Jump to content

Suggestions on how to improve AI squad building, transfers and manager skill


Recommended Posts

Firstly I would like to say how much I love the FM series and how many hours of fun (and frustration!:p) they have given me. So thank you to SI for that.

However there are some weaknesses with long term play-throughs, mostly relating to the AI. Instead of just adding to the existing threads complaining about this, I though I would open a new thread with ideas on how the areas of AI squad building, transfers and manager skill could be improved.

Team Building

Currently in FM2011, the AI struggles in the long term when building squads. It sometimes does not have right players for the formation the manager is using, not having a big enough squad, having too many players for a position and sometimes far too big a squad. It often just does not get the balance right at all.

It seems to me that the AI is lacking templates to build its squads too.

For each formation that an AI manager can use, there should be information about how many players for each position are required. There should be min and max values to account for different sized squads. The AI managers should then aim to recruit players to fill out positions which do not meet the min, or positions where most of the players available for it fall below average star rating for that team. So if all the DL for a squad are 2.5 or below, the manager would prioritise recruiting a new one. If the manager is recruiting to improve a position they should then aim to sell or loan out an existing player in that position if there are too many players in that position.

The templates would therefore provide a good guide on squad sizes and help regulate AI managers into having enough players to fill positions and not too many spare. Hopefully no more 7 strikers in a squad for a lone striker formation. The templates would also help when a new manager takes over and changes formation as the template would change and they would be aware of players who do not fit into the template and positions they need to recruit for the formation to work.

Roles to Guide Transfers

Whilst the templates would help the AI recruit the right numbers of players for the positions it has, I think the AI should also use player roles to guide what players it buys for each position. As each role a position already says what attributes are key for a position, the AI should therefore look for players with high stats in those attributes when buying players. Often AI managers buy players far too often based just on CA/PA and not on whther they have the right stats for the formation.

Combining the templates with role based buying should result in much more coherent and stronger AI teams as games go on, making sure the game remains tough throughout instead of getting easier.l

Form also needs to play more of a role with the highest average rated players in leagues more likely to be target for transfers, not just those in the league with big CA/PA values.

Manager Skill

Obviously having all of the above work the same for all managers doesn't provide much variety, so with all it feels like it would be good to have some degree of randomness and manager personality affect what goes on. A manager who favours youth should favour a younger player with slightly weaker stats over a better but much older one. One who is a wheeler dealer should aim for the lowest priced decent player for a position. Poor quality managers should be able to stuff up their transfers, and ambitious ones more willing to break the wage budget and spend more on players just to win.

Just some ideas I have had that I thought I would post as suggestions. Thought they might provoke some interesting discussions on these areas and aim at more constructive criticims with ideas for improvements.

Link to post
Share on other sites

Good post, rather than just complaining you bother to offer a solution. :thup:

Not that I particularly agree that the AI is that weak anyway, I certainly don't hold the view that they are particularly bad squad builders I think it is just that they are somewhat hamstrung. If you or I prefer a 442 but inherit a squad primed to play 41212 with no wingers we would probably adapt our tactic to fit a squad with no wingers rather than shoe-horn a player into wing - I haven't seen an AI manager alter away from their preferred formation which, in situations like this, mean they need to survive in the job long enough to buy enough wingers to make a decent 442.

Don't AI managers pick by role? In my save Delio Rossi has bought outstanding MC's for his 4312, not massively high star rated players but ones whose attributes fit the role perfectly. OL in my save have an absolutely amazing squad with 2 regular strong STRs fighting for the lone striker role with a similarly well-equipped youngster ready to step up when the time comes. Although I will concede that unless a youngster gets lucky enough to find a smaller club where they'll have game time the AI is very poor at developing talent - they make an effort to play them but I've seen many youngsters make only 3 appearances for 3 years and then get binned when they looked like a real good prospect.

My save is about to hit 2019 so I'm only just coming into the 'regen era' where these problems might become more pronounced.

Link to post
Share on other sites

It is definitely more pronounced in the regen era, but problems with too many players for a position can be seen at club like Man City and others before then. I have seen AI managers recruit players based just on CA/PA who don't have the right stats for their formation so they sit in the reserves and get sold again a few years later.

Link to post
Share on other sites

2 things that I believe would go a long way towards helping this issue:

1. Ownership preference- the owner/chairman of the team should have a preferred system, much like coaches. An example would be a club that historically has played with 3 strikers or a sweeper, etc. Clubs would also have a preference for established or up-and-coming coaches. This could even go so far as to accomodate clubs like Chivas, which only fields a team of players that are eligible to play for the Mexican national team. Owners should also influence the type of player that clubs look to sign (elite talent, big names, youngsters, etc.)

2. Manager preference- not just what formation a manager likes to use, it should also include the type of player that he looks for (emphasize size and speed, or mental awareness?) whether or not the coach favors developing talent in-house or signing free agents, etc.

How do to this? Introduce a couple of new rating categories and rate them 1-20, just like everything else. An example would be foreign players: a rating of 1 would indicate that the team never considers signing foreign players, while a rating of 20 would indicate a willingness to sign a player from anywhere (even small nations.) This could create an interesting dynamic where we could see a club chase after a big-name coach, only to see him play a different style of game than what the club is used to (Mourinho in Madrid comes to mind) and potentially causing friction. Or, if coaches could adapt their traits to fit the club they work for, you could see teams promote from within.

Link to post
Share on other sites

Shamelessly bumping this thread because I think it deserves more debate. I like the general gist of the OP's suggestions but I thought most of that was already in the game in some way. I think the key problem is more one of how it works in-game, especially in longer saves when regen's dominate, rather than there being something missing and it's how it works that needs work rather than anything else adding.

I know some people do various things in the various editors to improve the overall quality of the AI and their squads and I'd be interested in just how much of a difference it makes to the game. I'd also like to know how the regen manager's work too. I know that regen players are created from a template and assume that the AI managers are too, if the changes made in editors make a noticeable difference surely SI could incorporate those changes into the manager templates, no?

Link to post
Share on other sites

Shamelessly bumping this thread because I think it deserves more debate. I like the general gist of the OP's suggestions but I thought most of that was already in the game in some way. I think the key problem is more one of how it works in-game, especially in longer saves when regen's dominate, rather than there being something missing and it's how it works that needs work rather than anything else adding.

I believe the game currently does generate a lot of stuff based on existing squads/players/managers but obviously it needs improvement. My suggestions were just a way to end up with something that could be done fairly easily by SI that could help the long term game a lot.

Link to post
Share on other sites

I know some people do various things in the various editors to improve the overall quality of the AI and their squads and I'd be interested in just how much of a difference it makes to the game.

For me i dont know if its been inconclusive, i wonder if changing the managers has had as much of an effect as i think, or if the money i pumped into the clubs made the bigger difference over all.

Link to post
Share on other sites

Yep, AI Transfers need to be overhauled. In my file City have placed the likes of Balotelli, Nasri into the reserves...and bought way rubbisher players to take their positions.

On my game one of the best strikers of the dutch league (hamdaoui) is rotting in the ajax reserves, for almost 1 year now, and even though clubs are making offers for him, he simply refuses to leave, it's broken.

Oh wait, this wasn't my game, this was in real life. This sort of stuff can happen imo.

Link to post
Share on other sites

The debat on what SI can do for the long haul games will go on and on because there people who don't watch te matches have their assistant carry out most of the tasks and just basically keep pressing the continue button, racing through season after season. Then there are players who go into every detail making sure that the club is run the way they want and that it is financialy sound, and a season could take a couple of months to complete.

With that in mind some players may not come across these problems becuase the next installment of the FM empire is released and they move on. And Obviously some players have more time to play the game than others. It's all relative and SI have to weight up all the suggestions that we the customer puts forward and implement what they can in the time frame they have.

My personal opinion is that SI will not improve the AI to the level that the long haul players want because if they did and the game acted in real life with AI managers buying the right players adapting tactics etc. The players that play for the challenge to get low league teams to the top will take longer and therefore put off buying or playing the new FM until they achieve what they set out in the challenge. But like I said thats my personal opinion on the matter, I know there will people who disagree.

Link to post
Share on other sites

Clubs should also take into account whether the players they are buying are better than their current players. On my main save which is 7 seasons in only Arsenal have built a good squad which could challenge for trophies. Chelsea and Man Utd have wasted their money on average players who are not good enough for a Champions League team and are now mid-table teams. Man City have 7 strikers and 6 defenders when they play 4-5-1 so they have many top quality strikers on over £100,000 per week in the reserves.

In Spain, Barcelona and Real Madrid have only bought a few players so their best 11s are mostly the same as they are now, and as they have old squads Atletico Madrid is now the best team in Spain.

I am finding the game quite easy now as I have bought top class newgens but the AI is only interested in really average newgens and many teams in the Premier League have weaker squads than the did at the start of the game.

Link to post
Share on other sites

The debat on what SI can do for the long haul games will go on and on because there people who don't watch te matches have their assistant carry out most of the tasks and just basically keep pressing the continue button, racing through season after season. Then there are players who go into every detail making sure that the club is run the way they want and that it is financialy sound, and a season could take a couple of months to complete.

With that in mind some players may not come across these problems becuase the next installment of the FM empire is released and they move on. And Obviously some players have more time to play the game than others. It's all relative and SI have to weight up all the suggestions that we the customer puts forward and implement what they can in the time frame they have.

My personal opinion is that SI will not improve the AI to the level that the long haul players want because if they did and the game acted in real life with AI managers buying the right players adapting tactics etc. The players that play for the challenge to get low league teams to the top will take longer and therefore put off buying or playing the new FM until they achieve what they set out in the challenge. But like I said thats my personal opinion on the matter, I know there will people who disagree.

1. I play like this, I'm only just into season 5 in my Hereford save and...

2. I've run into Reading F.C. and their TWO, yes 2 senior defenders in the entire squad. I posted the screenshot in another thread...

So if I can come across these problems at the beginning of only season 5 then there is obviously an issue that needs resolution and it really needs to be dealt with as soon as possible IMHO. Surely you agree that 4 seasons of LLM before encountering this problem isn't really enough of a game for £30, is it?

Link to post
Share on other sites

I seriously hope not either. The AI issue is one of the reasons why I keep restarting my career once I've won all I can with the team I am. But the debate about the AI has been going on for a few years now, I'd rather see them get ride of the 3D and improve the AI because I see the improvement in the AI more important than being able to see the match.

Link to post
Share on other sites

I think i have said this in at least 4 different threads, but do you not think its a case of maybe they have done all they can so far with the AI? Its never going to be as smart as a human player, never ever, it will never stand up to the reasoning and forward thinking we put into the game. But who is to say with the level of technology and especially the level of technology available to most FM users that the AI is not at its peak? There are no games to test it against, no other game where you can directly test the smartness of the AI to the extent of FM, who's to say the technology is there to improve it to the point we would all like?

Be assured SI will definitely be aware of the AI failings and will be constantly working on improving both the new features and the AI. But it wont be a simple process, things have improved over the years and will continue too, but removing 3D wont in anyway shape or form improve the AI, or mean that SI will focus all of their attention to the AI. It has to be improved as part of the bigger picture.

Link to post
Share on other sites

I think i have said this in at least 4 different threads, but do you not think its a case of maybe they have done all they can so far with the AI? Its never going to be as smart as a human player, never ever, it will never stand up to the reasoning and forward thinking we put into the game. But who is to say with the level of technology and especially the level of technology available to most FM users that the AI is not at its peak? There are no games to test it against, no other game where you can directly test the smartness of the AI to the extent of FM, who's to say the technology is there to improve it to the point we would all like?

Be assured SI will definitely be aware of the AI failings and will be constantly working on improving both the new features and the AI. But it wont be a simple process, things have improved over the years and will continue too, but removing 3D wont in anyway shape or form improve the AI, or mean that SI will focus all of their attention to the AI. It has to be improved as part of the bigger picture.

Whilst there are obviously limitations to how good AI can be in games at the moments, and it will never be able to truly compete with a good human player in the long run, I definitely feel there is more that could be done to FMs AI to make it more competitive. There are always ways to make improvements to what is already there. Yes they take time and I'm in no doubt this is something SI is always working on. I started this thread as a way of suggesting some things that could be implemented to help take it forward.

Link to post
Share on other sites

oh yeah please dont think that was a holier than you post aimed at the OP, not my intention at all. I definitely hope there is room for improvement, i am one of the people activly editing managers to try and keep my game interesting, i just never buy into this nonsense that they intentionally hold the game back, or that by creating new they ignore things like the AI.

Link to post
Share on other sites

I agree with you milner. But it would be good for SI to inform everyone that IF at the moment they can only get out of the AI is what they are getting and HAS soon as technology does advance to allow for a improvement they will Implement it ASAP. So the debate like this will have an answer. (In an ideal world)

But in the real world of the Industry it finds itself in is that it has to make money so by not saying anything, the speculation of imporvements can lead to sales (I see it has SEGAs doing not SI)

Link to post
Share on other sites

I think i have said this in at least 4 different threads, but do you not think its a case of maybe they have done all they can so far with the AI? Its never going to be as smart as a human player, never ever, it will never stand up to the reasoning and forward thinking we put into the game. But who is to say with the level of technology and especially the level of technology available to most FM users that the AI is not at its peak? There are no games to test it against, no other game where you can directly test the smartness of the AI to the extent of FM, who's to say the technology is there to improve it to the point we would all like?

Be assured SI will definitely be aware of the AI failings and will be constantly working on improving both the new features and the AI. But it wont be a simple process, things have improved over the years and will continue too, but removing 3D wont in anyway shape or form improve the AI, or mean that SI will focus all of their attention to the AI. It has to be improved as part of the bigger picture.

I don't think they're anywhere near that threshold yet. I look at the 3D match (yes it has its flaws) but I can see how talented and passionate those that wrote that code are. Its by far the best match engine in any football sim I've seen, and I've probably seen and played them all at one point or another. So taking that on board, I believe if thats the level of talent SI has on its staff then they're well capable of improving the other game AI as well, that is, AI managers and their squad building ability.

And I don't think there's an agenda to produce a flawed game with no longevity so people continue to buy the next one. I just think the AI has been neglected for so long mainly because the vast majority of FM customers were short-term career players who A.. never knew it existed to the extent it does, and B.. didn't care anyway because it only really effected long term career saves - something they never played. If there's no pressure to improve something it won't get improved. If the masses call for feature A B and C, without realising fundamental aspect D isn't very good, then A B and C will get priority.

But there's been a very noticable paradigm shift over the last 18 months or so to the point where we're seeing more and more people understand the long-term flaws - and we're seeing a lot more in-depth posts about it too. Dare I say even the pathetic "if its too easy, manager a smaller team" excuse not being bandied about as much these days now those that regurgitated it wake up to the real underlying problem :thup: Maybe more people are playing longer games? I certainly think longer saves may become the norm when Manage Anywhere & Anytime is introduced.

Anyway, besides all that, I think SI are well capable of improving it, and I'm sure they do each year to some extent. Hopefully it'll become a top priority for next years game instead of some new feature they had planned? For starters one thing I'd like to see them change is AI managers stop putting so much emphasis on a players CA and instead shift the emphasis on to his position and attributes, and take it from there.

Link to post
Share on other sites

oh i didnt mean talentwise at all, sorry i should have made that clearer, but technology wise, who knows?

C'mon milner, you know better than that. SI haven't yet gone 64-bit, when they do they open up a lot more scope for improvement than they have available to them in 32-bit. However, I don't think they have exhausted that technology yet so there's still masses of room for improvement whether the game stays 32-bit or moves to 64-bit. One thing is for sure, they haven't gone as far as they can technologically just yet..

Link to post
Share on other sites

fair enough i didnt really think about the whole 64 bit thing, but still there is a reason the game is not already 64 bit, with the current level of technology they have been dealing with, is it possible to have a much more resource intensive AI?

I'd say not really, mainly because I've been running an experiment with FM trying to see if it will run on Windows 7 with 4GB of ram and zero swapfile and, although normal everyday stuff flew along quite nicely, once it got to actually playing a match it used up all the available ram and hung... (I maybe should have thought about that before I played the match, didn't save it pre-match and now have to go back a week or so of pre-season :p).

As I understand it, 32bit can only utilise 4GB ram so that's why it stalled and adding more ram would make little difference so creating a more resource intensive AI is probably not possible, improving the current AI should be (I think?). If users can edit improvements in, surely SI can program them in without using that much more resources than it currently uses or so you would think, wouldn't you? Is it really that unfair of us to expect them to improve the AI or even demand it (as some do)?

Link to post
Share on other sites

I am a little ignorant of the issue because of 2 reasons but I have a question... I have only had FM2006 and FM2009 - in FM2006 on my old computer I played 2 saves of 15 seasons each (after that it stopped, maybe because of PC limitations or something?), and 1 game so for of 10 seasons in FM2009 so I haven't gone super deep into a game.

My question is whether anyone knows if the problem in question is exacerbated or lessened by playing a small database or vice versa?

Link to post
Share on other sites

Shamelessly bumping this thread because I think it deserves more debate. I like the general gist of the OP's suggestions but I thought most of that was already in the game in some way. I think the key problem is more one of how it works in-game, especially in longer saves when regen's dominate, rather than there being something missing and it's how it works that needs work rather than anything else adding.

I know some people do various things in the various editors to improve the overall quality of the AI and their squads and I'd be interested in just how much of a difference it makes to the game. I'd also like to know how the regen manager's work too. I know that regen players are created from a template and assume that the AI managers are too, if the changes made in editors make a noticeable difference surely SI could incorporate those changes into the manager templates, no?

Giving all the big clubs in Europe Sugar Daddy surely boosts the amount of transfers.

Whether or not that is something that leads to better squads remains to be seen. The thing is that the AI clubs lack ambition. Whatever we do, we do it for one reason only; complete world domination! That's why we use the entire transfer budget every season until we have reached that goal. In addition to the budget we are given by the board, we increase it by getting rid of players we don't want. The AI don't do that. Just look at the coach advice every month - if you have a player with decent PA or Reputation transfer listed, they suggest that you remove him from the transfer list because he has something to offer to the club. This means that the AI wouldn't have transfer listed him, and therein lies the problem! The footballing world is a tough world, no place for sissies - a player that is not good enough will have to go... but in FM they stay until their contracts run out or they retire.

As the OP said (which is mostly the same I have said earlier many times), the AI doesn't seem to have templates with which to build squads. I agree with him that the AI managers need to look at key attributes in a more conscious fashion, but I would go farther than that: I would like -all- the attributes to be connected to player types/roles, not just the key ones.

Link to post
Share on other sites

As I've said so many times already, AI staff shouldn't just buy/sell players basing their decision on PA and reputation...

The trend looks like:

Players 16-23 are bought based PURELY on their PA, almost regardless of their actual current ability, attributes and "workability"

Players 24-27 are still bought because of their PA, with an eye on their reputation too... Usually they're decent but too often it's clear they're going nowhere

Players 28-31 are all about reputation... AI clubs will gladly sign players who have already peaked for silly fees, offering insane wages too

Needless to say, even the best AI managers will waste valuable money on 19yo kids with awful attributes just because they're 175PA (ignoring the current 110CA and the <10 key attributes) and/or a plethora of Average Joes who won't get any better... Even the odd Top Player will be a guy with 1-2 good seasons left in him, thus being a huge loss.

Nothing wrong with taking chances with younger players.. but Top Clubs should have HIGHER STANDARDS... Instead they'll just bring in ANYONE whose PA is high enough...

I can even live with them overlooking macroscopic flaws (like 5 workrate, 6 stamina, strikers with <10 attacking skills, horrible mental traits etc...) because if they see "potential" in a kid, it's fair enough they're willing to take a chance.

Yet again I refuse to accept the fact Barcelona or Arsenal would bring in half a dozen of CLEARLY UNWORKABLE players just because they have >150PA

Just like your scout will give a 1* rating to an average lower league player, the same should happen for all those "could-have-beens" who are 19-21 and still haven't made the transition from "great potential" to "actual ability"

E.g. there's NO REASON for Real Madrid to sign this guy here... yet they did (for 9M) and they had to get rid of him later on for a fraction of that fee... Which manager shouldn't already see he wasn't Bernabeu-material?!

mediocre.jpg

Uploaded with ImageShack.us

The thing is: Top-level managers should have more selective patterns when looking to sign new players... Not just a PA/Reputation standard, but also "ATTRIBUTE CRITERIA" to fulfill...

Say, Barcelona shouldn't even consider a player with <15 technical skills for an attacking role... Top EPL teams shouldn't give a second look at a CB with <15 defensive/physical skills...

And generally speaking, any player above 18 with <10 mental traits should get an automatic thumb down at the highest level, unless he's has Maradona's feet...

And such "pattern" should apply, toned down accordingly, to every rung of the football ladder...

Link to post
Share on other sites

I am a little ignorant of the issue because of 2 reasons but I have a question... I have only had FM2006 and FM2009 - in FM2006 on my old computer I played 2 saves of 15 seasons each (after that it stopped, maybe because of PC limitations or something?), and 1 game so for of 10 seasons in FM2009 so I haven't gone super deep into a game.

My question is whether anyone knows if the problem in question is exacerbated or lessened by playing a small database or vice versa?

If I had to guess, the problem would probably be worse with a smaller database since there would be even less talented players avaialble. But wether you have a massive database loaded and all the leagues too, it doesn't make the problem any better, simply because its the AI transfer logic and young player development that is flawed - the amount of good players available makes little difference imho.

RBKalle's post #28 sums it up. I've done a few tests myself trying to figure out what goes through an AI managers mind when he's transfer dealing and team building...and its not much :D, at its basic level all he's 'thinking' about is PA, and he has little or no thinking in regards to "do I really need to buy this player, and if so, how will I ustilise him in my system?". AI also put far too much emphasis on CA, even to the point were it means buying players whose attributes are woeful compared to the other players the AI manager already has in his squad.

Link to post
Share on other sites

Big problem im finding is that the rich clubs waste huge amounts on old players, like Man City just bought my turkish regen DC for like 30m even tho he is 31 soon to be 32

i know its all relative since he was prob the best in the game in his prime and was signed (before i joined the club) age 26 for 62m club record fee but still there is no value buying such a player, tbh i was looking to move him anyway and had i not got an offer i would have let him go for half that money

Edit - he is also being paid 205k per week and i dread to see his agent and signing fees, judging by my teams signing i would guess 10m+

Link to post
Share on other sites

Allowing the AI to asses players skills beyond PA/CA should be the top priority for SI.

A nice easy solution would be bundles of attributes.

I'll give an example of how it could work

The AI looks for a striker ( Target Man, this would only work if the AI can differentiate between different player roles in its formations )using CA\Reputation obscured by the scouts ability. Compiles a short list then checks a bundle of Target man attributes against a minimum standard set by the managers ability and the level the team is in.

Say the AI sets a minimum value of 78 for the target man role bundle.

It looks at each players Strength, Jumping, Heading, Bravery, Aggression, Work-rate totals the values up and if they come to less then 78 discards the player from the shortlist. If the value is greater he stays in a short list which is now sorted by the value of the players bundles.

it could be streamlined even more by giving ever player in the Database a field for role and a field for role bundle value.

Link to post
Share on other sites

Allowing the AI to asses players skills beyond PA/CA should be the top priority for SI.

A nice easy solution would be bundles of attributes.

I'll give an example of how it could work

The AI looks for a striker ( Target Man, this would only work if the AI can differentiate between different player roles in its formations )using CA\Reputation obscured by the scouts ability. Compiles a short list then checks a bundle of Target man attributes against a minimum standard set by the managers ability and the level the team is in.

Say the AI sets a minimum value of 78 for the target man role bundle.

It looks at each players Strength, Jumping, Heading, Bravery, Aggression, Work-rate totals the values up and if they come to less then 78 discards the player from the shortlist. If the value is greater he stays in a short list which is now sorted by the value of the players bundles.

it could be streamlined even more by giving ever player in the Database a field for role and a field for role bundle value.

It would be easier (from a non-programmer's point of view at least) to just extend the attribute requirements in the TC to -all- attributes, tie these to a much more sophisticated Player > Personal screen with player descriptions indicating what type of player they are.

So that this player:

guerral.png

Uploaded with ImageShack.us

guerrapersonal.png

Uploaded with ImageShack.us

... Has a Description saying "Vito Antonio Guerra is a world-class athletic, explosive classical poacher with a winner's attitude" or something. The attributes underpinning tht description should be available to the AI, since they can't really read.

Link to post
Share on other sites

For starters it would really be rad if manager's tendencies and skills played a major role in their signings...

Ie. a manager with 20 Defending and 5-4-1 as favourite tactic and Cautious approach should generally look for tough defenders (fullbacks included) and for hard-working-type midfielders

Or an adventurous manager fond of 3-4-3 should privilege pacey forwards and offensive backs/wingers

But that has more with Squad Building to do, not necessarily linked with the key aspect of the issue, ie. AI clubs signing crappy youngsters, mediocre filler players and aging expensive stars...

About the latter, the thing could be reworked rather easily (I guess)

IIRC every position has a different attribute weighting pattern, so the same principle could be used to rework the AI's transfer policy...

Just like your scout will tell you that 20yo striker "would be a decent Championship player" and "could be a decent EPL striker in the future", the AI staff should learn to read between the lines and realize the current attiributes (eg. 11 Finishing, 7 Composure and various <10 physical and mental attributes) wouldn't be a good enough platform for an eventual move to EPL top-half clubs...

It doesn't really matter if the CA is 110 or the PA is 155... what should count are the key attributes for his position, then the secondary attributes, then and just then the PA...

I don't think it'd be that difficult to implement... as suggested by whilewolf, some sort of "minimum attribute value" could work just fine... Basically instead of just telling the AI "if he has CA and/or PA >N, sign him" it could be "if the technical/physical/mental attributes are >N, sign him"

P.S. that would require a little overhaul of the newgen patterns... Enough already with DMs with low Bravery and Aggression, or with MCs who are lightning-quick but can't trap/pass a ball to save their life...

Link to post
Share on other sites

There doesn't seem to be any planning of sorts really - some screenshots like the Reading one suggest that the AI is unable to "take a step back" and look at the squad as a whole. It is unable to identify weaknesses but is able to identify good players, although not necessarily players that fix those weaknesses.

It doesn't seem to think about the future, either - fielding a youngster for short-term pain but long-term gain, for example. Teams invest in youngsters but let that investment rot. It can spot talented but flawed youngsters, but doesn't notice the flaws and doesn't fix them through tutoring.

I think the game will need something like:

- Views on players - The ability to see past CA and actually evaluate a player.

- Views on squads - What's the current weaknesses and strengths of my squad? Where do I need to invest, or where do I need to cut fat from? Can I promote a youngster instead of signing a new player?

- Manager squad style - i.e. Big Sam favours big defenders, whilst Wenger favours technical ones.

- Views on players in the future - Where is this player going in the future? Am I signing a good player with little upside for development, or a youngster with lots of upside? What do I need to do to get this ideal development? Is it likely, or going to need lots of work? Is it worth it? Will I lose the player anyway?

- Views on squads in the future - Basically the same, but for squads as a whole. Am I going to get 5-6 of my players retiring all at once, meaning although my squad is awesome, I need to smoothen the blow?

- Having a set of goals - A manager should have a set of goals, possibly after discussion with the board, with the idea of a vision for the team in the short-, medium- and long-term, based on where the club is now, and where they want to be in the future. The AI then needs to identify what needs to be done to achieve these goals, and actually go out and do it. It needs to take into account risks and options (i.e. it is very risky to sign 11 16-year-old wonderkids and field them as your first-team - surely signing 2 a year is just as likely to achieve a young squad in 5-6 years time?). It needs to constantly evaluate these objectives and identify what needs to be done.

The last one is probably the big one - if implemented properly, it will feel like there is method to the madness, especially if the in-game media catches on (i.e. "Clearly, Wenger has gone for a short-term measure here with the signings of Benayoun, Arteta, Park and Mertesacker - dare we say, very un-Wenger like indeed!").

Link to post
Share on other sites

There doesn't seem to be any planning of sorts really - some screenshots like the Reading one suggest that the AI is unable to "take a step back" and look at the squad as a whole. It is unable to identify weaknesses but is able to identify good players, although not necessarily players that fix those weaknesses.

It doesn't seem to think about the future, either - fielding a youngster for short-term pain but long-term gain, for example. Teams invest in youngsters but let that investment rot. It can spot talented but flawed youngsters, but doesn't notice the flaws and doesn't fix them through tutoring.

I think the game will need something like:

- Views on players - The ability to see past CA and actually evaluate a player.

- Views on squads - What's the current weaknesses and strengths of my squad? Where do I need to invest, or where do I need to cut fat from? Can I promote a youngster instead of signing a new player?

- Manager squad style - i.e. Big Sam favours big defenders, whilst Wenger favours technical ones.

- Views on players in the future - Where is this player going in the future? Am I signing a good player with little upside for development, or a youngster with lots of upside? What do I need to do to get this ideal development? Is it likely, or going to need lots of work? Is it worth it? Will I lose the player anyway?

- Views on squads in the future - Basically the same, but for squads as a whole. Am I going to get 5-6 of my players retiring all at once, meaning although my squad is awesome, I need to smoothen the blow?

- Having a set of goals - A manager should have a set of goals, possibly after discussion with the board, with the idea of a vision for the team in the short-, medium- and long-term, based on where the club is now, and where they want to be in the future. The AI then needs to identify what needs to be done to achieve these goals, and actually go out and do it. It needs to take into account risks and options (i.e. it is very risky to sign 11 16-year-old wonderkids and field them as your first-team - surely signing 2 a year is just as likely to achieve a young squad in 5-6 years time?). It needs to constantly evaluate these objectives and identify what needs to be done.

The last one is probably the big one - if implemented properly, it will feel like there is method to the madness, especially if the in-game media catches on (i.e. "Clearly, Wenger has gone for a short-term measure here with the signings of Benayoun, Arteta, Park and Mertesacker - dare we say, very un-Wenger like indeed!").

For once I agree completely with x42 :) The way the AI works needs a complete overhaul, IMHO, and something like this idea would be a massive improvement (providing the benefits outweigh the obvious bugs that will occur...).

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...