Just found something amazing!!! Must see!!!

[VMX]

This is, by far, one of the most interesting threads we've ever had here. All thanks to Ljuba

Now I have to say that to find a fix, first we need to identify why we have this problem. Also, as it's been said, there's no "easy" way of fixing this without making a major change to the system. I'll try to give some ideas here.

It seems that, in previous CM versions, there was no attribute 'weighting' when it came to calculating the CA of a player.

I don't know the exact numbers, but let's say that regardless of the player's position, one point of CA spent would mean that the player would improve one of his attributes for a fixed amount of points. So, if the player was a ST, it would consume the same number of CA points to improve 1 point of tackling than to improve 1 point of finishing.

It may look perfectly ok though... why should it be 'more expensive' to learn some attributes than others? Especially considering that a striker should be more suited to learn striking abilities, wouldn't he?

But it arised some problems:

A striker with 20 in half of his attributes (the most offensive ones) and 1 in the other half of his attributes (the most defensive ones), would have the same CA than a striker who had 10 in ALL of his attributes. This might seem a bit ridiculous, although it doesn't necesarily have to be a problem: he spent all of his ability in learning new stuff, but he never specialized in anything in particular. So he's an overall better player, while the other striker is an specialist. Managers should then choose which one they prefer (but we agree that everyone would go for the 'specialist' if they're looking for a striker).

Now, the actual problem here was that the AI will RATE players based mainly on their CA. So they would consider the first one just as good as the second one. Which, for a striker, is wrong. So the problem here was the way the AI judged the players... but I can't see anything wrong in that system regarding player development, right?

But, instead of improving the way AI values a player (well, they probably did improve things, but not significantly), the main 'fix' they developed was to 'include' the position a player plays in inside the CA, in the form of attribute 'weighting'. In such a way that if a player has such a high potential, he "must" be good at his position, because he's "probably" used much of his CA points to develop some of the most "expensive" attributes for the position he plays at. However, the AI will probably still be fooled if they find a player with high CA, but who has spent most of his CA points to get all 20's the most pointless attributes for his position. PaulC says they do look at some of them, but still, it's not very realistic, and causes other major problems as we've seen.

So, what I really think that's needed is giving the AI a new way of rating players, while also changing the way CA is calculated right now. Attribute weighting should be completely outside of a players CA, because it's only needed for AI teams to judge players. We humans have our own brains...

Possible CA/PA system to be used

I think what could actually be done is the same that's done in RPG games... that is: the higher a skill is, the more CA (=experience in RPG games) points it would require to improve it! This way, if a player has 10 in all areas, still wouldn't have such a high CA. While if he has 20 in half of them, even if the rest are crap, he would be better CA-wise.

Also, it would seem logical to me that it's easier to learn basic stuff about something, but the higher level you get to, the more difficult it is to become a 'world-class' player at something.

This would be a much more logical approach than the current philosophy of: "A DC can easily learn finishing without spending much CA, while an actual striker will spend most of it to achieve the same".

Rating of players by the AI

As I said, the CA shouldn't be used to judge the ability of a player by the AI. It should only be a tool to manage players development. Only scouts should try to see the difference between CA and PA, so that they know if a player is to improve or not. But CA shouldn't be a reference of how good a player is.

So what the AI should ideally do is to use the same "weighting" system that's currently used to calculate CA, but only when they want to sign a player.

So if they're looking for a striker, they should obtain a "number" which will be calculated by weighting all of his attributes, but giving more importance to the offensive attributes than the defensive ones. Then they should assign that player an internal "rating", just like they store scout reports or whatever.

It wouldn't matter if the player can play at other positions. If they're looking for a striker, they should try to "weight" him as a striker. The ability to play in multiple positions should be an added value, just like the fact that he's cheaper, or whatever.

Don't you think something along these lines would work much better?

[matt_132]

I definitely agree with VMX, you should apply for a job at SI

[Majkee]

It seems good VMX, nice idea to have CA only for player development in RPG style. But have you taken in mind retraining possibility? Because this should do some damage when for retrained player will be the attributes recalculated.

[HerrKaputt]

Possible CA/PA system to be used

I think what could actually be done is the same that's done in RPG games... that is: the higher a skill is, the more CA (=experience in RPG games) points it would require to improve it! This way, if a player has 10 in all areas, still wouldn't have such a high CA. While if he has 20 in half of them, even if the rest are crap, he would be better CA-wise.

Also, it would seem logical to me that it's easier to learn basic stuff about something, but the higher level you get to, the more difficult it is to become a 'world-class' player at something.

This would be a much more logical approach than the current philosophy of: "A DC can easily learn finishing without spending much CA, while an actual striker will spend most of it to achieve the same".

This looks like a great idea. In a way, the player's "natural talent" will be his starting attributes (if a regen is "born" with 15 finishing, he will almost surely be a great finisher, even if trained little in that category), and the training he gets throughout his career will determine how much he progresses/regresses in each attribute, up to his potential. Not only this solves the problem Ljuba alerted us to, but it also is closer to real life.

Rating of players by the AI

As I said, the CA shouldn't be used to judge the ability of a player by the AI. It should only be a tool to manage players development. Only scouts should try to see the difference between CA and PA, so that they know if a player is to improve or not. But CA shouldn't be a reference of how good a player is.

So what the AI should ideally do is to use the same "weighting" system that's currently used to calculate CA, but only when they want to sign a player.

So if they're looking for a striker, they should obtain a "number" which will be calculated by weighting all of his attributes, but giving more importance to the offensive attributes than the defensive ones. Then they should assign that player an internal "rating", just like they store scout reports or whatever.

It wouldn't matter if the player can play at other positions. If they're looking for a striker, they should try to "weight" him as a striker. The ability to play in multiple positions should be an added value, just like the fact that he's cheaper, or whatever.

Don't you think something along these lines would work much better?

Again, a great idea, especially the part about the AI rating the player according to the position they are looking to reinforce.

With these two ideas it would be possible, in principle, to fully retrain a player to a new position, including attributes. I'm not sure whether this is a problem or a bonus: in real life, players DO retrain, sometimes with great success. I think that if the game makes learning new attributes (including regressing the stats which are not trained much) fast for young players and slower as they age, this would not be a problem. And it makes perfect sense if you get a regen who is a natural DC but has not only good jump, tackling and heading but also decent passing and creativity, that you retrain him (positionally, not in stats) to be also a MC or DMC. With VMX's ideas it would not improve one single bit his ability as a DC, and if you train him more in passing you will train him less at tackling, and he will be less of a specialist but will be still quite good on both positions, which again is perfectly realistic.

[VMX]

Yeah, I believe everything would be much more logical this way. It all spins around the fact that all the attributes should 'weight' the same no matter which position you play in, but it should as well be harder to master any of them than learning the basics.

Also, I don't know if the time it currently takes for a player to learn an ability has a relation with the number of CA points required to improve that ability, but... if that's the case right now (which I don't know), it would still look realistic enough, because:

Players who play in lower leagues would not improve too fast even if they have low attributes, because usually the training facilites at those teams are much worse. On the other hand, a player who has really good attributes would have additional reasons to 'feel the urge' to switch to a better team, because it would be harder for him to improve his 15's 16's and 17's if he's in a team with poor facilities. Which is just what actually happens in real life.

Of course, for this to work it needs to be possible that the AI judges players the way I commented above, and only SI can know if that's possible...

<thinking out loud>

* SI comment would be much appreciated *

</thinking out loud>

[ngm]

I think a quite easy way to temporarily fix this problem is assigning a factor to the weight each position has in relation to the attribute.

For instance, say marking, the fact that a player is a DC is more important in order to consider how many points of CA this attribute should weight rather than if the player is a ST.

Take a player with DC 20 ST 15 with a 15 in marking: You assign a ratio of 1 in relation to marking if a player is a DC and a ratio of 0.1 if the player is a ST.

With the current system, the amount of CA points that marking "uses" is something like (15 * 10 *20 + 15 * 1 * 15) / 2.

With the ratio is something like that (15 * 10 *20 * 1 + 15 * 1 * 15 * 0.1) / 1.1.

This way, you increase the CA points used with marking.

There is a problem with that, tough, that players with multiples positions are going to have worse attributes and that's the opposite problem, so you should adjust those ratios until you find a better overall situation than the current one, hence the temporary easy fix.

[ngm]

I think this image represents what I'm trying to say.

Sorry for the double post, but I can't edit the previous post at the moment.

[chronic187]

So let me get this straight.

If i go into the editor and change Fernando Torres to a

Striker 20 and

Defender center 14

And put his marking and tackling and bravery down to 3.

His striker stats should shoot up ?

[VMX]

So let me get this straight.

If i go into the editor and change Fernando Torres to a

Striker 20 and

Defender center 14

And put his marking and tackling and bravery down to 3.

His striker stats should shoot up ?

Exactly.

But moreover, you don't need to use the editor.

Just train him to be a DC and, assuming he doesn't have good defensive attributes, his attacking stats will suddenly boost up. THAT is the real problem. No editors or 3rd party tools needed to cause this bug.

So, from now on, everytime you see a player who has excellent (maybe world-class) stats, but you notice that, strangely enough, his value or his scout rating is not too high, just make sure you check how many positions he can play in.

If you notice he's able to play in many different positions, then that's the cause why he's being rated poorly: his CA is too low for his quality. However, you can safely bypass the scout's rating and buy him because, as his attributes will show, he's a class player.

That's why we're suggesting new possible systems that could be used instead of the current one.

[Ljuba82]

I think a quite easy way to temporarily fix this problem is assigning a factor to the weight each position has in relation to the attribute.

For instance, say marking, the fact that a player is a DC is more important in order to consider how many points of CA this attribute should weight rather than if the player is a ST.

Take a player with DC 20 ST 15 with a 15 in marking: You assign a ratio of 1 in relation to marking if a player is a DC and a ratio of 0.1 if the player is a ST.

With the current system, the amount of CA points that marking "uses" is something like (15 * 10 *20 + 15 * 1 * 15) / 2.

With the ratio is something like that (15 * 10 *20 * 1 + 15 * 1 * 15 * 0.1) / 1.1.

This way, you increase the CA points used with marking.

There is a problem with that, tough, that players with multiples positions are going to have worse attributes and that's the opposite problem, so you should adjust those ratios until you find a better overall situation than the current one, hence the temporary easy fix.

Remember that by having several positions on which player can play his CA an ability on the pitch can be smaller than if he had only one position!!! So che could be worse player, not only better necessary!

So that means that you could add aditional handicap to players that are handicapped by this anyway.

But nice thinking

So let me get this straight.

If i go into the editor and change Fernando Torres to a

Striker 20 and

Defender center 14

And put his marking and tackling and bravery down to 3.

His striker stats should shoot up ?

Yes!

But you don't have to do this in editor.

I don't know how much Torres is versatile, but if you give him no DEFENDING training, while giving him instructions to learn new position (DC), eventualy, when he reach DC rating of 11 (not 14), his attacking attributes will shoot up immediately!!!

More players learn new position (rating from 11-20)...his attacking attributes will increase more!!!

This is, by far, one of the most interesting threads we've ever had here. All thanks to Ljuba

Possible CA/PA system to be used

I think what could actually be done is the same that's done in RPG games... that is: the higher a skill is, the more CA (=experience in RPG games) points it would require to improve it! This way, if a player has 10 in all areas, still wouldn't have such a high CA. While if he has 20 in half of them, even if the rest are crap, he would be better CA-wise.

Also, it would seem logical to me that it's easier to learn basic stuff about something, but the higher level you get to, the more difficult it is to become a 'world-class' player at something.

This would be a much more logical approach than the current philosophy of: "A DC can easily learn finishing without spending much CA, while an actual striker will spend most of it to achieve the same".

Rating of players by the AI

As I said, the CA shouldn't be used to judge the ability of a player by the AI. It should only be a tool to manage players development. Only scouts should try to see the difference between CA and PA, so that they know if a player is to improve or not. But CA shouldn't be a reference of how good a player is.

So what the AI should ideally do is to use the same "weighting" system that's currently used to calculate CA, but only when they want to sign a player.

So if they're looking for a striker, they should obtain a "number" which will be calculated by weighting all of his attributes, but giving more importance to the offensive attributes than the defensive ones. Then they should assign that player an internal "rating", just like they store scout reports or whatever.

It wouldn't matter if the player can play at other positions. If they're looking for a striker, they should try to "weight" him as a striker. The ability to play in multiple positions should be an added value, just like the fact that he's cheaper, or whatever.

Don't you think something along these lines would work much better?

First have to say, that if I didn't get this "magic tool" from my mate here, I wouldn't get any idea about this.

Also special thanx are reserved for all you who discussed this with scientific aproach, cause we discovered more and more thing while discussion was developing.

We didn't only discovered about this positional/retraining bug, we discovered some bugs with "weaker foot" (example with AML and his unlogic CA when we change his weaker foot...). We also discovered "scouting bug", cause it seems to be that scouts are useless when talking about this multi-positional players!!!

...about your way for solving this problem...

...I'll give another example to show why I think this is no good way...

...let's say that player have only 2 attributes in game (tackling, finishing)...and that maximum rating for attribute is 20 and that each attribute have same weight (like you suggested). Let's say that each point in attribute uses 1 CA point.

So...maximum CA for every player is 40 CA pts.

...now look at this...

Player1.....tackling 15, finishing 15...................so CA is 30 (PA 30)

Player2.....tackling 1, finishing 19.....................so CA is 20 (PA 20)

It's more than obvious that player 2 is better ST than player 1 and that he will be much more expensive at start of the game.

But with training shaping...and after many years (I said years cause you said that the higher attribute rating is, the higher amount of time you need to train this attribute)...you will eventually raise Player1 finishing to 20, while droping his tackling to 10. That means that he will be much better (and also more expensive) cause scouts will (like you said) look at his attributes, not at his CA.

This is also unlogic...cause both players have reached their PA at the start, but in 10 years time the one who was much better at start will be much worse at the end.

One more thing...I'm not sure if we should look for solution for this, it's a job for SI, I just want to see what they have to say about this bug that makes the game unplayable!!!!!!!

[Evon]

Maybe this is the reason for the attribute changes in the game from the database? Ie. Silva and Elano's attributes being boosted for some reason.

[PODSY12345]

Exactly.

But moreover, you don't need to use the editor.

Just train him to be a DC and, assuming he doesn't have good defensive attributes, his attacking stats will suddenly boost up. THAT is the real problem. No editors or 3rd party tools needed to cause this bug.

Hi All,

Does anyone know if this works for FM 2009 Patch 1, or just the demo?

Also, apart from marking and tackling, what other defensive attributes can be kept low without altering his attacking performance?

By the way, thanks to Ljuba82 for sharing this! Really great work.

[phnompenhandy]

Hi All,

Does anyone know if this works for FM 2009 Patch 1, or just the demo?

Since SI weren't aware of the 'exploit', it will be in FM08, FM09 demo, unpatched and patched, I'd say.

[Ljuba82]

Hi All,

Does anyone know if this works for FM 2009 Patch 1, or just the demo?

Also, apart from marking and tackling, what other defensive attributes can be kept low without altering his attacking performance?

By the way, thanks to Ljuba82 for sharing this! Really great work.

It works for 2009...and it will work for every version of 2009 no matter if you have a patch. This is a big bug and I'm afraid they can't fix it for this version cause it will change whole game, but I think that they will have lots of problems for FM10...

But fixing this and changing CA, PA system for FM10 is a MUST!

STILL WAITING FOR YOUR RESPONSE SI!!!

[PODSY12345]

Thanks Ljuba

So, to recap...

If I do the following (using a ST as an example)

1) Using FMRTE give my ST 15 in all attacking categories, and 1 in all defensive categories

2) Using FMRTE, or in-game, give him DC as a new position

3) Don't alter his training ie. keep it attack-minded.

4) His attacking attributes should increase

Is that a fair re-cap?

[Ljuba82]

Thanks Ljuba

So, to recap...

If I do the following (using a ST as an example)

1) Using FMRTE give my ST 15 in all attacking categories, and 1 in all defensive categories

2) Using FMRTE, or in-game, give him DC as a new position

3) Don't alter his training ie. keep it attack-minded.

4) His attacking attributes should increase

Is that a fair re-cap?

Yes................

[PODSY12345]

Cool. Thanks my man...

[Joor]

Heh, I just tried this with Ronaldinho. Made a new game, and then using Fmrte so he had 15 in the central defender position.

Most of the the technical attributes went up and some mental. First game in the pre season he made 16 runs, 2 assists and scored 2 goals --got 9.2

http://img80.imageshack.us/my.php?image=dingoasdefender1ue3.jpg

http://img80.imageshack.us/my.php?image=dinhofirstgamepi1.jpg

This is just wrong - everyone can retrain a wonder kid striker and make them extremely powerful

[ngm]

Remember that by having several positions on which player can play his CA an ability on the pitch can be smaller than if he had only one position!!! So che could be worse player, not only better necessary!

So that means that you could add aditional handicap to players that are handicapped by this anyway.

But nice thinking

I'm not saying that there is a perfect way of fixing this problem, but I think that my approach is the one that minimize the impact of the anomaly that you found.

Anyway, if a player have multiple positions and his attributes are not suited for those positions I think that he should be able to un-learn those positions and minimizing the handicap, although that is something that you can not do in the game at the moment. The biggest problem is when a player is very good at very different positions (like Luis Enrique, Barça captain who was a DR, MR, AM and sometimes even a second striker, although that is not in this game), but that’s something that doesn’t usually happen and I can’t think of a player who is a good defender and also a good attacker (and no, a full back who is also a good winger does not fall in this category).

You also said that you maximize the attacking attributes by making a player a DC, DL, DR and ST. That makes me think that in order to figure out what is the rating of a given attribute in CA points the game takes into account all the positions a player can play (I could even be able to prove it, because the way they calculate the CA is just a simple neuronal network (if a player can only play one position), and if you build that network and you feed it with the players attributes you can figure out the different ratings. But that’s just a lot of work and not a lot of benefit for me in doing it since I don’t play FM Live or FM 2008 / 2009 and FM Scout doesn’t seem to work very good for those versions of the game). I think that that’s a mistake and they should aggregate the similar positions: for instance, DR and DL, MR and ML and so on in order to figure out the rating. This way, you can minimize the impact of this bug.

I think the model they have for a player that can only play one position is very good, although I think it would be better to split the CA and PA into categories, such as mental, physical, technical and tactical, but that’s for another thread.

The problem comes when a player has multiple positions and that because they in order to calculate the rating they give the same weight for each position a player can play, and that is just a contradiction with the model for a one position player, where the position of the player determines the rating. But, if a player can play different positions they give the same weight to each position, which is in my opinion just wrong. That is what I was trying to say with my previous post

Something similar comes when the player is two-footed: those players suffer a loss in CA points due to this fact, and that’s just the opposite result of this anomaly (the attributes of the player decrease if the player is two-footed). Regarding this, I think that they changed the game engine in order to make two-footed players better than the one-footed, but that backfired in the way that the game calculate the different attributes and now you almost need to run an algorithm to figure out is a player is good enough or not when you look at his attributes.

In my opinion, these flaws can be corrected, and I think that with the approach I previously posted they can fix the positions problem, although it may be difficult to find the right balance, but I think that they can use a neuronal network to predict the new weights. For instance, in this thread they said that Fazio and Gamberini have basically the same attributes but they have a CA differential of 20 points approx. So they change the CA of Fazio to the CA of Gamberini and then just let the neuronal network figure out the new weights (Of course they would need a lot more data, but that’s just an example).

My “recommendation” to SI is that they need somebody that can be able to figure out a better mathematical model for the players attributes, and then use this model to build the game engine around. I’m saying this regarding the two-footed problem, because my opinion is that the fact that a player is two-footed shouldn’t take CA points that are used to determine all of his attributes (pace, consistency, …). But that’s just my personal opinion.

[juvefan4ever]

I've always thought of having multiple PAs. I don't know how hard it would be to code but I think it would be more accurate. For example, a player can have four (technical, mental, GK and physical) PAs or six (offensive and defensive technical, offensive and defensive mental, GK and technical) PAs. Appropriate attributes are linked to each of the PAs.

For example, CRonaldo could have attacking tech PA of 95, defense tech PA of 35~40, attacking mental PA of 85, defense mental PA of 40, GK of 10 and physical of 90. Nedved could be 80, 65, 85, 70, 10 and 85...you get the idea. I would love it if SI could make a future FM like this...

All attributes can be increased by tranining, but techincal attributes should have the most linear relationship to it. Mental stats will mostly increase as a player ages and becomes more experienced, but can also decrease in certain instances, such as if the coach benches him for a season. Physical stats will increase as a young player matures and peaks at around 28, and then slowly decreases. Other attributes can be increased or decreased with player talks, such as if you tell a player to be more/less aggressive. This can further distinguish different type of players and I think everyone would love it...

And btw, I don't think two-footedness should take up any CA, that just seems wrong. Instead, it should be natural and you can only improve it marginally via training.

[Deza]

We didn't only discovered about this positional/retraining bug, we discovered some bugs with "weaker foot" (example with AML and his unlogic CA when we change his weaker foot...). We also discovered "scouting bug", cause it seems to be that scouts are useless when talking about this multi-positional players!!!

One more thing...I'm not sure if we should look for solution for this, it's a job for SI, I just want to see what they have to say about this bug that makes the game unplayable!!!!!!!

Very interesting things to read in this thread. Thanks for sharing Ljuba, good work!

I´m also curious about what SI will say about these issues and how/if they will change/fix things. But I don´t really think this things make the game unplayable.

- For the "positional/retraining bug" the simple workaround is not to retrain a ST to get the ability to play as a DL/DC/DR. And normally no one will do a such retraining as it´s not naturally to do, in my opinion anyway.

- About the "weaker foot" thing, if I´m not wrong, the case is that the players who can use both their feet well don´t get as high attributes as a player(with same CA, same position) that is one-footed. Am I correct? But even if the two-footed player doesn´t got as good attributes as one-footed ones, he can still perform good or even better in the match as the ME gives him an adventage to be able of using two feet. This is for me a logically adventage and the thing we as mangers will do is not to only look for a players different attributes, we should also check his ability to use his feet, like we maybe check the height when looking for a DC. So this make sense for me.

- About the "scouts are useless when talking about this multi-positional players" I agree and will keep in mind to look once more on a player if he is cabable of playing in many different positions. He could be a really good player, even if the scout doesn´t say that The workaround for this issue is to keep in mind of looking extra at a player that can play in many positions.

If I got wrong or bad knowledge in something about this things, please tell

Thanks.

[halldors78]

vow how can anyone play the game like this??? i have one friend who plays it like this, every player with great unrealistic ratings and he wins everything for many years. its just so boring. the game just starts to be fun after 2-3 season when you have YOUR team, or when you take over a smaller team and make it big. it have to be a challenge just like in real live, otherwise its just cheating

[zeusbheld]

i think that, while the phenomenon is interesting enough, the hysteria over how it makes the game unplayable is comical and a bit sad. um. if you train a player at positions you wouldn't normally, for the sole purpose of improving some attribute, then just don't train them at positions they won't actually play.

looking for cracks in the game is interesting in terms of understanding how the game works under the hood but really i don't see the point of allowing yourself to do things a real life manager wouldn't and couldn't just to "win." allegedly we play this game to pretend we're football managers, don't we? or are there rich prizes for winning that no one told me about?

in other words, hurts when you do that? don't do that.

[United__FM]

we discovered some bugs with "weaker foot" (example with AML and his unlogic CA when we change his weaker foot...).

How does this one work? An AML with a strong right foot gets better attributes than the same CA one with a strong left foot?

[matt_132]

i think ljuba misinterpreted vsm's idea

if there were 2 attributes, tackling and finishing, minimum 0 maximum 5

let us say PA is converted to a certain amount of "points" which can be used to advance attributes of either skill depending on the training schedule

0 is the standard stat

to get 1, 5 "points" are required; total 5

to get 2, 10 more points required; total 15

to get 3, 18 more points required; total 33

to get 4, 30 more points required; total 63

to get 5, 47 more points required; total 110

therefore maximum points in this instance is 220

a player with 110 points can have 5 for one skill, 0 for other

a (better) player with 126 points can have 4 for both

What i want to know is, is the CA and PA decided by how good a player is at his best position, or if a player has other skills, not related to his position, does he have a higher ability potential

[achurchill]

I've not read through the entirety of this thread, but I had to say one thing: it's not entirely unrealistic that a striker, when trained as a defender would become a better striker because of it. The reason is quite simple. As a defender you begin to learn to "read" striker movement better, you begin to think like a defender. Then, when you go back to playing as a striker, you can "read" what the defender is thinking, and makes it easier for you to disguise your plans, do something unexpected and throw him off.

Personally, this happened to me. When I started playing football originally I was crap, and was always put in goal, and eventually got quite good. When I then went out on the pitch again, I was actually a pretty damn good striker because I knew what the GK or defence was expecting and so I would always be able to do the unexpected and put the ball in the back of the net time and time again.

Of course, this is schoolboy level, not professional level, and I wouldn't expect the results to be so significant as you are showing in the research here, but it's not totally unrealistic to think that a player trained in multiple positions would become better in their natural position when returned to it simply because they can read the game better.

[United__FM]

I've not read through the entirety of this thread, but I had to say one thing: it's not entirely unrealistic that a striker, when trained as a defender would become a better striker because of it. The reason is quite simple. As a defender you begin to learn to "read" striker movement better, you begin to think like a defender. Then, when you go back to playing as a striker, you can "read" what the defender is thinking, and makes it easier for you to disguise your plans, do something unexpected and throw him off.

Personally, this happened to me. When I started playing football originally I was crap, and was always put in goal, and eventually got quite good. When I then went out on the pitch again, I was actually a pretty damn good striker because I knew what the GK or defence was expecting and so I would always be able to do the unexpected and put the ball in the back of the net time and time again.

Of course, this is schoolboy level, not professional level, and I wouldn't expect the results to be so significant as you are showing in the research here, but it's not totally unrealistic to think that a player trained in multiple positions would become better in their natural position when returned to it simply because they can read the game better.

I see what you're saying and it's a good point, but then how the game does two-footedness is inconsistent with your point.

Apparently, being able to use both feet well means that a player isn't as good at using just one foot. And one of the justifications given that I've seen is that such a player would have less time devoted to practicing individual feet.

Well, shouldn't learning a new position take time away from practicing the original position then? Or alternatively, a new position does add skills like you say, but then using two feet should have a similar effect. Knowing how to do some skills with one foot should make learning to do the same with the other foot easier, like the multiple positions example you give.

[el sid]

I've not read through the entirety of this thread, but I had to say one thing: it's not entirely unrealistic that a striker, when trained as a defender would become a better striker because of it. The reason is quite simple. As a defender you begin to learn to "read" striker movement better, you begin to think like a defender. Then, when you go back to playing as a striker, you can "read" what the defender is thinking, and makes it easier for you to disguise your plans, do something unexpected and throw him off.

Personally, this happened to me. When I started playing football originally I was crap, and was always put in goal, and eventually got quite good. When I then went out on the pitch again, I was actually a pretty damn good striker because I knew what the GK or defence was expecting and so I would always be able to do the unexpected and put the ball in the back of the net time and time again.

Of course, this is schoolboy level, not professional level, and I wouldn't expect the results to be so significant as you are showing in the research here, but it's not totally unrealistic to think that a player trained in multiple positions would become better in their natural position when returned to it simply because they can read the game better.

This happened in a Japanese TV show called "Hungry Heart" or "Wild Striker" in some places. The guy is forced to play as a defender and then when he plays as a striker he reads all the movements of the defenders and goes on a scoring rampage.

[VMX]

No official word from SI yet?

I sincerely believe that all the effort that Ljuba82 put into this, as well as other people who contributed to the thread, deserve a bit more feedback on such an important subject. It would at least be nice to hear someone saying that they're aware of this problem and that they're going to look into this for future versions... or something.

[phnompenhandy]

No official word from SI yet?

I sincerely believe that all the effort that Ljuba82 put into this, as well as other people who contributed to the thread, deserve a bit more feedback on such an important subject. It would at least be nice to hear someone saying that they're aware of this problem and that they're going to look into this for future versions... or something.

To be fair, Paul C did reply on the last page to say they would look into the matter but are snowed under at the moment. I don't think you can complain about that; their priority is fixing things for a second patch for FM09. Any analysis of Ljuba82's findings would go into a remodelling of FM10 at the earliest. So let's cut SI some slack and ensure they are gently reminded in the early part of next year.

[HerrKaputt]

To be fair, Paul C did reply on the last page to say they would look into the matter but are snowed under at the moment. I don't think you can complain about that; their priority is fixing things for a second patch for FM09. Any analysis of Ljuba82's findings would go into a remodelling of FM10 at the earliest. So let's cut SI some slack and ensure they are gently reminded in the early part of next year.

I agree. Since in this thread there appear to be more scientific-minded people (I am one as well), you should know reasonably well that patching up a working model usually takes very little time (i.e. reducing rate of injuries, for example, probably just involved changing one or two scalar values somewhere in the game code). Completely redesigning the model takes new coding, then testing, then further redesigning/testing cycles until the thing works reasonably well.

Plus, although I agree that the current model has its problems, it isn't a game-breaker to me (the game-breakers for me are the in-game advertising and the poor quality of the Portuguese research, in comparison to that this is quite a minor issue).

I would indeed like some more detailed comment from SI (something like "we looked into this and agree that it has this and that fault; we will open a thread for suggestions for new models" would be too good to be true), but given that they are currently still making the activations work smoothly, and are still under fire because of many unpatched bugs and in-game advertisement issues, that's probably not forthcoming.

[Thandui]

how long does this take to kick in? i mean after you've retrained your player to centre back how long before the attributes go up?

[PaulC]

I've had a brief look at it, but havent come to any conclusions as to the best way to deal with it yet - partly because its not considered a "must fix" issue at this stage.....although it will certainly be addressed in time.

[GerdMuller]

That is one of the threads i wish i never would have read. Destroys so much illusion of realism in FM. I am trying to forget about all this and try not to train my strikers as defenders. Also i hope for a fix, but i understand Paul, this is something that seems to need a major overhaul and cant be fixed immidiately. I guess its something for FM 10.

[jayahr]

Thank you ver much for the answer Paul

Good to see that you are looking into this. Also, I fully agree with the content of what you said as we have lived with this comfortably for a while now already.

[Thandui]

i've tried to test this out and it hasn't worked, i put someones marking and tackling to 1, then re-trained them as centre back yet none of their attacking attributes have risen, why?

[Thandui]

sorry for this but i wanted to try this out, can anyone tell em how?

[zeusbheld]

That is one of the threads i wish i never would have read. Destroys so much illusion of realism in FM. I am trying to forget about all this and try not to train my strikers as defenders. Also i hope for a fix, but i understand Paul, this is something that seems to need a major overhaul and cant be fixed immidiately. I guess its something for FM 10.

to me it is one of those things like player height. i always try to find tall center halfbacks even if the stat means nothing to the match engine, because that's how i'd imagine doing it as a manager.

i only train strikers as defenders if i'd have a "real" reason to--if they could actually play center half, for example. and then i'd have their defending abnormally high for a striker. as i understand it, under those conditions the effect won't happen. it is interesting, though, and i do enjoy reading threads like this.

as for knowing it exists... so? database editors exist. if one wanted to find an unrealistic advantage, one could just edit the data. playing that way doesn't make for a very interesting game though.

[sthptngomad76]

That is one of the threads i wish i never would have read. Destroys so much illusion of realism in FM. I am trying to forget about all this and try not to train my strikers as defenders. Also i hope for a fix, but i understand Paul, this is something that seems to need a major overhaul and cant be fixed immidiately. I guess its something for FM 10.

Me too. I actually kind of regret signing up for forums in the 1st place, as I always find something that spoils the illusion- some sort of crack within the game, like the diablo tactics of old, the corner bugs etc. Tbh, its almost as if FM has become so complex and unwieldy that SI have 'lost control' of the code- the explanation for the corner bug for eg (that they can't fix it because it would unbalance the entire game) suggests this to some degree, that they no longer know what effect potential changes have long term. But then again I'm not a programmer (just an observer) so I wouldn't really know...

[outlander]

I've not read through the entirety of this thread, but I had to say one thing: it's not entirely unrealistic that a striker, when trained as a defender would become a better striker because of it. The reason is quite simple. As a defender you begin to learn to "read" striker movement better, you begin to think like a defender. Then, when you go back to playing as a striker, you can "read" what the defender is thinking, and makes it easier for you to disguise your plans, do something unexpected and throw him off.

Personally, this happened to me. When I started playing football originally I was crap, and was always put in goal, and eventually got quite good. When I then went out on the pitch again, I was actually a pretty damn good striker because I knew what the GK or defence was expecting and so I would always be able to do the unexpected and put the ball in the back of the net time and time again.

Of course, this is schoolboy level, not professional level, and I wouldn't expect the results to be so significant as you are showing in the research here, but it's not totally unrealistic to think that a player trained in multiple positions would become better in their natural position when returned to it simply because they can read the game better.

It could be a good explanation but there is big "BUT".

The striker can actually become worse in skills after learning DC position. It all depends on the skills destribution. There is an explanation to this too: a player can't be be good at different positions and he should lose skills when he plays in many positions(O'Shea irl ).

But now we basicallly have it random with opposite consequences with different players in the same position re-trainings.

I now met this problem quite seriously in my game. Here is my regen

He was MR/AMR at start but left only foot. So re-trained him ML position. That alone ate a few CA points(Ljuba showed calculations on page 2 I think) and that is important for a team playing in Blue Square North. Now that he is slow for a winger I think about re-training him to MC.

But there are two problems. First, with his skills I think that will eat a bit of his CA too. And secondly and mre important it will mean that I just lost those CA points that were taken by ML position as I won't use him there.

I really think about "cheating" with real time editors to remove ML positions(and he didn't have them at start) and then training him in game to MC so I won't lose CA due to his accomplishment at left flank.

[zeusbheld]

Me too. I actually kind of regret signing up for forums in the 1st place, as I always find something that spoils the illusion- some sort of crack within the game, like the diablo tactics of old, the corner bugs etc. Tbh, its almost as if FM has become so complex and unwieldy that SI have 'lost control' of the code- the explanation for the corner bug for eg (that they can't fix it because it would unbalance the entire game) suggests this to some degree, that they no longer know what effect potential changes have long term. But then again I'm not a programmer (just an observer) so I wouldn't really know...

you guys sound like the sort of people for whom the movie would be ruined when you found out you were actually watching a movie, and that what you were seeing wasn't really happening

[zeusbheld]

It could be a good explanation but there is big "BUT".

The striker can actually become worse in skills after learning DC position. It all depends on the skills destribution. There is an explanation to this too: a player can't be be good at different positions and he should lose skills when he plays in many positions(O'Shea irl ).

But now we basicallly have it random with opposite consequences with different players in the same position re-trainings.

I now met this problem quite seriously in my game. Here is my regen

He was MR/AMR at start but left only foot. So re-trained him ML position. That alone ate a few CA points(Ljuba showed calculations on page 2 I think) and that is important for a team playing in Blue Square North. Now that he is slow for a winger I think about re-training him to MC.

But there are two problems. First, with his skills I think that will eat a bit of his CA too. And secondly and mre important it will mean that I just lost those CA points that were taken by ML position as I won't use him there.

I really think about "cheating" with real time editors to remove ML positions(and he didn't have them at start) and then training him in game to MC so I won't lose CA due to his accomplishment at left flank.

if it's like 2008, he'll lose ML eventually once you stop training him there. worth a try.

[Rob_]

Honestly, my preferred solution to the issue which would not be inclined to do harm to the game is that players should refuse and/or be unable to be retrained to a position that differs from their natureal one by more than two arrays, i.e. strikers could not be retrained to D/WB/DM positions and defenders could not be retrained to AM/S positions.

This way you could fully maintain the positive aspects of attribute weighing and get rid of the most severe impacts of the exploit Ljuba discovered. Players who are natural DL/AMLs would remain unaffected.

Sol Campbell started out as a striker. Chris Sutton has played both in defense and attack.

Perhaps all youth players (except goalkeepers) should start out positionless and come with a recommendation from the youth team coach but we can choose how to develop the youth players as we see fit.

[zeusbheld]

Sol Campbell started out as a striker. Chris Sutton has played both in defense and attack.

Perhaps all youth players (except goalkeepers) should start out positionless and come with a recommendation from the youth team coach but we can choose how to develop the youth players as we see fit.

that would, if nothing else, be fun, and whether or not it is more realistic, it would feel more realistic.

[Ljuba82]

My “recommendation” to SI is that they need somebody that can be able to figure out a better mathematical model for the players attributes, and then use this model to build the game engine around. I’m saying this regarding the two-footed problem, because my opinion is that the fact that a player is two-footed shouldn’t take CA points that are used to determine all of his attributes (pace, consistency, …). But that’s just my personal opinion.

I agree with you...and also agree that your solution is better than current one, but it also have some faults....hopefully SI will solve this...

Very interesting things to read in this thread. Thanks for sharing Ljuba, good work!

I- About the "weaker foot" thing, if I´m not wrong, the case is that the players who can use both their feet well don´t get as high attributes as a player(with same CA, same position) that is one-footed. Am I correct? But even if the two-footed player doesn´t got as good attributes as one-footed ones, he can still perform good or even better in the match as the ME gives him an adventage to be able of using two feet. This is for me a logically adventage and the thing we as mangers will do is not to only look for a players different attributes, we should also check his ability to use his feet, like we maybe check the height when looking for a DC. So this make sense for me.

You are welcome

Actually in my oppinion it's ok that in situation when you have 2 players with same CA but one player is one footed, while the other is 2 footed......it's logical that first must have better other attributes...I mean we can talk about how much 2 footedness is imortant and how much other attributes should be afected by this, but I think we all agree that it have to use some CA pts.

The "weaker foot bug" I explained on previous pages...but in short....

...all other attributes are same.

1st AML with right foot 20, left 19...........179 CA

2nd AML with left foot 20, right 19............174 CA

3rd AML with right foot 20, left 1..............145 CA

4th AML with left foot 20, right 1..............160 CA

5th AML with right foot 20, left 15.............172 CA

6th AML with left foot 20, right 15.............171 CA

....it's not logic that if 1st player is better than second......that 4th is better than 3rd...

vow how can anyone play the game like this??? i have one friend who plays it like this, every player with great unrealistic ratings and he wins everything for many years. its just so boring. the game just starts to be fun after 2-3 season when you have YOUR team, or when you take over a smaller team and make it big. it have to be a challenge just like in real live, otherwise its just cheating

aaaaa...people read a bit....I don't play like this...I just like statistical examination...and when I get this stuff....I just start to explore and explore and found this...

i think ljuba misinterpreted vsm's idea

if there were 2 attributes, tackling and finishing, minimum 0 maximum 5

let us say PA is converted to a certain amount of "points" which can be used to advance attributes of either skill depending on the training schedule

0 is the standard stat

to get 1, 5 "points" are required; total 5

to get 2, 10 more points required; total 15

to get 3, 18 more points required; total 33

to get 4, 30 more points required; total 63

to get 5, 47 more points required; total 110

therefore maximum points in this instance is 220

a player with 110 points can have 5 for one skill, 0 for other

a (better) player with 126 points can have 4 for both

What i want to know is, is the CA and PA decided by how good a player is at his best position, or if a player has other skills, not related to his position, does he have a higher ability potential

Yes....nice idea

I've had a brief look at it, but havent come to any conclusions as to the best way to deal with it yet - partly because its not considered a "must fix" issue at this stage.....although it will certainly be addressed in time.

..if this is not considered a "must-fix" I don't know what is considered. This is major problem that affect whole game!!!

I really think about "cheating" with real time editors to remove ML positions(and he didn't have them at start) and then training him in game to MC so I won't lose CA due to his accomplishment at left flank.

...I play a little with this example. It's true that it would be better if player is only MR/AMR.....you can save up to 5 CA pts if player loose his ML/AML rating from 20 to 10 (remember that there is no need to decerease this rating below 10, cause changes in weight for attributes happens only when player reach rating of 11 on some new position). But it's probably somewhere around 2 or 3 CA pts (from 15 to 10).

...but guess what? If you train him to MC your player will be better by 2 CA pts...that means that all other attributes will be scaled down for thatdifference.

And if you look at player...it's logic...one of the better player attributes are passing, first touch, creativity, work rate....all attributes that consume a lot of pts for MC position.

The best solution is to train him to DC or ST (if he is versatile), but use him as winger...in that way you can increase a liitle his poor winger att.

[aka.arrogantio]

I'd have thought the practical solution is actually much less radical than some of those described

1) Player's attribute weightings should be determined primarily, if not exclusively, by their "natural position". This is actually pretty simple to implement, even at the expense of lots of complaints to researchers that their versatile players have changed a lot since the previous version. The only complication comes from players with more than one natural position, but I understand that the research guidelines suggest this shouldn't happen (and I'm sure that for the exceptions with two or more "natural positions" some kind of hierarchy of normal position importance can be coded to identify which position should be used to determine RCA)

2) Secondary positions should never be able to decrease the weightings of a particular low attribute (ie. require an increase in all other attributes for a player who is poorly suited to a secondary position if his CA is to stay the same).

If you teach a player to play in a position that he is rubbish at (ie Ibrahimovic at defending) then it should never lead to a commensurate improvement in attributes in other areas to compensate. A world class striker who has learned to play miles out of position in central defence doesn't need doesn't need to get better at attacking, defending or anything else to compensate for the fact that he's severely lacking as a "competent" defender - he's still world class in his preferred position. There's a reason DC is a secondary position - he might be a lot worse in it than in his first choice. It's nonsense to suggest that a player needs to be better overall to compensate for being badly suited to a secondary position.

If this causes selection issues (ie. AI managers thinking that even though he's usually a world class striker a "competent" centre back with a tackling of 1 is better than the available DC backup) then use attributes or positional RCA more in selecting the team. Compensating for weaknesses which barely matter in a position a player doesn't normally occupy by rounding up attributes which are important in a player's preferred position is not a solution - in the original example Ibrahimovic will still be rubbish if played at centre back and there's no logical reason for his increased versatility to make him that much better up front.

3) If secondary positions have any effect on players attribute ratings, it should only be to increase the weighting of particular high attributes (ie. require an increase in CA for a player who is well suited to a secondary position to to maintain all his attributes if he learns to play there).

To give a simple example - if you teach an average fullback with an unusually high proficiency in finishing he rarely uses to become an attacking winger/forward - then he arguably becomes quite a bit more effective as a player (ie. he is better at using his attacking talents to score goals if you play him as an attacker) in which case his ability (CA) arguably should go up. If he's already at his peak then either (i)retraining will be impossible/ineffective or (ii)learning to play a different position will actually have a negative impact on his overall abilities - ie. he will actually lose the confidence in front of goal or drop some of his defensive duties.

-

In practice rather than calculating RCA as a [badly] weighted average of all attributes' relative importance to all positions a player has a certain capability to play in, RCA should be calculated based simply on the weightings for a selected primary position.

If, and only if, the player has a higher RCA for secondary positions (1 < player's position rating <20) then the difference between RCA[primary] and highest, weighted RCA[secondary] should be added to RCA[primary] to give the actual RCA value used in game.

In game, an increase in CA should be needed in order to acquire aptitude to play in a secondary position where RCA[secondary] > RCA[primary]. If an increase is not possible but the game still determines that a point needs adding to the secondary position (where RCA[secondary] > RCA[primary]) then overall attributes should be recalculated (reduced in some places) to come to a value which adds up to the actual RCA value as described above

[sthptngomad76]

i think that, while the phenomenon is interesting enough, the hysteria over how it makes the game unplayable is comical and a bit sad. um. if you train a player at positions you wouldn't normally, for the sole purpose of improving some attribute, then just don't train them at positions they won't actually play.

looking for cracks in the game is interesting in terms of understanding how the game works under the hood but really i don't see the point of allowing yourself to do things a real life manager wouldn't and couldn't just to "win." allegedly we play this game to pretend we're football managers, don't we? or are there rich prizes for winning that no one told me about?

in other words, hurts when you do that? don't do that.

I dunno personally I'm getting a little tired of being told I cant do this or that in game or else the entire thing falls apart. I'm not allowed to allow myself to tinker with the different corner settings or else I get defenders scoring 30+ a season. Now I'm not allowed to fiddle with the training settings or else I get this. Shouldn't a game of this age be able to handle these sort of quirks? Its as though the game has become a little too complex for it own good.

[conshaldo_bonaldo]

maybe players(and obviously regens) should have a different PA for each and every position??

[PaulC]

maybe players(and obviously regens) should have a different PA for each and every position??

Unfortunately, from a research point of view that is impractical.

What will happen is that we will try to stop the weighted attributes in any of the player's primary positions exceeding his CA. But this is unlikely to happen in the next patch, more likely FM2010 if I am being honest.

[arnold_kidd]

Argh and I was thinking about making the next Dion Dublin (DC/ST)