The impact of positional attributes

THIS THREAD HAS CA SPOILERS!! Be warned!!

I have always wondered about how important is the positional rating of a player to his performance in a given position. It is fairly difficult to answer that question empirically, because so many factors influence a player's performance in a match - I do not try to answer this question with the research I will describe to you... But I think a related question can be answered quite precisely - how much does the AI managers value the positional attribute vs CA.

Since FM09 we have been able to see the assistant manager/coach star assessment of players in positions where the player is only accomplished. I have noticed, as certainly most of you have, that when deployed at non-natural positions, coaches/assist managers lower the assessment of a players ability up to 2 stars. This led me to wonder to how much CA points does one point in positional attribute correspond to. To find this out I did the following:

1- Set a player to have 1 in every position, except 20 at AM C and 20 at S C. (David Villa at Valencia)

2- Changed my assistant manager JPA and JPP to 20,20 - in order to get the most accurate assessment possible.

3- Set Villa potential to 200 with FMRTE.

4- Moved Villa CA around to find the boundaries of the different star ratings (for instance 4 stars was given for CA=[169,175] and 2 stars for CA=[130,137])

5- Changed Villa rating as a striker to 19, and repeated step 4. Changed him to SC = 18 and repeated step 4, etc...

6- Having done this up until SC = 15, I averaged how much did the boundaries of the star assessments change in response to a change of 1 in the S C positional rating of villa, and came to the conclusion that 1 point in the positional attribute is equivalent to 6!!! CA points, at least in the way the AI sees a player.

A clear recommendation coming out of this analysis is for training a player for a position if you plan to use him on it, even if he is already accomplished at it. Good news in this regard is that is much easier to get better in a position since FM09, especially if the player is not only training but also playing in the position. It is even possible to get players to become natural in a position, unlike FM08 and before.

On my view though, this is a huge overrating of the importance of positional attributes. C Ronaldo has competence 14 as a S C, therefore, if my calculations are correct, from the point of view of the AI he is as good as a 154 CA SC playing at the striker position (eg. Tommaso Rocchi)... seems like too heavy a penalty to me.

Edit: In the original post I was off in the the starting CA of Ronaldo by one point.

Interesting.

You example of Ronaldo having 155 for the postion of striker sounds about right to me tbh.

But what about Fabregas only being 118 as a AM C (AM C=11), instead of 172 as a M C, or Rooney 142 as a AM L (AM L 14), instead of 178 as a F C?? Does it not seem like quite a heavy penalty?

But what about Fabregas only being 118 as a AM C (AM C=11), instead of 172 as a M C, or Rooney 142 as a AM L (AM L 14), instead of 178 as a F C?? Does it not seem like quite a heavy penalty?

the fabregas example does seem like a very harsh penalty.

Very interesting thread- yeah does all so seem harsh to me- world class forwars with the correct attributes should be able to play a number of positions and not be penalised too harhly for not playing in their 'so called' best position. And a change of 54 points from CM to AM is just ridiculous IMO- how can anyone defend this. Yes primarily he is an CM but also ventures forward and argubaly due to Arsena'ls inter changable positions/tactics then he should be a lot higher in AM. And yeah I agree about Ronaldo especially when comparing to the Italian player. Very good find indeed.

Note: Try and get player naming in the title or rather CA- should have spoilers I think.

I would say that probably Fabregas should be rated more than 11 at AM C, but my point is not to attack the researchers in this thread... I just think that people should be aware of how much positional attributes influence performance on the field, or at least how much SI thinks they influence, as they are making the AI pick teams according to these evaluations.

I would say that probably Fabregas should be rated more than 11 at AM C, but my point is not to attack the researchers in this thread... I just think that people should be aware of how much positional attributes influence performance on the field, or at least how much SI thinks they influence, as they are making the AI pick teams according to these evaluations.

Oh know I was not attacking the researher(s) it was just IMO and I'm sure there will be people saying the number of 11 for AM is fine or go the other way.

I'm guessing Fabregas is accompolished in AM?

No, he is just competent.

No, he is just competent.

Yeah I guess that will be why he only has 11 for the position.

thanks for the analysis. i am going to train my players to become natural for their positions.

I reckon the AI recognizes the impact of positional attributes. That's why if you enable the view to see player's new position, so many players are being retrained by the AI.

Also, the relative ease of retraining a player to natural probably means it's something which SI would recommend us to do. I also had a regen who went from accomplished to natural solely by playing in that position, without any retraining.

Before everyone goes to train players in natural positions, how will the extra positional training, which takes 10% for the whole duration the training is on, affect attribute and CA gains? I have tried positional training, and took them off, to see significant differences in attribute gains.

Even if that is the case John, I would still strongly recommend training players to be natural in the positions they are used, once CA reaches PA and no big attribute improvements due to training happen anymore.

Then it should only apply to players over 24, as training players in new positions will take off much of their attribute gains.

Not necessarily, as players bellow 24 can already be close to potential, and also the performance boost received by being good at a position (and its effect through the average rating on the development of a player), might be worth the loss of training time.

I don't think the experiment is entirely conclusive.

1. Did your asst manager have a CA of 200? Having an AM with CA 100 vs CA 200 might have an effect on the experiment, regardless of a JPA and JCA of 20.

2. Did you try it with a AM with JPA / JCA lesser than 20? That would act as a control to compare against.

3. The positional ratings were only adjusted down to 15. How about those below 15? Positional effects may not be linear. You could have a x loss in CA down to 15, then a y loss in CA down from 15. An 'accomplished' player may not suffer as badly CA-for-position-rating compared to an 'awkward' player

4. Tried it for players with a different PA? This will find out the limits of the JPA/ JCA system.

5. A better system would be to play a player with position rating at 15 in that position for a season, then compare him at position 20 for a season, and find out the difference in match ratings.

Before answering each one your comments, I should say that I state on the first post that the aim of this test is not to say that each point bellow 20 in the positional rating will make a player play at level of CA 6 points lower - this would be very hard to test in practice due to alot going on at the same time... the aim of the test is just to show that the ai managers (I suppose) and our assistant managers (for sure) think that is correct trade off between CA and positional ability. Now answering your comments:

1- No, I didn't set the CA to 200, but I have played around extensively with the Assistant Managers attributes, and the only ones who seemed to have any impact in the star rating given by assistant managers was JPA/JPP, determination, ambition and level of discipline (I have also set these last three to 20 during my experiment). If you do not believe me just load FMRTE, change the attributes of assistant managers around and you will see.

2- In fact I have done that too to see how important was the JPA/JPP ratings of assistant managers, and I have found that assistant managers with less JPA/JPP are less precise in their star ratings (which is expectable) and tend to bunch players more in the middle (eg., instead of spreading the stars, tend to give more 1.5, 2 and 2.5 stars).

3- Indeed I have not done anything bellow 15 - my comments on how much players with 14 or 11 at the positional rating would suffer were extrapolation. On my defense I should say that it isn't easy to get the star ratings the assistant managers would give to players with less than 15 in a position, because it no longer shows in the player profile. I agree it might not be linear, but I have no way (at least easy) of testing it. I should say though I have good reason to believe it continues to be linear, because it was fairly linear between 15-20.

4- I have not tried for players with lower PA. I didn't want PA to influence the results - I was just concerned with CA.

5- Definitely this would be a better way to see the influence in the match engine... it would also take 1000x more work to do a test with any kind of statistical significance.

Your idea is interesting. It's just that there's too many limitations in the method and results, plus a lack of comparison between the effects of positional gain vs training loss. Based on my personal observations, training (and retaining the training of) a player in a new position has a significant effect in attribute gains. What advantage is gain in positional training may or may not make up in attribute loss.

Hence the only thing I would agree on is that positional training is 'safer' in players over 24, where they can't gain much in CA, or in players who have already reached or are very close to their PA. Not everyone should happily go retrain their players in new positions unless necessary.

I agree with your last sentence, not because of training, but because as Paul C confirmed here (http://community.sigames.com/showpost.php?p=4405568&postcount=450) in FM10 to avoid a exploit related to players versatility (present in FM09 and 08) players that are versatile have a small penalty on their attributes. Therefore I would only recommend to train a player to a position if you plan to play it there regularly. That's what I do myself.