Tactical experiments

[Exius]

I decided to make a small experiment to check some tactical instructions and player roles. The thing is that most of the instructions are not explained clearly in the game or even in the manual - this is something that should be improved considerably in the next iterations of the game. The best thing to get clear with something is to try it yourself and see the difference. But the problem is that the difference itself can be influenced a lot by the opposition team instructions, by attributes, morale and condition of the players, by various small factors and even pure luck. So, to prepare clear conditions for the experiment I took my U21 and U19 teams, changed all the players so that all their attributes are 20, they are of the same height and weight, have excellent morale and tactical familiarity and able to play in any position (I also removed all their traits). I decided to use my junior teams to remove any home/away side advantages or disadvantages and I'll be able to control both the teams. I'll use no team-talks and make no changes during the match. The players are in excellent condition and have excellent stamina, and I even changed the weather, so it is a warm and dry evening. In other words, I tried to minimize any factors which I can influence on team performance aside pure tactics.

The first thing I decided to check is Tempo instruction. It defines how fast players try to make decisions and move the ball, thus influencing how fast the ball moved to the goal.

Both teams play the same formation - 4-2-3-1 with 2CDs, WBs, AP-A and DLP-S in CM strata, AM-A, two wingers and Poacher as a striker. U-19 has no TIs and plays with Ultra-defensive mentality. U-21 plays with positive mentality and has the following TI: Play out of Defence, Pass shorter, Higher line of engagement, Counter-attacks, Higher D-line, Distribute to centre-backs. I will change tempo of U-21 team from game to game and see how is this reflected on the team performance. Since a single game can depend on various random factors, I'll play at least three games to see, if there are any trends. So, I'm playing 3 games with the lowest tempo, 3 with standard tempo and 3 with the highest tempo.

And here are the results (average positions and match stats). U-21 team (which tempo I'm changing) on the left and U-19 team (no TI) on the right.

Low tempo:

 

Medium tempo:

 

High tempo:

Frankly speaking, the results are very confusing. It seems either adjusting tempo doesn't change anything, or players with max attributes perform the same no matter what tactical instructions are. I expected to see more passes with the highest tempo, more attacks, more shots. Only 1 game of 3 with the highest tempo resulted high number of passes, while the highest number of shots was achieved with medium tempo. And the second highest number of passes was achieved with the lowest tempo. After playing these 9 games I don't see any patterns affected by tempo instruction at all.

Another conclusion from this experiment is that playing with ultra-defensive mentality not always will result your team position very low on the pitch. In fact, in most of the games average positions of the teams is very close to each other. In a couple of games the team with ultra-defensive mentality was even positioned higher than the team with positive mentality.

And the third conclusion is that the team is capable to play even without any instructions, changes or team-talks during the match. They know themselves when to go forward and when to sit back. I watched all these games in Commentaries only mode, looking at the average positions screen. I saw after conceding a goal or couple of goals, the losing team went forward and tried to equalise, even though their mentality was ultra-defensive, while the winning team was forced to sit back despite their positive mentality. Of course it could be affected by the fact that every of 22 players on the pitch was the best possible player in the game - in technical, mental and physical aspects.

But what worries me the most - that I couldn't find any differences by changing tempo. So, does it really affects anything? Or maybe I made some mistakes while preparing this experiment? Please share your thoughts and suggestions what can be tested further.

[wixxi]

3 hours ago, Exius said:

I decided to make a small experiment to check some tactical instructions and player roles. The thing is that most of the instructions are not explained clearly in the game or even in the manual - this is something that should be improved considerably in the next iterations of the game. The best thing to get clear with something is to try it yourself and see the difference. But the problem is that the difference itself can be influenced a lot by the opposition team instructions, by attributes, morale and condition of the players, by various small factors and even pure luck.

Are you trying to say that this is not the case in real life too? The game is a simulation after all and in real life the way a team plays is influenced by such factors that you have listed except they are slightly more abstract in the sense that we can't really give a real life player attributes in a numerical way

[Exius]

11 minutes ago, wixxi said:

Are you trying to say that this is not the case in real life too? The game is a simulation after all and in real life the way a team plays is influenced by such factors that you have listed except they are slightly more abstract in the sense that we can't really give a real life player attributes in a numerical way

Oh gosh, my whole post was about completely different thing, and you took the least meaningful sentenсe out of it... Of course not. I explained why it's not always relevant to make experiments against AI team - they can react to your tactic and what you'll see - might be a result of AI changes but not yours tactic. You need to create more "sterile" conditions for the experiment, and I tried to make them. But it doesn't seem to work.

[wixxi]

Really don't get the point of these type of experiments where you create a lab within a save since you cannot draw any meaningful conclusions that apply to the game everyone else plays, since nobody playing the game normally will have this type of situation where their players have max attributes. While the best way to understand the differences within the game are definitely to try it yourself, why bother trying it with a situation like this when you won't be able to apply anything to your regular saves since the context will be completely different?

[HUNT3R]

While the experiment was fine, imo, you're looking at the wrong thing, made wrong assumptions and so you drew completely wrong conclusions.

Tempo just affects how long a player has the ball before making a decision. If I wanted to see the effect of this, I would literally just watch the match to see the difference between the tempos. 

Tempo isn't pass direction. So they could play backward or sideways still. More attacks could have been started, but the higher tempo could have also had them make more mistakes. The slower tempo could have been too slow, resulting in no options but to shoot. The tactic could have been poor so no matter what the tempo, you will have similar stats from it.

2nd conclusion is also pretty wrong. The average heat map only shows so much. Possession was about equal, so that tells me the defensive team had the ball quite a bit. So they were able to keep hold and work it forward somewhat. The attacking team could also have struggled to break through the defense, so they're also going to be in a similar average area. Again, watching at least some of some of the matches should show you where the teams are, how far they do get forward or back.

 

[sporadicsmiles]

12 hours ago, Exius said:

I expected to see more passes with the highest tempo, more attacks, more shots

More shots yes. Why more passes? Higher tempo means you do everything faster. So assuming the opposition does things the same (which you tried to ensure) faster attacking phases means more chance for the opposition to keep the ball. I would not be surprised to see lower passes and lower possession in that case. Conversely a slow tempo should result in more passes because buildup play will be, well, slower.

The problem with things like this is you would need a truly huge sample size to draw any meaningful conclusions. Football is a chaotic system. As you see, identical starting conditions can lead to wildly different finishing points. It really does not make sense to compare few results in a chaotic system. You need hundreds so you can start to try to spot trends, do clustering analysis, etc. So I would not worry too much about not seeing any trends on such a small sample. Plus tempo is a bit esoteric. How do you measure tempo from those stats? I'd say you cannot. You would need to measure something like the average time between a pass. Which is not something you really want to do for an entire football match unless you are actually a data analyst! I'm not trying to criticize you trying, just wanted to point out a few flaws in the design. An easier experiment to do would be work ball in box vs shoot on sight, for instance. The outcome can be measured with different shot types and locations, which is all easily accessible.

[herne79]

12 hours ago, Exius said:

changed all the players so that all their attributes are 20

Just a note on this, that's not how it works.  Whilst you can indeed change all attributes to 20 on screen, the game won't recognise it.  That's because changing all attributes to 20 would exceed the maximum allowed Current Ability (CA).  If you moved the game on a few weeks you would notice those same players suddenly have all attributes much less than 20 as the game would adjust and rebalance attributes back to within CA limits.

You also haven't mentioned anything about hidden attributes such as Pressure, Consistency, Big Matches and so on.  All of those will have every bit as big an impact on things as visible attributes do.

One other point:

12 hours ago, Exius said:

for the experiment I took my U21 and U19 teams

Players don't try as hard during friendlies as they do in competitive matches.

To an extent I agree with your sentiment that some descriptions in game could be clearer.  The problem can become one of information overload - there is so much to do and we get thrown in at the deep end that it can become bewildering, so trying to keep the sheer volume of information within acceptable levels is almost an impossible task.  Which is kind of where this forum comes in handy.  So overall I can see where you're coming from in this experiment, however I think all you're really doing is confusing yourself even more.

If you want to know more about the game, read some of the guides in the pinned thread at the top of the forum; watch rashidi's videos on youtube (bustthenet channel); or just ask .

 

[Exius]

Thank all of you for the answers. All of you have a point but the main problem here was in something else. We all missed one thing - what actually a player with all attributes 20 would do in a match. I think nobody knew it, because nobody tried. I tried to watch at least one game in full mode and I was surprised with what I saw. The was one more problem with the results I provided - this is my regular tactic, and when I play it with my regular team I usually complete around 400-500 passes. And here it was only around 200 passes completed by perfect players. I thought the problem was that all of them are perfect in defending, so they just don't let to play each other, but it's not the case. The problem was that every player just tried to dribble to the opposition area - even defenders. They literally took the ball out of the defence and dribbled to the opposition area, without passing it to anybody. And then they either tried to shot, or passed the ball, or lost it. I tried to make some tactical changes and add more TI and PI which can increase number of passes - dribble less, pass much shorter, be more disciplined, work ball into the area, etc., but it didn't produce any effect. And then I realised that the problem was in those "20" attributes.

It seems, that players are actually "aware" of their attributes! I mean, they know that they have 20 Dribbling and try to dribble more even though they don't have such trait (I removed all traits). And 20 Flair (and maybe some other mental attributes) made tham ignorant to my tactical instructions. So, if Aggression is some sort of natural pressing intensity, than Flair is a natural creative freedom. So, a player with high Flair does what he thinks is best (and what he is best in) but not what you tell him to do. It's not a problem when it is a midfielder with 20 Dribbling and Flair, but defenders are not supposed to have such attributes, so it ruined the game.

I changed all players attributes to 12 and after that I saw them really following my instructions (in terms of passing and mentality). But changing tempo still doesn't reflect on team stats... I'll share new results a bit later.

10 hours ago, wixxi said:

Really don't get the point of these type of experiments where you create a lab within a save since you cannot draw any meaningful conclusions that apply to the game everyone else plays, since nobody playing the game normally will have this type of situation where their players have max attributes. While the best way to understand the differences within the game are definitely to try it yourself, why bother trying it with a situation like this when you won't be able to apply anything to your regular saves since the context will be completely different?

The point is to minimize any other factor which can affect player's performance than tactics. Of course it's not possible to do completyely, but I tried to. And why I initially chose to set all attributes to 20 - because a mistake is also a random factor and I wanted players to make less mistakes, thus minimizing this random factor. But it worked in the way I didn't expect.

4 hours ago, HUNT3R said:

Tempo just affects how long a player has the ball before making a decision. If I wanted to see the effect of this, I would literally just watch the match to see the difference between the tempos. 

Tempo isn't pass direction. So they could play backward or sideways still. More attacks could have been started, but the higher tempo could have also had them make more mistakes. The slower tempo could have been too slow, resulting in no options but to shoot. The tactic could have been poor so no matter what the tempo, you will have similar stats from it.

I still don't get it. How can changing tempo have no effect at all on the stats? More mistakes - than it should be a higher number of passes or shots attempts. I tried to do this with a 4-2-3-1 preset without any TI and PI at all for both the teams, so it shouldn't be the tactical issue. And I tried to watch, but I don't see any considerable difference. Yes, with lower tempo players take a bit more time with the ball to make a decision - they can make a small pause with the ball (like dwelling on a ball), but I don't understand why it doesn't affects match stats. If they stay with the ball longer - than it should be fewer passes made, but it's not!

 

5 hours ago, sporadicsmiles said:

An easier experiment to do would be work ball in box vs shoot on sight, for instance. The outcome can be measured with different shot types and locations, which is all easily accessible.

I agree with you - 3, or even 10 matches is very little sample size. But unless you are a SI developer and can perform automatic tests, I'm afraid there is no way we can do something else here. But it would be very interesting indeed!

Yes, I'm planning to test these TIs as well, along with some others (play out of defense, hit early crosses, compare players stats with various roles and duties).

 

4 hours ago, herne79 said:

Just a note on this, that's not how it works.  Whilst you can indeed change all attributes to 20 on screen, the game won't recognise it.  That's because changing all attributes to 20 would exceed the maximum allowed Current Ability (CA).  If you moved the game on a few weeks you would notice those same players suddenly have all attributes much less than 20 as the game would adjust and rebalance attributes back to within CA limits.

You also haven't mentioned anything about hidden attributes such as Pressure, Consistency, Big Matches and so on.  All of those will have every bit as big an impact on things as visible attributes do.

I used an unoficial ingame editor to do this. It turns a player to superstar by setting all his atributes (including hidden except injury proneness and controversy) to 20, and his CA and PA to 200. Anyway I already know that this is not needed and actually ruins my tests.

5 hours ago, herne79 said:

If you want to know more about the game, read some of the guides in the pinned thread at the top of the forum; watch rashidi's videos on youtube (bustthenet channel); or just ask .

Of course all of these are great guides, but, you know, you learn something much better when you try and see it yourself than just read about it. I'd like to see an exact effect of some tactical instructions. I'm laso struggling with understanding some player roles and their differences and that's something I'd like to test in this experiment as well.

[HUNT3R]

2 hours ago, Exius said:

I still don't get it. How can changing tempo have no effect at all on the stats?

It absolutely can. There's a lot of context that you could miss, as I've said.

2 hours ago, Exius said:

More mistakes - than it should be a higher number of passes or shots attempts.

Maybe. Maybe not. Again - context. Slower tempo could result in fewer pass attempts, but being caught in possession more. Higher tempo could result in the same number of passes because a 'move' still only goes from dlp to ap to winger and lost. Stats only tell you so much and in the case of tempo, as I've said, you're looking at the wrong thing.

2 hours ago, Exius said:

I tried to do this with a 4-2-3-1 preset without any TI and PI at all for both the teams, so it shouldn't be the tactical issue.

This doesn't make sense, at all. A preset isn't perfect. It's literally just a preset for a style of play. No taking into account team or players or anything. Did you pay attention to how it played out?

2 hours ago, Exius said:

And I tried to watch, but I don't see any considerable difference. Yes, with lower tempo players take a bit more time with the ball to make a decision - they can make a small pause with the ball (like dwelling on a ball), but I don't understand why it doesn't affects match stats.

So you did actually SEE the effect of tempo. That's all to be taken from this experiment. 

 

2 hours ago, Exius said:

If they stay with the ball longer - than it should be fewer passes made, but it's not!

Why? This is again just an assumption. You would be able to see in the matches why it isn't. What if they have more time to recycle the ball? What if there's more time to attempt a pass? 

 

There's a tactical element that you're ignoring. Tempo could have an effect on match stats. In this case, it didn't. You did see what tempo did though. So what happened? If I had to guess, I'd say that the ball goes to the playmakers and they pass to a winger and you lose it there. Whether it happens fast or slow, the pattern remains the same. Tempo doesn't affect that - that's tactical. But you watched the match. You can observe what patterns of play develop, how quickly the ball is played, to which player(s), etc.

This is very much like the "roam from position" instruction issue that I ran into. Sure it sounds great to have all your players roam to find the free space, until you find that 2 or 3 of your players are trying to find the same space - because that's what's available. I had that with a 442 diamond. I realised that while it sounds good, it's more a case of applying that instruction with more thought and purpose. It's the same here.

Tempo is just what it is - how long players have before making a decision with it. That's all it is and you saw it too.

A slower tempo can see players get more time to make a proper decision, but also more time for the defense to get into position or close down the carrier. Faster tempo could catch a defense out of position, but there's less time to act, so it requires faster thinkers.

[sporadicsmiles]

6 hours ago, Exius said:

I agree with you - 3, or even 10 matches is very little sample size.

I agree it is interesting, and you are aware. It is just a caveat that sometimes you will not be able to draw conclusions. I am always interested in people making little experiments with the game, so I shall continue to follow!

[Exius]

21 hours ago, HUNT3R said:

Tempo is just what it is - how long players have before making a decision with it. That's all it is and you saw it too.

A slower tempo can see players get more time to make a proper decision, but also more time for the defense to get into position or close down the carrier. Faster tempo could catch a defense out of position, but there's less time to act, so it requires faster thinkers.

Yes, but the problem is, what I saw - is not a clear pattern. I made another test - I took a Control possession 4-2-3-1 preset with default TI for both teams. For one team I set minimal tempo, for another - maximal. Yes, sometimes players with the lowest tempo dwelled on the ball, but sometimes players from the team with the highest tempo did it either! I watched the whole match, but if I didn't know the tempo of both teams - I couldn't tell it only by watching the game. The only difference which allows you clearly say who's playing with a higher tempo - when there are set pieces, players with high tempo don't wait for the ball and run to teturn it to the spot themselves. That's all. Anything else - they do in both teams. Players with high tempo can take time with the ball while players with low tempo can perform quick actions with the ball (like first time shots or passes), and they do it quite regularly. So, if there is such a small difference between the highest and the lowest tempo - then what's the point in this instruction at all?

I think that's because the tempo should define how fast a team tries to get to the goal. When we say IRL that a team plays with high tempo - we mean that it rushes to score, makes quick transitions, a lot of sprints, etc. In FM it's controlled by mentality and passing directness, while the tempo controls comething I can't clearly see.

Another thing is that tempo varies naturally through the game. Players can't play all 90 minutes in high tempo. Sometimes they slow things down to take a break, and in such pauses the team with slow tempo may hurry things a bit. So, after all, they all look pretty similar through the whole match.

If you think that it's poor choice of tactic to see the effect of tempo, could you please recommend me a tactical preset which should be affected more by changes in tempo? Or maybe change some other conditions to see the difference more clearly (make one team weaker or less motivated, or smth else)? 

[Exius]

Here are results of several more games. As I said, 4-2-3-1 control possession preset was used for both teams with default TI and no PI or OI.The only difference is that the team to the left plays with the slowest tempo, the team to the right - with the highest.

As you can see, tempo doesn't affect stats in any way at all. It doesn't affects, shots, possession, passes.

The only useful conclusion which I can carry out of this experiment is that tempo doesn't affect passing accuracy. You can play short passes + high tempo and the accuracy will be the same as short passes + slow tempo. It doesn't seem realistic to me, but that's how it is in the game.

[CARRERA]

maybe one thing you might concider is that there is also creative freedom where players are allowed to make their own decisions. so properbly remove that freedom and you will have your players stick more closely to your instructions.

[herne79]

43 minutes ago, Exius said:

As you can see, tempo doesn't affect stats in any way at all. It doesn't affect ... possession ...

Except it did in every single example.

But why expect to see big swings in possession when you are telling both teams to play a possession based game?  Both teams have plenty of support duties, both teams use shorter passing (which also reduces Tempo btw), both teams are pressing more urgently.  So both teams are aiming to control possession.  Yet the team you tell to reduce tempo gains more possession overall than the team you tell to increase tempo.

Possession in FM is defined not by the number of passes (as in real life) but by time in possession of the ball.  And the team with lower tempo had more possession.

Also, as I said above, you are playing a friendly and players don't try as hard during friendly matches.  And if you are persisting with messing around with player attributes, CA and PA, what are you doing now?

[Exius]

I played a bit more games (10 overall), so here are updated results:

What can I say? Indeed, a team on a lower tempo has a bit more possession (in 5 games they had +8% possession, in rest of the games possession was equal or nearly equal). An even when they don't have considerably more possession, they still make more passes (players find more time to move into position to receive pass?) - in 7 games from 10 the team with lower tempo made 10-15% passes more (I mean number of passes, but not accuracy percent). So, tempo alone may have some influence on possession but it doesn't reflects on any other stats. Lower tempo = more passes + a bit more possession.

1 hour ago, herne79 said:

But why expect to see big swings in possession when you are telling both teams to play a possession based game?  Both teams have plenty of support duties, both teams use shorter passing (which also reduces Tempo btw), both teams are pressing more urgently. 

Could you please suggest a more suitable tactic to test tempo influence? I mean not only influence on possession, but any other stats?

1 hour ago, herne79 said:

Also, as I said above, you are playing a friendly and players don't try as hard during friendly matches.

Yes, in friendly matches players don't try as hard, but, as I explained, I tried to exclude home/away advantage/disadvantage, which I believe may have much more influence. And there is no other way to control both teams and play on neutral ground other than to play a friendly between my junior teams.

1 hour ago, herne79 said:

And if you are persisting with messing around with player attributes, CA and PA, what are you doing now?

Now I've set all attributes to 12, with CA 120 and PA 150. As I said, the editor I'm using makes it automatically - you only need to choose player level you'd like to achieve - from very poor to superstar. The reason why I'm doing it is because I'd like to exclude the difference in players attributes from equation.

[Exius]

On 09/08/2019 at 10:29, sporadicsmiles said:

An easier experiment to do would be work ball in box vs shoot on sight, for instance. The outcome can be measured with different shot types and locations, which is all easily accessible.

Here you are. Yet again I used 4-2-3-1 control possession preset for both teams with default TI. U21 team has Work ball in box by default. U19 team has Shoot on sight instruction. I also set Be more disciplined instructions for both teams to make sure they follow the tactical instructions.

I played only 5 games because there are no surprises here. The team with Shoot on sight instruction usually had more shots but many of them were outside the box, while the team with Work ball into the box instruction had less shots but most of them were inside the box. This is pretty obvious and expected. There was one game where U19 team had less shots, and there was a game where U21team had most of their shots outside the box despite their respective instructions. I didn't watch the games (I was watching Napoli-Barcelona friendly at the moment) but I can assume what happened there - things went wrong for the teams in that particular games and they struggled to get to positions to shoot. U19 team had produced some shots due to their Shoot on sight instruction (maybe they would have had much less shots without it), but most of them were outside the box. U21 team was forced to shoot outside the box even though they had work ball in box instruction, but without this instruction they would have had even more shots, still outside the box because they failed to get into better positions.

So, Shoot on sight instruction will increase the number of shots. Many of them will be outside the box. If you struggle to produce the shots, you may start producing some (outside the box mostly). But that instruction alone doesn't mean that all your shots will be outside the box. If your players succeed to get to positions to shoot inside the area - they will shoot from there. They just may have some more shots outside the box as well.

[Warrenwwr]

These experiment threads are great. I've seen some impact from every setting I change in this yrs fm.

Try the tactic template for tiki taka and vertical tiki taka but keep all position settings at default I guarantee you will notice the difference in those extra notches of tempo.

That is granted the players are available to execute possession based tactics.