putting together a squad & being able to ensure the club has enough quality players who meet the homegrown rules, the letter was a major flaw with FM11 as the average number of HG players at top clubs was 2 or 3.

There are still some anomalies cropping up though, currently March 2021 & C.Ronaldo is apparently not good enough for the Real Madrid squad but is OK to draw a "120K pw pay cheque.

http://api.ning.com/files/iySBkmGbHe0uv*K7X2cyD0f5DKN7RtHC3iPoSrvVKdEe5P0CIzA26UIJbdG9WnF-A-*Y6aj3VGHQSl1WkgYp-oTphy7Evnnb/CristianoRonaldoOverview_Attributes.png