If A.I doesn't improve, then FM will fail.

[kingking]

we need to improve the A.I in terms of

• Squad management
• Squad rotation
• Tactical adaptability, tactical risk taking etc
• Squad registration
• Transfers 

I feel like the A.I can improved if the developers try.

Edited October 19, 2020 by kingking

[forameuss]

Counter-point, it's easy to see a situation where if the AI improves too much, FM will also fail.

Not that I'm saying it shouldn't improve, as it should, but it's an interesting balancing act FM will always have to perform.

[Mars_Blackmon]

9 minutes ago, kingking said:

we need to improve the A.I in terms of

• Squad management
• Squad rotation
• Tactical adaptability, tactical risk taking etc
• Squad registration
• Transfers 

I feel like the A.I can improved if the developers try.

Hot take.

But I’ll engage. How would you improve the AI. Specifically what would you like to see done in every area you have listed?

Give examples on how none of the things listed aren’t optimal. 
 

That conversation would be more helpful to the game.

[XaW]

22 minutes ago, forameuss said:

Counter-point, it's easy to see a situation where if the AI improves too much, FM will also fail.

Not that I'm saying it shouldn't improve, as it should, but it's an interesting balancing act FM will always have to perform.

Exactly, if the AI is good enough to counter player setups, then I can imagine the rage in here...

[Wakers]

6 minutes ago, XaW said:

Exactly, if the AI is good enough to counter player setups, then I can imagine the rage in here...

In a sensible world that ability would be linked to manager stats.

 

So you might expect to get out maneuvered by Jose but not Pardew.

[Pasonen]

One step to a more realistic and challenging direction for player would be what kind of players AI managers sign and use. Depending what's the preferred style of the manager. Sitting deep and counter -> slow but tall dc's, hardworking cm's, quick st's etc. Playing flanks -> tall st's, good crossers and dribblers.

[Wakers]

Honestly it feels like SI have forgotten that managers have stats. The only thing that matters at all to the AI is reputation, which determines which jobs they end up in. Nothing else I've seen in the game bears any relation to their stats. Even preferred formations they tend not to use rather than use.

[Neil Brock]

We're always working on improving the AI, every single version I've ever worked on (well over 10 years) has had improvements in these areas.

Adding more calculations to the AI adds more work for the game to do. If every AI manager is constantly considering all the points listed above down to the minute detail it will use more processing power and memory meaning the game would become slower and slower.

So we improve it as much as we can each version whilst not affecting the overall playing experience for users. Taking into account all our users, whether they're on the very latest tech or on lower-spec machines.  

[anagain]

47 minutes ago, kingking said:

we need to improve the A.I in terms of

• Squad management
• Squad rotation
• Tactical adaptability, tactical risk taking etc
• Squad registration
• Transfers 

I feel like the A.I can improved if the developers try.

Whilst I agree that the AI could be improved, I will say that I don't think you udnerstand the limitations on AI in games. If you play enough games ytou will understand that AI is often limited in relation to the human. That is why in games such as Civilization the AI often needs a major boost to compete with the human player. Watch anyone who is even halfway decent at Civ and you will see that they can destroy the AI at most difficulties.

I am going to agree wholeheartedly with @Mars_Blackmon. Tell us how you would improve the AI in each header in your OP. If it is that easy then you'll have plenty of ways to do so.

I am sure SI work extremely well on improving the AI of FM, as does every gaming development studio. I'm pretty sure I have heard AI referred to as the ultimate frontier - or words to that effect - in gaming. I'm pretty sure it is limited by technology too. More calculations per second by the AI would need more computing power. I'm no expert, just a passionate gamer. It's all algorithms and coding designed to help the AI learn.

Like I said, and for what it is worth, your desire is a worthy one. It is surely shared by SI. The game won't fail if it's not improved in the short term though.

[Mcfc1894]

Feel like home teams always had a bigger advantage than they should have because of the negative tactics AI managers always set up for away games 

[kingking]

2 hours ago, Mars_Blackmon said:

Hot take.

But I’ll engage. How would you improve the AI. Specifically what would you like to see done in every area you have listed?

Give examples on how none of the things listed aren’t optimal. 
 

That conversation would be more helpful to the game.

@anagain 

• Lack of Risk. Too Often A.I Mangers play too defensively or cautiously against teams that are better then them. I would like to see more risks and confidence to play aggressive and more attacking.
• Poor Registration & Squad Management. A.I Managers do not try hard enough to sell their players so there squad ends up being to large with many players not registered. And sometimes A.I buys too many players.
• Unnecessary Transfers & Loans. Sometimes A.I would loan players into the team but they don't use them.
• A.I Manager struggle to adapt to the situation and find a way to win . Sometimes the A.I managers struggle to find a way to break Defensive/Cautious/Park the Bus tactics. I often see strong teams managed by A.I struggle to win matches that they should actually win. I see top teams managed by A.I struggle meet their own aims and objectives 

Edited October 19, 2020 by kingking

[GianniM]

9 minutes ago, kingking said:

I often see strong teams managed by A.I struggle to win matches that they should actually win.

[kingking]

47 minutes ago, GianniM said:

Yeah big teams can slip up and struggle but in FM I think it happens too often however i might be wrong.

do you agree with me on 

• Excessive & Unnecessary transfers by teams such as PSG
• Poor Team management where a team buys to many players and do not sell enough.
• Weaker sides being too predictable and not play enough aggressive and attacking tactics against teams that are stronger then them?

2 hours ago, Neil Brock said:

We're always working on improving the AI, every single version I've ever worked on (well over 10 years) has had improvements in these areas.

Adding more calculations to the AI adds more work for the game to do. If every AI manager is constantly considering all the points listed above down to the minute detail it will use more processing power and memory meaning the game would become slower and slower.

So we improve it as much as we can each version whilst not affecting the overall playing experience for users. Taking into account all our users, whether they're on the very latest tech or on lower-spec machines.  

Yeah the A.I has actually improved alot and thank you for your involvement in FM.

I guess it's hard to find a balance between A.I Intelligence and Processing Power and Memory.

My main issues are the

• unnecessary and poor transfers by teams such as PSG. Also teams loaning players into the team without even using them .
• Poor squad management where a team has too many unregistered players because they buy too many players and do not sell enough players.
• Occasionally I would like to see risk and attacking & aggressive tactics from teams with a weaker side against teams with a stronger side.

Edited October 19, 2020 by kingking

[anagain]

16 minutes ago, kingking said:

@anagain 

• Lack of Risk. Too Often A.I Mangers play too defensively or cautiously against teams that are better then them. I would like to see more risks and confidence to play aggressive and more attacking.
• Poor Registration & Squad Management. A.I Managers do not try hard enough to sell their players so there squad ends up being to large with many players not registered. And sometimes A.I buys too many players.
• Unnecessary Transfers & Loans. Sometimes A.I would loan players into the team but they don't use them.
• A.I Manager struggle to adapt to the situation and find a way to win . Sometimes the A.I managers struggle to find a way to break Defensive/Cautious/Park the Bus tactics. I often see strong teams managed by A.I struggle to win matches that they should actually win. I see top teams managed by A.I struggle meet their own aims and objectives 

 

Ok, some good points there. I agree with you.

Caveat. That's what you want to be better. How do you make it better? I am sure that SI know these points, but the proof of the pudding is getting the AI to a stage whereby it can manage all this on a consistent basis. As Neil Brock said, SI have been working on this every edition for years. For every line of code you add the more you make for the hardware in your computer to handle. You are left with a choice of whether to make FM super demanding, and alienating a lot of casual gamers, or doing the best you can to have the game run on a variety of systems.

The one that is probably easiest to improve is the tendency of opponents to sit back, but this is also because human players are very fond of being attack minded. In a way you could argue that the AI is learning. It knows that the human likes to attack.

I'd argue on the idea of managers not selling players. This is true in a number of situations, but it's not universal. I just bugged a team in Slovenia seriously weakening themselves by selling players for way under what they could get and leaving themselves very short. On the other hand, when I got the Swindon job in a past save I found the squad littered with useless talent. This is also down to it being hard to offload players at lower levels of the game. I sometiems wonder if SI could add in some way for un-manageable clubs to more often buy players not needed at playable teams. There is also a tendency for clubs to refuse sending players to clubs that are not all that far below their level. In real life Bournemouth loan a lot of youth players to local clubs. In FM Bournemouth won't enertain a loan deal that is to clubs of that stature. That needs to change.

Here's to improvements in all these areas.

 

[GianniM]

3 minutes ago, kingking said:

Yeah big teams can slip up and struggle but in FM I think it happens too often however i might be wrong.

 

Your screenshot above is only after nine matches. Things might even out still.

And Arsenal actually finished 8th last year so sitting 9th is pretty accurate...

 

4 minutes ago, kingking said:

Excessive & Unnecessary transfers by teams such as PSG

Again, look at Barcelona's transfers in the past years

[Freakiie]

I mean, IRL a lot of clubs are pretty terrible at squad management.

Schalke in FM20 has like 5 strikers? All roughly at the same level and all expecting playtime...

Barcelona? Do we even have to go there?

Real Madrid that decided they don't need to replace Ronaldo's goals and have basically been competing with Barcelona in a game off "who can be the shittier champion, since we'll always be better than at least 17 clubs in the league anyway"?

Man City... No comment.

The issue is just that the player is too good at squad management, because it's very hard for a player you buy to completely flop. If you buy youngsters you immediately know their personalities, injury risks and so on. Sure their PA might not be as high as you'd like it too, but you can be pretty much sure that every player you buy will get close to his PA because you avoid anyone with a poor personality. As for big name signings? Well most of us don't even make those and when we do we can again see their exact stats, we can even know about potential adaptability issues (while the guy never left his country), injury risk even though he never got badly injured and so on, so again, it's highly unlikely to buy an actual flop.

The only part of AI squad building that I would consider needs a straight up fix is them making offers for multiple players in one position (which makes sense), but then not cancelling offers once they get their main priority in so they end up buying all 3 players for one position, with the other 2 being offloaded in the next window since they never played. Especially noticeable with Bayern, who has a lot of money and a relatively small squad in FM20, so they buy like 3 replacements for Gnabry (He better be buffed in FM21, so Bayern stops dumping him), while they only "need" 1. Also often seen them buy both Hector and Wendell as back up left backs, only to offload Wendell right away again in the winter since there's no way he accepts being third LB with 0 playtime. Wendell to Bayern and then half a month later back to Leverkusen for half the price is probably one of the most common FM transfer combinations I've seen.

I do agree that I'd love to see managers (and clubs) to change things up and not always just go "oh higher rep, time to park the bus". In general though, reputation is a massive elephant in the room that has (imo) way too much influence all around in every aspect off the game, which also ties in to squad management with high reputation ageing stars being preferred over young talent because of their rep, even though they really aren't good enough anymore. Would love to see manager personalities mattering much more and some managers as a result just going aggressive even as massive underdogs.

In the end, I'd much rather see SI making squad building more difficult for the human player, as a realistic AI simply can't compete with a player, as the player doesn't play in a "realistic" manner. We can offload entire squads and buy together a new one with no penalty, we can play without a single experienced player without serious drawbacks, we'll be on Brazilian regens the moment they spawn into the game snatching up every single one that looks like he has half decent personality and potential. Of course not a single one of those will be a massive flop that we paid way too much for, because we know after a week of scouting that this 16 year old is professional, highly adaptable and has no injury issues! While I love the part of finding the next Messi, it currently is far too easy because we can just snap up any half decent 16 year old and just watch to see his CA skyrocket because he's professional and the development curve is messed up too where until a player is like 22-23 he'll easily get 20-30 CA a season and then once he's 23 you can basically forget developing him much further, so you just offload your "bad" 22 yr olds to the AI who at that point still pay a ton of money for him because they see him as a young talent.

Somehow this turned into a much longer rant than I thought... oh well, good job on anyone actually bothering with my ramblings.

[Mars_Blackmon]

I don’t think the player will sacrifice enough to compete with the AI. Im sure a lot of people don’t use attribute masking and use the player search to look for specific ratings. That in itself creates an huge advantage and if that was made difficult in the game for the users, people wouldn’t like it. It’s a fine line between making the AI smarter and realistic as mention even real life clubs make stupid decisions.

I personally would like the AI to be more aggressive in the transfer market. I usually can get the players that I want without any worry as long as they like my offer and I rarely over pay.

[FlorianAlbert9]

4 ore fa, Neil Brock ha scritto:

We're always working on improving the AI, every single version I've ever worked on (well over 10 years) has had improvements in these areas.

Adding more calculations to the AI adds more work for the game to do. If every AI manager is constantly considering all the points listed above down to the minute detail it will use more processing power and memory meaning the game would become slower and slower.

So we improve it as much as we can each version whilst not affecting the overall playing experience for users. Taking into account all our users, whether they're on the very latest tech or on lower-spec machines.  

Neil, a step at the time will be possible? 

For example: in a first step you could boost AI only for Manager with >165CA so that at least at a certain level we could have a real challenge without have to demand much more from processor. 

While in poor nation league you could use at least some shortcut like chairman with big money arriving in your rival team. (I always play in poor nation like Hungary and soon (max After 5 years) there is no more challenge 'cause you have too Money and stronger player that any other - i could win League even with my U19)

Edited October 19, 2020 by FlorianAlbert9

[Heywood JaBlowme]

The AI doesn't have the luxury of ignoring it's own "advice'. The advice from my backroom staff ranges from excellent to horrid.  

[andu1]

Actually i find the big teams were harder to beat in FM 19 and 20 than in the previous versions.

In my long term game, Real Madrid managed by Zidane in 2032 won around 6 or 7 Champions League and he also had an excellent squad at his disposal with many regens of course.

What AI managers are lacking in FM in my opinion is the adaptabilty to try different formations when they underperform or to adapt to a new tactical trend.

For Example, ZIdane's AI team always used the formation 4-3-3 DM with wingers when winning so many Titles, but no other AI manager tried to copy his style. Instead the AI became obsessed with 4-2-3-1 for some reason ( bug perhaps )

[Ruh Roh]

If you go into the editor, all the different attributes and PPM-type "tendencies" stuff you can fill in for managers, there's so much stuff there. I'm sure there are people in SI who know it all works and at least the top 50ish big name managers are doing what they're supposed to do.

At the same time I remember there was a years-long issue about some researchers getting "jumping" and "jumping reach" mixed up. And that was with reports on the forums going like...this player is 6'5", why is his jumping a 10? But all the manager stuff is basically hidden, so you're not getting too much feedback if there's a research issue. For a manager, does a 20 at the "attacking" tactical attribute mean you're great at it, or is it basically another tendency?

Edited October 19, 2020 by Ruh Roh

[Heywood JaBlowme]

21 minutes ago, andu1 said:

 

 Instead the AI became obsessed with 4-2-3-1 for some reason ( bug perhaps )

2051 in my save. I would estimate at least 95% of the AI managers use it. (4231 Wide)

[andu1]

1 minute ago, Heywood JaBlowme said:

2051 in my save. I would estimate at least 95% of the AI managers use it. (4231 Wide)

Yeah, i hope this bug is fixed in FM21. Although i will definetely run a test save to check it out.

[FlorianAlbert9]

9 minuti fa, Ruh Roh ha scritto:

 For a manager, does a 20 at the "attacking" tactical attribute mean you're great at it, or is it basically another tendency?

Tendency. It's his mentality style in the tactics.

At now, paradoxically, AI Managers don't have many quality attributes and they haven't attributes about their tactical quality (a parte the generico tactical knowl)

[destmez]

For example: in a first step you could boost AI only for Manager with >165CA so that at least at a certain level we could have a real challenge without have to demand much more from processor. 

"Smell of burning pc 😂"

[Bruce Drundrige]

16 hours ago, kingking said:

 

I feel like the A.I can improved if the developers try.

...what do you think they have been doing? Not trying?

[santy001]

As a researcher, I have to point out that we set the tactical preferences for managers. We also set some other stats that build their profile. This results in the decisions they will make as the AI applies their profile.

It's difficult to get right. 

[sporadicsmiles]

15 hours ago, anagain said:

. That is why in games such as Civilization the AI often needs a major boost to compete with the human player.

Yeah it seems a lot of people do not realise that "good AI" in some games is actually just regular AI with a bunch of cheats. Total War is good at this. Harder difficulty levels on that is the same AI with huge buffs to economy and units. I'm not sure people would be happy with this in FM. 

14 hours ago, kingking said:

• Lack of Risk. Too Often A.I Mangers play too defensively or cautiously against teams that are better then them. I would like to see more risks and confidence to play aggressive and more attacking

Part of the problem here is the way defensive football is translated into the ME. It leads to weird situations where the AI just passes the ball endlessly in their own half. However to counterpoint this, the AI will sit back for 75 minutes and then go for it in the last 15 when level or only down a goal. This is pretty decent in terms of strategy, depending on who you play. I do think some of the bigger sides setting out for a draw could be toned back.

14 hours ago, kingking said:

• Poor Registration & Squad Management. A.I Managers do not try hard enough to sell their players so there squad ends up being to large with many players not registered. And sometimes A.I buys too many players.

I'll add that the way the AI picks its players leads to squad imbalance. This is a problem, because it does mean that after a while the AI sides are a little weird sometimes. Now it is of course difficult to have them analyse like a player, because I would like my game to progress at a rate of more than a day per day. However maybe there is some way they can get around this? Or maybe something like this is already used. Each manager has a specific way of playing, and thus a specific set of player attributes he will want for each role. Perhaps rather than using CA/PA or reputation (or whatever is used), have player archetypes. So players are rated by some weighted average of the important attributes for the role being searched for. As a way to refine the selection of players. That way a team needing a creative midfielder will not buy a DMC destroyer, for instance. Just an idea.

14 hours ago, kingking said:

• Unnecessary Transfers & Loans. Sometimes A.I would loan players into the team but they don't use them.

I do that too. In fact I will use loans as a way to get myself backup players more often than not. I imagine AI managers are also frustrated with me not playing their players. 

14 hours ago, kingking said:

• A.I Manager struggle to adapt to the situation and find a way to win . Sometimes the A.I managers struggle to find a way to break Defensive/Cautious/Park the Bus tactics. I often see strong teams managed by A.I struggle to win matches that they should actually win. I see top teams managed by A.I struggle meet their own aims and objectives 

Not sure about this one. Some tactical styles are not well suited to the current ME, or simply cannot be well implemented into it. Such as a proper Guardiola style, which is incredibly hard to replicate even for a human. Couple that to the way defensive sides keep the ball in the own half basically encouraging a high press (and so rewarding teams who play that way), you can see certain teams struggle regularly. However not every team can meet their expectations in a given season anyway. Take your screenshot. Arsenal and Leicester being midtable after 9 games is not terribly unusual based. 

[Xavier Lukhas]

Il y a 14 heures, Freakiie a dit :

In the end, I'd much rather see SI making squad building more difficult for the human player, as a realistic AI simply can't compete with a player, as the player doesn't play in a "realistic" manner. We can offload entire squads and buy together a new one with no penalty, we can play without a single experienced player without serious drawbacks, we'll be on Brazilian regens the moment they spawn into the game snatching up every single one that looks like he has half decent personality and potential. Of course not a single one of those will be a massive flop that we paid way too much for, because we know after a week of scouting that this 16 year old is professional, highly adaptable and has no injury issues! While I love the part of finding the next Messi, it currently is far too easy because we can just snap up any half decent 16 year old and just watch to see his CA skyrocket because he's professional and the development curve is messed up too where until a player is like 22-23 he'll easily get 20-30 CA a season and then once he's 23 you can basically forget developing him much further, so you just offload your "bad" 22 yr olds to the AI who at that point still pay a ton of money for him because they see him as a young talent.

To be honest, even without the very gamey parts (for example, knowing that 22-24 yo are basically past it), human's squad building is more or less pragmatic and based on the data available in game... assuming you're not using external tools. The scouting revamp from two or three years ago has very much nerfed the human player's ability to just scout the planet since it's tied a lot more to the money you have to pay for scouting packs... or pay for assignments outside your scouting pack, as well as delaying the time it takes for the staff to report on scouted players: technically you can still access what a scout has gathered over a day of work, but you no longer specifically have an option to get a report in your inbox after one day (subtle, I know). It makes less rich clubs less likely to have an expansive (and expensive) scouting network, which when you play in smaller nations is basically the first thing you invest in since signing the players that you need right now is prohibitively expensive in most cases; better wait and develop a youngster instead. And to be fair, that's essentially what big clubs in smaller countries do: spot youngsters and try to flip for a profit instead of buying the players that already are at the level of the first squad.

In other words, you can say that in theory the AI should notice how good a potential signing is just as easily as you do, and therefore should compete harder with you on those signings. It's not as if they have less resources than you, they use them not as well as you do. It's not as if scouting youth competitions and NTs is particularly gamey or unrealistic: real clubs do that all the time. Therefore I don't think limiting the player's ability to scout even further makes a lot of sense if the AI doesn't become any better at spotting players. I know that sometimes the AI has out of the intake instant knowledge of certain youth players that would take you weeks to learn about, but FM is still a game of numbers: if the report is good, it's also good for other similar AI clubs. You don't have access to something they don't have.

The AI also needs to manage their squad better: I cannot count the times where Mourinho's Man Utd squad in FM17 or 18 would sign 4 strikers and have 8 players that could do the job as striker even if they were natural elsewhere despite playing a 4-2-3-1 Wide. Only one guy up top, but they unsettle half of the top strikers in Europe that they don't already possess. The game pressurises you in getting at least three players per position for all positions of your tactics. Fine, but the game also complains when your third or fourth option, which often can be a youth player, isn't on par with your second or main option. Of course they're not, they're still developing, leave me alone, I don't need to have Haaland, Icardi and Aguero in the same freaking squad and neither does Man Utd.

It ties in the fact that unlike the player, the AI is also pretty crap at having any sort of fluidity between the squads. Your own staff tells you that you should make a promoted youngster still available for the B squad. Or best, you would often, like IRL, call youngsters for one cup match but leave them in the Reserves or U19s. The people who had the miserable experience of managing B teams such as Spanish ones know all too well that in order to honour the "rule" that you need three options for one spot in your squad, will remove that player from the B squad and let him rot on the bench of the main squad. Your own staff tells you not to do that, and your staff is generally not very good at giving you valuable advice!

TL; DR: I think the human player is mostly pragmatic than "too good"... and AI clubs are just not really good at anything squad building. Nerfing the player's ability to scout wouldn't help much: the previous scouting revamp did do that, it just makes scouting more frustrating.

Il y a 16 heures, Wakers a dit :

Honestly it feels like SI have forgotten that managers have stats. The only thing that matters at all to the AI is reputation, which determines which jobs they end up in. Nothing else I've seen in the game bears any relation to their stats. Even preferred formations they tend not to use rather than use.

Amen to that. When you see the impact Reputation has on every aspect of the game, you cannot unsee it. Its bonkers how teams will kneecap themselves due to Reputation. I know that it helps maintaining the balance of the game, but it does so in a fairly egregious manner that really can kill immersion. How many times you meet a team that plays one hell of a rotten partition away and smack your face at home. I know home bias is a thing, but the way it works is incredibly cliché to say the least.

Edited October 20, 2020 by Xavier Lukhas

[FlorianAlbert9]

I agree with irrealistic squad building by us. 

I myself tend to have an almost complete different team in few years, a team stronger, younger and less paid. 

Cause it is too easy use young. 

"Decision" should be the most important attributes and the most of youngs should have it very low. (Now, "Pace" Is as Always overrated, Is there very very few part of the game where a player can reach his top pace and pace without agility/balance is pretty useless )

Adaptation should have more impact (but thats starting from data research: setting CA for a foreign should take in account that he could have failed for poor adaptation without then lowering his CA)

And then, if you can win a Champions League with a 100CA player don't means you are axgreat managet, but that there is an huge issue in the ME. 

[forameuss]

6 hours ago, Bruce Drundrige said:

...what do you think they have been doing? Not trying?

I'd say that's exactly what a lot of people believe, bafflingly.

[zyfon5]

3 hours ago, FlorianAlbert9 said:

I agree with irrealistic squad building by us. 

I myself tend to have an almost complete different team in few years, a team stronger, younger and less paid. 

Cause it is too easy use young. 

"Decision" should be the most important attributes and the most of youngs should have it very low. (Now, "Pace" Is as Always overrated, Is there very very few part of the game where a player can reach his top pace and pace without agility/balance is pretty useless )

Adaptation should have more impact (but thats starting from data research: setting CA for a foreign should take in account that he could have failed for poor adaptation without then lowering his CA)

And then, if you can win a Champions League with a 100CA player don't means you are axgreat managet, but that there is an huge issue in the ME. 

Squad building is obviously an issue but fixing squad building alone will not fix the problem. I managed Leicester city which placed 10th last season. The only transfer that I brought in is a new starting striker while I sold off their starting winger and promote a young winger with potential but slightly worse to the starting position. These are the only changes I made. I played with the same 4231 formation they used last season but fit my tactics around the players I have. With basically pretty much the same team that finished 10th last season under the AI manager, I brought them to win the premier league (expected finish for both seasons is 7th).

The conclusion that I have drawn from my journeyman save is the AI transfers although could be improved is not completely ****. The real difference is how much more a competent human player can get the maximum output from the same team compared to the AI.

[Pav_Makarov]

37 minutes ago, zyfon5 said:

The conclusion that I have drawn from my journeyman save is the AI transfers although could be improved is not completely ****. The real difference is how much more a competent human player can get the maximum output from the same team compared to the AI.

This. I have same experience, but I'll just add, that AI might need to develop it's youngsters more. Unless AI controlled team have wonderkids, ready to perform, they almost never develop own youth. Another exception is homegrown rule

[Mars_Blackmon]

As far as the human element, I think a proper fog of war would help. Imagine having all attributes hidden and having to rely on reports i think this could work now because of data analysis. I don’t really like the current attribute masking feature.

[FlorianAlbert9]

46 minuti fa, Mars_Blackmon ha scritto:

As far as the human element, I think a proper fog of war would help. Imagine having all attributes hidden and having to rely on reports i think this could work now because of data analysis. I don’t really like the current attribute masking feature.

This is my idea ( 2 years ago). 

Now, i will add others mental attributes in the hidden list (like composure)

Edited October 20, 2020 by FlorianAlbert9

[isignedupfornorealreason]

9 hours ago, FlorianAlbert9 said:

And then, if you can win a Champions League with a 100CA player don't means you are axgreat managet, but that there is an huge issue in the ME. 

I disagree.

I won the CL with Donald Love as my primary RB (or LB I forget) and his CA was something like 110 at the time. But his attributes were phenomenal for his role, to such an extent he was in position, anticipating play so well that he never had to use his rather poor physical and technical skills. And what few skills he did have, he was great at. I.e. tackling.

 

Anyway, one problem with the demands for better AI Squad building is that people fail to take into account that we need to simulate managers making mistakes and bad decisions. If we improve the AI to the extent that they manage squads well, we lose the opportunity to see teams sliding down the table due to mismanagement etc. 

[FlorianAlbert9]

10 minuti fa, isignedupfornorealreason ha scritto:

I disagree.

I won the CL with Donald Love as my primary RB (or LB I forget) and his CA was something like 110 at the time. But his attributes were phenomenal for his role, to such an extent he was in position, anticipating play so well that he never had to use his rather poor physical and technical skills. And what few skills he did have, he was great at. I.e. tackling.

I stay in my idea. 

Poor ME. 

I bet nobody in data research gave a CA of 110 to someone who won CL. 

So that create a grave discrepancy. 

And then no AI manager in top team will use that player, so another big problem in the sense of the game.

 

[turnip]

AI managers - and real life managers - will sometimes buy players who turn out to be complete donkeys.

And sometimes they'll sign players who perform really well in a particular role within a particular system, but don't perform nearly as well in a different tactical setup. 

 

It's quite unusual in the game to be able to sell a complete donkey for lots of money, because a player will usually only attract a lot of attention if he's playing often and playing well, but you can certainly sell some distinctly average players for good money if you can figure out how to get good performances out of them.

[Wavelberry]

The fact that CA is not the be all and end all and the better numberz don't guarantee the win is actually one of FMs greatest strengths IMO. Football is unpredictible and this quirk, if you wanna call it that, elevates the game beyond the spreadsheet element.

Also it is something that is a deviation from the norm rather than something that always happens so it ain't the biggest issue.

[FlorianAlbert9]

Il 21/10/2020 in 00:08 , Wavelberry ha scritto:

The fact that CA is not the be all and end all and the better numberz don't guarantee the win is actually one of FMs greatest strengths IMO. Football is unpredictible and this quirk, if you wanna call it that, elevates the game beyond the spreadsheet element.

No. 

Football Is not so unpredictable. 

Bayern

Liverpool

Real Madrid x 3

Barcellona

Real Madrid

Bayern

Chelsea

Barcellona

Inter

Barcellona

The winner of CL in the last 12 years. 

As you see there is not Swansea, Benevento or Valladolid. 

In the last final the the only player of Bayern <150 was Alphonso Davies (127) but 'cause he was a rookie and i bet in FM21 he would gain a great upgrade.

So yeah, a Lewandowski could perform better in CL then Messi. 

Jamal Lowe no. Cause hardly he even could plays in the CL

If you could win CL with Jamal Lowe, it's not thanks to your great manager ability.

There was not sense to pay so much top players wages if One could win with Championship level ones. In real life as in the game. 

Edited October 24, 2020 by FlorianAlbert9

[DementedHammer]

The importance of CA and PA is massively overrated by some people in my opinion. Now, granted I play as far down the leagues as possible, but one of my most effective players in FM19 was a winger with a CA and PA of 1 out of 200. He was reasonably pacy (around 13 for pace and acceleration), and had reasonable crossing (Maybe 9 or so). The rest of his attributes were very poor, but that was all he needed to beat his man several times a game and get a decent cross in that my target man was able to take advantage of. 

Another example I can think of was an anchorman that, on paper, would have looked terrible. He could mark and tackle well, and had reasonable anticipation and concentration. The rest of his attributes were abysmal, but he was an absolute rock for me, as all he was expected to do was to protect the defence, win the ball and play a short simple pass off to a more talented player. 

With the right attributes, and in the right system, you can definitely get a lot out of a supposedly inferior player. The trick is having a system that plays to their strengths and doesn't expose their weaknesses. 

Edited October 24, 2020 by DementedHammer

[Novem9]

В 19.10.2020 в 14:37, kingking сказал:

• Squad management
•  

it much better in fm20 then few years ago. Play journeyman saves and check yourself. Or change your mind about some things because AI managers have different preferences and you need to adapt some clubs for own style longer/shorter 

[FlorianAlbert9]

1 ora fa, DementedHammer ha scritto:

The importance of CA and PA is massively overrated by some people in my opinion. Now, granted I play as far down the leagues as possible, but one of my most effective players in FM19 was a winger with a CA and PA of 1 out of 200. He was reasonably pacy (around 13 for pace and acceleration), and had reasonable crossing (Maybe 9 or so). The rest of his attributes were very poor, but that was all he needed to beat his man several times a game and get a decent cross in that my target man was able to take advantage of.

That's impossible. 

Pace cost a lot of CA points, so a player with 13 in pace cannot have 1CA and he for sure is one of the best  (if not the best by far) in the lowest league

Edited October 24, 2020 by FlorianAlbert9

[DementedHammer]

27 minutes ago, FlorianAlbert9 said:

That's impossible. 

Pace cost a lot of CA points, so a player with 13 in pace cannot have 1CA and he for sure is one of the best  (if not the best by far) in the lowest league

I know it to be true. I was working on my 11 tier NZ league editor file at the time, so I was keeping a very close eye on the attribute differences between divisions. 

At the time I was disappointed, as I was hoping a 1CA player would have been worse than that particular player was, so that there would be even worse players in the lower divisions. 

Edited October 24, 2020 by DementedHammer

[Freakiie]

Pace, at least until you reach close to 20, uses surprisingly little CA on a winger. Technique though? Damn that eats CA. You can actually make a 14/14/15/17 crossing/dribbling/accel/pace winger in the editor while ignoring everything else and that'd produce a 1 CA player.

[herne79]

4 hours ago, FlorianAlbert9 said:

If you could win CL with Jamal Lowe, it's not thanks to your great manager ability.

It's 100% due to your managerial ability because the AI managers are nowhere near as good as a human manager.  We have all the advantages such as squad building, player management and tactical ability.

The Match Engine may well have issues, but to say the ME is flawed because we can win the CL with some lower CA players is incorrect - it's an AI weakness (or "flaw" if you will) because the AI is simply nowhere near as good as us humans.  Personally I can take West Ham and win the Premier League first season with no personnel changes simply because I'm that much better than the AI (big up my bad self ).  Remember - the ME cannot tell the difference between human and AI managed teams.

Further, the hidden CA value is not a measure of how good a player is.  It never has been and is not why it's in the game.  It just gets (incorrectly) used by some people for that purpose.  Stop thinking about hidden CA and things really do start to look much clearer .

You also mention all those top teams in real life who win the CL, stuffed full of high CA players.  And you're right, they do.  If you let FM play out with no human manager challenging for the CL, those same teams win the CL too.  It's only when a human manager gets involved that suddenly Naff United get themselves involved - because that human manager is at an advantage against the AI.

[DMaster2]

Il 19/10/2020 in 13:40 , forameuss ha scritto:

Counter-point, it's easy to see a situation where if the AI improves too much, FM will also fail.

Not that I'm saying it shouldn't improve, as it should, but it's an interesting balancing act FM will always have to perform.

The AI regarding to transfers and squad building is just awful. I'm italian hence i usually do a save per game with the Italy national team, and watching AI teams is sooo bad. Squads of literally 50+ players in the Serie A rotting because no one is loaning anyone. I mean do the SI guys even checked a single transfer market in Italy? Ton of loans of young players so they can get some first team football, especially since the finances of clubs of Serie C and D are a disaster so those loans help both.

[Lexis]

Lower end specs Neil? :)))) You mean toasters? The 15 year old specs you are afraid to phase out?

[Mars_Blackmon]

3 hours ago, DMaster2 said:

The AI regarding to transfers and squad building is just awful. I'm italian hence i usually do a save per game with the Italy national team, and watching AI teams is sooo bad. Squads of literally 50+ players in the Serie A rotting because no one is loaning anyone. I mean do the SI guys even checked a single transfer market in Italy? Ton of loans of young players so they can get some first team football, especially since the finances of clubs of Serie C and D are a disaster so those loans help both.

How many leagues are you loading as playable and how far down pyramid is Italy set to active?

these type of things often happen when few leagues are set to playable.

[DMaster2]

42 minuti fa, Mars_Blackmon ha scritto:

How many leagues are you loading as playable and how far down pyramid is Italy set to active?

these type of things often happen when few leagues are set to playable.

All the italian leagues plus top leagues of england, france, spain, germany, portugal, netherland, belgium, brazil and argentina. I see honestly no justification for this kind of behaviour of the AI regarding transfers.

Edited October 24, 2020 by DMaster2