Jump to content

Player development makes zero sense to me in this game.


Recommended Posts

3 hours ago, Freakiie said:

Researchers make mistakes. The game doesn't. A researcher might wrongly judge the peak of a player, the game won't. The whole "but researchers buff players" is really not all that relevant to the discussion of player development. Not to mention it also depends on the researcher and lets face it, more than a couple have a tendency to overhype the players they're rating (no surprise since it's largely done by fans of said clubs), meanwhile a Lewandowski needed to win world player of the year and a new set of researchers before someone looked at him and went "yeah, he really has improved these past few years" :lol:

Anyway, back to player development. Players in reality hit peaks, that's simply how it is. The issue in FM is that development is very one dimensional resulting in basically everyone going through the same process. Gain massive attributes till you either hit PA or hit 23 and then development is mostly gone. You won't see a player that's solid at like 25, but has PA to spare, so he slowly but surely grows towards an actual world class player in his later years. That's where the issue is.

If you want to translate the examples people are using here to FM, in FMs eyes they always had the PA to reach that level. A De Bruyne, Lewandowski, Pique, whatever other people named here, as newgens in FMs would always have had their massive PAs. However, due to how development worked, they wouldn't continue developing until late into their careers, even their early thirties in the case of Lewandowski, they'd either hit their PA before their development stops or they'd hit 23 and then stop massively improving and remain stuck at the level they hit early in their careers.

Now as for the whole "a player playing well should develop", why? Should Messi be improved because he plays well? Or does he play well, simply because he's the best footballer around? Also, lets say you change PA so that players playing well can always improve, then they improve, play even better the season after, should they then improve again? In the end, PA does what it's supposed to do; it indicates the peak a player can reach. Players having a peak is not the flaw in FM in my eyes and if people want to get into an argument about a peak or PA not existing in real life, well I disagree. Talent is a real thing and plenty of players branded as massive talents never got anywhere, as their talent was nothing more than them having developed quicker than the people around them, but then quickly hit their ceiling when they actually become a pro.

In the end, the issue in FM remains the absurd pace of development before a player hits 23. It makes everything else meaningless, as you can just build a squad of 22 year old world class players. Obviously nobody will ever see a latebloomer, as we never even bother with buying players that old in the hope they will develop. Why bother with that, when you can buy an 19 year old, that has the same attributes, of which you know that if he has PA left to spare he'll develop? We also won't see a Vardy story, because we can see his exact stats. If he'd be good enough to play in a higher league, he'd have the stats for that from the get go and thus he would have been poached from his non league club way earlier.

Long story short, PA is perfectly fine, the way how players reach their PA in FM is massively flawed.

Agree a lot with this. PA may not be perfect, but it works fine as is. A better dynamic PA system may be possible, but the current one does the job well and avoid weird results (you have to account there is thosuands of data points in thousands of saves, so potentially creating an unrealistic loop where a player plays better, gets better stats, plays better due to that and continues like that to the extreme is something that would happen).

 

The problem is totally on CA, been it very inflexible itself. It either hits PA very easilly and early on the players career or doesn't at all. And then when a certain age hits it simply goes down. There is a degree of exchange at that stage of physicals for metnals, thats true. But as you said at the end of the day its all the same for every player. The same process. You have players who reach their potential do so mostly at 22-23. Other players despite having good personalities and potential left at once they are 25+ gain maybe one point of attribute a season, even if they have playin time, great personality and training quality. And then you have 30+ players all behaving the same too. As in my example. Different ages and personalities yet they all consistently lost 1-3 points in at elast half the physicial attributes, with pace acceleration, stamina, agility and strength been the most consistently hit in that order. The first 3 more consistently.  Its a small sample size but it makes me doubt you can really have older players maintaining their physicials more or less with a more dragged out decline in game. Its easy to see it in the good current old players. The likes of Ronaldo and Messi suffer this same effect after just one season having lost around 8-10 points worth of physicals. I doubt they are that much hit in the changes between versions of the game.

Of course, its not supposed to be a percect system that has to match whatever the researchers are gonna give the next year to that player. But the aim is to be a plausible and sesible simulation that tries to follow what could realistically happen. And what I find funny is that in theory we already have things in game that should be able to dictate and differentiate how players develop or decline. Age should be a factor, but not such a sharp one as real players have demonstrated. We have a lot of variables like personalities, facilities and coaching quality, game time and performances, injuries suffered that could help making this process be variable yet sensible. SI itself says this things should matter. I guess they are progressively working on it but its not something they have achieved just yet. Specially given what others report about older versions.

Link to post
Share on other sites

  • Replies 275
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

4 hours ago, enigmatic said:

Agree with most of your post but FM has considerably slowed the average pace of early development in the last couple of versions.

It's still easy to build teams of very good young players, but that has more to do with the rest of your paragraph (you can see exactly which players that are unusually good for their age even if they're at a tiny overseas club or the AI doesn't play them)

 

It already is, and more variable than your example. I'd quite like a variable researchers can set that nudges players in a particular direction though

 

But the original purpose of the thread was arguing that a one or two point improvement in one particular key attribute over a period of sustained excellence wasn't enough. Now apparently we're in agreement that despite progressing from having his place in a mid table side under question to captaining a record breaking side, a one point attribute increase in most key areas to hit a hard PA cap of 159 is absolutely fine for Jordan Henderson.

Leadership, Determination and Work Rate can all easily be improved in senior players by three or four points in FM, FWIW 

 

Because, shockingly, games with millions of data points and many possible outcomes do not produce exactly the results people who like to complain about games on the internet expect. Of course it's also clear from these complaints that researchers are not increasing Bayern's attributes by a few points as a reward for winning a title

SI have said they're happy that in their soak tests, Man Utd on average come third and Spurs are on average battling for Europe. Also, gameplay matters as well as stats, so whilst it would be possible to get Kane scoring more goals for the AI by making him fast, Harry Kane is simply not fast, scoring lots of goals in previous seasons will not make him faster and so the researcher has avoided rating Kane in a way which makes him accelerate past defenders in the ME.

Agree or disagree with the outcome, it's not true that researchers do not consider match engine performance when assessing players.

You missed, presumably deliberately, the example of Rashford as a player who had a fine season and did not change by much, because scoring more goals did not convince the researcher that he had developed new areas of his game or that a previous attribute was a mistake.

More to the point, a great season IRL may be evidence some attributes set in the database might be too low, but a great season in game is definitely not evidence that a player's attributes are too low, because the attributes cause the performance

 

CA is not used in the match engine at all, and is just a weighted sum of the attributes. 

It's probable that you're missing all the good points of the defender who can't tackle and weak points of the tailor made CBs, and that faster teammates are being outpaced because they're tired or not trying to run at full speed.

This is fantastic.

So there definitely already are different development curves baked into the game? That’s wonderful news!

Link to post
Share on other sites

5 hours ago, enigmatic said:

But the original purpose of the thread was arguing that a one or two point improvement in one particular key attribute over a period of sustained excellence wasn't enough. Now apparently we're in agreement that despite progressing from having his place in a mid table side under question to captaining a record breaking side, a one point attribute increase in most key areas to hit a hard PA cap of 159 is absolutely fine for Jordan Henderson.

The purpose isn't just that a one point improvement in one attribute isn't enough, it's also that it will happen up to ages 23/24 and stagnate afterwards (see point 3 of OP). Henderson is an example of someone whose non-physical attributes increased across the board by 2 points after age of 25. This is something that the game doesn't replicate well enough.

5 hours ago, enigmatic said:

Because, shockingly, games with millions of data points and many possible outcomes do not produce exactly the results people who like to complain about games on the internet expect. Of course it's also clear from these complaints that researchers are not increasing Bayern's attributes by a few points as a reward for winning a title

Hence changes in player attributes aren't made to correct in-game performance or to work better with the ME. They're judged subjectively on IRL and then the ME is made around them to try and simulate IRL.

5 hours ago, enigmatic said:

Agree or disagree with the outcome, it's not true that researchers do not consider match engine performance when assessing players.

I never said that they don't do that at all, but that the change in attributes is driven predominantly by IRL as they cannot know how the changes will be interpreted by the new ME that they don't have access to.

5 hours ago, enigmatic said:

You missed, presumably deliberately, the example of Rashford as a player who had a fine season and did not change by much, because scoring more goals did not convince the researcher that he had developed new areas of his game or that a previous attribute was a mistake.

How about we don't presume that you can read my mind? I didn't comment on Rashford because his example is not what we are talking about. He is younger than 23 and his attributes between FM20 and FM21 have been increased in the amount in which the game would do it after a good season.

Actually, now that I took a better look at Rashford between FM20 and FM21, technicals are the same except for a 2 point bump in Passing and 1 point bump in Technique, phyisicals are pretty much the same, slight bump in Agility, Balance, Stamina and Strength (and apparently he lost 7kg of weight) that the game also does, Mentals however

Anticipation: +2

Bravery: +6

Composure, Decisions: +1

Leadership: +6

Off The Ball: +1

Vision: +3

While I might see Composure, Anticipation, Decisions  and OTB increased by 1 after a good season, I have never seen a +3 increase in Vision or a +6 increase in Bravery over 1 season in a player who is 21 over the course of a single season. I've never seen a +6 increase in Bravery over the course of one season in any player.

Quote

but a great season in game is definitely not evidence that a player's attributes are too low, because the attributes cause the performance

That's not what the argument is. The original argument is that PA is too rigid and that peak CA is hit too quickly and then stagnates, and the supporting argument is that if a player had a great (or a bad) season IRL, that would be reflected in the change of attributes. Going by OP's example of Bastoni, if the scenario he is describing happened IRL, his CA, PA and attributes would get bumped up.  This is something that the game doesn't do. The game also doesn't do the opposite, a player might score 30 goals at age 22, reach peak CA, and then score 8 goals in each of the next 5 seasons and his CA won't take much of a hit, if any at all. There's no regression to the norm. In-game there's little difference between players in mid 20's who fully developed to the same peak CA by age 23/24 but afterwards one had mediocre seasons and the other had great seasons.

Edited by goranm
Link to post
Share on other sites

32 minutes ago, goranm said:

Hence changes in player attributes aren't made to correct in-game performance or to work better with the ME. 

OK, you know best and all the researchers and SI employees posting in the Data Issues forum are the ones who don't know how the process works. :lol:

 

58 minutes ago, goranm said:

I didn't comment on Rashford because his example is not what we are talking about. He is younger than 23 and his attributes between FM20 and FM21 have been increased in the amount in which the game would do it after a good season.

Sure. He's not what you're talking about because his example is one of many which contradicts the idea that someone with a high natural ability and potential for their age like Bastoni gets radical overhauls of CA and PA every time he has a good season. If I was a betting man I'd put money on him not radically changing between 25 and 29 either.

 

34 minutes ago, goranm said:

Going by OP's example of Bastoni, if the scenario he is describing happened IRL, his CA, PA and attributes would get bumped up.  This is something that the game doesn't do.

Except that 

[i] if Bastoni has a similar career to Pique in an FM21 save, he will improve his attributes by a similar modest amount to hit a similar PA. Hell, I've even got an FM21 save with a 33 year old Bastoni whose attribute profile looks like Pique's current one, partly because of the significant ball playing increases and partly because he signed for Barcelona :D . Pretty sure Pique didn't improve much in his late 20s in FM research either...

[ii] if Bastoni performs well in FM, this is not evidence his attributes were set too low in FM. Therefore attribute bumps because a player just had a good season is something the game doesn't need to do.

 

40 minutes ago, goranm said:

The game also doesn't do the opposite, a player might score 30 goals at age 22, reach peak CA, and then score 8 goals in each of the next 5 seasons and his CA won't take much of a hit, if any at all. 

Why would it need to? The 8 goals they scored are not evidence the attributes may have been set incorrectly, they are the direct consequence of the attributes a player had at the time he was [not] scoring.

Of course attributes can drop prematurely if a player is unprofessional, not playing or injured, but it would be ridiculous to design the player development model on the assumption that if the attributes cause a player to perform well they might have been mistakenly set too low or vice versa

 

The causality works the opposite way round: in game natural development improves key attributes which allows players to perform better.  I'm not sure why people are so determined to argue this should be reversed.

Link to post
Share on other sites

I think some of the people have missed one part of my post. I have mentioned "FUN"

I am not saying that the less strict development would be more realistic. I am saying it would be more fun.

It would help with motivation in long term saves, it would make bonding with some player for many years also more fun because you could play longer just to see your favourite player maxed at age of 32 for example ( of course according to his form, potential, your results, his entire career, with proper training, etc )

If there is so little realistic things in the game, I don't know why development must be so strict for older players than 23.
- Are tactics in the game realistic? No
- Are transfers realistic? No
- Is winning Champions League with Lower League teams realistic? No

And that is all GOOD. If anything was realistic we would not be able to have fun, nor to win trophies with lower league teams. So, I don't see why development also would not be more fun. ( of course keep the realistic parts of development, but let players develop further as it was until 2015 I think )

Link to post
Share on other sites

46 minutes ago, enigmatic said:

OK, you know best and all the researchers and SI employees posting in the Data Issues forum are the ones who don't know how the process works. 

I doubt the process is how you're describing it. It is easy to generate empirical data that shows how changes in the attributes between versions of FM did not correct in-game performance to a significant amount, since the game will not only fail massively at predicting IRL (which is a huge task), it will keep making the same teams or players over/underpowered for several versions in a row (which is not as huge of a task to correct, but still challenging).

46 minutes ago, enigmatic said:

Sure. He's not what you're talking about because his example is one of many which contradicts the idea that someone with a high natural ability and potential for their age like Bastoni gets radical overhauls of CA and PA every time he has a good season. If I was a betting man I'd put money on him not radically changing between 25 and 29 either.

What is the point of one cherry-picked example though? There are examples of players that DO get radical overhauls after good seasons (and one could argue that Rashford is such an example with two of his attributes getting a +6 bump and one getting a +3 bump). The point is that generally the game doesn't represent this well.

46 minutes ago, enigmatic said:

Except that 

[i] if Bastoni has a similar career to Pique in an FM21 save, he will improve his attributes by a similar modest amount to hit a similar PA. Hell, I've even got an FM21 save with a 33 year old Bastoni whose attribute profile looks like Pique's current one, partly because of the significant ball playing increases and partly because he signed for Barcelona :D . Pretty sure Pique didn't improve much in his late 20s in FM research either...

[ii] if Bastoni performs well in FM, this is not evidence his attributes were set too low in FM. Therefore attribute bumps because a player just had a good season is something the game doesn't need to do.

I don't see how this is an "except that", since I'm talking about what would happen in the next FM if certain events IRL transpired, and you're talking only about what would happen in FM.

46 minutes ago, enigmatic said:

Why would it need to? The 8 goals they scored are not evidence the attributes may have been set incorrectly, they are the direct consequence of the attributes a player had at the time he was [not] scoring.

The point is that they'll have the same attributes when scoring 30 and when scoring 8. This would not happen IRL between versions of FM (e.g. Morata).

46 minutes ago, enigmatic said:

The causality works the opposite way round: in game natural development improves key attributes which allows players to perform better.  I'm not sure why people are so determined to argue this should be reversed.

Because (1) it's not only the attributes that contribute to a better performance, and (2) the development is too static. After a certain age there is no regression or progression around the norm after neither a bad nor a good season.

Edited by goranm
Link to post
Share on other sites

I also need to dispel the myth that signing for Barcelona increases your CA & PA. It does not.

There can be a coming together of factors whereby a club researcher for Team A realises they have underrated a player, in the build up to the interest of Big Club Name Z and should a transfer eventually materialise the player has higher attributes. 

If a player is performing badly, with his current attributes in the game, that doesn't mean he needs to have a CA collapse to further perform badly. That just makes no sense, he's already performing badly, so why does he need to get worse to keep performing badly?

The opposite is true, if a player scores 50 goals a season with the attributes he has, why do they need to get better? He's already playing well, that doesn't mean he suddenly needs to get better defensively, or faster, or stronger. He's already performing well and his attributes are sufficient enough to enable that. It also doesn't guarantee that he has the potential to do more. 

Link to post
Share on other sites

15 minutes ago, santy001 said:

If a player is performing badly, with his current attributes in the game, that doesn't mean he needs to have a CA collapse to further perform badly. That just makes no sense, he's already performing badly, so why does he need to get worse to keep performing badly?

The opposite is true, if a player scores 50 goals a season with the attributes he has, why do they need to get better? He's already playing well, that doesn't mean he suddenly needs to get better defensively, or faster, or stronger. He's already performing well and his attributes are sufficient enough to enable that. 

A point is that the same player will not take a significant hit or a significant gain to his CA if he is playing badly or if he's having a dream season relative to his norm. If a player has an average season 1, but a very good season 2, I'd expect a progression from the norm. If a player has a very good season 1, but an average season 2, I'd expect a regression to the norm. However, in-game after a certain age this doesn't seem to matter (and if it is actually doing this, it is not telling us). The same player with the same attributes can play both above and below the norm, and his attributes will be pretty much kept static.

Edited by goranm
Link to post
Share on other sites

1 minute ago, goranm said:

A point is that the same player will not take a significant hit or a significant gain to his CA if he is playing badly or if he's having a dream season relative to his norm. If a player has an average season 1, but a very good season 2, I'd expect a progression from the norm. If a player has a very good season 1, but an average season 2, I'd expect a regression to the norm. However, in-game after a certain age this doesn't seem to matter. The same player with the same attributes can play both above and below the norm, and his attributes will be pretty much kept static.

This is where a short-term reputation system would be useful - CA and PA should be reflective the player's actual current and potential ability, not how he is currently performing, which could be heavily influenced by the team around him (think Barcelona Messi vs Argentina Messi).  Short term reputation, however, would influence how a player is perceived by scouts and managers based more on a couple of season's form than long-term metrics like CA and the game's current (long-term) reputation system. Movement of attributes should be more to do with the personal qualities of the player, game time, injuries and quality of training rather than anything form related.

 

Link to post
Share on other sites

What i really don't understand is how a player's physical attributes are linked to current ability. For example, you set a player to have 20 for pace in the editor but, because he's 16 you also set his CA to, say, 100. As a result, his pace will be about 10 or 11 when you start, only reaching 20 as his PA maximises.

To me, pace etc. should have no bearing on ability. If you're fast, you're fast - probably more so between 16 and 23. 

Link to post
Share on other sites

2 minutes ago, martplfc1 said:

What i really don't understand is how a player's physical attributes are linked to current ability. For example, you set a player to have 20 for pace in the editor but, because he's 16 you also set his CA to, say, 100. As a result, his pace will be about 10 or 11 when you start, only reaching 20 as his PA maximises.

To me, pace etc. should have no bearing on ability. If you're fast, you're fast - probably more so between 16 and 23. 

No, if you set a player with 20 for pace in the editor he can have 20 for pace regardless of what his age and ability is. But CA is a weighted sum of all the attributes, so if you set all the attributes to add up to more than the number the CA allows, they will all be reduced.

And pace is obviously an important component of footballing ability (IRL and even more in the FM ME) as anyone who watches a team with Salah and Mane up front should be able to easily see.

 

 

1 hour ago, goranm said:

I doubt the process is how you're describing it. It is easy to generate empirical data that shows how changes in the attributes between versions of FM did not correct in-game performance to a significant amount, since the game will not only fail massively at predicting IRL (which is a huge task), it will keep making the same teams or players over/underpowered for several versions in a row (which is not as huge of a task to correct, but still challenging).

I'm lost now, the first part of your attempts to insist I'm wrong involved insisting researchers were making huge changes all the time, and now to try to contradict me on another point you're insisting it is easy to prove they don't change them enough? :seagull:

For the avoidance of doubt, I am not suggesting that SI's research is perfect, I am suggesting that it aims to ensure players perform believably in the ME (and transfer market), and that researchers take in game performance into account when making their adjustments (more so than people haranguing them to increase a rating because the player scored goals, or fantasy-football style rankings). So instead of conducting some bizarre empirical experiment on whether the data matches what you think it should be, you could just ask a friendly researcher who's popped in:

@santy001, do you directly and indirectly take how players behave in the match engine (and selection/transfer AI) into account when [re]rating players or not? 

 

 

1 hour ago, goranm said:

The point is that they'll have the same attributes when scoring 30 and when scoring 8. This would not happen IRL between versions of FM (e.g. Morata).

*yawns*

come on, you're surely just trolling now.

I simply refuse to believe that someone capable of understanding concepts like regression to the mean is genuinely too thick to understand that if match engine calculations based on a player's attributes cause a player score few goals in a season, the attributes do not need to be lowered to permit them to not score many goals.

Link to post
Share on other sites

2 hours ago, enigmatic said:

*yawns*

come on, you're surely just trolling now.

I simply refuse to believe that someone capable of understanding concepts like regression to the mean is genuinely too thick to understand that if match engine calculations based on a player's attributes cause a player score few goals in a season, the attributes do not need to be lowered to permit them to not score many goals.

Surely you can converse in a civilized manner without calling me stupid and a troll? A common thread in your replies is that you're replying only to parts of statements and omitting others; this time you are intentionally omitting key parts of what I'm saying  just so you can be toxic - it's not that "if the ME causes a player to score few goals in a season, then the attributes need to be lowered further" (which I've never said), it's that "the ME allows the same set of attributes to score 30 and to score 5 goals". Do you understand how wildly different those two statements are?

2 hours ago, enigmatic said:

I'm lost now, the first part of your attempts to insist I'm wrong involved insisting researchers were making huge changes all the time, and now to try to contradict me on another point you're insisting it is easy to prove they don't change them enough? 

Uh, no. First of all, please quote where I'm "insisting researchers were making huge changes all the time". You can't because that's not what I said anywhere, you're creating a strawman of what I'm saying. Second, I'm not saying that "they don't change them enough", I'm saying that if the changes are to correct in-game performance, then the changes are not working properly because in consecutive versions we see the same teams under/overperforming. You understand the premise and the conclusion and how they differ from what you say that I'm saying? What I've also said is that the predominant driver of attribute changes is IRL performance, and not corrections to ME. Pay attention to adjectives.

2 hours ago, enigmatic said:

So instead of conducting some bizarre empirical experiment on whether the data matches what you think it should be, you could just ask a friendly researcher who's popped in:

You mean the bizarre empirical experiment of playing the game? The game is literally advertised as a simulation, so it's not that I think the data should match IRL, it's how the people in SI/SEGA are selling it.

Edited by goranm
grammer, other clarifications
Link to post
Share on other sites

7 hours ago, Marko1989 said:

I think some of the people have missed one part of my post. I have mentioned "FUN"

I am not saying that the less strict development would be more realistic. I am saying it would be more fun.

It would help with motivation in long term saves, it would make bonding with some player for many years also more fun because you could play longer just to see your favourite player maxed at age of 32 for example ( of course according to his form, potential, your results, his entire career, with proper training, etc )

If there is so little realistic things in the game, I don't know why development must be so strict for older players than 23.
- Are tactics in the game realistic? No
- Are transfers realistic? No
- Is winning Champions League with Lower League teams realistic? No

And that is all GOOD. If anything was realistic we would not be able to have fun, nor to win trophies with lower league teams. So, I don't see why development also would not be more fun. ( of course keep the realistic parts of development, but let players develop further as it was until 2015 I think )

Fun is subjective 

Link to post
Share on other sites

I disagree that player development is broken in FM - and I think the CA/ PA model works pretty well considering the limitations. There are a few issues though which could certainly be improved.

You should be able to "shape" players better - still under the limitations of them hitting their PA. For example, if I was to put a McTominay on shooting training (and if he has the right determination & professionalism), over time I should see significant improvements in those metrics and not just a +1 increase across the board. 

Now this is not impossible but it seems quite hard to achieve unless there is a large enough gap between CA & PA - I did get the passing attribute up for my DM (a new-gen) from 8 all the way to 13 over two seasons by focusing on that as it was a particularly weak area. That was encouraging to see. 

I also agree that physical attributes decline too quickly with older players, although it seems to be a little slower than last year now - but it is certainly not reflective of real life.

Link to post
Share on other sites

8 hours ago, enigmatic said:

No, if you set a player with 20 for pace in the editor he can have 20 for pace regardless of what his age and ability is. But CA is a weighted sum of all the attributes, so if you set all the attributes to add up to more than the number the CA allows, they will all be reduced.

And pace is obviously an important component of footballing ability (IRL and even more in the FM ME) as anyone who watches a team with Salah and Mane up front should be able to easily see.

Pace shouldn't depend on anything. You either have it or you don't (with shades of grey in between). I don't really get why it's linked to anything else. It's an important 'attribute' in football, but not a component of 'ability' in my view.

Particularly, I don't get why it has to be weighted depending on current ability. You can still be fast or not, regardless of other ability factors.

Edited by martplfc1
Link to post
Share on other sites

10 hours ago, goranm said:

The point is that they'll have the same attributes when scoring 30 and when scoring 8. This would not happen IRL between versions of FM (e.g. Morata).

Because (1) it's not only the attributes that contribute to a better performance, and (2) the development is too static. After a certain age there is no regression or progression around the norm after neither a bad nor a good season

Im unsure why you insist so much on this. As in, what would this achieve in game?

The problem I see its that the values meaning is different IRL and in the game. The starting database values are the subjective opinion/evaluation of a researcher that is trying to quantify something relatively abstract in a 0-200 scale. This means that performance is very likely to have an effect because its very likely to impact that subjective opinion notably. But even then, a notable change on the researcher assesment and resulting skills in the database doesn't neccesarilly mean neccesarilly that the player skill itself has varied that much. The change can totally be caused by the researcher reevalauting his previous assesment, because after seeing the new performance may realize thathis ratings were before too low, or too high.

 

Now move to FM. Attributes there are, in the other hand, a very objective thing. For the ME having 172 dribbling, or whatever attribute, is what it is, it means a very specific thing inside its its calculations. Now, it doesn't guarantee a given level of performance because there is a degree of randomness, and because there is many more factors involved on the final performance of the player in an area (other related attributes, tactics, morale, personality, opposition attributes and tactics..), but no matter if the performances were good or bad the rating was never "wrong" because the ME doesn't really understand about right or wrong. It doesn't need to adjust the attributes just because of performances. As you yoruself said, performances can vary a lot for the same attributes, and that for me is an achievement of the ME, as long as overall it sill stands true that better attributes will lead to better results if everything else is kept the same. 

So, I dont understand yet what the simulation will gain from focusing attribute changes so much around perfromance as you seem to intend. The system's changes attributes when there is a cause for such. And it does already seem (Im not fully sure, bu I beleive it does) to account for performance to some extent, at least to evaluate the quaity of playing time. Having CA  (or even worse PA) been too dependant in performance feels like it could spiral very easily. If good or bad results shifted attribtues notably by themselves it would likely lead to more good or bad results respectively which would compont to extreme changes in some players too often on a questionable basis.

Going back to the initial question, what would be achieved by having performance drive attributes up or down more notably? More variance and extreme changes in players attribtues through their career, thats for sure, but I doubt it would be "good" for the game experience, specially when measuring performance in game can be iffy, as we have seen off lately with the ratings issues.

The game its supposed to simulate organic growth and decline based on reasonable factors, not the subjective changes on the assesment of a player's skill by researchers that may happen between versions of the game.

To be honest, it even does simulate even the latter part already, as scouts/coaches already adapt their idea of how good a player is and can get based a lot on things like their performance even if their attributes haven't changed that drastically yet. But the point is that this is compeltelly unrelated to the actual attributes themselves and the effect on the ME.

Link to post
Share on other sites

12 minutes ago, martplfc1 said:

Pace shouldn't depend on anything. You either have it or you don't (with shades of grey in between). I don't really get why it's linked to anything else. It's an important 'attribute' in football, but not a component of 'ability' in my view.

Sorry, post happened while I was writing mine so I couldnt add the quote in the previous one.

I dont see why you would say that. Like the way you put it you could say that for any attribute?

Not only it would break balance in favour of players with good pace, its not really justified. Pace is an ability like any other physical which has a very tangible contribution to the performance of a player an that can be trained and improved, or decline over time depending on several factors. Of course you may say there is a talent factor, and no matter how much you train you may not be able to be the fastest. But that's true for all the attributes.

Edited by Jervaj
Link to post
Share on other sites

2 minutes ago, Jervaj said:

Sorry, post happened while I was writing mine so I couldnt add the quote in the previous one.

I dont see why you would say that. Like the way you put it you could say that for any attribute?

Not only it would break balance in favour of players with good pace, its not really justified. Pace is an ability like any other physical which has a very tangible contribution to the performance of a player an that can be trained and improved, or decline over time depending on several factors. Of course you may say there is a talent factor, and no matter how much you train you may not be able to be the fastest. But that's true for all the attributes.

I get that it contributes to a player's overall game but I don't agree with the concept that a player has to have achieved a certain current ability level to maximise any physical attribute, not just pace, nor that if you want to allow improvement in one area, it has to be taken from the total 'pool' of attributes available, e.g. pace has to be reduced for others to go up.

In my view physical attributes should be fixed depending on age, injuries and other factors which affect such attributes, which includes the mental attributes to some extent - professionalism, determination etc. 

Appreciate that is probably a programming nightmare but I'm not sure the current method is particularly realistic, albeit it functions well enough.

In the same sense, I don't agree that if you can play in multiple positions or have two strong feet, your attributes would be lower than somebody who can play in a single position and has a terrible left foot, for example - all other things being equal.

Link to post
Share on other sites

I like how the system works, SI has it perfectly covered, stop moaning and learn how to develop your players.

If you want a player to have a 18 marking starting in 15, just train them as a Central Defender with Individual in defensive positioning. If you train him as a ball playing defender you will be diluting the gains (CA) into more atttibutes. Is as easy as that.

The AI is rubbish at developing players though, and we all know the AI needs a huge overhaul as this game is too easy, but player development is very easy to crack, watch Youtube videos, there are plenty of them.

Edited by Sharkn20
Link to post
Share on other sites

32 minutes ago, martplfc1 said:

I get that it contributes to a player's overall game but I don't agree with the concept that a player has to have achieved a certain current ability level to maximise any physical attribute, not just pace, nor that if you want to allow improvement in one area, it has to be taken from the total 'pool' of attributes available, e.g. pace has to be reduced for others to go up.

In my view physical attributes should be fixed depending on age, injuries and other factors which affect such attributes, which includes the mental attributes to some extent - professionalism, determination etc. 

Appreciate that is probably a programming nightmare but I'm not sure the current method is particularly realistic, albeit it functions well enough.

In the same sense, I don't agree that if you can play in multiple positions or have two strong feet, your attributes would be lower than somebody who can play in a single position and has a terrible left foot, for example - all other things being equal.

Ah, ok. I think I didn't understood your point fully last time.

The problem is that having something so impactful as physicals decoupled from CA would make balancing very difficult. Like, players that have good physicals will very easily match player without them in the rest while keeping the notable physical advantage. And also the opposite, once a player reaches their PA you could train him just in physicals to further improve his performance above his expected level.

29 minutes ago, martplfc1 said:

Also, if the game favours players with pace, that's a match engine problem that should be looked at.

I didn't mean it in such a way as in been pacey is OP right now. I meant that if pace was to be decoupled from CA, then players with high pace would be very strong because they could easiy be equally good in the rest of attributes and keep that pace advantage which is rather impactful.

The only attributes that seem to have no CA impact right now are those that are have a double edged component to them and that rely a lot on other attributes to give you a net benefit. So its not a simple "this player is much better because he has high X and the rest he can just train as he has CA/PA to spare".

Edited by Jervaj
Link to post
Share on other sites

1 hour ago, Jervaj said:

Im unsure why you insist so much on this. As in, what would this achieve in game?

It would better reflect player development, as in players wouldn't be so static and predictable in development after age 23/24.

1 hour ago, Jervaj said:

but no matter if the performances were good or bad the rating was never "wrong" because the ME doesn't really understand about right or wrong. It doesn't need to adjust the attributes just because of performances. As you yoruself said, performances can vary a lot for the same attributes, and that for me is an achievement of the ME, as long as overall it sill stands true that better attributes will lead to better results if everything else is kept the same. 

The game its supposed to simulate organic growth and decline based on reasonable factors, not the subjective changes on the assesment of a player's skill by researchers that may happen between versions of the game.

The rating is never wrong, but that doesn't mean that the player can not improve or decline due to: injuries, training, or match experience. The issue is about development - as it is currently in-game, players generally develop until 23/24 and then stagnate. A player is already pretty much defined at that age and there is little flexibility on how his profile will look like in the future. In reality player development isn't so static, players adapt to new circumstances and change their game often - it's not that the game should necessarily mimic the subjective changes made by the researchers, it is that these subjective changes are evidence that players develop differently to how the game is developing them.

 

Edited by goranm
Link to post
Share on other sites

@enigmaticHow players perform in the match engine isn't a huge part of it on the front end, but there is always a consideration. In the past I've had to re-do Peter Crouch's stats because in one particular version of the ME he was far too effective (despite being at around 130CA or lower). That was the only way around it. But then in one edition of the game when Ryan Shawcross was in the 140's CA range, top sides kept buying him and I didn't touch the player because I raised what felt a valid point to me, why were these teams who were well known for employing build-up play from the back buying Ryan Shawcross? At his peak, he was an elite, no-nonsense CB. Fast, strong and good in the tackle, with a good sense of positioning. 

I will always advise people who raise general stuff about attributes to look at how they play in the game, because sometimes you can find a player does certain things with attributes you might not expect. Coming back to the Crouch example, there are often multiple ways you can recreate a player for FM depending on exactly how you think about it. For example, a player who can bring the ball down well - you could look at say First Touch, Technique, Off the Ball & Anticipation and you can then be in two minds about maybe you go higher on First Touch or Anticipation depending on exactly how you feel. Sometimes you can go higher on attributes you feel if you pull it back on other attributes like decisions & consistency. 

Link to post
Share on other sites

1 hour ago, goranm said:

It would better reflect player development, as in players wouldn't be so static and predictable in development after age 23/24.

The rating is never wrong, but that doesn't mean that the player can not improve or decline due to: injuries, training, or match experience. The issue is about development - as it is currently in-game, players generally develop until 23/24 and then stagnate. A player is already pretty much defined at that age and there is little flexibility on how his profile will look like in the future. In reality player development isn't so static, players adapt to new circumstances and change their game often - it's not that the game should necessarily mimic the subjective changes made by the researchers, it is that these subjective changes are evidence that players develop differently to how the game is developing them.

 

Well then we more or less agree on the issue, but diverge on what should change to adress it.

I dont feel like the current limitations of the development system radicate on match performance not driving progress or decline enough.  The reason why players are defined and stagnate at 23/24 is because:

- Growth seems to be slowed down too sharply at 24/25+ years

- Players overall hit their PA too easily and too quickly (usually around that 22/23 mark), which limits further growth to shifting some stats around at best and compounds with the previous point to cause that stagnation.

- Decline is too tied with age. I have seen some minor fluctuations but those are rearelly of any noticeable magnitude. Players for the most part are either going up, stable or going down, and every phase basically happens at very similar ages for all of them.

So summing up, in my opinion, the system at its core its good enough as it is. The issue is that the crazy fast development in the early years which makes them world class at 22 yo should be the exception rather than the rule, requiring very specific conditions and some luck to happen. And that age should be a milder and more progressive factor when it comes to speed of growth and/or decline, so others can counteract its effect for longer periods of time.

I dont feel like increasing performance impact its neccesary to fix the current issues, and could come with new ones as I already mentioned earlier.

Link to post
Share on other sites

8 hours ago, goranm said:

 it's not that "if the ME causes a player to score few goals in a season, then the attributes need to be lowered further" (which I've never said), it's that "the ME allows the same set of attributes to score 30 and to score 5 goals"

Fine, let's skip over the fact you keep saying things like  a player might score 30 goals at age 22, reach peak CA, and then score 8 goals in each of the next 5 seasons and his CA won't take much of a hit, if any at all.   then, because it's 'toxic' of me to assume you were maintaining a consistent argument, and maybe you don't think players should take a CA hit for not scoring goals in the ME now. Great, perhaps we'll end up agreeing :) 

Yes, there is absolutely no doubt that the same set of attributes allow 30 goals and 5 goals under different circumstances. This is not in dispute.

There is also absolutely no doubt that these goals [not] scored are the direct result of the player's attributes being used in match engine calculations also involving teammates, tactics and luck, and a player's performance in the ME is therefore categorically not evidence the wrong values have been set in the database. 

Likewise, it seems fairly difficult to disagree that goals [not] scored IRL are totally unconnected to what an SI researcher has entered into a database. And since we also agree SI researchers are capable of error, it seems possible that a researcher will sometimes conclude that a player scoring very few goals over the course of a season IRL implies ratings made on the assumption they were the sort of player that would consistently score goals might be wrong. Especially if, like Morata, it involves different researchers.  So Morata's attributes/CA/PA lurching up and down from season to season as researchers try to figure out whether he's an ill-used massive talent or an average squad player don't tell us very much about how player development should look for an inconsistent player in game. Same goes for researchers making quirky adjustments like Rashford's Bravery (the only area in which Rashford got massively braver last season was politics!) as well as naturally rounding out the player a little, or Leicester's squad getting overhauled not because winning the league meant its ageing players learned new skills but because it was clear some of them had been very underrated in a game you'd need to sell them to mount a title challenge in.

So since FM's player progression model doesn't have to worry about the possibility of uncertainty, error, differences of opinion or having more or less freedom to increase CA based on SI guidelines and not making the AI do weird things, it shouldn't really resemble a research update model that is heavily influenced by those and really isn't evidence SI's development role is doing anything wrong. And yet when we look at cases like Pique's where their potential was accurately predicted, researchers really haven't needed to change attributes more than SI's development model changed Bastoni over a similar win-everything career journey in FM.

Are we in agreement yet? :) 

 

 

2 hours ago, martplfc1 said:

I get that it contributes to a player's overall game but I don't agree with the concept that a player has to have achieved a certain current ability level to maximise any physical attribute, not just pace, nor that if you want to allow improvement in one area, it has to be taken from the total 'pool' of attributes available, e.g. pace has to be reduced for others to go up.

In my view physical attributes should be fixed depending on age, injuries and other factors which affect such attributes, which includes the mental attributes to some extent - professionalism, determination etc. 

Appreciate that is probably a programming nightmare but I'm not sure the current method is particularly realistic, albeit it functions well enough.

In the same sense, I don't agree that if you can play in multiple positions or have two strong feet, your attributes would be lower than somebody who can play in a single position and has a terrible left foot, for example - all other things being equal.

There isn't a concept that a player has to have achieved a certain current ability level to maximise any physical attribute, not just pace, nor that if you want to allow improvement in one area, it has to be taken from the total 'pool' of attributes available in how players progress in game. The game decides that a player improves his speed (which is influenced by the sort of factors you talk about), and his ability rises to match. It works that way round. Or the game determines the player doesn't have the potential to improve and they don't get quicker.

The only time people run into 'pool of attribute' problems is if they are trying to do illogical things in the editor like make a player faster and better on his weaker foot without changing the hidden value that says how good he is overall. 

Link to post
Share on other sites

6 hours ago, enigmatic said:

Fine, let's skip over the fact you keep saying things like  a player might score 30 goals at age 22, reach peak CA, and then score 8 goals in each of the next 5 seasons and his CA won't take much of a hit, if any at all.   then, because it's 'toxic' of me to assume you were maintaining a consistent argument, and maybe you don't think players should take a CA hit for not scoring goals in the ME now. Great, perhaps we'll end up agreeing :) 

Do you not see that in the highlighted part I'm pretty much saying that the same set of attributes allows a player to score 30 and to score 5? Please understand that the "not scoring goals in current season" is tied to "scoring many goals in previous season" - I am not saying that a player should take a hit to CA if he doesn't score much, I'm saying that if his performance drops relative to previous season, his CA should take a hit (and vice-versa, overperforming relative to previous season should see a CA gain). CA is Current Ability (yes, I know it is a weighted sum of attributes), so why should a player who previously scored 30, but now is scoring 5, have the same Current ability, assuming nothing else changed? Also, the toxicity was calling me a troll and stupid, not assuming I was maintaining a consistent argument.

Quote

a player's performance in the ME is therefore categorically not evidence the wrong values have been set in the database. 

I never said it was. What I said was that how a team/player performs in the ME is not a major driver for attribute changes in subsequent FM's, which I think santy001 confirmed in one of his posts here.

Quote

Are we in agreement yet?

For some things yes, for some no. The point isn't that player development should mimic the subjective changes made by researchers, it's that it should be performance based. The gist of it is that good performances should accelerate development, average performances keep it at some constant pace, and bad performances negatively impact development. The way development is currently set-up is that the quality of performing doesn't rally matter, as long as a player is getting match time in a good league, they'll generally develop by the age 23/24.

Link to post
Share on other sites

57 minutes ago, goranm said:

 I am not saying that a player should take a hit to CA if he doesn't score much, I'm saying that if his performance drops relative to previous season, his CA should take a hit (and vice-versa, overperforming relative to previous season should see a CA gain). CA is Current Ability (yes, I know it is a weighted sum of attributes), so why should a player who previously scored 30, but now is scoring 5, have the same Current ability, assuming nothing else changed?

Because as several people including me have pointed out, his Current Ability is what made him score those current goals or miss those current chances, so it's clearly at exactly the right level for those performances. And indeed the most common non tactical reason for a player playing differently one season from another is that attribute changes which already happened cause performance changes. 

For the avoidance of doubt, I'm not saying performances should never affect any attribute at all. Existing mechanisms built into the game ensure that players who get game time at a suitable level they weren't getting before will tend to develop and players that get frozen out will stagnate or go backwards, more so if they have weak mentalities.  It's probably realistic to assume that players [occasionally] improve certain mental attributes simply from confidence boosts of past performances, but generally they play better because they improve, not the other way round. That's how the real world works, and its even more the case when player performance is literally the output of a bunch of ability numbers.

 

57 minutes ago, goranm said:

What I said was that how a team/player performs in the ME is not a major driver for attribute changes in subsequent FM's, which I think santy001 confirmed in one of his posts here.

What happened is that I made a generic and obvious statement that players don't get attribute changes just because they have good seasons, they get attribute changes if and when its evidence they're capable of performing in excess of their FM ability.

For some reason this triggered you to repeatedly insist that I was wrong because researchers "don't have access to the game for which they are doing the research", which is going to come as a particular shock to those working on the mid season database update at the moment. Apparently even a researcher confirming he plays the game, considers in game impact in his research and has even been directly asked to made specific changes to a player solely due to an individual match engine quirk won't convince you otherwise, so I'll leave you claim your definitely-not-trolling Internet victory if it makes you happy.

 

1 hour ago, goranm said:

 The point isn't that player development should mimic the subjective changes made by researchers, it's that it should be performance based.

But attributes cause performances. So it doesn't make any sense for performances to also be the main cause of attributes.

Since you seem particularly hellbent on not believing me, try believing one of the other people who have made this point in this thread. As I've said more bluntly above, I'm finding it hard to believe that cause and effect is genuinely too complex for you to grasp. 

It's one thing to conclude that the balance of player development isn't quite right which is a pretty common opinion sometimes even acknowledged by SI, quite another to keep insisting the real problem is that the effect isn't also the cause...

 

P.s. actually impressed that Man Utd v Wolves was duller than this back and forth :D

Link to post
Share on other sites

3 hours ago, enigmatic said:

Because as several people including me have pointed out, his Current Ability is what made him score those current goals or miss those current chances, so it's clearly at exactly the right level for those performances.

Well, no, it's not clearly at the right level since scoring 30 goals and scoring 5 goals are not the same levels of performance, they are opposite extremes, hence factors other than CA had a far greater influence on those stats than CA did. The first season might have been a freak season (or the second one might have been a freak season), but the game will not recognise that since it wont take into consideration that random factors had more effect than the attributes.

3 hours ago, enigmatic said:

For some reason this triggered you to repeatedly insist that I was wrong because researchers "don't have access to the game for which they are doing the research", which is going to come as a particular shock to those working on the mid season database update at the moment.

I wasn't triggered. If anyone was triggered it was you, since you felt the need to call me a troll and stupid without ever apologising, and if anyone was insisting that they were right, it was you - if I remember correctly, and I do, you were the one who @ a researcher to confirm what you were saying.

3 hours ago, enigmatic said:

Apparently even a researcher confirming he plays the game, considers in game impact in his research and has even been directly asked to made specific changes to a player solely due to an individual match engine quirk won't convince you otherwise, so I'll leave you claim your definitely-not-trolling Internet victory if it makes you happy.

I know it's hard for you to admit your description of how research works was not correct and that you'd want to spin it your way, so I'll just quote what the researcher said: "How players perform in the match engine isn't a huge part of it on the front end, but there is always a consideration." Which is what I was saying all along, I can even quote what I wrote before the researcher responded: "What I've also said is that the predominant driver of attribute changes is IRL performance, and not corrections to ME." I'm literally saying that the ME "isn't a huge part of it", without dismissing that in some circumstances it will be taken into consideration. You don't have to spend a second in the Data Issues forum to figure out that, you just have to play the game.

3 hours ago, enigmatic said:

But attributes cause performances. So it doesn't make any sense for performances to also be the main cause of attributes.

Attributes are not the only thing that cause performances. The whole point of the game is to make 11 players work together according to some instructions. Sometimes players are played in different position which causes them to use different skills that they previously didn't, so they'll lose the sharp edge on some skills and gain a better understanding of some other skills. It makes complete sense for other circumstances to affect attributes.

3 hours ago, enigmatic said:

Since you seem particularly hellbent on not believing me, try believing one of the other people who have made this point in this thread. As I've said more bluntly above, I'm finding it hard to believe that cause and effect is genuinely too complex for you to grasp. It's one thing to conclude that the balance of player development isn't quite right which is a pretty common opinion sometimes even acknowledged by SI, quite another to keep insisting the real problem is that the effect isn't also the cause

I'm not hellbent on not believing you, you keep misinterpreting and misunderstanding what I'm saying and I'm making a wasted effort in clarifying since in the end you just keep calling me a stupid troll. Jervaj got what I was saying on first try. There's nothing wrong with effect feeding back into the cause, that's the basis of all recursive frameworks like reinforced machine learning and essentially the foundational principle behind DeepMind.

Edited by goranm
Link to post
Share on other sites

On 27/12/2020 at 14:08, SeaCarrot said:

Quite simply the game as a whole is too rigid. It never changes from the first year to 2200. The game world is static. 
 

it lacks so many things that make the OOTP baseball games amazing. Dynamic changeable potential is one. If a player has a ripper season they can improve their potential, a bad disappointing season and they can reduce their potential. Means the game world is alive. The same person in a different save game can be a flop or a new world beater, who knows. Different every time. The world changes, new rules or old rules changed every few years, teams fold, new teams get created and leagues expanded. It’s a real dynamic simulation of the world of baseball into the future.

FM is the exact same game every year of your save besides seeing different teams rise and fall now and then. Dull. 

I get this but if the SAME player was different in every save game, how would this be realistic? It would actually be unrealistic sooooo.

 

Edited by Preveza
Link to post
Share on other sites

On 28/12/2020 at 12:08, Freakiie said:

Now as for the whole "a player playing well should develop", why? Should Messi be improved because he plays well? Or does he play well, simply because he's the best footballer around? Also, lets say you change PA so that players playing well can always improve, then they improve, play even better the season after, should they then improve again? In the end, PA does what it's supposed to do; it indicates the peak a player can reach. Players having a peak is not the flaw in FM in my eyes and if people want to get into an argument about a peak or PA not existing in real life, well I disagree. Talent is a real thing and plenty of players branded as massive talents never got anywhere, as their talent was nothing more than them having developed quicker than the people around them, but then quickly hit their ceiling when they actually become a pro.

 

Yes Players playing well should see both an increase in PA and maybe a lesser increase in CA. It should be proportional to the players CA in comparison to the competition hes playing in and relative to the achievement eg. If a below average player in the premier league ends the year with an high 7 or 8 average then his increase should be a lot better than a star player achieving the same feat the star player should get a minimal boost maybe even need to achieve a concoction of accolades to just increase by 1 PA.

Link to post
Share on other sites

It's has been said lots of times, but I will say it one more time, PA is the best quality that a player could reach in his life, if he reach his PA and he continues with good ratings, why should he gain more PA? It simply means that at his better quality of his life he is playing well, but that not means that he has to improve. PA is a concept that by his own meaning can't change during the lifetime of the players during a savegame. 

However, in different savegames I think that PA should be more variable, because researchers obviously are not vident and they can't know the exact quality that players are going to reach. Because of that, I use a database that randomise the PA of all players sub23, (searching in FM famous webs you will find it) but it has no relation with the concept of the static PA, because PA has to be static in each save when it starts! 

Edited by pejocho
Link to post
Share on other sites

After reading through this thread and reading the things Seb has said in separate threads it is clear to me that there are no major issues with player development.

Also the claim that players stop developing at 23/24 is simply not true, it depends on the player, however as is true in real life most players will have fulfilled most of their potential by 24/25.

PA being static is fine, but how and when and if players reach their PA should be variable, and this is how it is in game. Think of PA as the maximum physical and mental capabilities of a person. E.g. I could never jump as high as Ronaldo as Im not tall enough, I couldn't spot a pass as well as Fabregas as our brains are wired differently, I couldn't kick a ball as far as Buffon as my leg muscles wont be as strong (due to the chemical makeup of my body). Maxing out your capabilities will depend on training, morale, tactics and personality.

Solely playing well shouldn't lead to a bump in CA or PA as CA, tactics and morale lead to the player playing well, so making the player even better 'just because'  makes no sense. CA increases should only be a part their development towards their PA and form plays a part in that anyway. The only thing that should possibly lead to an increase in PA is a personality change and this should be very rare anyway.

 

Edited by francis#17
Link to post
Share on other sites

Think of PA as the maximum physical and mental capabilities of a person. E.g. I could never jump as high as Ronaldo as Im not tall enough, I couldn't spot a pass as well as Fabregas as our brains are wired differently, I couldn't kick a ball as far as Buffon as my leg muscles wont be as strong (due to the chemical makeup of my body). Maxing out your capabilities will depend on training, morale, tactics and personality.

Link to post
Share on other sites

16 hours ago, goranm said:

Well, no, it's not clearly at the right level since scoring 30 goals and scoring 5 goals are not the same levels of performance, they are opposite extremes, hence factors other than CA had a far greater influence on those stats than CA did. .

Sure, the inputs that produced the outputs are clearly not at the right level to have produced those outputs :rolleyes:

Yes, other factors affect match outcomes: the CAs of teammates and opponents, tactics and tactical changes, fitness fatigue and morale on match day, referees, weather, home and away status and luck. Even altitude is simulated, I believe. None of these sound like important factors for player development currently missing from SI's model

(I mean, opponent ability is - broadly - taken into account inasmuch as getting more game time at suitable higher levels helps improve players anyway. Revising it so players benefited more from the sort of outcomes they'd get playing easier opponents certainly wouldn't be a step forward.)

 

17 hours ago, goranm said:

I know it's hard for you to admit your description of how research works was not correct...

Well if you're back for the victory lap - because you're really not trolling - you could always entertain yourself by listing each individual statement I made which is incompatible with santy's post. Happy hunting  :p 

I had hoped that the detail he volunteered might help you (and others convinced that Niang is a transfer to PSG away from being considered a smart player!) understand that there's a bit more subtlety to attribute bumps than 'because he had great seasons time and time again' style rewards, but never mind. 

 

16 hours ago, goranm said:

There's nothing wrong with effect feeding back into the cause, that's the basis of all recursive frameworks like reinforced machine learning and essentially the foundational principle behind DeepMind.

Except it's not an ML model updating an estimate to converge on a true (or optimal) value. It's exactly the opposite: the starting point is the true value which caused the outputs, and so all your process finds is the other variables unconnected with that particular player's [at the time] current ability. Most of which are, from the point of view of player development, pure noise.

More practically: the two most loathed characteristics of SI's development model are the super rapid developing wonderkid and the super rapid decline, as mentioned several times in this thread.

This is the exactly the sort of outcome attribute improvement based on match performance is most likely to produce (both are probably influenced by less explicit positive feedback loops in SI's existing model). Player signed for relegation fodder is unable to put in effective performances, therefore the model concludes his attributes are too high, which makes his performances (and the teams') less effective still etc, which causes to make the attributes fall more etc. Poor old West Brom would start the season with players only good enough for the Championship and end them with players only good enough for League One.  

(Fortunately, as I've tried in vain to explain, the actual research process doesn't work like this and takes into account the fact West Brom are already not good enough in FM and Liverpool are already the highest rated team in the Premier League.  Player development shouldn't work like that either.)

Link to post
Share on other sites

4 minutes ago, enigmatic said:

Well if you're back for the victory lap - because you're really not trolling

 

You keep replying as well, as well as calling me stupid, but you're not a troll and I am? You are doing the same things that you're accusing me for. But I am the troll here.

14 minutes ago, enigmatic said:

Yes, other factors affect match outcomes: the CAs of teammates and opponents, tactics and tactical changes, fitness fatigue and morale on match day, referees, weather, home and away status and luck. Even altitude is simulated, I believe. None of these sound like important factors for player development currently missing from SI's model

So let me get this right - your opinion is that the quality of teammates, the quality of opponents, and in-match experience (as in carrying out the tactical instructions and adapting to changes) do not sound as important factors for player development?

18 minutes ago, enigmatic said:

(I mean, opponent ability is - broadly - taken into account inasmuch as getting more game time at suitable higher levels helps improve players anyway. Revising it so players benefited more from the sort of outcomes they'd get playing easier opponents certainly wouldn't be a step forward.)

Obviously performances in lower levels wouldn't have as much impact as in higher levels.

7 minutes ago, enigmatic said:

Except it's not an ML model updating an estimate to converge on a true (or optimal) value.

I never said it was. I said that there's nothing wrong with outputs being fed back into the input, since you've said that it's a "quite another [thing] to keep insisting the real problem is that the effect isn't also the cause". There's nothing "quite another" about it.

Link to post
Share on other sites

1 hour ago, goranm said:

So let me get this right - your opinion is that the quality of teammates, the quality of opponents, and in-match experience (as in carrying out the tactical instructions and adapting to changes) do not sound as important factors for player development?

What I actually said was that they are important factors for player development not incorporated into SI's model. Not sure how that bit eluded you bearing in mind you even quote my next post about how opponent quality is - loosely - factored into SI's development model.

 

But no, I don't think the relationship between those factors and the in-match performance of a player of a given ability remotely resembles the relationship between those factors and their capacity to progress/regress

Players might well achieve dramatically higher pass completion from being instructed to play simple passes, but it would be absurd for this to increase the attribute which improves their range of passing. Players might win a higher percentage of headers against short opponents, but clearly it would be absurd for playing mostly against midgets to improve the attribute which makes them better at soaring high into the air to head the ball away. Having teammates that can't hit a barn door with a banjo will reduce your assists, but is unlikely to actually impair your crossing ability.

Fortunately, we have development boosts for match experience at a suitable higher level, the ability to learn skills and become more mentally resilient from teammates and role-specific training to incorporate the effect of tactics on attribute distribution anyway, so we don't need to clumsily kludge side effects 

 

1 hour ago, goranm said:

I never said it was. I said that there's nothing wrong with outputs being fed back into the input, since you've said that it's a "quite another [thing] to keep insisting the real problem is that the effect isn't also the cause". There's nothing "quite another" about it.

There's plenty wrong with feeding the outputs back into the input to drive player progression in FM, as already explained.

Edited by enigmatic
Link to post
Share on other sites

2 minutes ago, enigmatic said:

What I actually said was that they are important factors for player development not incorporated into SI's model. Not sure how that bit eluded you bearing in mind you even quote my next post about how opponent quality is - loosely - factored into SI's development model.

I know what you said, I'm asking if that is what your opinion is. I'm not sure how the question mark "?" indicating a question, eluded you.

9 minutes ago, enigmatic said:

Players might well achieve dramatically higher pass completion from being instructed to play simple passes, but it would be absurd for this to increase the attribute which improves their range of passing. Players might win a higher percentage of headers against short opponents, but clearly it would be absurd for playing mostly against midgets to improve the attribute which makes them better at soaring high into the air to head the ball away. Players might well score more goals if their teammates are really good at putting them through on goal. Having teammates that can't hit a barn door with a banjo will reduce your assists, but is unlikely to actually impair your crossing ability.

Clearly the game already tracks key passes, pass directness, risk of completing a difficult pass etc. so why would only short simple passes have to be accounted for? Completing a short simple pass would carry much less weight than completing a difficult pass. Similarly, the game tracks key headers, player height/jumping reach etc. so why should headers only against short opponents be accounted for? The game already works in this way - players completing more difficult tasks such as key passes, headers, chance creation etc. are rewarded with a much better rating than those playing it safe. The only thing absurd here is that it seems like you're not playing the game at all.

21 minutes ago, enigmatic said:

Fortunately, we have development boosts for match experience at a suitable higher level, the ability to learn skills and become more mentally resilient from teammates and role-specific training to incorporate the effect of tactics on attribute distribution anyway, so we don't need to clumsily kludge side effects 

And yet the end result is an inflexible and predictable development system past ages 23/24.

5 minutes ago, enigmatic said:

There's plenty wrong with feeding the outputs back into the input to drive player progression in FM, as already explained.

Again, there's nothing wrong with reinforced recursive systems. Some state-of-the-art AI works on that principle. FM is not special, there's nothing in FM that would prevent incorporating a well thought out reinforcement system.

Link to post
Share on other sites

3 hours ago, Overmars said:

I was curious, so I decided to spoil the fog of war effect and track CA over 3-4 months in FM for one team. The team was in the second tier in Scotland and had 22 players receiving similar match time and training and no serious injuries (the two long-term injuries were removed from the analysis). The changes in the team's CA over this time span were:

  • 18-19 year olds (includes a goalkeeper who turned 20 during the experiment):
    • 12 players
    • 36 total CA point increase
    • 8 of 12 players increased
    • 4 of 12 players did not change
    • 0 of 12 players decreased
    • Average increase = + 3.0 CA points
  • 20-23 year olds:
    • 7 players
    • 0 total CA point increase
    • 0 of 7 players increased
    • 7 of 7 players did not change
    • 0 of 7 players decreased
    • Average increase = 0.0 CA points
  • 24+ year olds:
    • 3 players
    • -2 total CA point increase
    • 0 players increased
    • 1 player did not change
    • 2 players decreased
    • Average increase = - 0.7 CA points

To me, that's too skewed in favour of teenagers. If the argument is that a 20 year old needs to be at a bigger club in order to improve then that's really limiting from a gameplay perspective. I remember being able to develop players as old as 25-27 in older FM versions, but that's gone now.

This is good for a start but we can't draw any conclusion from this. A much bigger sample size (more teams) and a longer test time (more than 3-4months)  is needed before we can draw any conclusions.

The team you chose could have, by chance, had all 20+ players already at or very close to their PA, or maybe CA increases, on average, take slightly longer than for teenagers (which makes sense), or maybe lots of the players just had a CA increase before you started the experiment. Lots hasnt been accounted for and a bigger sample size over a longer period of time will help to account for these things.

Link to post
Share on other sites

1 hour ago, francis#17 said:

This is good for a start but we can't draw any conclusion from this. A much bigger sample size (more teams) and a longer test time (more than 3-4months)  is needed before we can draw any conclusions.

The team you chose could have, by chance, had all 20+ players already at or very close to their PA, or maybe CA increases, on average, take slightly longer than for teenagers (which makes sense), or maybe lots of the players just had a CA increase before you started the experiment. Lots hasnt been accounted for and a bigger sample size over a longer period of time will help to account for these things.

Yes, a bigger sample would be good. The problem is that it's really tough to control for all of the variables in a bigger sample like I can when managing one club myself. Also, it's a chore to track CA changes on a large scale.

I can confirm that both the teenager group and the 20-23 year olds in my experiment all had room to grow with their PA exceeding their CA by a healthy margin. The 20+ year olds just didn't tap into that potential at all.

Link to post
Share on other sites

I had a decent regen striker at Dulwitch Hamlet in fm19. I loaded up the save at age 16, 22 and 32. I felt that he became better every single year, or kept pace with the promotions. He even managed to get a green circle at Complete Forward for a single season around age 29, but then lost it again (the green circle suddenly became fuly green for a season, then complete black/red again). I took picture of his stats. Did he really peak at 23 like you guys are suggesting?

Here is the earlies save I found with him in the squad a few months after he came to the club:
OHZ7RW5.png

 

Here is his stats at age 22:
OpPPSwP.png

EDIT: I found a picture of when he learned to play as a Complete Forward at his peek performance:
cpYHzBC.png

And here he is at age 32: (he was 2.5 star rated before I sold him for 5.5M) the attributtes was identical at age 31, but didnt want to post too many picures.
spPyOZc.png

He almost played 500 matches for the club, Mainly as Advanced forward, except for the single season around age 29 when he was able to be a complete forward.
dmlDA0Q.png

Edited by Speedyol
Added the peek performance image
Link to post
Share on other sites

On 27/12/2020 at 15:44, RocheBag said:

PES has the best development system of any sports game. FIFA's dynamic potential is ok but it still falls short of PES.

In PES, players have no potential rating. Every player gets assigned a "soft peak" and a "hard peak" age. Players will develop with game time, rapidly until their soft peak, the  slower until their hard peak.

Form and playing time impact development a lot. If a player plays often and plays well, he will develop fast. If he doesn't play much he will develop much slower. If they hit a purple patch of incredible form their development is boosted massively for a short time. Even players past their hard peak can develop from this form related development.

This leads to a game where every file a player can end up widely different because their development is so dependent on how they play rather than a predetermined number.

So naturally if you have a 21 year old player and an 18 year old player with the same current rating, you can assume the 18 year old will probably end up better because he has more time to develop. So you can use your common sense to figure out a rough potential of players. But it can vary so much with gametime and form that it leads to a very dynamic world.

As an avid PES fan I agree to some extent.

But the issue is match performance rating systems are broken. Keep a clean sheet and defend well - 2 CBs and Gk are often 6 and below cos PES has an illogical player rating system in games. Players getting 4 assits in 1 game - gets a 7. Madness.

On topic - I think a players mentals and personality are pivotal to development. As is coaches, facilities, game time, level of opposition....probably missing others out but all of those have to be at their best for a "wonderkid" to reach the 190pa level from say - 160pa. 

Link to post
Share on other sites

I think this is an interesting topic to discuss because the player development has a great impact to gameplay and immersion. I would like to discuss with all of you, using also examples related to football IRL.

First of all I want to make clear my opinion about the current PA / CA system:

- PA exists IRL and I think it is implemented correctly in the game, because everyone of us born with a certain DNA and through the first years of life he start to show and develop specific skills and talent, since the players and newgen in the game are already 15-16 yo, they already went through childhood and develop somehow coordination, technique and so on, besides their current attributes (and so their CA), they have already a certain potential. I have been playing football since I was 6 yo, immediately it was clear that I wasn't a new Pelé or Maradona, when I was 15 yo I knew for sure I could not be a professional, at that age there is a certain potential for physical, technical and mental attributes;

- CA is defined by the mix of physical, technical and mental attributes related to different positions on the pitch: I think also this is correct already and IMHO it is already fine tuned, it works well.

The real issue is the way physical, technical and mental attributes are changing across the years in the game. There is a pattern of player developing fast from 15-16 to 23-24 yo and then almost nothing happening until 30-31 when there is a decay of physical attributes more or less dramatic. This kind of player development is making almost all of us playing with a certain pattern: build up a team of talented youngsters and wait season after season until they reach their full potential. I think this is the area which need a lot of improvement and I think we don't need to wait for FM22 or FM23, it can be fixed with a patch, of course it requires fine tuning and testing.

NOT ALL PLAYERS DEVELOP IN THE SAME WAY AND FOLLOWING THE SAME PATTERN AND THE "ENVIRONMENT" USUALLY HAS HUGE IMPACT ON PLAYER DEVELOPMENT: I strongly believe that this is the starting point if we want to improve the player development in the game. Now I start to describe how it works IRL and how should also work in the game, IMHO.

Lets consider a youngster, 15-16 yo in our team, he has certain traits:

  • personality

  • CA ( physical, technical and mental attributes)

  • PA (it can be fixed or can be included in a certain range as -8, -9 and so on)

IRL the player will be affected by the environment (team reputation, league reputation, number of apps in first team or B team of youth teams, skills and personalities of team mates, ability of coaches and training facilities) and I think this is taken into account by the game and is implemented quite well, however I think there is room for improvement.

Physical, technical and mental attributes should improve with training, however for 70-80% of players the growth of mental attributes should not be that dramatic until 23-24, there should be room for improvement in this area going from 24 to 34.

It would not be realistic to have players becoming more fast or strong from 30 to 32 for instance, but there can be an improvement in mental attributes. Technical attributes should not vary dramatically, however the right training should put some +1 or +2 in certain areas.

The personality of player could be set up to change not only because of “environment” but also because of the age, as almost all of us become more wise and mature when becoming older.

The improvement of mental attributes in conjunction with improvement of certain hidden attributes such as consistency, important matches etc. should be more dramatic from 24-25 yo, this will add a lot in terms of gameplay: it will make sense to buy and sign players which are already 27-28 yo or even older than 30 yo, because they will still have the possibility to improve as players and to bring positive attitude to the team. IRL mental attributes and personality have huge impact during the match, furthermore personality have impact also outside the pitch. This latter aspect is already present in the game, but it can be made more relevant.

Now I would like to bring to you the example of some players and how they developed IRL.

  • Leo Messi: he had a development really “FMish”, since he reached his full potential already when he was 22-23yo and kept doing great until now, he is 33 and since a couple of years we noticed is playing different as he is not fast and explosive as he was in his prime (of course he still a top player);

  • Cristiano Ronaldo started his career as winger and with the time he lost some of acceleration and pace (he's great but not as 4-5 years ago) but improved enormously some mental skills such as composure and movement and became a n outstanding forward. Here we see another aspect: the ability when already 28-29yo to learn a new role and make it become the natural role;

  • Another example similar to Cristiano Ronaldo is Paolo Maldini: transition from left back to center back in the last years of his career and he kept a status of top player, winning two Champions League;

  • Zlatan Ibrahimovic had a development similar to development of Cristiano Ronaldo, he didn't change the position but role and duties, furthermore he had a change in mentality and his “positive leadership” made him really important for AC Milan nowadays;

  • Ronaldo Nazario da Lima was a fast and explosive player in his first years, he made a change after, not because of age but because of the injuries he incurred: he became more “a striker” and won Champion League and World Cup with outstanding performances;

  • Luca Toni had a slow development, he didn't play in top teams until he joined Bayern Munich when he was already 30 yo. The same with national team, he has never been in it until he was 27 yo...and when 29 yo won the World Cup being really important for the team;

  • Antonio Cassano was a diamond, due to certain traits of his personality he had never reach his full potential (the potential of a top player), however the magic was still there and when the conditions (environment) where the right ones he performed exceptionally well after 25yo, in that certain moment the “CA” was the one of a top player.

What I would love to see in the game is some more variety in the way different players develop, in conjunction with improvement of mental attributes being more dramatic from 25 yo (not for everyone, but lets say for 80% of players) .


 

Edited by Delvi
Link to post
Share on other sites

2 hours ago, Speedyol said:

I had a decent regen striker at Dulwitch Hamlet in fm19. I loaded up the save at age 16, 22 and 32. I felt that he became better every single year, or kept pace with the promotions. He even managed to get a green circle at Complete Forward for a single season around age 29, but then lost it again (the green circle suddenly became fuly green for a season, then complete black/red again). I took picture of his stats. Did he really peak at 23 like you guys are suggesting?

Here is the earlies save I found with him in the squad a few months after he came to the club:
OHZ7RW5.png

 

Here is his stats at age 22:
OpPPSwP.png

And here he is at age 32: (he was 2.5 star rated before I sold him for 5.5M) the attributtes was identical at age 31, but didnt want to post too many picures.
spPyOZc.png

He almost played 500 matches for the club, Mainly as Advanced forward, except for the single season around age 29 when he was able to be a complete forward.
dmlDA0Q.png

That's for FM19. I recall being able to develop players into their mid and sometimes even late 20s in FM15. The situation changed in FM20 from what I can tell. Now it seems you need to have a high reputation league to eek anything out of 20+ year olds, and even then, you're not going to get the same development you did in older FMs.

Link to post
Share on other sites

19 minutes ago, Speedyol said:

@Overmars Sorry, I only read the OP and I thought he included fm19 into this by the statement "This FM version is after 3 versions finally one that is fun to me, but there is something almost game breaking for me. ( not just in this version, this was in last couple of years)"

I think it's relevant to discuss other versions to see how this has evolved. I'm currently testing the development in FM20 since I am now back to playing that version until the next update to FM21.

Link to post
Share on other sites

1 hour ago, goranm said:

why would only short simple passes have to be accounted for?

huh, where did I say that? ah I didn't say that

What I did say is tactical instructions to play simple passes would tend to dramatically increase pass completion a player with a particular Passing attribute achieves in the ME. An example of the effect of tactics on ME outcomes, which you were keen to solicit my opinions on

My opinion is that a player completing significantly more of his passes in the ME because of a tactic is not an indication that his passing attribute should change. 

Alternatively the development model could be designed so this ME statistic could cause Passing to go up or down. But making Passing go up in this instance is clearly silly; the only reason his pass completion is good is because fewer of the passes tested his passing range. If anything, persistent low risk passing tactics are likely to attenuate passing ability, but it'd be even sillier to have a mechanism which causes Passing to decline due to pass completion statistics increasing.

I don't think throwing in other metrics like key passes and other factors like opposition pressing and whether the intended pass recipients were too slow or stupid to get the ball makes it any easier to conclude that the Passing attribute the player started the match with should change 

So I'm pretty sure SI are better off continuing to replicate the effect of managerial preferences on a player's passing ability through training role templates than trying to divine it from the noise of the match engine output.

 

17 minutes ago, goranm said:

 so why should headers only against short opponents be accounted for? .

huh, where did I say that? ah I didn't say that,

Now what I did say is that playing against [more] short opponents would result in winning a higher percentage of headers.  It's that 'opponent quality' factor we were talking about, with the relevant quality in this case being the Jumping attribute.

So the opponents being short is certainly reflected in the headers won percentage (and probably total headers, key headers, headed goals and match ratings) but it's just noise which tells us absolutely nothing about it helping or hindering a player's progression with respect to heading. Mostly it just tells us their opponents weren't great in the air.

Personally, I'd rather SI stuck to more sensible assumptions of assuming that Jumping would change largely in response to height, physical conditioning and dedicated training.

 

1 hour ago, goranm said:

The game already works in this way - players completing more difficult tasks such as key passes, headers, chance creation etc. are rewarded with a much better rating than those playing it safe

ah yes, the famously accurate rating system which absolutely nobody thinks already has too much in-game effect.

IIRC the original discussion was whether we could get evidence for how individual attributes like Bastoni's marking should increase or decrease from the ME. And I believe you were also keen to see Ronaldo-esque late career jumps in long shot accuracy?  how do you get either of those from average rating 7.32? 

 

45 minutes ago, goranm said:

FM is not special, there's nothing in FM that would prevent incorporating a well thought out reinforcement system.

I think the "well thought out" bit is the problem. :lol:

If the end goal is to better represent the factors which cause a real world professional footballer to mature over time, and the proposal is to evaluate the relationship between inputs and outputs of a series of instantiations of the SI match engine using a reinforcement learning process, any sane AI engineer is going to run a mile. 

Link to post
Share on other sites

2 hours ago, Delvi said:

What I would love to see in the game is some more variety in the way different players develop, in conjunction with improvement of mental attributes being more dramatic from 25 yo (not for everyone, but lets say for 80% of players) .


 

This is basically what I would like to see too :thup: Great player examples btw.

It just gets a bit boring trying to build up a good team (especially once you're competing for the Champions League etc) because you know exactly what sort of traits/attributes to target, and at what age.

Maybe I should just get one of those "fog of war" skins that removes attributes and go purely by recruitment analyst data to make it more interesting for myself.

Edited by ajt
Link to post
Share on other sites

 

6 hours ago, Overmars said:

Yes, a bigger sample would be good. The problem is that it's really tough to control for all of the variables in a bigger sample like I can when managing one club myself. Also, it's a chore to track CA changes on a large scale.

I can confirm that both the teenager group and the 20-23 year olds in my experiment all had room to grow with their PA exceeding their CA by a healthy margin. The 20+ year olds just didn't tap into that potential at all.

A larger sample would actually mean that we would be able to see how players CA changes on average for all age groups in the game (instead of only testing how well you can develop players), while taking into account all these game world variables that will affect development. When the sample size is 1 then we have no idea if you were just unlucky because of the game world variables.

Anyway as you say its a chore for one person to do this. So far I havent had any issues developing 20+ players in low reputation leagues though so I dont think you are right about them not developing.

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...