AI Buying Philosophy. Is it a bug?

Currently in my third season as Cardiff City and in the Premiership, coming to the end of the season and i have noticed that most of the teams have the same players that they had at the start of the game. In three years time, most of the Premiership teams, as well as teams in other divisions will have at least 6/7/8 new players. Is it a bug? It's not very realistic and makes the game a little boring.

Clubs buying new players? That's Madness!

I think he means clubs aren't buying new players

Maybe i did not phrase my point. I will start again.

Why after three seasons in the game do most clubs have the exact same players in the exact same positions as at the start of the game?

Result of poor game AI. They don't/won't buy players for the positions they actually need, and rarely are any bids for the big-time players made. It's why Utd never replace Carrick, Berbatov and don't give youngsters like Wellbeck a chance. A team like City, for example, will 5 or 6 AMRLC's, but neglect central defence.

It's an illogical and poor AI system, that's all.

That is one of the biggest problems with longterm games I think, AI teams hang on to their old players for too long and don't build for the future. Young prospects that are still developing and don't have a first team spot can be bought for peanuts. It is way too easy to find promising players that don't have real contracts or precontracts that you can sign for a million or two in compensation. There seems to be absolutly no coding in place to make the AI consider stuff as home grown status or foreign player limits, in my save 20+ seasons in there was less than a handful players with italian home grown status that was good enough to even consider for a champions league squad in the entire game, when I took over Sampdoria they had 0 home grown players in the squad that had been playing in europe for a couple of seasons.

The poor A.I really ruins the game long term imo, it makes it much too easy.

Result of poor game AI. They don't/won't buy players for the positions they actually need, and rarely are any bids for the big-time players made. It's why Utd never replace Carrick, Berbatov and don't give youngsters like Wellbeck a chance. A team like City, for example, will 5 or 6 AMRLC's, but neglect central defence.

It's an illogical and poor AI system, that's all.

So basically, the AI is now Arsene Wenger.

No because it won't buy promising youngsters either.

No because it won't buy promising youngsters either.

They buy promising youngsters, they just don't know how to make them into players they can use...

They buy promising youngsters, they just don't know how to make them into players they can use...

They buy them and then either sell them onwards for peanuts or fail to develop them.

If someone has a long-term save going they should post some screen shots of their current AI-managed Champions League holders, AI-managed Premiership, La Liga, and Seire A championship squads to show just how poor they become compared to how they start out in season 2010/11.

Maybe then SI would take note of how flawed the core of this game really is.

I've thought about doing it myself - including a massive write up with stats and graphs showing just how weak the entire gameworld becomes with each passing season, but to be honest I really don't have the energy anymore because I know SI won't bother looking, and if they did they still probably wouldn't comment on it. They never do when it comes to AI team building and transfer policy.

Its seems its always destined to be a game for short-term players. Play 5 or 6 seasons, quit and start a new game... rinse and repeat. I'm hoping this new feature manage anywhere will encourage more players to keep playing their saves longer, then more people will notice the flaw and complain about it.

In fact, thinking about it, manage anywhere may be the biggest own-goal SI have scored - they may have to finally fix a crucial problem thats been lingering for years - fingers crossed

So basically, the AI is now Arsene Wenger.

hahahaha so true.

Can always upload a screenshot of the Premiere league winning Man City.

Actually the clubs want to buy new players. The problem was that this FM version AI has (what i call) a bug. 99% of the offers made to clubs are 2x the presented value (at maximum) and, since most of the players are valued by there clubs for a value superior to the 2x value, no club can buy new players since all their offers are rejected.

Cheers mimland.

Those 20 players have a combined net worth of around £132M.

Man City in season 2010/11 have FIVE players worth almost that.

I know it should be judged on ablity (attributes etc) and not just transfer value, but still, it gives an indication of how strong the players and clubs are at the beginning of the game to how much weaker they get the longer you play.

Good thread this. I will also make a contribution

I've already made some threads like this before:

http://community.sigames.com/showthread.php/253299-Why-clubs-never-spending-big

http://community.sigames.com/showthread.php/254086-Is-there-going-to-be-something-done-about-the-transfer-system-in-patch-11.3

http://community.sigames.com/showthread.php/248738-11.2.1-Can-their-please-be-something-done-about-this-transfer-system

Unfortunately, no usefull response from SI. It gives me the idea, how much I respect them, that they just don't care about it. And stick to their own, in my opinion unrealistic, view of the AI-transfer system.

I now almost got rid of this by just looking at a few save games and looking at the standard "crappy-players" that topclubs wil buy. Names as Nasief Morris, Fernando Baiano, Sneijad Ibricic, John Carew, Abou-Trika etc. Who are in real life all "crappy" (at least no player material for top teams) players, but in FM end up at the top clubs, and just lower their CA/PA with the FM-editor. This kind off "solves" the issue, but of course the top teams will still find somewhere some unneeded player to buy.

The whole problem behind this AI-buying philosophy:

After the economic crisis in Europe, SI thought last year that this would have a major impact on the world of football. Due to this, they insert "something" in the game, that in the first years (except for teams who have a sugardaddy) wil not buy big names for much money. This causes the whole unrealistic transfer thing, and yes, of course, if you're compare the big teams after five or ten years you will see that they've lost quality compared to the start of the game.

Small example of the problem:

If I start a game with lets say Napoli, and I will offer one of my most crappy players (lets say Aronica) to other clubs I get interest from Man Utd, Arsenal, Liverpool, Chelsea, Juventus, Milan, Fiorentina, Tottenham, Aston Villa and Sevilla. Same with Heskey at Villa if I would offer him to clubs, and same with even the likes of Arouna Koné if I start a game at Sevilla. Top teams will buy unneeded players from smaller clubs, and those smaller teams will replace them with unneeded players of even smaller teams! Its the football world upside down.

How to solve this problem?:

Again, as much as I respect SI, they're premise of the economic late-2000s financial crisis having an impact on the football world, is just wrong. There are coming more and more clubs who get took over by large investment groups or some rich Sheikh from the Middle-East who's uses the real life world of football as his personal version of Football Manager. And look at Aston Villa how buys a Darrent Bent for a stunning 25 million pounds, a Man Utd who is willing to pay 22 million for Jones, a Atletico who is prepared to pay 38 million (!) for Falcao, a Juventus who splend 15 million on Matri, a Roma who buys Lamela for 17 million, a Napoli who buys a Inler for 15 million, a Henderson who moves for almost 20 million to Liverpool, should I give more examples?

Next to that, if clubs don't sign players (due to financial problems or not), they literally don't buy players for their first team, look at the Chelsea at moment (leaving their possible comming transfer action to one side for this moment). But they not going to start buy a handful of unneeded crappy players who are in no way a boost to the teams strength.

SI needs to get this "flaw" out of the game and let clubs spend money again. And most of all, look at what players (top) teams spend their money! I think it would be easy to say for example that clubs with a reputation of 6800 - 9999 will not buy any players who are older than 20 and have a PA lower than 145 or 150. And especially with players who are older than 25, so mainly will not progress much more in the game, for them there should be set a minimum CA of 150/155 before a top team will even show interest in such a player.

Concluded: In my opinion this problem would be solved if the reputation of clubs will be connected to the players CA/PA.

The young player buying madness issue:

A thing you will also see in the game, is that a lot of teams, again, mainly the top teams, buy a lot of young players who will never make it to the top. This is not even that unrealistic, but what makes it unrealistic, is that many of those "young" players, are not that young anymore. The "young" players that top teams buy are usually in the age of 17 - 23. Where I would say that in real life most top teams, chase youngsters of 14-17.

And again, also here, is maybe a more specific look at the players PA needed, how older the player, how more AI-clubs should look at the players PA. Otherwise they will buy a lot of "young" 21-year old players with a CA of 106 and a PA of 135, and this does happen a lot now in Football Manager 2011. For this I would say that if a top teams buys a player of under 17 they should be able to make some mistakes cause at that point in the young players development, cause its difficult to see if a player of that age will make it to the top. But if a top team buys a player of 18 years or older, they should have at least a PA of 150. Of course, exceptions on this rule can be there, and the clubs need to have the freedom to buy a total misbuy. But lets say that 70% of the players who are 18 years or older and get signed by a top team should have a PA of 150 or higher.

I think its also the AI's over-simplifcation of how it looks at transfer targets. For example, I created 2 players in a test database, gave 1 player 20 attribute points in every department - effectively making him the highest skilled player in the entire game. I gave him 150 PA and 200 CA.

I duplicated the player and reduced the duplicates skills to 10. Basically making this 2nd player half as good as the first player. But I gave him 200 PA and 200 CA.

Checking with FMRTE to see how the differing PA would round down the attributes, they became approx. 18 to 19 across the board for player A, and 13 to 14 across the board for player B. [The difference between an all 18 attribute player and an all 13 attribute player is pretty much the differnece between signing a Lionel Messi and signing an average championship player!]

So...

Player A - average of 18 attributes in all departments. 150 PA.

Player B - average of 13 attributes in all departments. 200 PA.

I started a new game and put both players up for sale for £0. Player B attracted 11 bids, including one from Man Utd. Player A attracted just 2 bids (both from average Russian clubs). I reloaded and repeated this several times - always the higher PA player attracted the most bids and bids from bigger reputation clubs too.

So in effect, Man Utd and all those other clubs that placed bids ignored a "Messi" in favour of a "Barry Robson" just because Barry Robson had a higher PA

Ok this was a very simple and basic test, and other factors are no doubt involved, but most people already know the AI bases its transfers too heavily on PA. And this streamlined example gives an insight of how silly the AI transfer system is.

AI clubs should always choose who they want to buy based on how good (attribute-wise) a player is. The reason why they buy some players they never really need and often never even use is because they buy them based soley on their PA and not on their "skill", or how they'll fit into their squad.

It seems PA is too big an influence when AI clubs target players. Couple that with their poor dealings with young players - failing to blood them or train them efficiently, often letting them rot, and add the poor quality of regens too then its no wonder why AI clubs weaken season after season - leaving the human controlled club to eventually dominate.

I'd like to see an AI managers attributes have more effect when he's transfer dealing.

For example, the better the AI manager is, the more accurate his assessment of a taget players skills are. That way we'd see the best managers in the game bidding for the best players, not the best PA's !!! And poor managers would make poor descions and buy players that just don't cut it or players they don't need or use (exactly like it is now in the game!!).

Cheers mimland.

Those 20 players have a combined net worth of around £132M.

Man City in season 2010/11 have FIVE players worth almost that.

I know it should be judged on ablity (attributes etc) and not just transfer value, but still, it gives an indication of how strong the players and clubs are at the beginning of the game to how much weaker they get the longer you play.

One of the things that make me even more bothered with this is that the prices on players have skyrocketed over the past years. To buy a good quality player that is actually playing first team football it is not unusal to see bids at 50 million. So for them to have a squad value at 130 mill is far from good.

Tbh what ****es me off most about AI behaviour is the sheer lack of squad building and planning. It is pretty simple. The below listing is a raw simplification of the mindwork I put in.

FM 2012 AI manager license (read carefully and put into practice!)

1. Choose a primary formation

2. Choose a primary playing style

3. Assess the weak parts of the squad (where do you need more quality? do you need more quality to achieve your goals?)

4. Assess the age structure of the squad (if too young, look for experience / if too old, look for youth)

5. Assess the club finances and possible targets, for how many players should the transfer budget be sufficient?

6. Does your team want immident success or are you building up a stronger squad for coming years?

7. Find possible targets based on

- player position and role (see point 1. - does the club need another wing-back?)

- availability (see point 5. - can the club afford the player?)

- current ability and potential ability (see points 4. and 6. - more weight on PA if the squad needs youth and is only starting to build a squad and vice versa)

- strengths and weaknesses (see point 2. - does he fit your style?)

- personality (would he have difficulties blending in?)

In real life teams and their fans recognize what the team needs to improve. Lack of a creative midfielder, a clinical striker or a goalkeeper up to standard won't go unnoticed. I'm not asking for the world here, am I?

Apart from the problems that are already said in this thread, I also find the nationality a bit of a problem.

About 20-30 seasons in Belgium, almost every team has 2-5 Belgians in their first team (and I don't mean the first eleven). Even though at some point when I became manager of Genk, I couldn't pick 18 players for competition due to nationality rules.

Its pretty commond knowledge that the AI is flat terrible in squad building and newgen development. Which makes me think about FM12...

So they update some of the contract workings and such. Great, all fine and good, but it's all pointless if the AI is still terrible in these areas in terms of long terms playing.

While it was simple I think Erimus1876's test was quite telling. In fact I think the simplicity is what shows it's weight most of all. I mean PA should only matter at all to players that are in their teens. Not just CA but points in the PROPER skills ofr the players position should be the primary area the AI looks at first.

Why the AI can't figure out what it needs for the team confuses the hell out of me though. Like for me an asst man usually doesn't have great advice, or give good recommendations for players (in most cases) but it generally is at least very acurate with my teams weakest position. Why can't the AI use that same calculation, then all SI need to fix is the AI determining what a good player and a poor player is.

Will just have to see if FM12's AI is any better. But I don't expect much considering it wasn't good in 10 or 11.

Will just have to see if FM12's AI is any better. But I don't expect much considering it wasn't good in 10 or 11.

I don't think they've even looked at it. If they had you can be sure it would have been mentioned in the interviews and hype they start spinning this time of year. Its such a crucial part of the game, fixing and improving it should make headline news. It hasn't, so I don't hold out much hope.

I'll wait until patch 3 in Feb/March before deciding wether to get this version.... and only then if they've improved the AI....

Is this linked to the fact there are a lot of spare DMCs - some free agents and others on the transfer list or being offered to clubs. It is this position more than any other - that clubs managers dont' have the smarts to either play tactics with decent DMCs.

Also I'm in the end of my third season and Ramsay and Wilkshire can't get a game for Arsenal. I'd put in a cheeky bid if it wasn't an area I'm already strong in.

Editing the most well-known managers to have Squad Rotation 20 helps. They don't know -when- to rotate, though.

Si really need to sort out the AI with regards to their squads.

You have a manager, the manager has their playing style.

That's what the squad should be based around.

With increased squad size restrictions these glaring errors creep further into the game.

In places like the A-League I've seen squads end up with 1 keeper and 9 midfielders.

Some beautiful posts in this thread, specifically by LeandroDavide- What a well reasoned and eloquent post

Is there anything that can be done in the editor to go some way to rectifying this issue? I know AI is a coding thing, but If we could somehow get managers to buy proper players and the like....

Some beautiful posts in this thread, specifically by LeandroDavide- What a well reasoned and eloquent post

Is there anything that can be done in the editor to go some way to rectifying this issue? I know AI is a coding thing, but If we could somehow get managers to buy proper players and the like....

Only solution i've found so far and its nots 100% better but dose work slightly is play with transfer budget off 1st window

now almost got rid of this by just looking at a few save games and looking at the standard "crappy-players" that topclubs wil buy. Names as Nasief Morris, Fernando Baiano, Sneijad Ibricic, John Carew, Abou-Trika etc. Who are in real life all "crappy" (at least no player material for top teams) players, but in FM end up at the top clubs, and just lower their CA/PA with the FM-editor. This kind off "solves" the issue, but of course the top teams will still find somewhere some unneeded player to buy.

But that's true to life... they're generally signed as back-ups.

Some examples... Martin Fulop to Man City. Ross Turnbull to Chelsea. Paul Konchesky to Liverpool. Kevin Doyle rumoured to be joining Liverpool or Arsenal soon....

But that's true to life... they're generally signed as back-ups.

Some examples... Martin Fulop to Man City. Ross Turnbull to Chelsea. Paul Konchesky to Liverpool.

Two of those are back up keepers, who are all complete gash anyway. Konchesky did a job for Roy at Fulham and so it is understandable. Carew to a 'top club'(I presume a top 4 club), will never, ever happen. My issue is not so much who the clubs sign(though invariably s**t), it's who they don't sign. There is zero aggression in the market, apart from City. They will buy 6 Amrlc's however and neglect CD.

Why are the likes of Pastore, Ganso, Neymar etc never bought by clubs? I've haven't seen that happen once since 11.3 came out. Then we have Hazard rejecting both Barca and Real due to pay demands, lol!

Utd don't buy, Barca don't buy, Chelsea don't buy, Tottenham NEVER buy(apart from some god awful free transfers), and it goes on and on..

Kevin Doyle rumoured to be joining Liverpool or Arsenal soon....

Emphasis on rumour. Though one could envision 'Pool bringing Doyle in as some sort of back-up.

That is 1 of the reason why i'm putting off buying FM12 until the 3rd patch. Not going to make the same mistake.

Anyway, its impossible for the AI to code them like real life managers. I think SI should look into making the AI follow basic squad building.

SI should code the AI in a way that AI finds squad balance. Meaning if they only have 1 right-back with poor CA, they will prioritize right-back during the transfer window.

So if the club has a reasonable amount in the bank, they might pay over the odds for a right-back on the last day, if they can't find a right-back by then.

Just for a start they should code the AI in that way. Because at the moment, the AI is quite terrible in squad building and its just frustrating to watch.

Why are the likes of Pastore, Ganso, Neymar etc never bought by clubs? I've haven't seen that happen once since 11.3 came out.

Odd. Chelsea buy Pastore in roughly 90% of my 11.3 savegames.

Neymar played for Barca for most of his career in my game, Pastore at Man u and Ganso at AC Milan all signed in within the first 3 seasons.

The AI signed a England u19 left back for Fulham from Matlock (the side I was) during the game and that was a good 6 years in but then they did loan him back to me and after that let him rot in the reserves so it could do with a slight tune up.

Just decided to take a look at Chelsea in my game and they don't have a single player worth 10 mill and only one player valued over 5 mill.

Just decided to take a look at Chelsea in my game and they don't have a single player worth 10 mill and only one player valued over 5 mill.

They have lost reputation then.

im not going to stick up for the AI squad building it is poor, but looking at values of players does not give any indication really as to where a club is. As Biggus says reps can go down, which in turn means they sign poorer players, comparing any squad 10 years into the game to the starting squads is almost completely pointless unless all of the clubs reps have stayed almost exactly the same through those 10 years.

Yes I just find it interesting that a club that big would drop that far, they are just and just considered a mid table team at this point. I know values are a bad thing to look at but considering the inflation when it comes to the cost of a decent first team player looking at the values gives a rather good indication to wheter they have signed anything of quality recently. Transfer fees over 20 and 30 million isn't all that uncommon anymore. I do support that some teams move up and some teams move down in the world but with decent transfer policies the big teams wouldn't make total crashdives.

There is no inflation in FM that im aware of? Rooney would cost the same at the start of the game as he would 30 years in. Again im not fully sticking up for the AI, it does need a lot of work, but look at Arsenal, three years ago you have put them as one of the main title contenders, now you say they will do well to break the top 6, who's to say if they sack Wenger or he leaves that it could get worse for them, before long they could end up in the same position Chelsea are in that game.

See Miles reply on Twitter :

Disappointing !

See the reply of Miles on Twitter :

gregianjohnson Greg Johnson

@milessi with the AI's ability to build teams & squads in the short/long term can it analyse strengths/weaknesses like assman 'team reports'

50 minutes ago Favorite Retweet Reply in reply to ↑

@milessi

Miles Jacobson @gregianjohnson the AI has the same tools that human managers do. Always has had.

whats new with that, we always knew that was the case, its just that the AI is not and will never be as smart as we are, all we can hope is that they can improve the AI intelligence.

whats new with that, we always knew that was the case

whats new with that, we always knew that was the case, its just that the AI is not and will never be as smart as we are, all we can hope is that they can improve the AI intelligence.

Still, if I compare the buying of the AI-teams in FM 11 to other versions, its never been this bad.

go back, compare the squad building in FM08 or 09, especially considering the issues with regens in those games it would be very difficult to say it has gotten worse, it hasnt at all, as Neil says in the other thread, they have improved this area, given the AI more tools than they had before, it will always be an ongoing problem and the AI will NEVER come close to matching the human player, not with avaliable technology anyway.

That's the point! Miles has basically dodged the question, suggesting that there is nothing new. @gregianjohnson was looking for something new in FM12 - and it doesn't look like it's there.

you should know better, Miles will not reveal anything on twitter that isnt already public knowlege, it would be a very very silly thing to do. All he has done is pointed out the obvious warrented, but what more would you exepct? He doesnt know the people tweeting them and he certainly doesnt have to reveal any FM secrets to them.

That's the point! Miles has basically dodged the question, suggesting that there is nothing new. @gregianjohnson was looking for something new in FM12 - and it doesn't look like it's there.

@gregianjohnson is my twitter name and i wasn't necessarily asking about AI development for FM2012 but more whether we'd be seeing any developments on that front in the near future. The stuff said in and around these forums at times about SI deliberately freezing development to sell more product and expectations for quick fixes as if AI is an easy area to push on with are crazy.

you should know better, Miles will not reveal anything on twitter that isnt already public knowlege, it would be a very very silly thing to do. All he has done is pointed out the obvious warrented, but what more would you exepct? He doesnt know the people tweeting them and he certainly doesnt have to reveal any FM secrets to them.

Unless, of course, they are not, hence dodging the question.

Dodging the question never suggests a good thing.

who's to say it wont come in a later annoucement? Si never reveal every update in one go. Dodging the question on a public social network site tells me nothing at all i wouldnt expect any info from that source that i havent already read in an offical annoucement.

who's to say it wont come in a later annoucement? Si never reveal every update in one go. Dodging the question on a public social network site tells me nothing at all i wouldnt expect any info from that source that i havent already read in an offical annoucement.

The "Always has had" bit suggests that "there never was a problem in the first place".

Miles is suggesting that the AI always had the tools - so why don't they use them in, say, FM11? We know they didn't, because the AI "stockpiles".

that could as much be down to my phrasing and twitter's 140 char limit though.

Miles should be given the benefit of the doubt, especially on Twitter.