Tactics worse when the team knows it?

[Karnack]

So I've noticed that most of my tactics work alright up until the moment when my team is starting to become competent with all parts of it.

Could it be any logic in thinking that my tactics are so bad that they only work decently when the team doesn't really know it and thus don't really play it the way I've told them to?

This last save of mine I took over the dominating champions in the Swedish Premier League  and tried a new tactic (asymmetric for the first time ever) and we went 6 wins and 1 draw in 7 games, about that time they were at least competent in all areas and accomplished in some. Since then I've had 7 straight losses. Coincidence? If so, it's seems to be a common theme for me.

[bommen]

Im fm17 the ai gets to know your tactics aswell. So if you play the same over and over you get dominated. Tweek your tactics depending on your opponents weaknesses.

[Karnack]

10 minutes ago, bommen said:

Im fm17 the ai gets to know your tactics aswell. So if you play the same over and over you get dominated. Tweek your tactics depending on your opponents weaknesses.

I know thats not true, they only play more defensive if my reputation increases for example, but the team I'm playing have 6 consecutive league wins, doubt anyone is underestimating them. They don't learn my tactic.

[bommen]

Mate. Why would I Lie ?

 

 

[bommen]

Read this if you dont believe me

 

https://thesetpieces.com/gaming/get-better-fm17-someone-knows/

[Karnack]

It's a myth thats been going on forever that the AI learns the players tactics and then adapts.

But nvm, wasn't really relevant to my question anyway.

[bommen]

😳

[herne79]

@bommen Karnack is correct.  The AI does not "get to know your tactics", but the AI is quite capable (and actually pretty good at it) of adjusting how they play against you depending on your reputation, performance during the season or even how you do during an individual match.  If anyone anywhere says differently I'm afraid they're talking out of their proverbial and just prolonging the myth.

@Karnack  Tactical Familiarity feeds into something called "Blend".  This is part of under the hood calculations and is a Match Engine modifier.  Thus the better the familiarity the better the Blend should be and therefore a reduced impact on the ME.  If you are seeing your team play better when you have lower familiarity that's a new one on me.  However, if your team are underperforming when they do have decent tactical familiarity that might indicate a tactical problem.

[lemeuresnew]

2 hours ago, Karnack said:

Could it be any logic in thinking that my tactics are so bad that they only work decently when the team doesn't really know it and thus don't really play it the way I've told them to?

depends on your tactic. if you are convinced that it is happening purely because of your tactical familiarity, then the only logical conclusion would be that your tactics are so bad that once the team follows them they arent playing as well as they should be and had done previously

the more logical conclusion, if this was in my save, would be over confidence. then it is easier to keep a run going than it is to stop the rot. if you are talking to your players like you expect to walk the league, as you are in this post, then that would definitely be the main problem

[Svenc]

With 7 straight losses of a dominating side, that sounds like a horrible, at least experimental, tactic. One that is guaranteed to make for inconsistency. Unfortunately the AI despite all FM myth, is very limited in exploiting the holes in horrible or experimental systems, and mostly does so by random chance. Whichever tactical traits any AI manager has edited into the db may draw it all bust. Some approaches may expose the holes more, some less, and that may influence such inconsistency. Hopefully one day AI will be more intelligent. The more curios I have ever seen was a player fielding an experimental formation for a 2 legged Cup tie. All over the save, the AI had no clue how to respond to it. THe first leg resulted in a 7-1 win. In the return however, the AI switched to 3 forwards, its attacking formation edited into the db, to at least attempt to get back into the tie. The player's experimental formation resulted into all kinds of gaping holes in his backline. Those 3 forwards wreaked havoc, the AI won that match like 8-1, game over.

Generally, as long as matches on this as in football remain to be settled in pitifully key seconds of 90 minutes, never read too much into results alone.  You can go onto a good streak despite making a lot of questionably decisions, as the game tries to struck the balance between players and tactics and also squad management. THerefore, even with getting the decisions spot on, the reverse can happen. At some point, your side may have a bad day too, hidden consistency traits etc.. Tactics never "stop working". In particular if they aren't overly one-dimensional and attack purely down the middle, or anything. However, if you don't roughly know what you are targeting, you may never know what is tactical and what isn't. The spoilers are unsurprisingly in the end products on each end of the pitches, which is the nature of the shots. However, you may also find it a struggle to get into the opposition box from play if they simply have the superior players on the day, etc.

 

4 hours ago, herne79 said:

@bommen@Karnack Karnack is correct.  The AI does not "get to know your tactics", but the AI is quite capable (and actually pretty good at it) of adjusting how they play against you depending on your reputation, performance during the season or even how you do during an individual match.  If anyone anywhere says differently I'm afraid they're talking out of their proverbial and just prolonging the myth.

Which SI do in the above too, mind. [Shameless promotion piece, though that article reads fantastic mind!].

 

 

 

[Karnack]

I'm going to assume it's a tactical problem after reading your posts then. Not surprising at all since I'm horrible at creating tactics

But it might force me to do all of my tactics-experimentation in FMT from now on since I can't trust the results until the team truly gets familiar with my experiment. Just wanted to confirm that this could actually be true.

I'm guessing over-confidence might be part of it too, seeing as at least 3 of the losses against me the opposition only hade 1 or 2 shots on goal whilst I rarely produces less than 10. At least half of them were good I would say, 1-3 CCCs according to stats but I reckon there were a few shots that should have gone in with an attacker in decent form.

[bommen]

Asked the man himself Miles - ☺️

[Bobby_Zamora]

1 hour ago, bommen said:

Asked the man himself Miles - ☺️

That is very different than *learning* how you play.

AFAIK, the AI built in will use the tools available ingame to do this, such as scouting, OIs, etc. There is no AI learning. Otherwise, this would be a huge feature and we would see some crazy tactics from time to time

[bommen]

Thats the Way football works via scouting, but maybe we missunderstand eachother. It is commonly known that supertactics doesnt exist in fm anymore, thats my whole point

[Bobby_Zamora]

29 minutes ago, bommen said:

Thats the Way football works via scouting, but maybe we missunderstand eachother. It is commonly known that supertactics doesnt exist in fm anymore, thats my whole point

Erm, FM is a computer game -- I was just pointing out how I think the game probably handles "knowing" your tactic. There's no machine learning as far as I know.

The reason supertactics (probably) do not exist is due to reasons that are not disclosed, i.e. statistics dictating outcomes of the match engine, toning down factors coded in the match engine, etc. Has to do with great coding.

So, again, I guess that there is no proper learning taking place in FM.

[themadsheep2001]

Statistics don't dictate the outcome of the match engine, nor are there toning down factors. 

The AI does adjust to your rep and the dominating tactical formation trend. Which I assume is what miles means by learning 

[Bobby_Zamora]

10 minutes ago, themadsheep2001 said:

Statistics don't dictate the outcome of the match engine, nor are there toning down factors. 

The AI does adjust to your rep and the dominating tactical formation trend. Which I assume is what miles means by learning 

It does not? There are no likelihood calculations in the match engine?

I assumed the match is played 'kick by kick', thus we got to have a factor there to balance the success of something to someone, even beyond the attributes, otherwise 1 player might always win. I thought real life statistics, as the many stats released by FM's partner ProZone per match, could have been build in this calculation.

[Cap'nRad]

Probably the ai getting familiar with it's own tactics, and thereby getting better at defending yours.

[Svenc]

6 hours ago, themadsheep2001 said:

Statistics don't dictate the outcome of the match engine, nor are there toning down factors. 

The AI does adjust to your rep and the dominating tactical formation trend. Which I assume is what miles means by learning 

It also has had historically limited ways by responding to formations via shouts, and so on. Arguably, FM 2017 has the closest to a "truly" supertactic there was in more recent iterations, as outside of one or two semi-regularly used AI formations, AI cannot defend it if the central spaces are flooded with players. That's naught to do with stats tweaking though. It's that poorest of players visibly pop up in tons of rather undefended spaces for shots that are hard to defend and eventually score a few of goals, regardless of that players' quality. Personally I think such are mostly unfortunate and also highly undesirably from a developer's perspective, as they stop thousands from grasping what the game is trying to do/be (or getting much of an understanding of their results), but yeah. This kind of adapting only happens by chance though, which is why you can suddenly have really weird matches, purely statistically. As soon as you stop looking at stats, and click through each shot in the analysis, you'll note most of them are **** poor though (second half of the post).  Rest assured there is a remarkably difference between such a match and this one (which required one human controlled team to have a horrible defensive shape AI don't engage in, but space is space, and no space is no space).

@Karnack Which such a run with such a dominating side (Malmö?) you could be sorted pretty quickly. You could do it then yourselves, in a couple seconds. Provided that was tactical, and you not fielding every player out of position, including the forward in goal.

[lemeuresnew]

23 hours ago, Karnack said:

I'm guessing over-confidence might be part of it too, seeing as at least 3 of the losses against me the opposition only hade 1 or 2 shots on goal whilst I rarely produces less than 10. At least half of them were good I would say, 1-3 CCCs according to stats but I reckon there were a few shots that should have gone in with an attacker in decent form.

what type of goals are you conceding? i always found if i was letting in there only shot or 2, it was because i was overly aggressive and thus easily hit on the break. nothing wrong with that when youre scoring, but if your strikers have a bad day then 1 or 2 counters can really hurt... quality always pays off better than the number of shots

i wouldnt trust the CCC stats, always very dodgy   

 

[Karnack]

1 minute ago, lemeuresnew said:

what type of goals are you conceding? i always found if i was letting in there only shot or 2, it was because i was overly aggressive and thus easily hit on the break. nothing wrong with that when youre scoring, but if your strikers have a bad day then 1 or 2 counters can really hurt... quality always pays off better than the number of shots

i wouldnt trust the CCC stats, always very dodgy   

 

Stats would suggest through balls but I would say individual mistakes to a large extent as well. One goal was my CB just standing with the ball close to the box and allowing an opposing player to run from almost the half line to take it of him and score. There was no pressure on him, he could easily pass another CB or one of the FBs or even the GK. 

I take CCCs with a large pinch of salt, but I've had several good quality chances where the striker just have to tap it in and it either go straight over or right into the arms of the GK.