Human-assisted AI squad building

One of the toughest challenges games developers face is creating an AI that can challenge the human mind. Its probably an impossible task at the moment. However, I think instead of trying to make their AI compete against the human mind, they should instead look at it from a different perspective.

Instead, why not just directly copy human thinking and tap into that vast resource of creativity and ingenuity?

If the AI can't be significanlty improved to challenge the human manager any time soon in FM I suggest giving us the option to disable AI Tactical Tendancies in the editor and the reputation-based transfer policy in game, and allow those of us who wish to to import human-created squad templates and player-made tactics for the AI to use instead.

The human-made squad templates would be something along the lines of;

• adequate cover for all positions
  
• a mixture of experience and youth
  
• each player skilled with the correct balance of attributes for that position
  
• each player at least adequate for the league level he's playing at
  

- basically everything we humans look to do with our own squad building.

All the above would be moulded together by the human player over time and once he's satisfied that its working and is proving successful in his/her save, that basic squad template he's built could be optionally made available for AI squads to adopt (or imported to AI squads via the editor). Crucially these templates could be shared and downloaded by anyone who seeks a challenge and imported to their own databases & games so we could build up a vast and diverse library of human-made squad templates and tactics that the AI clubs could use. The AI wouldn't then be so predictable to play against.

The AI would see the templates as just numerical values, obviously. But would try and match those numerical values as exactly as possible over time.

Once a human-made template is adpoted the AI would no longer look at things like player reputation when squad building, instead its prime focus would be on finding a player that is the closest possible fit to each position in the template - based directly on the criteria listed in the bulleted points above. Forget randomness and tendancies, CA & rep - just cut straight to the chase. Copy what works. Copy from the human mind. Of course it would take into account available budgets etc, and still have some emphasis on PA. Youth development would be focused on finding players that have the best PA to fit that template too, instead of the randomness it currently is (or seems to be).

Tactics would follow in a similar vein... i.e the human creates some tactics and player roles that work well with a particular squad template, he then associates those tactics with that template via tick boxes or whatever. Now the AI, having decided on its template, adopts those tactics too.

Effectively there is very little pre-determined or random AI decisions involved in the process anymore since the AI is strictly following a human script - human thinking. Thinking thats based on hours of trial and error & countless tweaks - something the AI could never do.

What it boils down to is basically something like what we have in the Tactics forum where people share their tried and tested tactics with each other that they they then go and use to destroy the already disadvantaged AI with. With this idea, its the opposite premise as we'd have people dedicated to creating AI tactics and templates designed to challenge the human player instead. Yes people will exploit it and make weak templates and tactics for their AI opponents so they can dominate them even more, so what? There are plenty of fantastic tactic makers and squad buliders out there that will love to see their creations decimate their fellow human players :thup: And its all optional.

Would this make AI managers redundant?

Not totally. From now on the AI manager decision making would be soley concentrated on what strategies he uses during a match.. i.e. what shouts he uses and when to switch tactics etc. This is an area where the higher-attributed AI managers could stand out from the poor AI managers, and be noticeable by us for once.

Can this work in FM?

I'm not sure. But I remember an old game from the Amiga days called Kick Off - Return to Europe. On that game it became pretty easy to beat the AI early on. But crucially there was a tactics editor and any tactics you made could be incorporated into AI teams so they would use them against you. I remember making some really tough AI tactics to play against that took the game challenge to a whole new level, added years of longevity to the game, and eliminated the need to use skill levels. This was back in 1990, I'm sure it isn't beyond SI to implement something similar 20 odd years later.

In essence, an AI which is developed in line with what humans think it should do is what we already have.

Also I cannot believe that enough people would actually like making the required huge effort to make assigning the necessary coding time to this viable.

Unfortunately it doesn't work. The AI tries to emulate human thinking. I say it should directly copy it - at least until it can be significantly improved.

Also there wouldn't be a huge effort involved since players would play their game as normal, and if they build a successful squad and tactics to match, they can simply share them with other players (like in the tactics forum). You don't have to change the way you play one iota if you don't want to, since its all optional. Unless you meant SI assigning the necessary coding time to make the changes? In which case I'd probably agree unfortunately.

The thing is, does everyone really use a set template for squad building? I know from my own point of view i dont really. Yes i make sure of basic things, cover for positions a few youngsters but things like this are already possible for the AI. I dont follow any defined patterns for signing players, i find a squad weakness and i improve it, the only difference really between myself and the AI is i am thinking 5 years ahead, the AI doesnt always do that.

I actually think what your suggesting would limit the AI more than anything, what works for managing Man Utd probably wont work for managing Boavista or Cheivo, so if you start making the entire AI work to the formula your using at United, your going to end up with a horrible game world with squads all over the place.

AT the moment i think the biggest advantage we have is our ability to see beyond rep and CA/PA, the AI seems to struggle with that, improve that part of the game and i think the AI squad building could be fine, at the moment it seems a bit too reliant on a couple of numbers being high or low enough and not enough about the actual player in question. But you dont need to limit the AI to a certain way of thinking to achieve that. You also need to make sure and not remove the AI's ability to sign the wrong player, something that is very important if you want to keep a realistic gameworld going.

The thing is, does everyone really use a set template for squad building? I know from my own point of view i dont really. Yes i make sure of basic things, cover for positions a few youngsters but things like this are already possible for the AI. I dont follow any defined patterns for signing players, i find a squad weakness and i improve it, the only difference really between myself and the AI is i am thinking 5 years ahead, the AI doesnt always do that.

I actually think what your suggesting would limit the AI more than anything, what works for managing Man Utd probably wont work for managing Boavista or Cheivo, so if you start making the entire AI work to the formula your using at United, your going to end up with a horrible game world with squads all over the place.

AT the moment i think the biggest advantage we have is our ability to see beyond rep and CA/PA, the AI seems to struggle with that, improve that part of the game and i think the AI squad building could be fine, at the moment it seems a bit too reliant on a couple of numbers being high or low enough and not enough about the actual player in question. But you dont need to limit the AI to a certain way of thinking to achieve that. You also need to make sure and not remove the AI's ability to sign the wrong player, something that is very important if you want to keep a realistic gameworld going.

Personally I do have a strategy that I use almost all the time when it comes to squad building my own teams, but I know others don't and don't follow a set pattern. I think thats the beauty of FM, we all seem to play the game differently, which leads to an almost unlimited diversity. So I don't think it'd limit the AI. It'd be as diverse as you like, the more templates and tactics incorporated, the more diverse it would be. The AI doesn't have to be restricted to just one template either. For example once a large enough library is available and general player knowledge of what templates are out there grows, people will find and recommend templates that complement each other, and they could be some how linked so the AI has some diversity and options within that aspect too.

I think what I'm getting at is there seems to be far too much randomness involved in AI squad building that it becomes a mess after a few seasons, its obvious its not working well at all in its current format, mainly because of the things you mentioned. Stream-lining AI decision making by having it copy stuff that does work could be a temporary way around things until games companies come up with much more efficient and logically 'thinking' AI.

The easiest fix to challenging game is to allow us to make AI tactics. Forget about templates and AI squad building. If we were given the abiltity to make AI tactics we could make the AI a lot tougher to play against. But thats too quick of a quick fix and takes away 95% of the rest of the game - making it almost redundant imho.

I agree the AI's ability to sign duds shouldn't be eliminated. That is something that could be directly connected to the AI managers attributes (it probably already is, but other factors make it happen far too often at the moment).

There are a few issues when it comes to using this sort of method for computer-controlled entities. If you're using a script, the computer follows that exact script, as you have rightly said. But, by taking away the pseudo-randomness that usual AI currently operates on, if the player finds a way to combat those techniques set out in the script, there is a greatly reduced chance of failure present within the game. If the AI is only going to use tactics created by players, and nothing else ever, the player needs only to develop and implement the correct counter-strategy for each game since they will know exactly what the AI is using, as opposed to what we have now where we can only make educated guesses and are forced to adapt on instinct. If the AI follows scripts to assist in squad-building, you will likely know what players they will target and have enough of an advantage to act accordingly, successfully or not.

As an option, however, it may not be a bad addition, and may appeal to some like-minded players. I can get behind that. Programming a challenging AI can be extremely difficult even for just a single monster in a game if it has enough to do. With Football Manager, there are so many variables that need monitoring that it isn't surprising that the AI is somewhat lacking in certain areas and inconsistencies arise as the game goes on. Off the top of my head, I'm trying to think of another workaround, but I don't know exactly what is already in place. Something about emphasised guidelines, rather than exact scripts, but that is only theoretical, and not sure how I'd get that working. Might do some research on the matter when I have time and see what I come up with.

Personally I do have a strategy that I use almost all the time when it comes to squad building my own teams, but I know others don't and don't follow a set pattern. I think thats the beauty of FM, we all seem to play the game differently, which leads to an almost unlimited diversity. So I don't think it'd limit the AI. It'd be as diverse as you like, the more templates and tactics incorporated, the more diverse it would be. The AI doesn't have to be restricted to just one template either. For example once a large enough library is available and general player knowledge of what templates are out there grows, people will find and recommend templates that complement each other, and they could be some how linked so the AI has some diversity and options within that aspect too.

I think what I'm getting at is there seems to be far too much randomness involved in AI squad building that it becomes a mess after a few seasons, its obvious its not working well at all in its current format, mainly because of the things you mentioned. Stream-lining AI decision making by having it copy stuff that does work could be a temporary way around things until games companies come up with much more efficient and logically 'thinking' AI.

The easiest fix to challenging game is to allow us to make AI tactics. Forget about templates and AI squad building. If we were given the abiltity to make AI tactics we could make the AI a lot tougher to play against. But thats too quick of a quick fix and takes away 95% of the rest of the game - making it almost redundant imho.

I agree the AI's ability to sign duds shouldn't be eliminated. That is something that could be directly connected to the AI managers attributes (it probably already is, but other factors make it happen far too often at the moment).

Surely tho the more you throw at the AI, the less its going to be able to cope with the data? You could install say 3 templates made up by yourself or another, and you could end up with these 3 tempates fighting against each other and leaving the AI with no idea as to where it goes. I do get where your idea is coming from, but i think your looking for a quick fix rather than a long term solution. Templates are well and good for specific situations, but would you expect the AI to be able to distinguish between these situations?

As for tactics i would completely disagree again. Set tactics are exactly why we can beat the AI so easily sometimes, the AI is not dynamic enough to counter situations, throwing more templates at it is not going to make the AI more sturdy, the AI needs to be more responsive to each game. Again i see that suggestion as a quick fix rather than a solution.

I know its difficult for SI to get it right and I know they work on it with each incarnation but the AI seems to be being left behind year after year. I doubt it'll ever catch up to players who are experienced with this game now.

So the whole idea is a quick fix. But an optional quick fix.

Its pretty clear

I can quick fix my own game already by improving AI squads via fmrte, but as you know its time consuming and risky. I think the time for an SI quick fix may be needed sooner or later - as long as its optional!

There are a few issues when it comes to using this sort of method for computer-controlled entities. If you're using a script, the computer follows that exact script, as you have rightly said. But, by taking away the pseudo-randomness that usual AI currently operates on, if the player finds a way to combat those techniques set out in the script, there is a greatly reduced chance of failure present within the game. If the AI is only going to use tactics created by players, and nothing else ever, the player needs only to develop and implement the correct counter-strategy for each game since they will know exactly what the AI is using, as opposed to what we have now where we can only make educated guesses and are forced to adapt on instinct. If the AI follows scripts to assist in squad-building, you will likely know what players they will target and have enough of an advantage to act accordingly, successfully or not.

As an option, however, it may not be a bad addition, and may appeal to some like-minded players. I can get behind that. Programming a challenging AI can be extremely difficult even for just a single monster in a game if it has enough to do. With Football Manager, there are so many variables that need monitoring that it isn't surprising that the AI is somewhat lacking in certain areas and inconsistencies arise as the game goes on. Off the top of my head, I'm trying to think of another workaround, but I don't know exactly what is already in place. Something about emphasised guidelines, rather than exact scripts, but that is only theoretical, and not sure how I'd get that working. Might do some research on the matter when I have time and see what I come up with.

Yeah I've been wracking my brain trying to think of how the AI can become challenging without SI taking a couple of years off and purely concentrating on that aspect alone, because I honestly think thats what it'll take to bring it up to speed. Incremental improvements year after year aren't cutting it imho.

I think thats why I opted for the quick fix solutions since SEGA get their yearly release and players looking for a better challenge at least get it - albeit a "fake" one in some respects. The way its going thats all I see happening tbh.

I'm not sure about us setting templates, but I'd certainly like to see the AI copying the strategy of the human when they are being successful.

An obvious example came about in my game in Northern Ireland. I was able to make my way from the bottom to the top very quickly by bringing in foreign players, but the AI clubs continued to restrict their transfers to the UK. A few clubs have started to branch out as they grew, but they still conduct a lot of their transfers within the UK and remain miles behind my club in terms of squad quality.

This is one of the problems in the long term, clubs seem to be very set in their ways and refuse to try and copy what other successful clubs are doing.

I'm not sure about us setting templates, but I'd certainly like to see the AI copying the strategy of the human when they are being successful.

An obvious example came about in my game in Northern Ireland. I was able to make my way from the bottom to the top very quickly by bringing in foreign players, but the AI clubs continued to restrict their transfers to the UK. A few clubs have started to branch out as they grew, but they still conduct a lot of their transfers within the UK and remain miles behind my club in terms of squad quality.

This is one of the problems in the long term, clubs seem to be very set in their ways and refuse to try and copy what other successful clubs are doing.

AI strategy is very predictable. I still think human-made templates & tactics will remove some of that predictability. In fact looking at it from another angle, it'd be a way of bringing the diversity of a muliplayer game to a single player game with the AI standing in for another human player, but adopting human created tendancies and strategies.

As for AI clubs learning to follow a successful long term strategy like your N.I. game, thats something I'd love to see in-game one day. Its another human advantage that needs to be emulated (somehow!)

I agree that the AI needs to move over to player-style tactics and use that as a basis for squad building. In fact, I suggested something similar to this a few months ago. However, I'm not sure overriding the default AI behaviors would be the most efficient way to implement it. As milnerpoint said, just tossing on an alternative, player-defined system would have all kinds of unpredictable knock-on effects that could easily just create several game-breaking problems. It would be better just to replace the default AI behaviors, adjust the other game mechanics to work with the new transfer/tactics modules and simply reduce the influence of reputation rather than eliminating it altogether.

While reputation does frequently lead to mistakes, it's important to keep in mind that the goal isn't to have the AI thinking like a player of Football Manager. The goal is to have the AI thinking like an actual football manager, and while reputation isn't the fundamental force of nature that it is in FM, it still has some role to play and should still occasionally have managers go after players based solely on reputation. Certainly, FM players themselves often sign players hyped up in real life (Neymar! Hazard!) while ignoring less glamorous players who would actually fit their system better.

Another mechanism that should be implemented is tougher restrictions on assigning "squad status," automatic changes to squad status based on how often a player is actually used and more consequences for having it changed. Essentially, neither the AI nor players should be able to sign players with the promise of being a "First Team" or "Key" player only to have that player content to be used as an occasional rotation option. This would prevent some of the more absurd player-hording like, for example, a recent save of mine in which AI Harry Redknapp signed 4 first team-quality leftbacks in the first January transfer window.

I agree, reputation still needs to play a role, but much more of a backseat role than it already does. Sometimes, if the price is right, a club might sign a player that is high-profile for the level the club is at in order to bring in extra money through merchandising because of the interest the player generates.

Just wondering if there is a way to add weight to certain behaviours to each manager entitiy in a bid to influence them to act out certain behaviours more often and allow them to make more intelligent decisions, yet maintaining enough pseudo-random traits to not have those exact behaviours happen all the time and make everything stupidly predictable. Such weightings would have to be dynamic to show progession or regression as a manager. It could be argued that staff attributes already do this to an extent. If this isn't how it's done already, it ould be expanded on into specific situational behaviours, rather than what I would probably dub as "Global behaviours", which, from my observations, I reckon might be how the game works currently. Feel free to correct me.

I messed around in FM11 with Manager attributes (creating super-managers with maxed out attributes in certain catergories) trying to influence thier behaviours in regards to their squad building, but I didn't see any significant trends develop or any situational behaviours that I can recall. I would say they continued to follow what you called global behaviours. I'm not sure if anything has changed in FM12 in that regard but I really don't see any significant difference between a highly attributed manager and a lowly L2 newgen manager with some of the descions they make in targeting players, or what they do during a match to influence its outcome. I'm not saying it doesn't happen, its just not noticiable enough. I have a feeling manager attributes don't do what they should do to the extent they should.

I messed around in FM11 with Manager attributes (creating super-managers with maxed out attributes in certain catergories) trying to influence thier behaviours in regards to their squad building, but I didn't see any significant trends develop or any situational behaviours that I can recall. I would say they continued to follow what you called global behaviours. I'm not sure if anything has changed in FM12 in that regard but I really don't see any significant difference between a highly attributed manager and a lowly L2 newgen manager with some of the descions they make in targeting players, or what they do during a match to influence its outcome. I'm not saying it doesn't happen, its just not noticiable enough. I have a feeling manager attributes don't do what they should do to the extent they should.

I have found something interesting doing the same as you. I have edited managers in the EPL in my game to have 20's in what i thought were important attributes, use of subs, buying players ect, but the really interesting one was the buying players, if i put it up to 20 i find the AI only going for the best players, in my save i did it to the Man City manager and in 3 seasons he signed 5 players, but spent close to £180m on them, but went through a season with a squad of 18 players, all great but a tiny squad which lead to them not competing because of injuries and stuff. Just for interest i changed his buying players to 5, the following season they signed 10 players, some of them a bit crap, some of them brilliant, but they then had a squad and suddenly City are challenging at the top again. It seems having that attribute too high means the AI will only go for the very best, if they cant get them they are not interested, put it a bit lower and you see the AI taking more risks, which does lead to them wasting money, but it means they carry full squads to get through a season.

Thats interesting. Is this on FM12?

I may try this as it could pinpoint a few things that can improve longer term saves. The only drawback is if I find the magic formula of what to set managers buying attributes at in the database its still only a temporary fix since they'll all eventually retire anyway, and its back to square one. I suppose fmrte may come in handy here though?

Yeah im doing it with FMRTE on FM12 which i guess you could argue might not be that accurate. You could also argue that the manager finally decided to buy players after a few years but im pretty sure its linked. Its something i thought i had noticed before with my long term scottish save but never really thought much about it, but with this game i was getting annoyed at City having a tiny expensive squad even tho i thought i gave their manager the best attributes for avoiding that, it seems the opposite has more of the effect i was looking for, i dont mind them having poor squad players, as long as they have cover for each position.

What ive also found is my super managers dont always do that well, sometimes the managers already in the game are more succesful, i guess because they have more rounded attributes whereas i tend to throw 20's at whatever i can.

Yeah im doing it with FMRTE on FM12 which i guess you could argue might not be that accurate. You could also argue that the manager finally decided to buy players after a few years but im pretty sure its linked. Its something i thought i had noticed before with my long term scottish save but never really thought much about it, but with this game i was getting annoyed at City having a tiny expensive squad even tho i thought i gave their manager the best attributes for avoiding that, it seems the opposite has more of the effect i was looking for, i dont mind them having poor squad players, as long as they have cover for each position.

What ive also found is my super managers dont always do that well, sometimes the managers already in the game are more succesful, i guess because they have more rounded attributes whereas i tend to throw 20's at whatever i can.

Thats what I basically noticed on the FM11 test. Maybe there's a 'magic number' for each category to get the best out them. If I get time I may revisit that old test at some point.

Thats what I basically noticed on the FM11 test. Maybe there's a 'magic number' for each category to get the best out them. If I get time I may revisit that old test at some point.

It could be how the attributes work with each other and obviously there isnt much documentation as to what each attribute does, if i set say "offside" to 20, does that mean the mangers insists on his team playing the offside trap? Does setting "use of substitutes" to 20 mean he will use them well or use them more often? Hardness of training as well, is it better to have this higher or lower? I think tho it could be one thing we will never fully understand, not unless a kind SI employee would like to enlighen us

Going by what you experienced with the Man City manager I'd lean toward the numbers being like a slider, with 10 being 'neutral', and the higher values (like 20 in your case) being heavily weighted towards transfer restraint, and the lower values (like 5) weighted towards more transfer flamboyance and risk taking.

If those numbers in the catergory list are sliders and not quantative it could explain a lot of things in the test I ran last year.

It could be how the attributes work with each other and obviously there isnt much documentation as to what each attribute does, if i set say "offside" to 20, does that mean the mangers insists on his team playing the offside trap? Does setting "use of substitutes" to 20 mean he will use them well or use them more often? Hardness of training as well, is it better to have this higher or lower? I think tho it could be one thing we will never fully understand, not unless a kind SI employee would like to enlighen us

Manager attributes only control how frequently an AI manager does something. The only attribute that determines how well an AI manager does something is, presumably, "man management." Actually programming the AI to be better or worse at something like squad building or tactics would be well beyond the limits of even the best funded developers. As others have mentioned whenever difficulty levels are discussed, almost all strategy games, with the exception of chess games, have one set of AI behaviors that can be enhanced with various bonuses/cheats associated with different difficulty settings.

So having a high "Buying Players" or "Use of Subs" just means he's more likely to use his entire transfer budget whenever he gets it (which would explain why they immediately go after more expensive players as well as opposed to spending bit by bit here and there as the need arises) and make substitutions rather than keeping tired players in the game. Hardness of training, again, just determines the workload assigned to the players. There's not necessarily a better setting so much as there is a trade off (more improvement but higher likelihood of injury... or less improvement but less likelihood of injury). There are also other factors at work in the form of manager tendencies (e.g., "likes to have a large squad," "uses entire transfer budget on one player").

In terms of tactical attributes, again, each one either denotes the frequency of doing something (i.e., using the offside trap) or just directly translates to a baseline slider setting (sitting back = time wasting, attacking = baseline mentality)... the only one that isn't quite straight forward is "Depth" since it is the opposite of the player's defensive line setting. A higher "Depth" actually means a deeper defensive line as opposed to a higher one.

Now that would make a lot of sense actually. If the numbers act like that it would definiately explain what i have seen with my edits.

Manager attributes only control how frequently an AI manager does something. The only attribute that determines how well an AI manager does something is, presumably, "man management." Actually programming the AI to be better or worse at something would be well beyond the limits of even the best funded

So having a high "Buying Players" or "Use of Subs" just means he's more likely to use his entire transfer budget whenever he gets it (which would explain why they go after more expensive players as well) and make substitutions rather than keeping tired players in the game. Hardness of training, again, just determines the workload assigned to the players. There's not necessarily a better setting so much as there is a trade off (more improvement but higher likelihood of injury... or less improvement but less likelihood of injury). There are also other factors at work in the form of manager tendencies (e.g., "likes to have a large squad," "uses entire transfer budget on one player").

In terms of tactical attributes, again, each one either denotes the frequency of doing something (i.e., using the offside trap) or just directly translates to a baseline slider setting (sitting back = time wasting, attacking = baseline mentality)... the only one that isn't quite straight forward is "Depth" since it is the opposite of the player's defensive line setting. A higher "Depth" actually means a deeper defensive line as opposed to a higher one.

But then with a very high buying players attribute the managers seem to buy less, but better players, with the attribute lower, they tend to buy more, but with an overall less quality....

But then with a very high buying players attribute the managers seem to buy less, but better players, with the attribute lower, they tend to buy more, but with an overall less quality....

Admittedly, I haven't ever experimented with Buying Players, so it could, in fact, be reversed from what one would intuitively assume it to mean (sort of like the "Depth" attribute) or it could reflect a combination of behaviors that can be further fine tuned by the manager tendencies ("Large Squad", "Buys Youth Players," "Uses entire budget on one player", etc.) So a high number could represent a "big spender" who likes to use his budget on extravagant purchases whereas a low number could represent a "wheeler-dealer" who favors bargains (and thus, will be more likely to have resources to maintain a full squad rather than a depleted squad of expensive superstars).

So if that's the case, a "20" would be Roman Abramovich, a "1" would be David Moyes... in which case, if you're looking for more efficiency, then yes, a lower number would be better though it may result in a lack of big player transfers (which could negatively affect the top teams over time).

It'd be nice if SI explained exactly what these numbers represent and how they effect the gameworld. The editor is a powerful tool but the documentation for it is non-existant afaik. If we got an insight on stuff like the manager attribute panel, we could have a better understanding on how to improve the database in regards to squad building and AI tactics.

I'm not sure about us setting templates, but I'd certainly like to see the AI copying the strategy of the human when they are being successful.

An obvious example came about in my game in Northern Ireland. I was able to make my way from the bottom to the top very quickly by bringing in foreign players, but the AI clubs continued to restrict their transfers to the UK. A few clubs have started to branch out as they grew, but they still conduct a lot of their transfers within the UK and remain miles behind my club in terms of squad quality.

This is one of the problems in the long term, clubs seem to be very set in their ways and refuse to try and copy what other successful clubs are doing.

However, may the actual problem with that not be that these foreigners irl would not join a Northern Irish side in the first place?

I read a thread about lots of good foreigners being happy to play for BSP sides for a pittance. Irl it certainly needs a certain minimum for people to move abroad for a job. 300 pounds a week may (no offense intended) be enough for Albanians (due to the average salaries paid there), but for instance not for French (who moved to the BSP in numbers in the example I read about).

However, may the actual problem with that not be that these foreigners irl would not join a Northern Irish side in the first place?

I read a thread about lots of good foreigners being happy to play for BSP sides for a pittance. Irl it certainly needs a certain minimum for people to move abroad for a job. 300 pounds a week may (no offense intended) be enough for Albanians (due to the average salaries paid there), but for instance not for French (who moved to the BSP in numbers in the example I read about).

That's true, but it wasn't really the point I was trying to make. It was just an example of how the AI could adjust when the human is getting the upper hand.

Anyway, the NI Premiership is now ranked #4 in Europe on my game, you'd think things would have changed by now.

That's true, but it wasn't really the point I was trying to make. It was just an example of how the AI could adjust when the human is getting the upper hand.

Anyway, the NI Premiership is now ranked #4 in Europe on my game, you'd think things would have changed by now.

Each nation has transfer preferences which appear to determine which nationalities their clubs will attempt to import, so that might have something to do with it.

Though editing them may exacerbate issues with the AI failing to account for homegrown rules...