AI Transfers/squad building in FM24

[Cloud9]

This was one of the promised features on this FM, and I've been continually disapointed by the quality of the AIs decision making in the transfer market this year, which has been far poorer in my own saves than in previous editions. 

Here's a link to the "smarter transfers" outlined for this year:

https://www.footballmanager.com/features/smarter-transfers-squad-building-and-finance

Here are two examples from how I've seen City spend their entire budget in the past few seasons:

Spoiler

Spoiler

Is anyone else experiencing really poor AI in their saves this year? I understand that teams make bad calls IRL, but 258m pounds by a team w/one of the best track records for successful transfers feels like shoddy work on a promised feature for FM24. Riccardo Orsolini is also a fairly low quality Premier League quality player (73ca/76pa via FMinside), and having signed him for 73m pounds he's played 4 times and is now transfer listed for 9m.

• For context: the world record transfer is still Neymar at 198m pounds. I believe Madrid's failed bid for Mbappe was around 200m euros last summer, roughly 170m pounds (for the best player in the world). 

Disappointing that this has not been addressed (if it is an issue outside of my own saves) in the Winter Update. Not only have these issues not been addressed, but there's no match engine update (which is needed to continue to tune the excellent but a bit OP positional play feature), and the data update itself has been poor. Three of the biggest star players outside the big 6 (Neto, Paqueta, and Kudos) are all without upgraded CA/PA's to reflect their undeniable quality. Neto has been in talks to be the PFA player of the year as a dark horse and Paqueta is clearly one of the best midfielders in the league to anyone actually following the football. Since the game focuses primarily on and around English football it's a bit of a let down for FM24's biggest update.

Edited March 8 by Cloud9

[NineCloudNine]

For those two transfers it looks to me like the AI has been coded to place greater weight on performance rating and potential rather than just CA, but it’s been over-tuned.

[Zachary Whyte]

Hello @Cloud9

We made a lot of changes for FM24 to AI Transfers/Squad Building as highlighted by the blog you linked. And they have made a positive impact on the game.

It's a system we can tweak further but we need good examples of poor transfers with save files from before & after to examine further.

Do you have a save file from before that Riccardo Orsolini transfer place?

[NineCloudNine]

52 minutes ago, Zachary Whyte said:

Do you have a save file from before that Riccardo Orsolini transfer place?

It’s great to have you chipping in to threads, thank you.

I’m a little surprised that it’s the Orsolini transfer that has caught your eye. +£70m for a high-performing winger doesn’t seem wildly out of line with real life (Nicolas Pepe, Jadon Sancho, Antony…) whereas £258m for Rodrygo is very hard to rationalise.

[Ein]

43 minutes ago, NineCloudNine said:

I’m a little surprised that it’s the Orsolini transfer that has caught your eye. +£70m for a high-performing winger doesn’t seem wildly out of line with real life (Nicolas Pepe, Jadon Sancho, Antony…) whereas £258m for Rodrygo is very hard to rationalise.

The manager doesn't even attempt to play Orsolini unlike the mentioned real life cases who amassed 25 appearances or more in their first season. Something like that would only make sense if the player suffered a long-term injury or if the manager was replaced right after the transfer.

[Zachary Whyte]

59 minutes ago, NineCloudNine said:

It’s great to have you chipping in to threads, thank you.

I’m a little surprised that it’s the Orsolini transfer that has caught your eye. +£70m for a high-performing winger doesn’t seem wildly out of line with real life (Nicolas Pepe, Jadon Sancho, Antony…) whereas £258m for Rodrygo is very hard to rationalise.

Not really bothered by that example too much, While the fee is high Rodrygo is a phenomenal player who City could play for 10+ years. 

By the looks of it, after some good form City signed Orsolini given him a few games, but then Rodrygo's become available and they've splashed out on him because he's a massive upgrade making Orsolini a backup player.

As mentioned above there's real life precedent with someone like Sancho being signed for big money, underwhelmed then they spent big on Antony and Sancho barely got a look in.

But a save would be good to see if City could have looked to sell Orsolini instead of keeping him.

[numbas2]

I’m in 2035 and AI squads are poor. And players with 5* potential are wasted. Same story as previous FMs unfortunately. 

[Cloud9]

4 hours ago, Zachary Whyte said:

Do you have a save file from before that Riccardo Orsolini transfer place?

Hi @Zachary Whyte, I'd be happy to provide a save file. 

A few questions on my end:

• How long before the transfer are we talking? I think I should have a few backup files to pick from. 
• Where should I submit the file to?
• Finally, and most importantly, I manage as myself in my saves (as usual ). Is there a way to remove myself as manager before submitting for privacy reasons?

I just noticed the transfer for Orsolini, if that's the outlying transfer for you, is more egregious than I previously reported. 73m pounds is the initial fee paid by City, with add-ons seeing the transfer rise to 98m pounds. 

[Litmanen 10]

12 hours ago, Cloud9 said:

 

  Reveal hidden contents

  Reveal hidden contents

Is anyone else experiencing really poor AI in their saves this year?

Yes, but this is something that has been in the game practically always.

As I don't play long-term saves basically at all, the particular issue you raised isn't something I have noticed. What I have noticed is real passivity from the AI in terms of squad building and strengthening their teams. After the first season in La Liga for example, even the promoted teams only signed 1-3 players, and big teams like Real Madrid were not doing anything.

This on the other hand means that human players can freely build their teams in a way they want as there is no competition. Combine this with the AI's lack of squad rotation and you will soon have plenty of unhappy players transfer listed and available for peanuts. 

[turnip]

With these kinds of threads, you always need to post how your save is set up, i.e. which leagues, playable/viewable, database size, etc.

The fewer active leagues you have and the smaller your database, I'd argue you get more outliers like the Rodrygo transfer because if you've got (for example) 3 active leagues and a small database, there aren't that many top quality players for Man City to buy who can improve their squad. So when they identify someone like Rodrygo, they know that they'll have to pay a lot of money (because Real Madrid can't replace him easily either) and things ratchet up. 

 

I've got a massive database and lots of active leagues, and if I look at the biggest transfers from the previous couple of seasons most of them look fairly reasonable. There are a couple of exceptions - Liverpool dropped £189m on a right-winger, and a season later are now using a formation that doesn't include wingers  - but overall, the transfers I'm seeing look sensible, albeit subject to the usual price inflation that occurs in long-term FM saves. 

[NineCloudNine]

24 minutes ago, turnip said:

The fewer active leagues you have and the smaller your database, I'd argue you get more outliers like the Rodrygo transfer because if you've got (for example) 3 active leagues and a small database, there aren't that many top quality players for Man City to buy who can improve their squad. So when they identify someone like Rodrygo, they know that they'll have to pay a lot of money (because Real Madrid can't replace him easily either) and things ratchet up.

This is true. I'd add however that the game makes it easy to set up a small game and even encourages it with the performance rating it suggests. There is no description or warning of any negative effects of running a smaller database. Depending on how the player wants to play, there's also often no reason to run multiple countries. Given this, the transfer/squad building code should be able to adapt.

In this case I don't agree that £258 for Rodrygo can ever be justifed. He is

Spoiler

168 CA / 178 PA

in the database - world class but in no way worth 25% more than the current highest ever transfer. The only time numbers anywhere near the Neymar record have been mentioned is for Mbappe, who is currently a

Spoiler

188 CA / 197 PA player

in FM, which is on a different planet to Rodrygo.

Edited March 9 by NineCloudNine

[Cloud9]

On 09/03/2024 at 03:08, turnip said:

With these kinds of threads, you always need to post how your save is set up, i.e. which leagues, playable/viewable, database size, etc.

The fewer active leagues you have and the smaller your database, I'd argue you get more outliers like the Rodrygo transfer because if you've got (for example) 3 active leagues and a small database, there aren't that many top quality players for Man City to buy who can improve their squad. So when they identify someone like Rodrygo, they know that they'll have to pay a lot of money (because Real Madrid can't replace him easily either) and things ratchet up. 

 

I've got a massive database and lots of active leagues, and if I look at the biggest transfers from the previous couple of seasons most of them look fairly reasonable. There are a couple of exceptions - Liverpool dropped £189m on a right-winger, and a season later are now using a formation that doesn't include wingers  - but overall, the transfers I'm seeing look sensible, albeit subject to the usual price inflation that occurs in long-term FM saves. 

I don't think that should be an issue in my case!

Spoiler

Edited March 14 by Cloud9

[turnip]

I'd agree with that. That's a plenty big enough database. 

[SPE3D]

I can only comment on my save but I am reasonably happy with squad building by the AI. I'm now in season 2033 and the squads look pretty realistic to me at least in the leagues I play in.

[SPE3D]

For context I run a lot of playable leagues spread evenly throughout the world on a large database with 130,000 players loaded. Half a star performance but I ignore that and the performance has been good throughout. No complaints from me.

[Flohrinho]

I have noticed one absolute improvement. In FM 2023, squads seemed to grow out of proportion. In FM 2024 the squads seem to be quite small, more often than not around the 25 player mark. This means that changing teams in a long save seems to be less of a hassle if you are deep in the future. You don't inherit 40 player squads with two thirds of the players unhappy.

[DP]

Surprises me how the Rodrygo signing for £258m is ‘not really bothered about’ considering how City’s transfer record is £100m. Cant see any world where City will be buying players for over double their current transfer record, especially with P&S rules.

These inflated transfers have been unrealistic in FM for some time but maybe it’s a mechanic to try to catch up with what the human player can feasibly do?

Edited March 10 by DP

[Cloud9]

8 hours ago, DP said:

These inflated transfers have been unrealistic in FM for some time but maybe it’s a mechanic to try to catch up with what the human player can feasibly do?

Unfortunately because of FFP, the AI spending this high on a transfer hugely hamstrings them in the save (FFP impacts all clubs regardless of who is in charge). Since the Rodrygo transfer in 2024 Man City have only been able to spend 111.5m pounds 2 years on (that's 4 player's across those transfer windows) while moving out 141m pounds worth of player's in that time (a negative transfer spend of 29.5m pounds since 2024).

From what I can presume, they're now unable to sign new player's due to using up most of their FFP on Rodrygo (although Orsolini can't have helped either). Rodrygo is sitting on a nice 425k pounds a week at City which is an additional 22m a year on top of the transfer fee. I don't believe FM uses amortization (which would be a work around on the problem by spreading the transfer fee over the length of the contract), it's on a simple 3 year rolling period instead. That makes the transfer expenditure 368M pounds at least (this excludes whatever the signing on fee/agent fee/bonuses would have been) for a 5 year contract, but almost all of that is front loaded to the first 3 years of FFP. Essentially I could start seeing Man City spend again next year at the earliest, as they appear to be bang out of any FFP.

• Scouted Rodrygo to see his full contract. I think appearance fees, signing on fees, and bonuses add on 30-34M pounds, making it roughly a 400m pound transfer for FFP concerns. That's a 352m pound loss from one transfer that will apply to the first 3 years of FFP after the initial transfer fee is spent.

Even if he performs well, it's still a disastrous decision for the trajectory and competitiveness of a club that should be a behemoth in my save (domestically and in Europe). This is reflected in their performance on the pitch, where Spurs beat them to the Premier League title this year and they were knocked out of the Champions league all three years (first in the Group Stages, then the Semi Finals, and most recently the round of Round of 16).

Edited March 11 by Cloud9

[numbas2]

"I’m in 2035 and AI squads are poor. And players with 5* potential are wasted. Same story as previous FMs unfortunately. "

 

I stand by what I said before and I've got like 30 leagues running and a dB of 208k players. 

90% of those playing FM don't play long term saves so far as I understand. So who cares what we think. 

 

[StatboySpurs]

The realism mod goes a long way towards making things more, well, realistic in terms of transfer values.

 

 

[Cloud9]

On 08/03/2024 at 06:07, Zachary Whyte said:

Do you have a save file from before that Riccardo Orsolini transfer place?

Following up on this!

I'd also like to highlight some questionable transfers out, particularly the sale of Doku 35.5 (48.5m) a two footed 18 pace winger (73ca/84 PA, arguably a more dangerous player because of these attributes than a Rodrygo) and the sale of Nunes for 22m (75/81) to fund the signing of 1 Pedro Goncalves (75/79) who is an older slightly more offensive version of Nunes. They also released a Tommy Doyle for free at age 22, who is now my club captain and worth 39m-51m pounds after one full season lol. 

Man City splashed 53m pounds to bring in Nunes and 55m pounds on Doku IRL.

Spoiler

This by the way, is Tommy Doyle:

Spoiler

He is Resolute, Reserved media handling, Consistent, and Enjoys Big Matches. His off foot is reasonable and he doesn't have issues with injuries. Even for a low reputation club like mine, he only asked for 36k pounds a week as a squad player. I ended up rejecting bids up to 45m for him this summer because he's so important to my midfield 3 (playing as a Mez(s) in a mid block counter attacking style). I'm not sure what his CA/PA are (I've been trying to play w/scout only signing and avoid looking outside of reporting for this thread), but he literally has no down sides.

It's worth mentioning he has developed in the one year I signed him so to be fair I've included his development as well. I do quite a lot around training but still, it's not like he was in a bad place at all when I picked him up:

Spoiler

If the AI can't figure out this is player to keep instead of release it's indicative of big decision making issues around transfers / squad building.

Edited March 14 by Cloud9

[Litmanen 10]

I hear you @Cloud9. The AI seems a little bit like "give me something and I will sell" at the moment. They are basically just doing transfers for the sake of it instead of trying to actually improve their squads.

Also if you look at players' willingness to change clubs, it becomes quite weird. I'm relying just on my scouts to find players but it seems that in same stature clubs like Valencia, Villarreal, Real Sociedad and Athletic Bilbao, almost all their regular starters would be at least somewhat interested in making sideway moves to my Betis. And when the AI is selling their players with very low prices, it basically messes the whole gameplay up. 

Would love to see a version of FM where the AI would not be so willing to sell their players and players might actually say no to you. 

With the current transfer and contract negotiation system it's basically like being a kid in a candy store on Saturday. Pick what you want and pay pennies. 

[endtime]

My biggest gripe with the transfer AI in FM24 is that it regularly splashes a lot of cash on players that it apparently doesn't need or want. These players either never play at all or make a handful of appearances and are sold at a loss or just sit in the squad until their contract expires. Some of the most egregious examples in my current game:

João Mário, bought by Arsenal for 43m, makes 6 appearances and is sold at a loss 6 months later

 

Spoiler

Joe Willock, goes on loan to Tottenham, makes only 3 appearances but is bought out for 26.5m

Spoiler

Jan Thielmann, bought by Arsenal for 37.5m, 8 appearances

Spoiler

Unai Simon, bought by Arsenal for 21.5m, makes one appearance and is sold at a loss 1 year later

Spoiler

Joey Veerman, bought by Arsenal for 43m, 3 appearances

 

Spoiler

Lewis Hall, bought by Newcastle for 28m, 4 appearances in 2 years

Spoiler

Enzo Barranechea, bought by Tottenham for 24.5m, 5 appearances

Spoiler

Jhon Duran, loaned in and then bought by Ajax for a combined 10m, 2 appearances in 2 seasons

Spoiler

I don't understand the logic behind these transfers, it's not even that these players were signed by an outgoing manager and the new manager then used a different formation that they didn't fit into. And this is with an editor file that increases the frequency of injuries, so theoretically there should be more opportunity for rotation.

[El Payaso]

Should be fairly easy for the AI to see whether the new players actually fit into the team and strengthen the current squad. And as the transfer market in general is what it is, those are going to be cheap bargains for the human players. 

[tezcatlipoca665]

3 hours ago, endtime said:

My biggest gripe with the transfer AI in FM24 is that it regularly splashes a lot of cash on players that it apparently doesn't need or want. These players either never play at all or make a handful of appearances and are sold at a loss or just sit in the squad until their contract expires. Some of the most egregious examples in my current game:

João Mário, bought by Arsenal for 43m, makes 6 appearances and is sold at a loss 6 months later

 

  Reveal hidden contents

Joe Willock, goes on loan to Tottenham, makes only 3 appearances but is bought out for 26.5m

  Reveal hidden contents

Jan Thielmann, bought by Arsenal for 37.5m, 8 appearances

  Reveal hidden contents

 

 

 

Unai Simon, bought by Arsenal for 21.5m, makes one appearance and is sold at a loss 1 year later

  Reveal hidden contents

 

Joey Veerman, bought by Arsenal for 43m, 3 appearances

 

  Reveal hidden contents

Lewis Hall, bought by Newcastle for 28m, 4 appearances in 2 years

  Reveal hidden contents

 

 

 

Enzo Barranechea, bought by Tottenham for 24.5m, 5 appearances

  Reveal hidden contents

 

 

 

Jhon Duran, loaned in and then bought by Ajax for a combined 10m, 2 appearances in 2 seasons

  Reveal hidden contents

 

 

 

I don't understand the logic behind these transfers, it's not even that these players were signed by an outgoing manager and the new manager then used a different formation that they didn't fit into. And this is with an editor file that increases the frequency of injuries, so theoretically there should be more opportunity for rotation.

What frustrates me the most about this (and it happens *a lot* in both FM23 and 24 for me) is that the players signed are given practically no chance before being discarded by the AI managers. Don't bother mentioning Kalvin Phillips as a real-life example btw before somebody points that out - it's clear Guardiola didn't want him and his lack of appearances has always been commented on as unusual, which is something that the media mechanic in these games isn't capable of highlighting (which makes sense, since this is not considered unusual or undesirable behaviour within the game overall). The optional/mandatory future fees for loans make no sense either when the teams don't even play these players during their loan periods before they sign them for the next season and, again after that, don't play them.

Another thing is that these transfers don't take into consideration any future fees owed to other clubs. Tammy Abraham joined Aston Villa in my game for £2.4m in the first season while he was injured, and they're not using him now he's fit. Not only does the transfer not make sense from Villa's point of view since they didn't need him and he wasn't even 'in-form' (he hadn't played a game), but Roma sold him at a massive discount while they're still scheduled to pay my team (Chelsea) £6.8m for each of the next three years for a player they don't have, and hilariously, now need again because Lukaku just returned to us from his loan period there.

[ViG1980]

There are loads of examples of the AI being crap. The problem is worsened a lot with how bad they are at bringing in youth.

A few starting 11s in 2028. They've done nothing....and there is no youth to be seen. 

This i

Edited March 28 by ViG1980