Could we get an official response - Do the AI figure out your tactics or not?

[John_Yuki]

It's a legend that has been passed down from year to year. There are players that swear blind that the AI always cops on to their tactics after a season or two, and other players who say that the AI doesn't learn like that.

I'm currently playing a Birmingham City save. In our first season in the Premier League (2018/19), we finished 13th - comfortably away from relegation - and won the EFL Cup, meaning we are in the Europa League this season. However, the second season has started awfully, with bad performance after bad performance. I spoke to some people and they also echoed the whole "the AI figured you out" sentiment. However I also have a Cork City save that I am in the 7th season of. In that save I have used the same tactic all 7 seasons, and have only improved my performances season by season winning the title every year, and advancing further and further in Europe. So why would the AI figure me out when I am Birmingham, but not Cork?

It would be really great to get an official answer on this one, as I have been hearing this "rumour" since I started playing FM in 2012, and no one seems to have an official source on it.

[Svenc]

The AI doesn't figure you out. Simply put, any AI manager in any match adapts to the next opponents it faces some. Basically, they target different results (convincing wins,  draws, not getting humiliated) and play more attacking or defensive football based on how they view any team -- including yours. Your statue in the game changes naturally over time, unless you don't ever under or overperform. If you overperform, more and more teams may pick more cautious approaches, which may mean you have to work for your space. If you'Re a top team from the go, you will face quite defensive teams every week, which AI managers of Barcelona, Bayern et all face likewise. This is meant to simulate what is going on in real football, as you rarely see the likes of Burnley going all gung-ho at Chelski. You are looking for official confirmation, of course, but will likely see this confirmed.

That aside, I'd personally like to  challenge your perception that finishing 13th must be down to good performance (or be primarily tactical). As matches on FM likewise tend to be settled in very slight margins, points will vary simply based on random chance, in particular from one season to the next. This happens in football too. Anybody looking at any of the more recent Bundesliga tables, where the difference between 6th and 16th place is oft just ~12-15 points (basically three wins and a couple scraped draws) and who doesn't see the randomness at play probably needs a goggle.   The German media meanwhile coinned the term  "Euro League Curse" based on such random variation, realizing that surprise teams making the top 6 rarely replicate that the following season. Firstly, for most of the teams getting there by "surprise" it is only to be expected that the only way to go is down, as they have reached their apex of what is possible.

Secondly: Realistically, finishing 6th place once in a league such as the BL doesn't mean much, and I wouldn't put too much into finishing 13th place in the EPL either -- in particular if there isn't a significant point gap to the relegation spots. It's one of the reasons why people (bookies, managers, analysts) trying to assess who really is a better team and the worse one don't much look at league points/tables, as those are going to vary greatly. That teams need to be leagues apart  for there to be point gaps is a micsconception, in-game, likewise. Over the course of a season, which is "just" 40 matches at best, some teams just win more and some teams lose more. This is ingrained into any sports DNA, which makes teams winners and losers every week. NO matter how tight a contest, there will be winners and losers. With teams more level, there tend to be less big wins though, which in-game too, shows in the goal differences in the simulation.

[HUNT3R]

They don't figure out tactics. In general, all they do is adjust to be more or less defensive depending on reputation and form.

 

Edit: If you want something more official: 

https://steamcommunity.com/app/482730/discussions/0/282992646978869621/?ctp=2

[XaW]

This one have been around the block before, but it keeps coming up again. What @Neil Brock states (in @HUNT3R's link) should be more than enough confirmation about this. Even with confirmation like this, there will be still be users clinging to their tinfoil hats refusing to listen.

I think this should be the new logo next to the SI one at the top

[sporadicsmiles]

It is not the AI "learning" your tactic (hell, I wish the AI in the game was that intelligent), it is a case of the AI adapting to your increased reputation (you are no longer potential relegation fodder to be attacked nor seen as an easy 3 points) and you failing to adapt to this.

More often than not, you were being gifted space as a relegation candidate as teams attacked you. Often, tactics are designed that are able to exploit space, and so you played well. Now that teams are more circumspect with you, they do not gift you space. You have to create it yourself, and that requires a different thought process (and often a slight tweek to what you are doing). As you go more attacking against newly defensive sides, you start to gift them the space that previously was gifted you. Net result, a down turn in performance.

[Svenc]

As of the info provided there is no evidence of a downturn in performance. Mind, FM doesn't ahve hugely feedback, and unlike in football you typically don't drop points in a row despite having the better chances, but still: Match results in isolation are a bad gauge of performance, and I am glad some of the "gurus" have picked up on this as well. Rashidi IIRC has worked out his own measuremeants of chance quality, and can roughly gauge where he was winning despite playing "badly" (which means he would need to fix a few things despite winning). Maybe he can give SI the feedback to improve their, well, in-game feedback. It's easy to fall for the trap that just because you lose, everything goes wrong.... and perhaps crucially, vice versa. After all, this is a game engine that lets 22 dots kick a ball-shaped thingamabob for 90 minutes, and but secondsn of it making the result... furthermore, most wins are by a goal or two scored in those key seconds, not completely unlike football after all. To trigger an emotional response, I'm going to argue he was just a bit on the lucky side finishing 13th in the first season (and likely, the point difference was but a neglible ~12 points, which is four close wins to the bottom anyway in matches that realistically could have gone totally either way). :P

But yeah in general that describes the process well. Of course there is a turning and tipping point. Firstly 13th place isn't like something that turns a teams reputation completely around, so it's not as if everybody was keeping it ultra tight now as they would against, say, Man City from the go. Secondly, after a while the rep decreases anyway so you are seen as the "cannon fodder to beat" again, which would happen at some point. A key indicator are the match odds prior, and how they treat you -- long shots, equal or favourite. As a side-note, one reason why the "myth" persists is that plenty players download "exploity" tactics someplace. In tendency though they are better defended by ultra defensive AI approaches, as chosing defensive formations/keeping masses of players behind the ball on defend duty plus dropping deep tends to crack over what is being exploited. Think of a hole, a physical one, and how much more likely it is filled some if there's a mass of soil thrown into it. As the "exploity" tactics also tend to vastly outperform player ability (due to the defensive hole exploited over and over) -- those players also now are in the situation every week where unlike a "truly" top team in FM, their players don't find space by individual brilliance alone as that just isn't there (e.g. Hazard completing 15ish dribbles per match FM15ish, etc.) Hence, the added struggle after a while. Hence one of the causes behind the "AI is cracking my tactics" myth.

 

 

[RodentofDoom]

One of the problems with FM is things often don't mean what you think they mean.

The mentality and shape settings being prime examples.

 

Many people have posted 'defensive tactics' and complained that they can't make them work.

Their tactic has not performed how they thought it would, because it's not actually very defensive at all.

A Defensive mentality setting does not make it a defensive tactic.

 

I think that what makes a tactic defensive or attacking is :-

• The duties you have assigned to the 10 outfield players.
• The overall shape of the formation.

 

If you want to build a balanced tactic then the duties given need to reflect the intent behind the tactic, whilst also considering that some formations are better suited to one style than another, the popular 4231 being a good example, with 4 very advanced player positions it is naturally suited to more attacking based styles.

Below for reference is usually how i try to balance my role assignmentsd based on intent for the formation im trying to build a tactic for.

                                                                                 

	Defend	Support	Attack
Defensive	6	3	1
Balanced	3	4	3
Attacking	3	1	6

[sporadicsmiles]

Mentality definitely needs a bit of a clean up with regards to how it is presented, although I can see why it is named this way - you do not want to make things complicated for casual users by added terminology that is needlessly complex. The descriptions need to make it clear, however, although you can get a feel for the differences based on the changes visible on the TI screen.

However, it is very easy to fall into the trap of "I want a defensive tactic so I pick defensive" as you note. The formation and the roles and duties are absolutely vital.

[Svenc]

Another thing about the "AI has cracked my tactics myth".

Confirmation bias. You read about someplace that the AI "cracks your tactics", which isn't hard, as it's an acknowledged FM Myth. So, basically, it must be true. (It doesn't, though it can appear such to a significant section of the user base, see previous post). From now on every time you struggle for results, it is evidence of cracking. Even if it might be just bloody randomness. That's also how and why fake news work, by the way -- a news piece is seen less critical the more it fuels one's beliefs. You can test this for yourself: Let's say you read about Ronaldo (or any other guy you personally don't like) that he is involved in say a massive match fixing scandal which has earned him billions throughout the years. The more suspicious you are about his,  the more likely you are going to accept it without further evidence. "Just knew this guy was an ass all along."

In games any such bias also tends to act as a means of self protection from "Harm". Hence also the many "AI is cheating stuff", even though it's fairly obvious the AI ain't all that invincidble, that is, if you know how to read and gauge the feedback the game represents. If a game cheats, you can't do much about it. The responsibility is all shifted to the code. If it doesn't cheat however, you can, so it must be something to do with yourself and how you play (which can't be, or can it?). Obviously there are players that are dead pan convinced that it does, as they run into the same "issues" over and over again -- in parts as they look for confimration of their beliefs every year, rather than inquiring players who obviously have none of such issues. The irony in all of this is that, as the game isn't all that upfront with some of its feedback, thousands of players actually outperform the AI massively without even realizing. [Ironically, such players oft rant the most about AI cheating].

However, top football management is plenty full of "conspiracies" itself, so everybody has good role models for playing the "blame game".

[Neil Brock]

Yeah the first paragraph from SvenC in his first post pretty much sums it up as far as I'm concerned. 

Another thing to take into account which may add to the 'myth' is that different managers have different formations they'll play in normal/attacking/defensive situations.

Say you're AFC Wimbledon and play Oxford a few seasons in a row. First couple of seasons let's say Jim Smith is Oxford manager. Unfortunately AFC tend to win every match home and away. Eventually Jim gets the boot and is replaced by let's say Karl Robinson. He has a different approach. Suddenly when AFC play Oxford they're facing different approaches and completely different tactics. Oxford start winning.

If the user was AFC Wimbledon they'd probably look at that and think that their tactics have been cracked. In actuality they've just come up against a different approach. 

[Snorks]

It is, as has been said, not te AI working out your tactic, but more the AI approaching a fixture against you - and this is influenced from many different things. Your team's reputation and form being the biggest influence, as Neil says, different managers from one season to the next, .

Your won tactic needs to constantly be reviewed and updated / tweaked as your reputation grows.

A quick look at the manager's profile page will give you the rundown on preferred formations, styles and stuff - but he may adjust slightly to face you because you are on a run of form,

[DementedHammer]

The thing that I don't get is that people actually complain about the AI cracking their tactics. I'd be absolutely delighted if the AI was advanced enough to do this and would consider it a massive step forward. It would add a completely new dimension to the game.

[FrazT]

Surely this is not a case of the AI manager cracking your tactic, but that he is looking at your team in a different way after playing you before and deciding that they need to adapt to how you play the game tactically.  Is this not what the art of football management is about?  All managers need to look at their opponents and pick players, a formation and a set up to get the best result against that particular team in their current form.

I often wonder how many users who post on here about the "second year or second half of the season syndrome" are still using the exact same tactic and set up that they made at the start of the game?  I suspect that % would be very high as we have all done it at some time.

[Svenc]

7 hours ago, DementedHammer said:

The thing that I don't get is that people actually complain about the AI cracking their tactics. I'd be absolutely delighted if the AI was advanced enough to do this and would consider it a massive step forward. It would add a completely new dimension to the game.

Interestingly, if that were to happen, it wouldN't merely affect the player. On FM, you are just dumped into a simulated game world -- it doesn't really give a hoot if you're there or not. The sole reasons human players see this "AI adapting" at the more extreme ends is that AI is incapable of overperforming to the degree a human player is. E.g. the statue of their clubs changes at rapid rates, whilst the average AI's not much at all. So what would happen down the linen is that AI would adapt more intelligently to each opposition, not merely the human team.

That aside I find the notion of tactics getting "cracked" a bit weird. Tactics isn't something hardwired written in stone, it's neither rock paper scissors either that can be cracked or canceled out. It's one factor amongst many that can somewhat increase and decrease the likelyhood of taking the points in a match, however, players, circumstance, the run of play and most of all chance would always play their part. The tactics forums has promoted "balanced", "multi-dimensional" tactics for a decade plus running mainly, as naturally, if you put an extreme focus on attacking/defending a specific way, that may become a bit less efficient unless you have a specific set of players available ; or the AI manager doesn't happen to employ something that draws it somewhat less effective (AI doesn't crack, it just plays more cautious or attacking and also each prefers different formations and has slightly bias for instructions or player types, e.g. target men). For a more obvious example, in the lower leagues it is traditionally oft not that hard to find forwards who have like an average +8 "jumping reach"/strength, etc. advantage over the average centre back on that level, which oft results in seasonal header completion rates above 70% (which is loads for a forward, in-game or otherwise). This can naturally be a bit utilized, say by channelling all play down the wings and bombing those forwards exclusively with crosses. However, at the higher levels that advantage fades, and the "one-dimensionality", the sole focus on that specific play becomes less effective. The download sub sections have worked a wee bit differently, naturally.

edit: I personally think it would be a massive step (and apparently big undertaking) getting AI to the point where they actually are capable of reproducing more varied playing styles, as far as possible anyway (let alone signing the players for it). Insert the casual yearly rants about how Barca et all in-game just don't have possession stats of 60%+, plus Ronaldo doesn't score his goal per match as AI unlike the real managers just don't channel all finishes to his (an all-time record worthy 7 average per match all throughout his Madrid career).  

[..Valhalla..]

I'm currently into my 7th season of using the exact same tactic for every match and every opponent, the only time I make a change is late in the game (very important matches) where we're only winning by 1 or 2 and I want to hang on. I'll change the mentality to contain and the duties all to defend where possible. Even if the AI have worked my tactics out, they have no answer for it  

[HUNT3R]

45 minutes ago, ..Valhalla.. said:

I'm currently into my 7th season of using the exact same tactic for every match and every opponent, the only time I make a change is late in the game (very important matches) where we're only winning by 1 or 2 and I want to hang on. I'll change the mentality to contain and the duties all to defend where possible. Even if the AI have worked my tactics out, they have no answer for it  

Once you're a top team (assuming you are) then your rep won't really change and your form will always be good, so there's nothing really more the AI can do.

 

Saying that though, I had someone on reddit who asked for tactical advice as his tactic that "worked" for a few seasons, falls apart against 1 specific team, every time in recent matches against them. Turns out, the AI copied his successful tactic (as can happen) but their version was a better one.   Just better roles and duties, as the AI still has "rules" for how many  A, S and D duties per mentality and used roles that suited the players. He was beaten by his own tactic.  

[Svenc]

54 minutes ago, HUNT3R said:

Once you're a top team (assuming you are) then your rep won't really change and your form will always be good, so there's nothing really more the AI can do.

 

Saying that though, I had someone on reddit who asked for tactical advice as his tactic that "worked" for a few seasons, falls apart against 1 specific team, every time in recent matches against them. Turns out, the AI copied his successful tactic (as can happen) but their version was a better one.   Just better roles and duties, as the AI still has "rules" for how many  A, S and D duties per mentality and used roles that suited the players. He was beaten by his own tactic.  

So he didn't beat them even once, not even by individual skill moves -- or, against all odds by blind luck (converting 3 direct free kicks --- vs. the opposition fails to convert 3 penalties)? After all, you can get a load of stuff wrong and still win regardless -- unless it is something truly horrible. Better dribblers rip open space simply by completing the additionally dribble, etc. etc. and not every goal is primarily something "planned" and tactical. Arguably, a load of goals aren't and shouldn't be (there's analysts that argue that a decent amount of goals in football are at least a good deal influenced by chance, e.g. deflections, etc).

As a side note, any (AI) that tends to always win (or seems hard to beat) is utilizing something that needs to be watched usually... (e.g. 3 forward switch, which has been something to look out for since FM 2015ish already, in parts as they just don't track back and are such an easy way for the AI or any team to counter. With each interception possibly springing an opportunity that has a reasonable high probability of being converted). The most remarkable upload I've personally seen in recent times of this ilk ("Impossible to win against!") was an AI that ALWAYS switched to 3 forwards at half time -- but only if the opposition team took the lead in the first half. That three forwards used to be the likes of Hazard, Lukaku and Aguero no less or something... in a sense, if that switch didn't come (early), the match was "easier" to win, so better not go up front early. IIRC it was de Boer, that is the SI AI version of him.

[HUNT3R]

14 minutes ago, Svenc said:

So he didn't beat them even once, not even by individual skill moves -- or, against all odds by blind luck (converting 3 direct free kicks --- vs. the opposition fails to convert 3 penalties)? After all, you can get a load of stuff wrong and still win regardless -- unless it is something truly horrible. Better dribblers rip open space simply by completing the additionally dribble, etc. etc. and not every goal is primarily something "planned" and tactical. Arguably, a load of goals aren't and shouldn't be (there's analysts that argue that a decent amount of goals in football are at least a good deal influenced by chance, e.g. deflections, etc).

It wasn't a great tactic, in truth. Or rather, it had huge weaknesses which any human opponent could easily exploit.

I think he was able to at least get a draw, but not wins. Wish I had a link to it!

[Tony Wright 747]

9 hours ago, FrazT said:

Surely this is not a case of the AI manager cracking your tactic, but that he is looking at your team in a different way after playing you before and deciding that they need to adapt to how you play the game tactically.  Is this not what the art of football management is about?  All managers need to look at their opponents and pick players, a formation and a set up to get the best result against that particular team in their current form.

I often wonder how many users who post on here about the "second year or second half of the season syndrome" are still using the exact same tactic and set up that they made at the start of the game?  I suspect that % would be very high as we have all done it at some time.

Absolutely agree with you here, I tweak my tactics for every opponent, this is the part of the game I personally find most enjoyable, e.g.  If a team I am playing against has two or three key players missing, I doubt I would use the same tactics as I would if they were available.  Surely this is what real life managers do.

[smajliss]

Hi guys. I understand theory of adapting. It makes sense until you find out that every manager has attributes which decides his tactical settings. Are you sure that manager with attribute Attacking 20 will play on counter attack against stronger team? What is the point of these attributes if it is all about adapting? :-)

[swansongs]

33 minutes ago, smajliss said:

Hi guys. I understand theory of adapting. It makes sense until you find out that every manager has attributes which decides his tactical settings. Are you sure that manager with attribute Attacking 20 will play on counter attack against stronger team? What is the point of these attributes if it is all about adapting? :-)

It's about degree of response. An Attacking 20 manager might drop to Control mentality in a tough fixture, while an Attacking 1 manager would be playing Contain w/ Timewasting from minute one in a similar matchup.

[smajliss]

Yeah, thanks. There is also attribute Flexibility. So the level of adjusting is more complex than just by reputation/form etc.

[llDracoll]

12 hours ago, DementedHammer said:

The thing that I don't get is that people actually complain about the AI cracking their tactics. I'd be absolutely delighted if the AI was advanced enough to do this and would consider it a massive step forward. It would add a completely new dimension to the game.

I think the ones that complain are not unhappy at the AI figuring out how to beat their tactics, but that, seemingly for no reason, their tactics are not working after some time. They don't consider that possibility.

[theonets]

On 11/09/2018 at 07:57, John_Yuki said:

It's a legend that has been passed down from year to year. There are players that swear blind that the AI always cops on to their tactics after a season or two, and other players who say that the AI doesn't learn like that.

I'm currently playing a Birmingham City save. In our first season in the Premier League (2018/19), we finished 13th - comfortably away from relegation - and won the EFL Cup, meaning we are in the Europa League this season. However, the second season has started awfully, with bad performance after bad performance. I spoke to some people and they also echoed the whole "the AI figured you out" sentiment. However I also have a Cork City save that I am in the 7th season of. In that save I have used the same tactic all 7 seasons, and have only improved my performances season by season winning the title every year, and advancing further and further in Europe. So why would the AI figure me out when I am Birmingham, but not Cork?

It would be really great to get an official answer on this one, as I have been hearing this "rumour" since I started playing FM in 2012, and no one seems to have an official source on it.

There's an actual easy answer to this.

 

In the EPL, all teams are top teams. That is, the difference between the top side and the relegated side is not that one can't control the ball while the other can, it's that one can do more things with the ball, or is more composed on it. 

 

Think of a top player at Norwich and look at a top player at Arsenal and compare stats. Even the Norwich player is a competent enough midfielder, and in the right tactics, could be lethal, even against your team.

 

But with Cork City, 7 seasons in, with you going further in Europe, you're buying and training players to a level that the rest of the country has no chance of competing with. The gap is literally growing every year. I think the prize money in the Irish League isn't even worth mentioning, compared to getting to the group stages, even

[..Valhalla..]

5 hours ago, HUNT3R said:

Once you're a top team (assuming you are) then your rep won't really change and your form will always be good, so there's nothing really more the AI can do.

 

Saying that though, I had someone on reddit who asked for tactical advice as his tactic that "worked" for a few seasons, falls apart against 1 specific team, every time in recent matches against them. Turns out, the AI copied his successful tactic (as can happen) but their version was a better one.   Just better roles and duties, as the AI still has "rules" for how many  A, S and D duties per mentality and used roles that suited the players. He was beaten by his own tactic.  

That's one of the best things I've ever read on here  

As for my save - yeah I'm managing Liverpool, have a huge number of home grown players (around 19 out of 25), most of which have been at the club for years so they're a very tight knit group who also all play together at International level as well. Probably the most well-gelled team I've ever had.