Current Ability and Atrributes Research

[Law_Man]

Originally posted by trebor1185:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content"> On 07 I did look at this, not in any strict mathematical manner but on that version some hidden attributes appeared to be separated from CA.

This seems to be the same in 08 from what i've seen in some other tests i've just done. </div></BLOCKQUOTE>

Yes and this would make sense to me really given the difference between ability in the generic sense (i.e. mainly technical, "skills") and attributes that are innate or inherent such as how well a player deals with pressure etc.

[kolobok]

First off kolobok since you do this for a living throw your manners out the window. If anything I post about the maths is completely wrong let me know since I'm working off the little bits that remain in my brain of things I studied 10 years ago but never actually used in practice. Plus I don't want to plug through it when my fundamental methods are completely arseways.

Not to derail the thread but out of curiosity, what do you work on?

Also does using matrices to solve sets of equations sound familiar to you? There's a part of my brain that's telling me there's an easier way to solve those equations using matrices rather than step by step substitution. But I don't know if it's my memory mixing things up. I've been on the net trying to find a specific link but can't nail it down.

That's good stuff in your analysis. It really helps to see some numbers that appear to back up the assumptions of position affecting weightings. Out of intellectual curiosity what was your 'little trick'?

Another issue which could screw up my method and came to mind as I was picking MCs for using in the equations. What if my assumption of 'natural position' as the controller is flawed and it is in fact a weighted combination of all positional ratings? I might just stick with my assumption and tests the results. If they're completely wrong then that may point to it being false (or me not really knowing what I'm doing Big Grin)

Sorry, I was busy for a couple days.

I did not see anything wrong in any idea you proposed. Sim. equations is always better method assuming you have a complete system, so if you can solve it for one group of players and confirm results by applying them to another group, your method is better than anything. The question is whether there is a random or hidden part of the equation (hidden attributes, personality, level of a team a player plays for, versatility, etc.). If there is one, then the results may vary depending on a group you chose. Though, I don't think we need the exact formula here, just close approximation. Btw, that's why I don't really care about making sure all classical statistical assumptions are in place for this experiment. Even if they are not, it means we would have a slight error in the coefficients, which is not important for approximation.

To me the most important parts would be to know:

a) which attributes are considered the most important for each position and in general (For example, I suspect pace is valued very high) - so that we could experiment with training and have better judgment for players by positions.

b) how CA translates to PA and (not sure how to model it, but anyway).

Time permit I will try to experiment next week and post my findings ASAP. It would be extremely interesting to compare results with yours.

To answer your questions: I develop credit risk models, and the little trick I did is assigning dummy variables by position (i.e. ST variable = 1 if position is ST, 0 otherwise).

[Law_Man]

So you're a risk analyst then Kolobok? That's a bit more my kind of language!

Cannot wait to see yours and isuckatfm's results and analysis! KUTGW

Btw, isuckatfm said he knows how to find out which attributes are important I think.

[trebor1185]

a) which attributes are considered the most important for each position and in general (For example, I suspect pace is valued very high) - so that we could experiment with training and have better judgment for players by positions.

This is what i've been looking at, and from what i've seen so far it appears that technical + mental attributes have a higher value than physical attributes.

I can show you some screenshots of the players I tested if your interested.

[kolobok]

Btw, isuckatfm said he knows how to find out which attributes are important I think.

Once he finds those weights in the equation, they will tell. The higher the weight, the more important the attribute. I should be able to find them as well. Then we compare results and will go from there.

[Hawshiels]

Originally posted by kolobok:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Btw, isuckatfm said he knows how to find out which attributes are important I think.

Once he finds those weights in the equation, they will tell. The higher the weight, the more important the attribute. I should be able to find them as well. Then we compare results and will go from there. </div></BLOCKQUOTE>

I don't want to spoil the fun for you, but something you may want to consider when calculating the weightings is this:

- Use the low CA players to make it easier

- Use 5-6 players with the same CA (e.g. 10) and then 5-6 players with te same CA (e.g. 11) and then a further 5-6 players with CA (e.g. 12)

Doing these first few steps will make some of the further calculations you need to do much clearer.

- Positions are not taken into account (for all outfield players)

- There are 2 or 3 of the attributes that have a floor higher than 1 (but you'll see this if you use simultaneous equation sets for each of the technical and mental attribute areas)

- You'll perhaps want to consider that the true CA is not clear until you take into account the other hidden attributes ( )

- There are 48 attributes and traits that are important if you want to get to your answer.

p.s. If you need the spreadsheets I used for this, I can upload them somewhere for you.

[Law_Man]

trebor - yeh go for it

Kolobok - cool, should be really interesting to see the results!

[isuckatfm]

Just to update I got some software (Maple 11 if anyone is familiar with it and can help let me know)to solve the equations but I'm learning it from the ground up and things aren't quite going to plan.

I ran the equations through and it gave me a unique solution for the weightings, but it gave negative values for some of them. So I need to figure out how to constrain the weighting values so that they are weightings i.e. greater than 0 and less than/equal to 1.

But if I do figure out how to do that, there is still a possibility that introducing the constraint will mean there is no exact solution. If this happens it means the original model and assumptions it's based on are flawed. In which case a statistical approach like kolobok's is a quicker approach and as he says will give a good enough approximation to improve understanding of the system without being 100% accurate.

originally posted by Hawshiels:-

I don't want to spoil the fun for you, but something you may want to consider when calculating the weightings is this:

- Use the low CA players to make it easier

- Use 5-6 players with the same CA (e.g. 10) and then 5-6 players with te same CA (e.g. 11) and then a further 5-6 players with CA (e.g. 12)

Doing these first few steps will make some of the further calculations you need to do much clearer.

- Positions are not taken into account (for all outfield players)

- There are 2 or 3 of the attributes that have a floor higher than 1 (but you'll see this if you use simultaneous equation sets for each of the technical and mental attribute areas)

- You'll perhaps want to consider that the true CA is not clear until you take into account the other hidden attributes ( )

- There are 48 attributes and traits that are important if you want to get to your answer.

p.s. If you need the spreadsheets I used for this, I can upload them somewhere for you.

Have you done this yourself? As much as I am intellectually curious about the exact model, if you have found something that gives a good approximation then let us know.

One thing that has made me wonder about which attributes are tied to CA is a couple of quick little tests I did. I set a player (MC natural)to CA 200 and bumped all of his attributes to 100/100 (20 in profile) using FM Modifier and within a couple of weeks the game had adjusted them down. I then took the same player and did it in reverse, set his CA to 100 and all of his attributes to 1/100. Again within a few weeks the game adjusted his attributes up to reflect his CA.

The interesting thing is that in both cases flair, determination, aggression and natural fitness did not change. This initially suggests to me that these attributes are not tied to CA at all. But at the same time in my current game I have a regen youngster whose flair has gone up by 2 points. How that happens I don't know

A secondary thing that I noted is that in the first test (CA 200, all attributes maxed) physical attributes acceleration, pace, balance and agility all decreased by less than all of the other attributes. In the second test (CA 200, all attributes 1) these same attributes increased by less than the other attributes. What this says to me is that they are weighted more heavily which ties in with something PaulC posted about particular physical attributes being 'inherent' so they are harder to improve and less likely to decline.

What is also strange in those two tests is that those 4 physical attributes (acc, pace, agility, balance) all increased/decreased by the exact same amount. All other attributes (except flair, determination, aggression and natural fitness which didn't change) increased/decreased by the same amount. This makes me wonder if there is actually an exact formula that links CA to attributes or if the effect of this CA/attribute link is in how it affects the development of attributes. There might be one used for the generation of players at their point of inception in the game but whether or not the same rules apply as a player develops is something I'm beginning to question.

Anyway if I hit a brick wall using the software and give up I'll post to let you know.

[VSinner]

Wow, this is a heafty posting to be reading after a bit of whiskey, but one thing that occured to me, and I may have missed it being mentioned by someone else, Why don't you use the editor to create a number of "test" players, instead of setting their ca to a set amount, set there PA, then when you start the game and the engine randomly assigns attribute values, you could then use MiniSE to view the CA that becomes associated with the attribute values? Math is not my thing and I only understood most of this conversation at the most basic level, but it seems to me that it would be a good base level to start from.

[Law_Man]

isuckatfm - well done for getting this far, let us know ho the Maple programme goes

Hawshiels: what isuckatfm said really, if you've got interesting results put them up!

[Hawshiels]

Originally posted by Law_Man:

Hawshiels: what isuckatfm said really, if you've got interesting results put them up!

[i'll do this in stages. I just wrote a long post and it failed to submit it properly and I lost what I wrote. ]

Each of the attributes actually has a value between 1 and 100. Although we only see 1-20, the game uses the ones in between to reflect training improvements, etc. So, for each 1 in the game, there are 5 further points allocated to the 'REAL' attribute score. This is important since there are some attributes with a floor that you cannot appreciate unless you look at the 1-100 scale rather than the 1-20 scale within the game. (Make sense ?)

Ok, firstly we have to identify certain attributes from the equation as having a higher floor than the other attributes (i.e. they start at 5 - not 1 out of 100).

They are:

Flair

Influence

Dirtiness

Bravery

Consistency

Aggression

Imp. match

Versatility

Note also that Versatility and Consistency are attributes where the value does not impact on the CA so it doesn't use up 'CA points'.

Next thing to remember to do (which I forgot initially and it through out my results quite a bit) is that each player has a goal keeper rating. The attributes in this area do actually 'take up' CA points. There was a good point made by VSinner in asking why not use the editor to set up the test players for this experiment. Well, for some reaon, if you do not set the goal-keeping attributes for the test players, each of the other attributes will be randomised even if you set them. But if you set all goal-keeping attributes (for a midfielder for example) to 20, the rest of the attributes will remain as you set them. This is actually quite a bit more complex than this but you'll see how it works if you try it.

It's perhaps also worth saying that CA and PA are independent in that the PA of a player has no obvious bearing on the CA attributes - other than obviously what they can reach. But there is no individual attribute that requires a high PA to achieve a maximum score in it.

You can also remove the other personality attributes from this as they take up no 'CA points'.

- Adaptability

- Ambition

- Controversy

- Loyalty

- Determination

- Pressure

- Professionalism

- Sportsmanship

- Temperament

Two 'attributes' that do strangely take up 'CA points' are:

- Left foot

- Right foot

Having said all this, we can now look at what takes up 'CA points' and how many 'CA points' each attribute takes up in relative terms ...

[Hawshiels]

The first consumer of 'CA points' is .... amazingly .... positions - or rather the number of positions a single player can play in.

So what you will find it that if two players have the same CA (e.g. 150), but one can play DC, DMC, MC, AMC, FC and the other is a pure MC .... the one that is a pure MC will have more of his CA points allocated to the attributes and not to a position.

This is the main reason I started looking into this in the first place because I would train a player in a new position and find that the CA had risen quite alot and yet the attributes hadn't appeared to have risen (relatively speaking). So, look out for using up CA points 'needlessly' in this way. It is usually better to find someone for that position and train the attributes as means of maximising the CA (towards the PA score).

[This makes sense in my head, but being more of a mathematician than a writer, I'm not sure I'm explaining it very well. Anyway ...]

[trebor1185]

So what you will find it that if two players have the same CA (e.g. 150), but one can play DC, DMC, MC, AMC, FC and the other is a pure MC .... the one that is a pure MC will have more of his CA points allocated to the attributes and not to a position.

Does being Accomplished, competent, etc. make a difference to how CA is used up by positions?

[Hawshiels]

Originally posted by trebor1185:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">So what you will find it that if two players have the same CA (e.g. 150), but one can play DC, DMC, MC, AMC, FC and the other is a pure MC .... the one that is a pure MC will have more of his CA points allocated to the attributes and not to a position.

Does being Accomplished, competent, etc. make a difference to how CA is used up by positions? </div></BLOCKQUOTE>

Yes, depending on the score for each position, the player will use up CA points. So an accomplished player will use more points to achieve this postional rating than a competent one.

[trebor1185]

Think i'll stop some of my youngsters training new positions then if that's the case then.

I'd rather they had better attributes then giving them a new position for the sake of it then.

[trebor1185]

Originally posted by trebor1185:

Think i'll stop some of my youngsters training new positions then if that's the case.

I'd rather they had better attributes then giving them a new position for the sake of it.

Thats better, I should read what I type before I post it.

[Drewstar]

Originally posted by Hawshiels:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by trebor1185:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">So what you will find it that if two players have the same CA (e.g. 150), but one can play DC, DMC, MC, AMC, FC and the other is a pure MC .... the one that is a pure MC will have more of his CA points allocated to the attributes and not to a position.

Does being Accomplished, competent, etc. make a difference to how CA is used up by positions? </div></BLOCKQUOTE>

Yes, depending on the score for each position, the player will use up CA points. So an accomplished player will use more points to achieve this postional rating than a competent one. </div></BLOCKQUOTE>

Great find Hawshiels!! I always wondered why my young players didn't develop as well as others I had seen. This might be the reason why!

[Law_Man]

This is really reall interesting stuff Hawshiels!

I take it the positions thing using up CA applies also to players who start with multiple positions?

[Hawshiels]

Originally posted by Law_Man:

This is really reall interesting stuff Hawshiels!

I take it the positions thing using up CA applies also to players who start with multiple positions?

Yes, and you can try this easily yourself.

1. Set up two new players using the editor.

2. Make them exactly the same (dob, CA, PA, etc).

3. Then make all of the attributes set to 20 (and don't forget to set the goalkeeping attributes also - this is sometimes missed)

4. Make both players MCs or any other position.

If you start the game like this, you will create two identical players - including all attributes being the same.

If you then edit the second of the players to be rated '20' in all of the positions (leaving the first player only 20 in one position), the game will automatically reduce the attributes of the second player downwards to reflect the fact that some of the CA points need to be taken up by the 'positions' he is good at.

Also, in general there is a weightings system that applies to each individual attribute (although there are only really 3 different weightings that I can see). This means that each individual attribute can rise or fall by varying amounts - for each increase/decrease in CA.

So, for example ...

Dribbling seems to take up the highest (equal)value, whereas marking takes up the least (equal) value. This means that an increase in dribbling takes up more CA points than an increase in marking. I am pretty sure that the attributes scores have changed in the latest patch which is why I am now working on all the numbers again, but it appears that it is much more linear now. The weightings seem to be as simple as 1/3 (one third) , 2/3, 3/3 depending on the attribute we are dealing with. I'll confirm more about this when I've played around with the latest version.

What I can tell you is that - having enjoyed playing around with this in all previous versions also - SI have created something very complex behind the scenes and I think I now understand how each attribute (or trait) impacts the other ..... and it's very clever. It goes into much more depth than ever before and alot more depth than I would have thought.

The two attributes I am checking out at the moment are teamwork and workrate because they appear to work slightly differently from the rest - and of course because of their importance I want to understand them more.

[kolobok]

Hawshiels,

Personally to me the whole point of this research is to improve my ability to judge a player ability and potential without using special soft, and I think Law_Man stated something similar in the opening post. So please don't find my comments below offensive in any way - I am just trying to find a way to get the best possible results on limited data

Each of the attributes actually has a value between 1 and 100. Although we only see 1-20, the game uses the ones in between to reflect training improvements, etc. So, for each 1 in the game, there are 5 further points allocated to the 'REAL' attribute score. This is important since there are some attributes with a floor that you cannot appreciate unless you look at the 1-100 scale rather than the 1-20 scale within the game. (Make sense ?)

Is there any way to see those 1-100 scale attributes in the game? If not, it's not very useful to build the analysis based on them. The reason why is pretty simple. If you don't want to use any soft like FMM or similar one, then there is no way to rely on the 1-100 scale, as well as on the hidden attributes. So if you want to get a clue about a player's CA without using special program, the model should rely on the available data only. If you want to use special soft, then you don't need a model.

Next thing to remember to do (which I forgot initially and it through out my results quite a bit) is that each player has a goal keeper rating. The attributes in this area do actually 'take up' CA points.

Good finding. One more thing to take into account.

It's perhaps also worth saying that CA and PA are independent in that the PA of a player has no obvious bearing on the CA attributes - other than obviously what they can reach. But there is no individual attribute that requires a high PA to achieve a maximum score in it.

Is this something you know for sure or it's more like an empirical conclusion from your research? I was thinking the following logic. Suppose that a good player for any position except GK requires high ratings for pace and decision. In addition DMC requires high ratings in stamina, teamwork, positioning, and first touch (just an example, don't consider it as my view ). If I see a DMC with position-specific ratings close to maximum, I can conclude that his CA is close to his PA at least for this position, and it would affect my decision to buy / sell player, etc. The actual difference between CA and PA may vary, for example due to his ability to be retrained as, say, ST, but this is something that not very important at least to me, as I wouldn't know the player's true PA.

The first consumer of 'CA points' is .... amazingly .... positions - or rather the number of positions a single player can play in.

So what you will find it that if two players have the same CA (e.g. 150), but one can play DC, DMC, MC, AMC, FC and the other is a pure MC .... the one that is a pure MC will have more of his CA points allocated to the attributes and not to a position.

This is really superb finding! If you think about in terms of "position specific attributes logic" it makes sense - to achieve a certain level in some attributes a player must sacrifice somewhere else. But this is very important to know as "attributes' weights" for MC only should be slightly different than ones for DC/MC. Something to think about.

[Hawshiels]

Originally posted by kolobok:

Hawshiels,

Personally to me the whole point of this research is to improve my ability to judge a player ability and potential without using special soft, and I think Law_Man stated something similar in the opening post. So please don't find my comments below offensive in any way - I am just trying to find a way to get the best possible results on limited data

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Each of the attributes actually has a value between 1 and 100. Although we only see 1-20, the game uses the ones in between to reflect training improvements, etc. So, for each 1 in the game, there are 5 further points allocated to the 'REAL' attribute score. This is important since there are some attributes with a floor that you cannot appreciate unless you look at the 1-100 scale rather than the 1-20 scale within the game. (Make sense ?)

Is there any way to see those 1-100 scale attributes in the game? If not, it's not very useful to build the analysis based on them. The reason why is pretty simple. If you don't want to use any soft like FMM or similar one, then there is no way to rely on the 1-100 scale, as well as on the hidden attributes. So if you want to get a clue about a player's CA without using special program, the model should rely on the available data only. If you want to use special soft, then you don't need a model.

Next thing to remember to do (which I forgot initially and it through out my results quite a bit) is that each player has a goal keeper rating. The attributes in this area do actually 'take up' CA points.

Good finding. One more thing to take into account.

It's perhaps also worth saying that CA and PA are independent in that the PA of a player has no obvious bearing on the CA attributes - other than obviously what they can reach. But there is no individual attribute that requires a high PA to achieve a maximum score in it.

Is this something you know for sure or it's more like an empirical conclusion from your research? I was thinking the following logic. Suppose that a good player for any position except GK requires high ratings for pace and decision. In addition DMC requires high ratings in stamina, teamwork, positioning, and first touch (just an example, don't consider it as my view ). If I see a DMC with position-specific ratings close to maximum, I can conclude that his CA is close to his PA at least for this position, and it would affect my decision to buy / sell player, etc. The actual difference between CA and PA may vary, for example due to his ability to be retrained as, say, ST, but this is something that not very important at least to me, as I wouldn't know the player's true PA.

The first consumer of 'CA points' is .... amazingly .... positions - or rather the number of positions a single player can play in.

So what you will find it that if two players have the same CA (e.g. 150), but one can play DC, DMC, MC, AMC, FC and the other is a pure MC .... the one that is a pure MC will have more of his CA points allocated to the attributes and not to a position.

This is really superb finding! If you think about in terms of "position specific attributes logic" it makes sense - to achieve a certain level in some attributes a player must sacrifice somewhere else. But this is very important to know as "attributes' weights" for MC only should be slightly different than ones for DC/MC. Something to think about. </div></BLOCKQUOTE>

I understand what you mean about using other software and respect your need to stay away from it for the sake of enjoying the game. I like to use Excel to store stats about my players so I am comfortable using various bits of software to get access to the real numbers. What you can deduce however is that the 1-20 scale is just a reduced version of the 1-100 scale so for each 1 in the game, you will know that the player has used 5 'virtual' points to get him there. You will sometimes notice that a player doesn't appear to improve in a particular attribute but this is not the case necessarily as the in-game scale will only show an increase when it reaches one of these milestones (i.e. every 5 virtual points).

I understand also your logic behind determining when a DMC (your example) is reaching his maximum PA. The way SI have done this is really complex so I can't say that I am 100% sure, but having done many many tests in this area myself (for the same reason you are interested in this), I am as sure as I can be that there is no bearing on the CA - regardless of the PA of the player. However, if you want to know when a player is at the stage of being 'maxed out' and hence at peak selling price (reputation aside), you will notice that trained categories (or rather the attributes within each one) will increase at the expense of other less-trained attributes. You will see this 'borrowing' of CA points happening more as the player reaches the PA.

And your final point about looking at the various positions a player is good at is important as you've correctly identified. I now tend to be much more mindful of what attributes can be shared between the postions the player is good in. However, this is where it becomes a little strange in a way because 'adaptibility' is something I always saw as a good attribute - in that it helped a player settle but also to retrain (AND RETAIN) positions better. But if I have a player that has used up CA points through positions, I now want him to forget the other non-required positions so I can use up this CA elsewhere. This is what I am trying at the moment because this could make a real difference as you can imagine.

[ntfc]

I have no results to offer but would just like to offer encouragement. This thread has been very interesting and I have enjoyed reading about your progress. Good luck in the future.

[Law_Man]

Originally posted by Hawshiels:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Law_Man:

This is really reall interesting stuff Hawshiels!

I take it the positions thing using up CA applies also to players who start with multiple positions?

Yes, and you can try this easily yourself.

1. Set up two new players using the editor.

2. Make them exactly the same (dob, CA, PA, etc).

3. Then make all of the attributes set to 20 (and don't forget to set the goalkeeping attributes also - this is sometimes missed)

4. Make both players MCs or any other position.

If you start the game like this, you will create two identical players - including all attributes being the same.

If you then edit the second of the players to be rated '20' in all of the positions (leaving the first player only 20 in one position), the game will automatically reduce the attributes of the second player downwards to reflect the fact that some of the CA points need to be taken up by the 'positions' he is good at.

Also, in general there is a weightings system that applies to each individual attribute (although there are only really 3 different weightings that I can see). This means that each individual attribute can rise or fall by varying amounts - for each increase/decrease in CA.

So, for example ...

Dribbling seems to take up the highest (equal)value, whereas marking takes up the least (equal) value. This means that an increase in dribbling takes up more CA points than an increase in marking. I am pretty sure that the attributes scores have changed in the latest patch which is why I am now working on all the numbers again, but it appears that it is much more linear now. The weightings seem to be as simple as 1/3 (one third) , 2/3, 3/3 depending on the attribute we are dealing with. I'll confirm more about this when I've played around with the latest version.

What I can tell you is that - having enjoyed playing around with this in all previous versions also - SI have created something very complex behind the scenes and I think I now understand how each attribute (or trait) impacts the other ..... and it's very clever. It goes into much more depth than ever before and alot more depth than I would have thought.

The two attributes I am checking out at the moment are teamwork and workrate because they appear to work slightly differently from the rest - and of course because of their importance I want to understand them more. </div></BLOCKQUOTE>

Goof work Hawshiels, just done that to see for myself! What would be really good is if you could try and translate your finding into practical, for those less mathsey like myself! Although I appreciate that given the subject matter that might be quite difficult!

One thing I just thought of, do different positions use up differing amounts of CA?

[Law_Man]

Originally posted by ntfc:

I have no results to offer but would just like to offer encouragement. This thread has been very interesting and I have enjoyed reading about your progress. Good luck in the future.

Cheers on behalf of Hawshiels, iscuckatfm and Kolobok - the three mathematical musketeers!

[Hawshiels]

Originally posted by Law_Man:

Goof work Hawshiels, just done that to see for myself! What would be really good is if you could try and translate your finding into practical, for those less mathsey like myself! Although I appreciate that given the subject matter that might be quite difficult!

One thing I just thought of, do different positions use up differing amounts of CA?

I take it that the 'Goof work' was a Freudian slip.

As for making it more practical ... I have approached this from the point of view of it being a puzzle that I enjoyed trying to solve, but there are some really good practical uses of the information we have. What I'll try to do is to be clearer in explaining how best to train players because with the information I have, I set up training regimes designed to get the most out of every player. This is perhaps the most practical application - although just knowing about the CA points being used up by positions is quite important too. I'll work on this more and send you what I have. I had already done the work on this but it has changed in 8.0.2 so I have some checking to do before posting it here.

As for your final point about different positions taking up different values of points .... excellent point. I hadn't actually tested this. But I will now!

[Law_Man]

Originally posted by Hawshiels:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Law_Man:

Goof work Hawshiels, just done that to see for myself! What would be really good is if you could try and translate your finding into practical, for those less mathsey like myself! Although I appreciate that given the subject matter that might be quite difficult!

One thing I just thought of, do different positions use up differing amounts of CA?

I take it that the 'Goof work' was a Freudian slip.

As for making it more practical ... I have approached this from the point of view of it being a puzzle that I enjoyed trying to solve, but there are some really good practical uses of the information we have. What I'll try to do is to be clearer in explaining how best to train players because with the information I have, I set up training regimes designed to get the most out of every player. This is perhaps the most practical application - although just knowing about the CA points being used up by positions is quite important too. I'll work on this more and send you what I have. I had already done the work on this but it has changed in 8.0.2 so I have some checking to do before posting it here.

As for your final point about different positions taking up different values of points .... excellent point. I hadn't actually tested this. But I will now! </div></BLOCKQUOTE>

Sorry pal, yep meant GOOD work! Between us we'll get there! I can think up the odd thing to test (to try and do my bit instead of it feeling like you guys having to doing all the work) and you guys who are in the know, can work your magic!

At the end of all this when we've got some good hard, useful conclusions, I'll try to collate everything you guys post and then produce (hopefully) a single document that's structured and can be of use to everyone. (That's much more my area of expertise than maths! )

[Law_Man]

Originally posted by Mitja:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by PaulC:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Arsenal71:

Paul, i think what the users are moaning about and quite rightly is that there seems to be something wrong with the code in terms of increasing some players attributes by sometimes 3 pts and decreasings by the same amount. A couple of examples are in 8.0.1 Silva and now in 8.0.2 Elano. An example of a player decreasing, Hleb in my case is a an unfortunate example. Their stats in game do NOT reflect the db attributes. We have discussed this in Sires, and its certainly code that is the problem.

The researcher side of this is getting looked into. I know you are supposed to see a preview of what the game is going to do in your editor and it looks like this maybe isnt matching up with the game for certain players..... </div></BLOCKQUOTE>

not for certain players, for many of them. it's the code when ME matches up CA and atributes. for example, if player has high CA and "low" atributes in editor, ME will give him + 1/2 even 3, for each key atribute. also the game decreases "all round" players like ronaldo or kaka and favours those who have some bad atributes, but good CA. like silva or aimar.

but basicly it's all about how CA matches up with total sum of atributes. for example if you want to have silva's stats in game just the same as in editor, you'll have to decrease his CA a lot. and that's where researchers failed... </div></BLOCKQUOTE>

What do we think of this quote (the last one) by Mitja? I don't think he's done any empirical research into it but its food for thought, even if its just his opinion.

[Law_Man]

Which bit?

"It is what it is" - means its not a bug, just the way the game works currently.

"Mini Editor" - its something that only applies to researchers who are seeing the game change their player data to match CA in an unexpected way.

Hope this helps Smile

Paul

This is also very interesting I think....

[Law_Man]

Final quote:

I dont want to go too deep into the technicalities on here to be honest. Its been covered in a fair amount of detail in this and other threads in the past.

Bottom line - the CA-attribute algorithm has been tuned so that CA is a more reliable guide to player performance in FM.

This was all posted by PaulC in this thread

[isuckatfm]

So I figured out how to use the software but unfortunately it's telling me that the set of equations I plugged into it are not feasible within the constraints of having the weightings fall between 0 and 1.

So to sum it up computer says no

Why that is is probably due to the assumptions in the model and what wasn't in it, some of which Hawshiels mentioned in his posts (the affect of multiple positional ratings, hidden attributes, left/right foot rating). I might expand the model and see what falls out. I tell you one thing it's not as simple as I thought it would be. TBH I don't know if I could figure out the exact model, and I don't know to what extent the software I'm using can do approximations.

Also after messing around with FM modifier and watching how players attributes change when CA alone is increased/decreased I'm fairly convinced that there's two CA models at work. One which controls values for a given CA and another that controls how attributes change as CA increases. I have a page and a half I've written on this in Word which I was going to post in this thread but I'll wait and see how kolobok and hawshiels results pan out.

@ Hawshiels

Have you found weightings greater than 0 and less than or equal to 1 that apply to attributes allowing them to sum to CA? If so can you give a list of attributes (whatever they are hidden, visible, positional ratings, foot ratings)? Did you get the weightings based on looking at the stored data of attributes between 0 and 100?

Did you find that these weightings were numerical values (0<x<=1) or were they functions themselves?

For example a player has a weighting applied to tackling based on his positional ratings beyond his natural one?

One more question, have you posted on this stuff before and I just missed it or have you been keeping all this knowledge and methods for training/getting the most from players CA for yourself

If you've got any excel workbooks with collated player data (visible attributes, CA, hidden attributes, positional ratings, feet strength etc.) any chance you could post them as I can use excel data as an input into the software I used for analysing the equations.

Anyway, looking forward to seeing what you and kolobok (are you to blame for the world's current economic woes? Yeah boss, put all your money into the sub prime market, there's no risk there )have.

[Hawshiels]

Originally posted by Law_Man:

Final quote:

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">I dont want to go too deep into the technicalities on here to be honest. Its been covered in a fair amount of detail in this and other threads in the past.

Bottom line - the CA-attribute algorithm has been tuned so that CA is a more reliable guide to player performance in FM.

This was all posted by PaulC in this thread </div></BLOCKQUOTE>

Thanks for posting this info here Law_Man. What Paul says makes perfect sense to me now in terms of the differences I've noticed between 8.0.1 and 8.0.2 and it could equally make sense of how SI as tuned the engine for 8.0.2.

It would appear (but requires more from me by way of testing) that they have tuned the engine - AS WELL AS THE DATABASE - to work more realistically. What I mean by this is that the attacking attributes of the players could either have been tuned down (in terms of their effectiveness) in the engine, or tuned down in the database (or both). It actually makes perfect sense though (I'll post more about this tomorrow) to have the attributes toned down in the database though and what SI have done is used the game itself to make these changes to the player attributes (when the game is loaded initially) rather than making wholesale changes to the original database. So, a player's attributes in the database may not necessarily (depending on the combination of high scoring attributes) end up looking the same within the game.

I hadn't fully appreciated the options available to SI in tweaking the engine until you raised this here. When I was looking at the attributes and how they affect each other within the game I hadn't appreciated that SI would/could use the database (and attributes scores therein) to tweak the game rather than just the engine itself but I suppose it makes real sense.

I will have more time tomorrow to have a better look at this. I won't post any spoilers but as we agreed, I will use this information to give you the information we need to help people in terms of 'maxing out the PA' and their 'training schedules'.

If possible, this thread just got even more interesting for me now. Nice one Law_Man.

[Hawshiels]

Originally posted by isuckatfm:

So I figured out how to use the software but unfortunately it's telling me that the set of equations I plugged into it are not feasible within the constraints of having the weightings fall between 0 and 1.

So to sum it up computer says no

Why that is is probably due to the assumptions in the model and what wasn't in it, some of which Hawshiels mentioned in his posts (the affect of multiple positional ratings, hidden attributes, left/right foot rating). I might expand the model and see what falls out. I tell you one thing it's not as simple as I thought it would be. TBH I don't know if I could figure out the exact model, and I don't know to what extent the software I'm using can do approximations.

Also after messing around with FM modifier and watching how players attributes change when CA alone is increased/decreased I'm fairly convinced that there's two CA models at work. One which controls values for a given CA and another that controls how attributes change as CA increases. I have a page and a half I've written on this in Word which I was going to post in this thread but I'll wait and see how kolobok and hawshiels results pan out.

@ Hawshiels

Have you found weightings greater than 0 and less than or equal to 1 that apply to attributes allowing them to sum to CA? If so can you give a list of attributes (whatever they are hidden, visible, positional ratings, foot ratings)? Did you get the weightings based on looking at the stored data of attributes between 0 and 100?

Did you find that these weightings were numerical values (0<x<=1) or were they functions themselves?

For example a player has a weighting applied to tackling based on his positional ratings beyond his natural one?

One more question, have you posted on this stuff before and I just missed it or have you been keeping all this knowledge and methods for training/getting the most from players CA for yourself

If you've got any excel workbooks with collated player data (visible attributes, CA, hidden attributes, positional ratings, feet strength etc.) any chance you could post them as I can use excel data as an input into the software I used for analysing the equations.

Anyway, looking forward to seeing what you and kolobok (are you to blame for the world's current economic woes? Yeah boss, put all your money into the sub prime market, there's no risk there )have.

I'll post more about this tomorrow, but just to let you know that all weightings (i.e. how many virtual points makes up a single CA point) are above 1 so maybe this is why the model doesn't work.

I haven't posted this before because I didn't realise anyone else would be interested. Nice to find out I'm not the only weirdo on here - although it was Law_Man that made me realise there were a few practical uses for the data.

I have though been using this data for a few years to create training schedules. And they work a treat !!!!

[As for the world's economic woes ... I didn't cause them, but lets just say I understand them and I'd now never have any money in any of the western markets - it is a ticking time bomb and I'd rather have my money invested elsewhere. I'm glad I left the financial services sector ]

[Law_Man]

Another thing I thought of that might investigate is (similar to do different positions use up more CA than others), does being two footed for example, use up more CA?

When (as apparently happens although I've never seen it myself to vouch for it) a player gains strength in his weaker foot/becomes two footed, does this use up some CA? Similarly, as players increase in height (again as I've seen posted that they do although I've never noticed this myself) does this use up CA?

What about personality (i.e. determination and professionalism scores) affecting which attributes have a higher weighting? i.e determined player - perhaps teamwork workrate etc could have a higher weighting specific to that player (in addition to his position)?

Lots of questions there!

@isuckatfm shame the computer said, no, but no worries and thanks for trying Looking forward to your further results.

@Hawshiels No worries, thought it might be of interest to you.... And well, I have to make up for my mathemetical inadequacies some how don't I!? Glad you're taking such an interest in this

I won't post any spoilers but as we agreed, I will use this information to give you the information we need to help people in terms of 'maxing out the PA' and their 'training schedules'.

Yeh I definitely think this is the right way to go, like I said, if you guys can do the maths, and then you can explain the results to me/I can actually understand your results lol, then I'll get something more "userfriendly" written up in the end hopefully. Just call me "the scribe" lol

ps

Anyway, looking forward to seeing what you and kolobok (are you to blame for the world's current economic woes? Yeah boss, put all your money into the sub prime market, there's no risk there Wink)have.

I heard Kolobok was a risk analyst for Northern Reck err I mean Rock...and that's why he's got a lot of free time on his hands to contribute to this project

However I did some trick (I am statistical analyst / modeler IRL, so trust me Wink)

Said Kolobok to a very annoyed head of the Collateralised Debt Obligations and Asset Backed Securities department....

[Law_Man]

I haven't posted this before because I didn't realise anyone else would be interested. Nice to find out I'm not the only weirdo on here

And I'm the biggest weirdo of them all I think, given that I actually can't stand maths! If only maths at school was Champ Man based.....

Re world economic woes and the financial sector, am much more comfortable here as I actually know something about it as opposed to statistical modelling and the like!

[isuckatfm]

originally posted by Hawshiels:-

I'll post more about this tomorrow, but just to let you know that all weightings (i.e. how many virtual points makes up a single CA point) are above 1 so maybe this is why the model doesn't work.

I think maybe we have our wires crossed, but it is kind of difficult to discuss these things on a forum as opposed to sitting in a room with pen and paper.

What I was looking at was CA as a sum of the attributes weighted in an equation that controls attribute distribution for a given CA (this would explain why player attributes started dropping after 8.02 despite CA not changing because SI increased the weightings):-

CA = sum(w(i)*a(i))

in which case it isn't possible for the weightings to be greater than 1 since CA is maximum 200 and attributes are stored at a maximum of 100. Unless of course CA is multiplied by some factor in the above equation. If you have figured this out then say so.

I think there is a difference between this CA model and the one that controls how attributes change for a given change in CA. That's my interpretation of it from messing with FM modifier.

Do you have a model you could post or is it something you know from looking at a load of players and how they develop? I know Law Man and kolobok don't want to know the ins and outs but if you could post it with a ********WARNING********* so they don't read it, I would like to see numerically what you've come up.

[isuckatfm]

originally posted by Law Man:-

I heard Kolobok was a risk analyst for Northern Reck err I mean Rock...and that's why he's got a lot of free time on his hands to contribute to this project

quote:

However I did some trick (I am statistical analyst / modeler IRL, so trust me Wink)

Said Kolobok to a very annoyed head of the Collateralised Debt Obligations and Asset Backed Securities department....

Now you see it, now you don't!! Kolobok the Magnificent and his magically disappearing millions (just pulling your leg kolobok )

[Law_Man]

Originally posted by isuckatfm:

[Do you have a model you could post or is it something you know from looking at a load of players and how they develop? I know Law Man and kolobok don't want to know the ins and outs but if you could post it with a ********WARNING********* so they don't read it, I would like to see numerically what you've come up.

Posting anything is cool by me, (as long as it doesn't infringe the forum rules: is there a ban on posting PAs? I've no idea), as long as there's a sufficinet ******WARNING PA's MENTIONED****** then it's all good. That way, people can choose to not look at it

[kolobok]

So I figured out how to use the software but unfortunately it's telling me that the set of equations I plugged into it are not feasible within the constraints of having the weightings fall between 0 and 1.

So to sum it up computer says no

Why that is is probably due to the assumptions in the model and what wasn't in it, some of which Hawshiels mentioned in his posts (the affect of multiple positional ratings, hidden attributes, left/right foot rating). I might expand the model and see what falls out. I tell you one thing it's not as simple as I thought it would be. TBH I don't know if I could figure out the exact model, and I don't know to what extent the software I'm using can do approximations.

Also after messing around with FM modifier and watching how players attributes change when CA alone is increased/decreased I'm fairly convinced that there's two CA models at work. One which controls values for a given CA and another that controls how attributes change as CA increases. I have a page and a half I've written on this in Word which I was going to post in this thread but I'll wait and see how kolobok and hawshiels results pan out.

First of all, great work anyway. You proved that CA (something that does not exist IRL) cannot be simply calculated from scratch unless you have complete information (which we don't). And that's good and consistent with Hawshiels's findings. I would be very disappointed with SI if it were such an easy problem .

Second, don't wait for my result, please. Tonight at best I will get a few raw text files with attributes (keep in mind, I am in the US, so add the time difference ). I hope to have something like 300-400 players from a few leagues, which should give me approximately 40 per position. Then I will need to merge them together, add CA and PA and it may take a couple days (I can only find a couple hours per day to work on this part, and it's if I don't play the game at all, which is difficult to resist ). The analysis part should be a bit faster as it's more fun plus I can do part of it at work during break time. But I will also need to write it up. So, I don't expect to post the results earlier than Thursday, and weekend seems more realistic.

So, post your findings as they may help me, Hawshiels or somebody else who maybe running his own analysis in the background to tune up the analysis and avoid mistakes or misinterpretation. Plus you will definiteley help all of us understand the game a little bit better.

[Hawshiels]

OK. I've thought about this - especially how to present the information in a way that makes it understandable/usable, etc.

So, lets start with how the CA points are used up by positions. And note that I am now using in-game CA meaning that each 1 point represents a value on the 1-20 scale - NOT the 1-100 scale.

Firstly, there is a difference in the number of CA points that are used up by each of the 3 areas of the park (i.e. DEFENCE, MIDFIELD and ATTACK). I'll give you these in a moment, but firstly we have to break down the costs of each pitch area in terms of MATT (mental attribute costs), TATT (technical attribute costs) and PATT (physical attribute costs).

Note that the table below shows relative points usage. In other words, these points usage numbers are based on a CA of 200. As the CA decreases, the cost of each position decreases in relative terms. So, for someone with a CA of 100, you can half the numbers below, etc.

[in the case of a player with a CA of 50 (or less), there is less of an obvious effect from within the game due to the 1-20 scale, but it is noticable on the 1-100 scale.]

<pre class="ip-ubbcode-code-pre">

AREA TATT-cost MATT-cost PATT-cost

Defence 4 4 8

Midfield 12 8 4

Attack 16 4 8

</pre>

There are also TATT, MATT and PATT costs for each side on the pitch you choose (i.e. LEFT/CENTRE/RIGHT). Here are the costs for the FIRST additional side.

<pre class="ip-ubbcode-code-pre">

TATT 8

MATT 8

PATT 0

</pre>

For the second additional side, the cost of CA points is this.

<pre class="ip-ubbcode-code-pre">

TATT 4

MATT 4

PATT 0

</pre>

So, if you have a player that can play DL/R/C, they will use up 12 TATT points and 12 MATT points more than a DC-only.

These TATT, MATT and PATT point increases seem to be consistent regardless of the area on the pitch (i.e. Defence, Midfield, Attack).

[Hawshiels]

Next thing to note is that the first 'foot' costs no points in the game (LEFT or RIGHT), but the second 'foot' is costly - in that it costs 8 points - probably reflecting that in FM08 more than ever, the player that can use either feet is a valuable resource and hence a higher calibre player. This applies to all positions equally as far as I can see.

Now for the attributes that do NOT use up points. Some are obvious, but some are perhaps a little surprising.

Obvious ones ...

Adaptibility

Controversy

Ambition

Determination

Loyalty

Pressure

Professionalism

Sportsmanship

Temperament

Although they do not use up CA points, they do cause the CA points in certain areas to move at different rates (up and down).

Now for the relatively obvious ones ...

Consistency

Aggression

Important Matches

Injury Prone

Natural Fitness

And the surprising ones perhaps ...

Penalties

Corners

Free-kicks

Long throws

These surprised me the first time I realised this because it meant that I could create training schedules for my 'maxed out' (i.e. CA is very close to PA) players to keep their other skills ticking over while they improved these areas. It makes the team much more likely to score goals from set pieces but doesn't cost any CA points.

[Note that what I am writing here is concerning outfield players only. It may apply to goalkeepers also but I haven't tested them - ever. Also, note that CA points are taken up by every player due to his 'Goalkeeping ability', but any increases to this thereafter appears not to cost any CA points. Strange I know, however if a player then becomes a goalkeeper the CA is the assessed based on these stats also. So I suppose this is how we would want/expect this to work.]

[trebor1185]

And the surprising ones perhaps ...

Penalties

Corners

Free-kicks

Long throws

These surprised me the first time I realised this because it meant that I could create training schedules for my 'maxed out' (i.e. CA is very close to PA) players to keep their other skills ticking over while they improved these areas. It makes the team much more likely to score goals from set pieces but doesn't cost any CA points. Eek

I also noticed this, at first I was surprised but having these stats maxed out isn't very useful if your player doesn't have good technique, composure, etc.

[Hawshiels]

Just in case I hadn't stated this already, the CA of a player is a real number and it is NOT affected by a player's age. In other words, a player aged 17 (with a CA of 100) is exactly the same as a player aged 35 (with a CA of 100).

[Hawshiels]

And now for the really cool bit - and probably the reason many people have found it hard to work out the CA formula.

There are actually attributes (other than the ones I posted above) that cost NO CA points depending on what position the player can play in. This is critical if you are going to create good training schedules without using up all the points towards the PA of a player.

So, here are the attributes that take up NO points for each position.

Defender

TATT

Crossing

Dribbling

Finishing

Long shots

MATT

Creativity

Off the ball

Flair

Teamwork

Midfielder (DM/M/AM)

TATT

Crossing

Heading

Marking

MATT

Flair

Influence

Bravery

PATT

Jumping

Attacker

TATT

Marking

Tackling

MATT

Flair

Influence

Bravery

[trebor1185]

I think teamwork doesn't take up any CA points for attackers, that's what i've noticed in some things i've tried out with the editor.

[Joor]

Originally posted by Hawshiels:

And now for the really cool bit - and probably the reason many people have found it hard to work out the CA formula.

There are actually attributes (other than the ones I posted above) that cost NO CA points depending on what position the player can play in. This is critical if you are going to create good training schedules without using up all the points towards the PA of a player.

So, here are the attributes that take up NO points for each position.

Defender

TATT

Crossing

Dribbling

Finishing

Long shots

MATT

Creativity

Off the ball

Flair

Teamwork

Midfielder (DM/M/AM)

TATT

Crossing

Heading

Marking

MATT

Flair

Influence

Bravery

PATT

Jumping

Attacker

TATT

Marking

Tackling

MATT

Flair

Influence

Bravery

This is really amasing

Im going to change my training schedules right away !

[Hawshiels]

Hopefully you can see the picture becoming much more simplified - but I'll now start to explain the CA formula a bit more.

Firstly, it is important to note that TATT, MATT and PATT are NOT independent. I have read on other threads that a certain CA allows a certain number of TATT points to be allocated, a certain number of MATT points and a certain number of PATT points. This is not the case.

There is a total number of points from the CA available and they can be used in any of the available attributes (attributes that cost CA points of course and not the 'free' ones). So, a player can look like this:

Player A

- Top rated TATT and MATT - but poor quality PATT ratings

Player B

- Top rated TATT and PATT - but poor quality MATT

Player C

- Top rated MATT and PATT - but poor quality TATT

Player D

- A good mixture of all 3.

Note that it is possible to have a single player with top ratings (i.e. 20 in every attribute) for any two of the areas - TATT, MATT or PATT. But of course the third area will suffer badly then from the lack of CA points available. This is important if you see a good player in lots of stats and yet he is relatively cheap in the game. I'll give you an example.

A defender has top rated (or thereabouts) MATT and PATT stats and yet he is very cheap. You may be tempted to buy this player (especially if he is young) in the hope that you can train his technical stats up. Beware because the only way this player can be trained ... is at the expense of the existing good stats. It is all relative to the PA of course, but this is critical in understanding how this works.

It is possible to have a player with a CA of 120 that is better than a player with a CA of 180 - because it depends on how the points have been 'spent' within the attributes. Equally, it is possible to have a player with a CA of 120 (PA 200) that will never be as good as a player with a CA of 100 (PA 130). It really does depend on the attributes that score highly for the individual players. More about this now.....

[Hawshiels]

Originally posted by trebor1185:

I think teamwork doesn't take up any CA points for attackers, that's what i've noticed in some things i've tried out with the editor.

Well done. You identified the one I missed. I'll take more care when I'm listing the attributes from now on but I have checked and I now believe this is the only one I missed.

Cheers

[Joor]

I really dont understand why Jumping, heading and marking do not cost any CA points for midfielders - Those attributes are so important imo.

[trebor1185]

Originally posted by Hawshiels:

And now for the really cool bit - and probably the reason many people have found it hard to work out the CA formula.

There are actually attributes (other than the ones I posted above) that cost NO CA points depending on what position the player can play in. This is critical if you are going to create good training schedules without using up all the points towards the PA of a player.

So, here are the attributes that take up NO points for each position.

Defender

TATT

Crossing

Dribbling

Finishing

Long shots

MATT

Creativity

Off the ball

Flair

Teamwork

Midfielder (DM/M/AM)

TATT

Crossing

Heading

Marking

MATT

Flair

Influence

Bravery

PATT

Jumping

Attacker

TATT

Marking

Tackling

MATT

Flair

Influence

Bravery

Just had another look at this and have a few questions, have you classed ML and MR as attackers, and crossing is an important attribute for full backs and does take up CA points from what I can see.

There's other ones such as wing backs and sweepers who have different important attributes to other defenders, or are you only concentrating on the main positions.

[Hawshiels]

Originally posted by trebor1185:

Just had another look at this and have a few questions, have you classed ML and MR as attackers, and crossing is an important attribute for full backs and does take up CA points from what I can see.

There's other ones such as wing backs and sweepers who have different important attributes to other defenders, or are you only concentrating on the main positions.

Sorry, I should have been clearer about this. I am only taking the central positions for midfielders because they work differently from the defenders and attackers.

For example, a midfielder on either the right or left gets a 'FREE' score within 'positioning' to make up for the loss in the free midfielder 'FREE' score of 'crossing' (which of course a right or left-sided midfielder would be charged for). I will post all of these though. I just thought I had a simple way of taking people through the process and introduce the complexities one stage at a time.