Jump to content

Detailed experiment of youth development - Troubling results


Recommended Posts

For anyone that is interested or have been reading my footbll Legends thread - there are some interesting developments.

Players with amazing PA almost reaching their CA by the age of 20

Very good attributes - Mario Kempes has 19 Pace and 20 Acceleration - stamina and strength are still low (12/13) but what do you expect at 20

Brazil and Argentina U20 teams full of future superstars (only entered South American legends so far) and Di Stefano has joined Man Utd and has broken into the national team by 19

All these players have random rep, a PA of between -8 and -10 (six players with -10) and have been put in their original club so basically they are on a par with the regens that are coming through - seem to be developing nicely icon14.gif

Link to post
Share on other sites

  • Replies 207
  • Created
  • Last Reply

Loving the thread here

Dont know how easy it would be to track this but...

Is it possible for you to find out how many new players there are into each field? for instance in the second save, players under21 you had 6 players of 190-200PA, the next year it was 7. Were 6 of the 7 the same from the previous save or did any of the 6 move up into 21-24?

That way we can see how many new U21's enter each field each time.

Link to post
Share on other sites

I'm wondering if we need a seperate thread about the decline in physical stats. While there can be some debate on whether the current crop of players is overrated and that's why future regens are less potent, there can be little doubt that physical stats decline to a seriously poor level after 10 years or so and it's a game destroying bug in many ways.

Link to post
Share on other sites

In past versions of FM, I always had the suspicion that the problem is less to do with the overall quality of regens - they've improved massively over the years, but with how the AI handles its squad with respect to young players.

In past games, it seemed to me that at the start I'd end up rounding up most of the talented youngsters whilst the big teams in would rather spend silly money on established players (high reputation). Whilst they did buy talented young players (low reputation), they let them rot on the benches instead of giving them playing time. Then, 2-3 seasons later, as they reached their early 20s, the players would be sold on to lesser teams.

They'd start playing for the lesser teams, but it was a case of too little too late. By then they were stunted in their development, and playing lower league football would only get them so far.

Things would change one the original players started to retire en mass (about 5-6 seasons in). At this point the AI was basically forced to start playing the newer players.

But as a human manager you could buck the trend by signing and PLAYING youngsters, who would go on to develop to be great players.

My suspicion is that the real culprit here is not the youth development or regens in general, but the AI's squad sense, transfer policies and youth development policies. I can't help feeling that it's a reputation problem. You've all no doubt noted that player's reputation seems to play too much of a role throughout the game - transfers, international call ups, scout reports, ass man reports etc. To me, this is an indication that reputation is used as a "crutch" to help the AI assess how good a player is. Unfortunately it's a case of:

Most high reputation players are good, but not all good players have a high reputation.

I suspect that the AI is selecting its team based on reputation too, as well as CA, and that this means that some poor unknown kid isn't given a chance to play whilst galactico supremo is in the squad. The AI needs to be more proactive in developing youths, and not buying up highly talented youngsters just to waste away in the reserves.

As for regen wonderkids, I haven't played FM08, but there were a fair few in previous versions. It'd be interesting to see whether things don't pick up again in your experiment once the real players have all disappeared and it's a more level playing field. Plus, you need to expect some sort of statistical fluctuations, otherwise we'd be back to CM2.

Link to post
Share on other sites

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">My suspicion is that the real culprit here is not the youth development or regens in general, but the AI's squad sense, transfer policies and youth development policies. </div></BLOCKQUOTE>

Good point

Sometimes it seems the AI has low intelligence

Link to post
Share on other sites

Really great experiment. Well done!

I would always love to see some SI input into this kind of thread. I understand they are probably (and justly) fed up with the forums by now, but it would be interesting to know their opinions on the subject.

Link to post
Share on other sites

A very interesting thread, and I can't say that I have been suprised by the results either. I'm only into the second season in my game, but I can see that very few of my youth players have developed much at all so far. And my young players who are playing in the first team have only made very small improvements as well, and these have all been playing very well.

Something that would be interesting would be if somebody did the same experiment, but with an unpatched version of the game. Unless my memory serves me badly, I seem to remember that my players developed better then they are since the patch has been released, so it would be interesting if this was the case.

Link to post
Share on other sites

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

Really great experiment. Well done!

I would always love to see some SI input into this kind of thread. I understand they are probably (and justly) fed up with the forums by now, but it would be interesting to know their opinions on the subject. </div></BLOCKQUOTE>

They really shouldn't be fed up of the forum.... they are a business selling a product. This is giving them valuable information to improve their product, ignoring the wishes of people who buy your product is surely commercial suicide?

I have stopped playing the game because i think it is a waste of time, even though i still do ok it just frustrates me at the amount of bugs. i'll give it another try when the next patch comes out and if it still isn't fixed i'll not be buying any future editions!

no moaning, no whining - just express my feelings with my wallet.

Link to post
Share on other sites

SiN8 you are as ABSOLUTE LEGEND for creating this thread. Firstly I would like to complement your on the logical layout of your statistics and the chosen feilds you have measured. This proves a few things that I have come to through while plying a save game for 7 seasons

1. FM 2008 manages to keep a constant amount of potential ability within the game.

2. New regens (i.e players under 18) on average have alot lower CA than players under 18 at the start of the game.

3. This means that althrough these regens have the high PA, due to lower CA they are less likely to fully their potential. I've only browsed the statistics but it looks like non of the rgens fully their PA.

4. Due to low CA and poor mental attributes this makes the regen even less like to full thier pontential.

4. Regens with high PA (i.e 170+) seem to be generated at random clubs. Whereas at the start of a new game players with high PA are normal at a large club; which has top training facilities and staff. This means the regen does not develop as fast and also is less to be spot by a big club. Also these regens probably have very low world rep, making diffucilt for scouts to find.

5. It is well known that any player in FM does his main development at the ages of 15-23. Now due to the random nature of regens unless you are lucky; you probably won't scout or find a high PA regen until they are 17-18. If they have been at a smaller club this is 2-3 years development they have missed thus the chance if fulling their PA is gone.

6. Even with good scouts they will not give the approiate amount of stars for a high PA regen due to such a low CA.

7. Clear on average all the regns have very poor physical stats.

8. Alot of regens seem to have random attributes. I.e a regen of PA 190 has:

17- long throws

18 - aggresion

Good attributesl back, but these are stats from my promising striker. At a start of the game; the players with high PA on average have rights attributes in the area's for their position. Thus even if they fully the PA, they still ain't as good as prevoius stars.

icon_smile.gif SiN8 you are as ABSOLUTE LEGEND icon_smile.gif

Link to post
Share on other sites

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

I have stopped playing the game because i think it is a waste of time, even though i still do ok it just frustrates me at the amount of bugs. i'll give it another try when the next patch comes out and if it still isn't fixed i'll not be buying any future editions!

no moaning, no whining - just express my feelings with my wallet. </div></BLOCKQUOTE>

Me too, Marks and Spencers made a mistake several years ago in believing that their customers wouldn't shop anywhere else, so M&S took advantage of this and instead of listening to what people wanted, they just told them what they wanted, the result was that people just shopped elsewhere. I hope SI dont make the same mistake.

Link to post
Share on other sites

One small cautionary note: CA/PA is not the be all and end all, I have only played FM 05 onwards, but I have seen MANY players with low CA/PA and even low STATS playing exceptionally well for the competition they are in(for example I had a right back in the PREMIERSHIP who had poor stats and only 13's in the important ones marking/natural fitness/ pace etc... and a CA of 120ish and he averaged 7+ for 3 seasons, there are many other examples I've seen like this). I'm hoping I'm not the only person who's noticed that quite a lot of players play better than their CA and stats would suggest.

Link to post
Share on other sites

as has been previously said this is a great thread with in-depth research

id just like to say very well done for the initiative

the results have made me really quite annoyed

what is the point of playing past the fourth or so season when all the regens are not as good as the established players

im someone that really only buys young players

i rarely buy someone over 24

so if all regens arent up to standard for my arsenal team what is the point of playing

i think until this is fixed i cant be bothered to play anymore because it seems like a waste of time to me now and im disappointed with si in the fact that the didnt check that regens were being produced of a high quality

i found nothing wrong with regens in 07 and had a number of games i played into the 20s with still a high quality squad

one possible reason ive come up with for the lack of regen quality maybe that they are trying to get them to reach their peak later in their careers as in 07 it seemed that most had peaked by 22 or so

again good thread and i hope ive added new food for thought for everyone reading this

Link to post
Share on other sites

I really like the fact that reaching your potential is much rarer in FM08, I always thought it unrealistic to be able to predict with 90% certainty which player will develop in a world class player by just looking at CA/PA levels at a certain age.

More worrying though is the complete lack of wonderkids, and my main concern is the astounding drop in physical ability of players. This would implicate that finding a fast player or a player with high jumping could tear apart the league even if they have no technique to speak of. This also means that you as a human player will have a huge advantage over the AI because you'll be able to compensate the lower physical stats by increased physical training, whereas the AI will probably not respond to the low physical stats, meaning your players will end up more balanced.

That's if you presume the problem is with the AI training and not with the regens themselves.

In short: I like what SI has done, but please a bit more 'wonder'kids and a higher physical standard, then everything would be perfect for me (in this department of the game, that is)

Link to post
Share on other sites

actually eldereth there were similar problems (and threads) about fm2007 but i think it was all fixed by the second patch. hopefully si get it fixed soon as well because this is a vital part of the game that is a key reason people play. i cant really understand why it isnt checked properly beforehand but o well.

Link to post
Share on other sites

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

hye sin8. u stopped this resarch? </div></BLOCKQUOTE>

I'm away on business this week. Don't have access to my home pc. icon_frown.gif I'll update it when I get back.

Link to post
Share on other sites

some of you should stop the moaning! its a good experiment, but some of you just use this thread as another "flame the si thread".

im in 2022 and players seems fine. maybee theres a lack of wonderkids and phys stats are not the best, but come on, its perfectly fine to keep playing.

"oh im gonna stop after 4 season because there are no good youngsters" thats bull

Link to post
Share on other sites

For me this thread proves that there isn't a big bug with the regens but a difference in what the researchers rate and what the regen system does.

Maybe this should be matched up a little for the future so that the difference isn't as large.

But since the regen world is pretty balanced out i think the system looks to be working well.

Link to post
Share on other sites

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by orion318:

For me this thread proves that there isn't a big bug with the regens but a difference in what the researchers rate and what the regen system does.

Maybe this should be matched up a little for the future so that the difference isn't as large.

But since the regen world is pretty balanced out i think the system looks to be working well. </div></BLOCKQUOTE>

The researchers are looking at players IRL and coming up with CA/PA. From this, they have come up with a ratio of good/bad players. The programmers have a different idea and, intentionally or not, created a regen system that had much different ratio that the researchers.

Judge the researchers as you well, but I think they do a good job. If they say X number of players IRL have >= 170, then the programmers should make sure the regen system stays consistent to that number.

Link to post
Share on other sites

  • SI Staff

This thread is an absolutley wonderful demonstration of how users can take steps to improve the games they love instead of jumping on to a forum and whinging about every little thing. SiN8, I salute you sir.

The one thing missing from this thread is a post with that lovely blue Sports Interactive logo attached to it. If ever a thread merited a response from the powers that be, it is this one. What does SI make of this research? Are they going to take steps to improve the youth development engine in the game? Perhaps this methodology could be used in future play-testing?

Before they even start thinking about makeing FM Live available to the masses, Sports Interactive need to sit up and take notice of things like this, since a MMO game lives and dies by its long term playablility. Things like this could be the difference between a World of Warcraft style success or a Matrix Onine style failure.

Link to post
Share on other sites

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

Great thread here, really great in-depth look at the regens. If you've got any of the save games avaliable, we'd really like to take a look at these so if you could upload them to the FTP we can use these statistics to have a good look at the regens. icon14.gif </div></BLOCKQUOTE>

Am i the only one that thinks it's weird that the testers didn't try this themselves?

I know there are more important rhings to test for but after the first handful of threads questioning the regens maybe someone should have tried it?

Let's have an open beta and get rid of more of these niggles!

Link to post
Share on other sites

  • SI Staff

Just because we don't reply to every thread does not mean it is ignored. Especially if it is constructive and makes good reading icon_smile.gif

Our testers have done similar analysis on the player progression during testing as seen here and will continue to do so in the future. It is indeed good to get more opinions on the player development matters and especially opinions with factual support as well. We will be looking at both our internal analysis as well as feedback such as this thread as we continue to finetune things in the code.

Balancing such a complex part of the game as player progression is not something that can be done overnight as the development of a player is affected by a lot of variables (clubs, coaches, injuries etc.) outside of just the starting ability and potential. Naturally the best we can aim for is to try and make the future generations of players in the game reflect the starting database as closely as we can. And like I said above, we do make similar internal analysis as well over the long term player development and we will continue to finetune things.

Link to post
Share on other sites

For me one of the major underlying issues here is that the AI does not know how to develop its players. That is what is at the heart of it. Players don't reach their PA because the AI does not know how to train and play youngsters in the right way.

The AI, for example, does not use reserve team/B-team to develop players between 19-22 years old at all, which means that much of their development is stunted... I've tried talking about this for a few versions of the game...

Link to post
Share on other sites

I think the points raised both here and in the transfer system are very similar in cause/solution. As some have mentioned, the AI is not selecting these potentially excellent young players adn so they never reach their potential. In the real world a promising player is easy to spot and they get played early so they can learn, even if they make mistakes. It's similar to the transfer problem - the Ai doesn't organise its squad well enough to decide who to buy.

I've done a little programming and I don't think it would be extremely tough to program in an AI selection 'personality'. This could vary depending on the manager - so some would prefer small squads of two players per position like Mourinho, some would prefer older cheap players like Harry Redknapp, others would prefer the Spurs approach of young, local players.

By adding this feature - which defines the values the AI manager gives to players, and the number of players per position an AI manager is aiming to have in his squad, you could drastically reduce the amount of issues people raise in a number of areas of the game.

Add in a factor about the CA, PA and form of a player (I think more players should be bought and sold on poor/good form rather than their hidden stats), so that an AI player can 'see' the places in their squad that are weak and replace them with better/more promising players, then sell the lower ranked players, would stimulate more realistic trades and give value to the mid-range CA players that currently can't be given away - if the lesser teams in the division or the higher teams in the division below can see this player isn't playing (which should be another factor in the algorithm) and is better than the player currently playing in the target's position, the lower team should attempt to buy them, and the player should want to move downward for first team football more than currently.

You could also put a higher priority on AI playing players in their strongest position to stop the 'Theirry Henry at LM' problems and avoid screwing up the assignments of players in each position - the AI manager should have a definite first choice position for each player and rarely consider them at other positions when calculating its transfer needs.

Add in a factor that an Ai manager should want to blood youngsters in lower-reputation matches and at the end of games and you start to solve the development problem too. In particular lower (ie not top6) teams should have far more preference for trying out youngsters than they do now. If a good youngster comes through at a team like Derby or Bristol City, you can bet they will play them a lot more than the AI does now (although their personality can change according to how much they want to risk the youngster, like Moyes with Rooney)

I feel the match engine, player stats and character has improved but without implementing a similar level of complexity to AI managers and the way they think, the game will continue to stagnate. It amazes me such an algorithm isn't already a key part of Football Manager.

Link to post
Share on other sites

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

Just because we don't reply to every thread does not mean it is ignored. Especially if it is constructive and makes good reading icon_smile.gif

Our testers have done similar analysis on the player progression during testing as seen here and will continue to do so in the future. It is indeed good to get more opinions on the player development matters and especially opinions with factual support as well. We will be looking at both our internal analysis as well as feedback such as this thread as we continue to finetune things in the code.

Balancing such a complex part of the game as player progression is not something that can be done overnight as the development of a player is affected by a lot of variables (clubs, coaches, injuries etc.) outside of just the starting ability and potential. Naturally the best we can aim for is to try and make the future generations of players in the game reflect the starting database as closely as we can. And like I said above, we do make similar internal analysis as well over the long term player development and we will continue to finetune things. </div></BLOCKQUOTE>

with all respect, and while acknowledging that getting the balance right is difficult, this is now the fourth version of football manager. the regen system isn't working for long term games, never has worked, and has actually gotten worse since fully patched fm07. how many versions of the game will it take?

Link to post
Share on other sites

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

For me one of the major underlying issues here is that the AI does not know how to develop its players. That is what is at the heart of it. Players don't reach their PA because the AI does not know how to train and play youngsters in the right way.

The AI, for example, does not use reserve team/B-team to develop players between 19-22 years old at all, which means that much of their development is stunted... I've tried talking about this for a few versions of the game... </div></BLOCKQUOTE>

I agree, in my game Arsenal won the league with about 4 games to spare, yet they continued to play a full strength side for the remaining games even though they were still in Europe as well. They even played a stong side in the League Cup all the way until they were knocked out in the semi's. The kids never got a look-in, so its no wonder were seeing a lot of "wasted talent"

Another thing that is stupid is that a team like Arsenal or Man Utd etc will pay £25 million plus for a regen on the game who plays in midfield for example who has CA145 and PA190

yet they wont give a chance to a regen in their youth team with CA115 and PA190. Changes need to be made.

Link to post
Share on other sites

I have read all the posts. Very good job from SiN8 and more of you. Now I'll tell my opinion:

-As you have said, PA is good, and CA of regens 30 surely will be ok. But CA for youngsters aged < 21 does not look like IRL. The game is playable, but it's weird to see lots of squads with dominating old players. Similar with physical attributes, with the difference that conference teams will have lots of chances against big teams in 2030 if their players have similar physical attributes.

-I think the problem is related with players improving. In the game players don't improve quickly when:

-They reach their potential. So they don't improve at all (well, if they improve in an area, they lose in another).

-They grow older.

-They have an injury.

-They don't have good facilities + coaches.

-They don't have good hidden attributes + determination.

This is true? I honestly don't know. Statistics from SiN8 say this weird? things:

-Players could raise more more CA points aged 23 than 17 with similar conditions.

-Players could improve very little being 17, playing and with nice coaches/facilities (unless that player trained in a very bad way).

In fm2008 players improve a lot comparing to real life (I don't have said they improve quickly, I just said they improve a lot). This is why existing players quickly reach their PA (look at saved games from 2009 and 2011) and because of this regens start with low CA: to prevent all of them to fill their PA. As I have said, it may work fine with regens at 30, but makes an awesome gap between players aged 20 and 30. IRL a team of players aged 20 could beat a team of players aged 30 (I mean 30, not 21-30) with similar conditions. Think about a team of best English players born in 1977-1978 against a team of best English players born in 1987-1988. Surely first team should win, but that match would be close. In FM in 2030 first team will crush the youngsters.

Players improvement must be revamped or tweaked at least, in order to have good CA regens.

Link to post
Share on other sites

I think several people have raised a very interesting (and probably crucial) point.

To me, it seems the reason the regens have such poor CA is the fact that the AI is poor at developing young talent. Just as the AI struggles to recognize that at some point very talented, lower rep national prodigies should be playing ahead of less talented, higher rep geezers(as this experiment has pointed out), club AI rarely forgoes an aging former talent for a wonderkid unless the superstar hits the dramatic part of the decline phase.

Granted, not all clubs/nations/managers should give the same opportunities to youngsters to develop. Similar to RL, clubs/nations that are struggling or underperforming should tend to afford youngsters more opportunities to break into the first team. And maybe as someone suggested earlier in this thread, managers should have a more distinct mentality in their team selections and transfers that biases them more towards certain age preference and development styles.

Link to post
Share on other sites

I'd like to applaud Darrkespur's excellent post. I agree completely with him that the AI needs to manage their squad in a much more realistic and intelligent way. At the moment the AI managers are pretty much stupid and it's sometimes painfully obvious.

This needs to be a high priority for SI, and I'd rather see one or multiple patches devoted to this than waiting for next year's game, so we can actually give immediate feedback.

Link to post
Share on other sites

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Crabby3:

great great thread,just wondering were did you find there PA and CA? </div></BLOCKQUOTE>

Using the editor or scouting programs (like geniescout and Mini SE

Link to post
Share on other sites

Have you noticed if any big talents show up in small clubs ?!... IRL that happens but because the bigger clubs in their area usually pick them before they get famous some people think only bigger clubs generate talented players (i mean with high PA)

Link to post
Share on other sites

chaps - going off at a tangent i was hoping some one could help me, i'm illiterate when it comes to downloads ive got the 8.01 patch ready at my disposal by there's a few to choose from, i need to download it as ive experienced similar problems to other members on the dec 2009 crash dumps which link do i click and then what happens after that ... i need a step by step foolproof guide or directions as to the where abouts of one if you would be so kind ...

Link to post
Share on other sites

One thing I've always wondered about as far as player development:

whenever you take over a new club, or ask the AI to automatically assign coaches, the default is to have every coach doing every training type they are allowed. The result is that you have a bunch of coaches training each type of trianing not very well. Any half-decent human player is going to specialize their coaches, getting much more out of each training category.

Is this how the AI does it when it manages a team? It's been this way since the new training mechanism was introduced, and would seem to me to partially explain the poor player development.

Of course, this ignores the other issues, like too few high PA players and too low physical stats, but that's another story.

Link to post
Share on other sites

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

One thing I've always wondered about as far as player development:

whenever you take over a new club, or ask the AI to automatically assign coaches, the default is to have every coach doing every training type they are allowed. The result is that you have a bunch of coaches training each type of trianing not very well. Any half-decent human player is going to specialize their coaches, getting much more out of each training category.

Is this how the AI does it when it manages a team? It's been this way since the new training mechanism was introduced, and would seem to me to partially explain the poor player development.

Of course, this ignores the other issues, like too few high PA players and too low physical stats, but that's another story. </div></BLOCKQUOTE>

I agree that is a problem, although they don't normally assign them to every training type but 2 or 3, which is rubbish, it means that although you have 2 coaches doing Aerobic training, the overall schedule only has 3 stars, whereas if each coach was specialising with only one training type they can get the full 7 stars.

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...