A suggestion for improving AI managers and long term squad building.

I have a suggestion how to improve the challenge of the long term game and give more unique characters to AI managers.

By adding five new attributes to manager profiles that influence which players they buy and sell two things can be achieved. AI managers will behave more like their real life counterparts in the transfer market. The squads they build will be better suited to the tactics they employ. Increasing the long term challenge.

The five Attributes I propose are.

Preference for Physical players 1-20

Preference for Technical players 1-20

Preference for fast players 1-20

Preference for professionalism 1-20

Preference for youth 1-20

I’m pretty sure that now when the AI decides its time for a manager to buy a player it uses an algorithm to choose one from the short list each club has. At the minute this algorithm is influenced by the “Judging player ability” and “Judging player potential” attributes and only seems to look at the CA and PA of players in the shortlist.

My suggestion would require the AI to look at the individual attributes of the players giving a greater likely hood of selection to players who have attributes that are preferred by the manager.

It would work something like this.

Tony Pulis

PF Physical players 18

PF technical players 4

PF fast players 10

PF professionalism 14

PF youth 4

When looking to buy a new CM and finding twenty CMs on the short list whose CA makes them a viable target. Tony Pulis would be more likely to bid for the one with a higher combined total for strength, stamina, Jumping, work-rate and team work.

And if the total of those attributes was very close for both players then he would be more likely to bid for the one with the highest total for determination, professionalism temperament ‘and treated as a negative number’ controversy.

Clearly this is a gross simplification and what is really needed is a complicated formula for weighing the PFs against one another and against Judging player ability. As well as a major rewrite of the AI so it can look at players individual attributes, but the benefits being two fold. Better modelling of managers and greater long term challenge I think it’s worth it. If you agree with me that it’s a good idea please post a manager and his PFs bellow. If this thread gets long enough SI will take notice.

I like it

brillaint idea and would deffo make it more of a challenge.

i very much like the general idea of this...

it certainly needs looking at as once all the starting managers disappear the game becomes ridiculously easy!

Gets the :thup: from me.

However the whole AI team building logic is very poor at the moment (probably because its based purely on CA ??). Its one of the reasons why the game gets so easy to dominate the more seasons you progress as it seems AI clubs tend to buy players just for the sake of it without checking to see if...

A - is the player (position) needed for our squad?

B - is he better than what we already have?

So your proposal would work well only when the AI team-building logic has been fixed/overhauled/implemented properly/or completely re-written.

You have the germ of a good idea there, but it needs to be a bit more sophisticated.

For example, Mancini likes his big 'ard lunks, but he still bought Silva and Tevez. Many managers might have preferences for certain profiles to fit certain positions.

That's not a real solution. The problem is AI managers not buying players they need to improve the squad. For example, Wenger never buys a GK, or Ferguson buys a crap GK, who's barely better than Kusczak, to replace VDS. You can add all the preferences you like, but it won't improve the challenge of the long term game.

Think its a great idea personally, and would be a fantastic addition to the game. I think some other people may be missing the point slightly though. The OP isn't stating that this will fix everything in the game long-term, so I don't see why saying it won't fix the issue of big teams buying poor goalkeepers, for example, is particularly relevant. We all know other things need to happen to make it more enjoyable as well, but this would improve the game quite a lot still.

There is already something like this in the game. If you go into the editor, you will notice that managers are rated for their tendency to sign youth and their tendency to domestic players, so I don't think this should be too difficult to implement.

With this idea you'd never see AI clubs evolve. Pulis, for example, has attempted to buy more technical player like Gudjohnsen and Tuncay IRL. With this method in place, He'd have a team of purely physical players, because his coding tells him that is his preferenc. It'd be imbalanced and would not solve squad building. But i do agree, the managers should look at more than CA. This part of the game drives me crazy, easily the worst aspect imo. Ferguson buys 5 new keepers, all sub-standard. Why can he not use his considerable resources to bring in world class keeper? What stops him from doing that?

That's why the tendencies would be rated 1-20. A manager wouldn't only buy players of a certain type, but if given a choice between multiple similarly rated players, they would decide based on their attribute preferences (which could be physical AND technical, just physical, just professional, etc.)

And as it stands, clubs don't evolve. They just randomly throw together players.

With this idea you'd never see AI clubs evolve. Pulis, for example, has attempted to buy more technical player like Gudjohnsen and Tuncay IRL. With this method in place, He'd have a team of purely physical players, because his coding tells him that is his preferenc. It'd be imbalanced and would not solve squad building. But i do agree, the managers should look at more than CA. This part of the game drives me crazy, easily the worst aspect imo. Ferguson buys 5 new keepers, all sub-standard. Why can he not use his considerable resources to bring in world class keeper? What stops him from doing that?

But a preference doesn't necessarily exclude the opposite wholly. So, if implemented well you'd have Pulis buying mostly physical types, with a couple of more technical players occasionally.

Brilliant idea, definitely good if you're going for the long hall!

But a preference doesn't necessarily exclude the opposite wholly. So, if implemented well you'd have Pulis buying mostly physical types, with a couple of more technical players occasionally.

Right but the level of AI the game currently has will not work in such a sophisticated way. Pulis will buy crap physical players and neglect technique. This idea is just changing, or defining where the coding will tell the AI to look. Hence, no evolution.

The idea is good, but I think it needs to be more sophisticated yes.

Instead of looking at "physical players" or "technical players", the managers need to look for "Tenacious midfielders", "towering strikers" or "fast wingers" to fit into the roles of the TC that he uses. In other words, each player in the game needs to be a certain type of player filling a certain role in a team.

This way, every manager in the game will look for certain types of players and consequently all of them will be building their teams by sending out scouts looking for them. The difference between a rookie manager and a good manager will then have to be that the good ones will buy players with CA of 3,5 stars or better as that will strengthen their team. This will not interfer with the other hidden traits of a manager, i.e looking for bargains or spending everything on one player.

This will require an overhaul of the transfer/scouting/agent system, as there are too few transfers in FM now to make a difference regardless of changes to the AI managers.

Right but the level of AI the game currently has will not work in such a sophisticated way. Pulis will buy crap physical players and neglect technique. This idea is just changing, or defining where the coding will tell the AI to look. Hence, no evolution.

Its a step forward because at the present a managers attributes can cause his teams to play in a direct style but if he buys players suited to that style is essentially random. He buys players based only on CA and PA so it is quite possible he ends up playing two small nippy poachers up front while his tactics are direct and focused on getting crosses into the box.

I think your right in your pessimism that my idea is unable to fix the issue with AI managers buying substandard players who are unable to improve the squad. But I think thats happens because its hard to create AI that realistically mimics human mistakes. It would be very easy to write a piece of code that ensured managers only ever bought players with a higher CA then the player already playing in their position, but as managers do make mistakes and do buy donkeys the AI has to as well.

The idea is good, but I think it needs to be more sophisticated yes.

Instead of looking at "physical players" or "technical players", the managers need to look for "Tenacious midfielders", "towering strikers" or "fast wingers" to fit into the roles of the TC that he uses. In other words, each player in the game needs to be a certain type of player filling a certain role in a team.

This way, every manager in the game will look for certain types of players and consequently all of them will be building their teams by sending out scouts looking for them. The difference between a rookie manager and a good manager will then have to be that the good ones will buy players with CA of 3,5 stars or better as that will strengthen their team. This will not interfer with the other hidden traits of a manager, i.e looking for bargains or spending everything on one player.

This will require an overhaul of the transfer/scouting/agent system, as there are too few transfers in FM now to make a difference regardless of changes to the AI managers.

Your idea would work as well. Using the newgen templates to define roles, all players in the database could be given a designation e.g target man or poacher. All preferred formations would have all eleven positions defined by their designation.

It does have a down side in that its more prescriptive. An AI manager with target man and poacher defined as the roles in his preferred formation would overlook a complete center-forward even if he had significantly higher CA then the TMs and poachers available.

You could get around that by adding a new attribute that governs how willing a manager is to disregard his preferred formation and buy a player who is of a different designation.

Ideally he would then be able to alter his Preferred formation for one that utilizes the new player but that would require a massive re-wright of the manager AI.

I think thats on my wish list for the future. AI Managers who can build teams round exceptional talents even if they have to change the formation and style of play to do it.

Its a step forward because at the present a managers attributes can cause his teams to play in a direct style but if he buys players suited to that style is essentially random. He buys players based only on CA and PA so it is quite possible he ends up playing two small nippy poachers up front while his tactics are direct and focused on getting crosses into the box.

I think your right in your pessimism that my idea is unable to fix the issue with AI managers buying substandard players who are unable to improve the squad. But I think thats happens because its hard to create AI that realistically mimics human mistakes. It would be very easy to write a piece of code that ensured managers only ever bought players with a higher CA then the player already playing in their position, but as managers do make mistakes and do buy donkeys the AI has to as well.

Your idea would work as well. Using the newgen templates to define roles, all players in the database could be given a designation e.g target man or poacher. All preferred formations would have all eleven positions defined by their designation.

It does have a down side in that its more prescriptive. An AI manager with target man and poacher defined as the roles in his preferred formation would overlook a complete center-forward even if he had significantly higher CA then the TMs and poachers available.

You could get around that by adding a new attribute that governs how willing a manager is to disregard his preferred formation and buy a player who is of a different designation.

Ideally he would then be able to alter his Preferred formation for one that utilizes the new player but that would require a massive re-wright of the manager AI.

I think thats on my wish list for the future. AI Managers who can build teams round exceptional talents even if they have to change the formation and style of play to do it.

You're too hung up on CA. A half a star player can put in consistently good performances just as well as any 4.5 star player.. my top scorer last season was rated at 1 star yet was the second highest scorer in the division (he isn't likely to improve either). I agree that the AI squad building is rubbish but it's not all about the CA/PA of the players exclusively, there are other factors to consider.