Are Newgens Too Good?

[michaeltmurrayuk]

On 13/08/2018 at 21:41, Weston said:

Also it looks like some of these guys have been generated at the start of the game so they may have been generated as first teamers to fill in some blanks rather than as real regens, though you'd generally expect those players to be slightly older (20-25 at generation).

So is this a bug, then?

Managed to check this from my game in 2018 and it looks like Argentina and other countries get their first youth intake during 2017 so from the history screen it looks like they got generated at start when in fact they were just part of the first years intake (if you check the youth intake transfer screen they should be listed if a proper regen).

Also as far as I am aware view-only leagues are treated the same as in-active leagues with the generation of regens - if you look at their youth intakes you'll likely find they only get one or two instead of the 16 you get in playable leagues. And as they are only producing a handful of regens on average they'd have a higher PA than the active leagues (mainly due to the fact that in-active leagues don't produce the low quality regens as they'd just waste game resources).

[Weston]

@michaeltmurrayuk Wow, that's an impressive amount of data, thanks for putting that together. So why are PAs so high for the first few years of newgens, then? Because they are trying to hype up the game, because they need to replace the current crop of high-CA players and they worry the AI and/or human managers will do a worse job nurturing talent than what happens in real life so they raise the odds to compensate, or because they actually lowball PA in real-life youngsters because they want to avoid looking like they got a big "prediction" "wrong"?

Seems problematic, though, that the game hides better-than-average PA newgens in inactive leagues as a sort of compensation. Seems like this doesn't really fix anything and can just be gamed, no?

[michaeltmurrayuk]

15 hours ago, Weston said:

@michaeltmurrayuk Wow, that's an impressive amount of data, thanks for putting that together. So why are PAs so high for the first few years of newgens, then? Because they are trying to hype up the game, because they need to replace the current crop of high-CA players and they worry the AI and/or human managers will do a worse job nurturing talent than what happens in real life so they raise the odds to compensate, or because they actually lowball PA in real-life youngsters because they want to avoid looking like they got a big "prediction" "wrong"?

Seems problematic, though, that the game hides better-than-average PA newgens in inactive leagues as a sort of compensation. Seems like this doesn't really fix anything and can just be gamed, no?

It's likely a combination of factors - the active leagues you select will influence PA - if you are running the English leagues your game is more likely to get high Potential kids than a game running just the Irish leagues for example, so the database will shift over time to ensure the game has the players for the active leagues. (how much it shifts will depend on your settings on startup).

For real youngsters as far as I am aware the researchers are largely free to set what PA's they want for players so if they feel John Smith should have a PA of 200 then they can give him that PA if they want, the lower PA's are more a case of it being hard to work out how good a kid will be from the limited data they'd have access to with most of them using the negative values to give the players a range (which would see their bracket move depending on the luck of the draw) obviously the game doesn't have this problem with regens as the game gets to play god (the staff PA star ratings are probably a better comparison to the researchers as that is based on a educated guess - it would be interesting to see how the star ratings of regens line up with the starting database as these tests don't show which players reach their PA - for example my figures showed an increase in players with the 181 bracket PA but the players with that CA stayed the same, so higher PA's becomes less of an issue if the amount reaching high CA's stays the same - and makes things harder if you have 6 from 37 players reaching 181 CA rather than 6 from 24 as you cannot just sign a random high PA players and expect him to develop).

For the game generating high PA players in non-active leagues this is just a consequence of database limitations - teams in active leagues produce ~16 regens a season, whilst teams in non-active leagues produce 1 or 2 - so if you are running all the English leagues that is ~150 teams, so ~2,000 regens of which only the Premier League teams on average are going to produce Prem Rated regens so you are looking at say 300 Prem PA regens which is 15% of the English regens, and if you take the average PA it's likely around a League 2 average (due to the non-league sides outnumbering the prem sides which brings the average down) compare to say an in-active Colombia which may produce 20 regens a season these players will all come through the top teams so you have cut off the majority of the conference level players which will increase the average to say Championship level but they may only produce five Prem rated regens but that is equivalent to 25%.

So Colombia produces 25% Prem PA players against the 15% for England, but England produced 300 against the 5 from Colombia - so England produces the better regens but because Colombia produces so few it's easier to find that one with Potential, which obviously makes things easier for the human manager as it's a quick job to manually scout the handful of regens produced by these countries against sorting for the diamond in the rough against the thousands of regens in active leagues. But there isn't really much way to stop this from happening as these teams need to produce regens of a good enough quality for the national team but the game cannot produce a dozen regens each season for every team to hide them as the database size would become unmanageable especially on lower spec machines.

[Weston]

Interesting, thank you for another detailed breakdown.

This is definitely another thing to keep in mind when I start a save and decide which leagues to make active. I typically only make European leagues active because I start my career in Italy and don't really see myself realistically leaving the continent in my career, but I always make other important leagues viewable because I wan't them simulated accurately due to their importance in the global footballing ecosystem.

Am I perhaps misunderstanding what "viewable" means, then? I was under the impression that status was closer to "playable" than being completely inactive, but you seem to be saying the opposite is true. I do a lot of scouting in South America, for example, so I want those leagues to seem "real" - should I make them "playable" even if I have no intention of every managing a club there? Does setting my database to the larger settings offset these issues a considerable amount? I wonder if there is perhaps some recommended template for how to create a realistically whole global footballing ecosystem for the average Serie A manager...

 

[Nacaw]

Yes, it's fair to say there are too many players who are fully developed age 19 to 21, if we strictly compare FM to real life. 

But I dread to consider the alternative. As it is, the AI consistently struggles to develop players at a rate that is comparable to human managers. If we make this process harder/longer, it will hurt the AI more, thus widening the gap. 

The19yo youngsters from Argentina need 130+ CA before moving to Europe because otherwise, they'd simply stop developing due to lack of game time. 

[michaeltmurrayuk]

22 hours ago, Weston said:

Am I perhaps misunderstanding what "viewable" means, then? I was under the impression that status was closer to "playable" than being completely inactive, but you seem to be saying the opposite is true. I do a lot of scouting in South America, for example, so I want those leagues to seem "real" - should I make them "playable" even if I have no intention of every managing a club there? Does setting my database to the larger settings offset these issues a considerable amount? I wonder if there is perhaps some recommended template for how to create a realistically whole global footballing ecosystem for the average Serie A manager...

 

 

View-Only works the same as not selecting the league, the only difference is you get to see the league table. If you check when creating the save view-only leagues don't add any more players into your game.

If you don't want to/cannot make the leagues playable then I think using the advanced options to retain players (rather than just increasing the database size) might make the teams that have been given full squads more active than they would be with near empty squads, though @Seb Wassell or @Neil Brock might be able to provide (or pester who knows :) ) a definite answer as to whether using the advanced options has any long term affects on teams not in playable leagues.

[Guest]

Retaining players in viewable league will correctly fill squads & clubs will run active squad shortlist so they will be active in the transfer market albeit with a simplified finance module & produce full youth intake cohorts.

The downside is the additional memory & processing that adds which means you might as well make the league active as the quick ME will not add much to the processing commitment & clubs will operate with a more realistic finance module.

[Weston]

So there's no real benefit to making leagues view only beyond just simulating a table? Good to know... I guess when FM19 drops I will experiment with setting all the viewable leagues I had to playable instead and see if this slows me down a drastic amount.

[Guest]

No, hasn’t been since add/remove leagues was introduced because that feature does pretty much same job behind the scenes.

[Weston]

Thank you for the insight.

In the spirit of #experimenting with #data #analysis would you be willing to provide the numbers for your newgens like we did above? I'm really curious to see a larger sample size.

[Guest]

7 minutes ago, Weston said:

So there's no real benefit to making leagues view only beyond just simulating a table? Good to know... I guess when FM19 drops I will experiment with setting all the viewable leagues I had to playable instead and see if this slows me down a drastic amount.

Do you usually add players to the view only divisions or something similar?

[Weston]

Do you mean just by checking the box that says something along the lines of "add players and staff to leagues..."? Because I usually click that, I believe.

[Guest]

Yep that’s what i meant, chances are you’ll end up with a similar size player pool so expectation should be very little addition cost in processsing time.

Will be interesting to read your feedback.

[Weston]

Wait, I'm confused and i want to make sure I'm understanding you correctly. So selecting a lot of view-only leagues + "add players" is similar to selecting a lot of playable leagues, or not? First I'm told it is no better than leaving them inactive and now I'm told there is very little difference to if they were playable. Are you saying it's roughly the same size player pool either way the only benefit is the clubs' financial modules? Is selecting "add players" what really matters regardless?

[Guest]

9 hours ago, Weston said:

Wait, I'm confused and i want to make sure I'm understanding you correctly. So selecting a lot of view-only leagues + "add players" is similar to selecting a lot of playable leagues, or not? First I'm told it is no better than leaving them inactive and now I'm told there is very little difference to if they were playable. Are you saying it's roughly the same size player pool either way the only benefit is the clubs' financial modules? Is selecting "add players" what really matters regardless?

The first part I’ve highlighted is my expectation, it’s not something I have tested but the theory is based on knowledge of the constituent parts of the processing & memory requirements

Second highlighted section is a yes, by manually adding players through the advanced DB option the user is doing the same task as the game code if the leagues were active & do tend to end up with similar sized active player databases & it has to be added that the user is more likely to create a database imbalance due to a lack of experience with player pool creation.

 

[Guest]

9 hours ago, Weston said:

Wait, I'm confused and i want to make sure I'm understanding you correctly. So selecting a lot of view-only leagues + "add players" is similar to selecting a lot of playable leagues, or not? First I'm told it is no better than leaving them inactive and now I'm told there is very little difference to if they were playable. Are you saying it's roughly the same size player pool either way the only benefit is the clubs' financial modules? Is selecting "add players" what really matters regardless?

The first part I’ve highlighted is my expectation, it’s not something I have tested but the theory is based on knowledge of the constituent parts of the processing & memory requirements

Second highlighted section is a yes, by manually adding players through the advanced DB option the user is doing the same task as the game code if the leagues were active & do tend to end up with similar sized active player databases & it has to be added that the user is more likely to create a database imbalance due to a lack of experience with player pool creation.

 

[Weston]

Okay. So is doing that a possible explanation as to why my numbers seem so inflated over time? It seems I might as well just make the leagues playable, but then should I keep "add players" checked even then or does it become redundant? Apologies for so many questions this just seems like a crucial part of creating a realistic save so I'm surprised it's not made more clear.

[Guest]

@michaeltmurrayuk has given a solid explanation on why your setup might be a contributory factor.  My rule of thumb is if you’re not sure what affect the add players options will have on long term save integrity leave the database at the default medium setting & active leagues only.

[BMNJohn]

On 16/08/2018 at 20:32, Weston said:

So there's no real benefit to making leagues view only beyond just simulating a table? Good to know... I guess when FM19 drops I will experiment with setting all the viewable leagues I had to playable instead and see if this slows me down a drastic amount.

I don't know how right or wrong I am, but I feel that it also makes leagues in View-Only a little bit more active on the transfer market than if they're not simulated at all. They tend to have a little bit more "actual" players instead of greys/fake ones. I find it a bit more advantageous to loan players where at least the league table is somewhat simulated since they get stats generated for themselves. Also, the scouts you send in remote places don't complain as much about their assignment being useless... even if they're mostly watching fake players.

But yeah, in terms of actual realism it's not there, but I think it's better than nothing at all; especially since the processing cost is very negligible if not inexistent.

[Guest]

1 hour ago, BMNJohn said:

I don't know how right or wrong I am, but I feel that it also makes leagues in View-Only a little bit more active on the transfer market than if they're not simulated at all. They tend to have a little bit more "actual" players instead of greys/fake ones. I find it a bit more advantageous to loan players where at least the league table is somewhat simulated since they get stats generated for themselves. Also, the scouts you send in remote places don't complain as much about their assignment being useless... even if they're mostly watching fake players.

But yeah, in terms of actual realism it's not there, but I think it's better than nothing at all; especially since the processing cost is very negligible if not inexistent.

Clubs in view only leagues are no more active in the transfer market than clubs in inactive leagues as transfer activity is determined by real player (non-grey) squad size, players also do not get any advantage being loaned to a club in a view only league as player match stats are calculated in the same way as an inactive league.

[BMNJohn]

That's interesting to note. What then would be the point of there being the option to add View-Only? I do understand that it simulates a league table, but what would be the point of simulating the table? If the option's there, it may serve a purpose instead of being completely graphic. Has such a purpose been identified? As a subsidiary question, what about loading every league with the lowest level of detail possible (which amounts to "none"), like in the picture below? FM's processing  performance is already bad as it is, so that'd be nice to know before I kill my CPU.

Regens/newgens being really good is something I thought was present, but mostly because the human player is way better at developing players than the AI. I wasn't sure if how FM generates players was an issue; although I do have an issue with some parts of it that aren't completely linked to this thread and should've been reviewed by the staff already. I especially suspect it because I run empty dbs which are dbs I enjoy testing. As such, they're filled with nothing but newgens. Obviously it's a very extremist way of testing something, but I would expect a spread similar to real life. However, I think, even after reading this thread, that because the AI isn't too good at developing youth, that FM generates a lot of players with great potential knowing very well that the AI won't be able to develop them all. Which makes it all the more jarring for the human manager. But it's only a supposition based on my experience, not something I can directly prove (much like anyone here unfortunately), as I'm not part of SI's dev team. Or maybe FM overreacts by creating so many potentially good players. Any supposition is valid and food for thought in my eyes!

Sorry to hijack the thread.

[Guest]

1 hour ago, BMNJohn said:

That's interesting to note. What then would be the point of there being the option to add View-Only? I do understand that it simulates a league table, but what would be the point of simulating the table? If the option's there, it may serve a purpose instead of being completely graphic. Has such a purpose been identified? As a subsidiary question, what about loading every league with the lowest level of detail possible (which amounts to "none"), like in the picture below? FM's processing  performance is already bad as it is, so that'd be nice to know before I kill my CPU.

 

View only was a nice feature to add a small level of simulation & promotion/relegation to unplayable leagues but since the introduction of the add/remove feature the game does that by default so imo all the feature adds now is the ability to view league tables.

As for adding all leagues with default processing there is a heavy processing cost & you can test it in your PC or find a comparable setup that’s already been posted by checking out the benchmark thread.

 

[BMNJohn]

I remember benchmarking FM17. FM reacts well to increases in CPU speed. You can find my post here. I'm not sure how useful it would be to benchmark FM18, but I'll keep that in mind. EDIT: As for the screenshot I provided earlier, everything is set to "Minimal" as noted in the upper right corner.

Also, I added a few things to my previous post in the thread we're in right now; do you have any information or guesses about what I wrote?

[F J]

I think the potential newgens pruduction should be lower in first years of save game, it's a bit anoying in second year of save game a newgen with 16 years in first squad.

In Brazil research we are very criterious with youth players to not create new "FM myths" but with it great part of "real" players retired early. Maybe for the first years the save game should "estimulate real players" even it lower all clubs level.

[Seb Wassell]

This was linked elsewhere, so for reference I would just like to post the below.

CA is the key here, please focus on that rather than PA. We are concerned with the current ability of players in future gameworlds, not how many could/should/would have made it. It is also not a good comparison with the starting DB, as researchers cull PA as players age. For a true comparison you would need to take every player in the starting DB and check their PA in the version of FM in which they were first introduced. In that data set you would see a massive spike in 170-200 PAs.

CA of the starting DB vs. CA of a future DB is what interests us in this discussion.

[FlorianAlbert9]

For my taste there are too many newgen with high CA.

But i have the same feeling at starting, indeed i always change CA-PA of the 100 best players in db. 

(I set 0 player in 186-200 range, 2 players, Messi and CR7, at 180-185, then 14 players in 170-179 and 77 in 160-169.

In the original db se have 2 players at 195, 8 in 180-189, 22 in 170-179 and 62 in 160-169)

[Seb Wassell]

2 hours ago, FlorianAlbert9 said:

For my taste there are too many newgen with high CA.

But i have the same feeling at starting, indeed i always change CA-PA of the 100 best players in db. 

(I set 0 player in 186-200 range, 2 players, Messi and CR7, at 180-185, then 14 players in 170-179 and 77 in 160-169.

In the original db se have 2 players at 195, 8 in 180-189, 22 in 170-179 and 62 in 160-169)

I'd be interested in knowing why you do this. With Messi at 185 are you saying that you believe a player will come along that is substantially better than him or do you believe that no player justifies 190+?

[FlorianAlbert9]

6 ore fa, Seb Wassell ha scritto:

I'd be interested in knowing why you do this. With Messi at 185 are you saying that you believe a player will come along that is substantially better than him or do you believe that no player justifies 190+?

I think Messi is an amazing player, of course. But...

...but i don't think is a game-changer and not even a player that can overcome the weakness of a team. 

He is maybe the best ever cherry on top. That he is.

I'm very nostalgic in some way and even if I love that now (in the last 13 years circa) that are a lot of protection on attacking players, that the balls now seem super-tele like model, etc.. i missing the the great battles defenders vs strikers.

The game were use to be more brutal. So i don't think Messi would be great also at 90s for example. 

The overprotection in Europe then is also the reason Messi cannot perform at World Cup where he a shadow even in 2014 ko stages. 

Winning title is not the Key. Performance is. And Messi cannot reach the level of Baggio in 94, Ronaldo in 98 or Zidane in 06 (talking about players that Lost final and that I saw).

And those player faced much stronger opponents. (Take Brazil-Holland in 98 to see how to stop Ronaldo, dutch had to be brutal - and there were also 2 penalties not seen for him)

Indeed with the current system you can set Messi at 195 with low Important Matches (that has no cost in ca points). But i don't know how much "Important Matches" affect the game.

I think "Important Matches" should be the most decisive factor.

The range should be more greater.

And great players (175+ with current system) would be very rare, so that would be harder get them.

(I would made a post about all that in the next day)

[enigmatic]

I've heard many criticisms of Messi, but this is definitely the first time I have heard it suggested that he is "not a game changer" 

[FlorianAlbert9]

8 minuti fa, enigmatic ha scritto:

I've heard many criticisms of Messi, but this is definitely the first time I have heard it suggested that he is "not a game changer" 

I don't mean the single 'game', but the game-football itself. 

To stop Ronaldo in Italy in 97-98 or Maradona in Naples opponents need one thing: refree that protect defenders (and that were paid to make Juventus win...).

With Messi, you have refree that protect him to let him perform.

It's an huge difference.

[Weston]

Wow - how did my thread about overpowering newgens devolve into "Messi is overrated because Juve pays refs" haha, amazing.

[Weston]

On 10/02/2019 at 06:50, Seb Wassell said:

This was linked elsewhere, so for reference I would just like to post the below.

CA is the key here, please focus on that rather than PA. We are concerned with the current ability of players in future gameworlds, not how many could/should/would have made it. It is also not a good comparison with the starting DB, as researchers cull PA as players age. For a true comparison you would need to take every player in the starting DB and check their PA in the version of FM in which they were first introduced. In that data set you would see a massive spike in 170-200 PAs.

CA of the starting DB vs. CA of a future DB is what interests us in this discussion.

Still, I don't understand how the PA of the youngest players in the DB at the start of your save would be so much lower than the PA of the newgens at their introduction. If you look at the first page of this thread there is a lot of data we pulled comparing how drastically the CA of 18-20 year-old players in the starting DB is in fact lower than the CA of the future DB.

[FlorianAlbert9]

1 ora fa, Weston ha scritto:

Wow - how did my thread about overpowering newgens devolve into "Messi is overrated because Juve pays refs" haha, amazing.

Where did i say that?

[YankeeRam]

I play in Finland the regens I get coming through AC Oulu are usually one really good one, a couple really bad ones and some that will do well in the Finnish Prem and others who might be good First Division ones. One year I had three regens who are now rocking it in my First team but the next year it was ok. This recent bunch I just got had a really promising GK, have fun breaking in dude when my 17 year old is rocking it, and one dude who I looked up his PA and it was 32. 

[Weston]

I don't even necessarily mean coming through one's own academy, but just in general. Once you hit 2022-23 pretty much every single team has 1-3 very young newgens in the starting lineup at all times when at the beginning of the save it feels like you almost never see such a high percentage of players that age good enough for that much playing time.

Edited February 13, 2019 by Weston

[santy001]

5 hours ago, Weston said:

I don't even necessarily mean coming through one's own academy, but just in general. Once you hit 2022-23 pretty much every single team has 1-3 very young newgens in the starting lineup at all times when at the beginning of the save it feels like you almost never see such a high percentage of players that age good enough for that much playing time.

See there could be another issue at play here, it's very often that clubs (and in particular) fans of clubs will feel a youngster is ready, is on the cusp of first team selection. 

What is more often the case is that the managers don't have the proclivity to trust youngsters and tend to do so more when their hand is forced. While there could be an issue with regens, from what you're saying it more felt immediately to me like perhaps AI managers were too likely to give youngsters a chance.

But then could this be because perhaps the more experienced heads of around 32 or so that a lot tend to gravitate towards when they need some "experience" are still fetching too high a price? 

I can't say I've looked at it myself, but feeling there are too many young players in the starting line-up feels like a symptom of another problem given that during the early years generally speaking there are much better distributed players in the actual database than the regens available at that time. They won't have had enough time to develop unless the first year batch but presumably you aren't considering 21 to be "very young".

[FlorianAlbert9]

8 ore fa, Weston ha scritto:

I don't even necessarily mean coming through one's own academy, but just in general. Once you hit 2022-23 pretty much every single team has 1-3 very young newgens in the starting lineup at all times when at the beginning of the save it feels like you almost never see such a high percentage of players that age good enough for that much playing time.

It's true. 

You can also build a team with only under20 and win league (at least in low nation).

Maybe mental attributes should be more important (and young should have low value in them)

[Weston]

9 hours ago, santy001 said:

See there could be another issue at play here, it's very often that clubs (and in particular) fans of clubs will feel a youngster is ready, is on the cusp of first team selection. 

What is more often the case is that the managers don't have the proclivity to trust youngsters and tend to do so more when their hand is forced. While there could be an issue with regens, from what you're saying it more felt immediately to me like perhaps AI managers were too likely to give youngsters a chance.

But then could this be because perhaps the more experienced heads of around 32 or so that a lot tend to gravitate towards when they need some "experience" are still fetching too high a price? 

I can't say I've looked at it myself, but feeling there are too many young players in the starting line-up feels like a symptom of another problem given that during the early years generally speaking there are much better distributed players in the actual database than the regens available at that time. They won't have had enough time to develop unless the first year batch but presumably you aren't considering 21 to be "very young".

 

5 hours ago, FlorianAlbert9 said:

It's true. 

You can also build a team with only under20 and win league (at least in low nation).

Maybe mental attributes should be more important (and young should have low value in them)

As I mentioned there is a decent amount of data in this thread about the quality of the newgens, and we haven't done any proper investigating into the number of teens (therefore, newgens) that actually feature in starting lineups a few years into your save vs at the start. Though I can say, anecdotally, the last match I just played, vs Stuttgart in the Europa League knockouts, I believe they had 3 in the starting lineup. They didn't even look *that* good, so perhaps that adds credence to the theory there is something else in play there. Though usually they are much better from what I've seen.

Perhaps part of it is very young players are more of an unknown quantity in real life whereas in FM their attributes are just as easily seen as a 35 year-old who has been around forever. That means it's easier to trust them in FM than IRL. Also, it's possible that as someone who watches almost exclusively Serie A, where young players are very rarely trusted over experience, the amount I see in-game more closely resembles reality in other countries and the AI just hasn't nailed down Italian-specific nuance, so it feels more drastic to me from my own perspective.

[Weston]

On 12/08/2018 at 21:27, Weston said:

Okay, using the editor I figured out how to search all players in the game ages 18-20 sorted by hidden attributes, and I compared two points, one at the current date of 6 June 2022 (so all newgens), and the start date of 3 July 2017 (so all real players). I'm using a pretty large database and can provide more details on that by request. Here is the spread:

CA - 2017 / 2022

200-191 - 0 / 0
190-181 - 0 / 0
180-171 - 0 / 0
170-161 - 1 / 3
160-151 - 2 / 12
150-141 - 14 / 28
140-131 - 49 / 81

PA - 2017 / 2022

200-191 - 0 / 1
190-181 - 2 / 15
180-171 - 5 / 60
170-161 - 53 / 55

Keeping in mind this is a small sample size of one (1) save game, it's pretty obvious FM is unrealistically inflating the abilities of generated players to make the game more exciting as it goes on and recognizable players die out, and every game I've ever played has felt this way. I mean, these are just staggering numbers! I'd be interested in seeing other people's results as well.

On top of this, it turns out I owned 5 of the top 6 highest CA players, which definitely bolsters the theory that I'm signing a disproportionate amount of wonderkids, but that's gotta be at least partially because of how many more there are to find! On the other hand, I only own 1 of the 16 players in the 181+ categories for PA, so I'm not doing anything too special when it comes to nurturing more uncovered/undeveloped talent.

Per your suggestion I will post these findings in the bug forum. I get that FM has to take certain liberties with realism in the interest of selling games, but this seems a bit much to me. It's simply inescapably silly that compared to real players there are FIVE TIMES more newgens with a CA above 150 and ELEVEN TIMES more with a PA above 170. I want to find wonderkids, but it kind of cheapens the experience to know how artificially saturated the game becomes with them...

Hello again! Almost exactly one year later I've decided to compare FM19 to FM18, adding up how many 18-20 year-old players fall into certain CA/PA categories both at the start of my save and at the end (note this is not a perfect comparison as the data set is 5 seasons on FM18 and 6 on FM19).

CA - 2018 / 2024

200-191 - 0 / 0
190-181 - 0 / 1
180-171 - 1 / 5
170-161 - 0 / 6
160-151 - 1 / 26
150-141 - 7 / 77
140-131 - 46 / 155

PA - 2018 / 2024

200-191 - 2 / 7
190-181 - 1 / 10
180-171 - 13 / 58
170-161 - 44 / 63

Again it seems pretty clear that the newgens are overpowered compared to the real-life players that populate the starting database. In fact, it seems that the top tier of newgens in FM19 are not have drastically stronger CAs than their real-life counterparts, but also have been boosted even further compared to the previous edition of the game. It'll be interesting to see if SI can get this balance a little better in FM20...

Interestingly, I did not own nearly as large a percentage of these top players this edition, so perhaps scouting has changed in a way that balances things out in a way. In FM18 I had 5/6 top CA 18-20 year-olds, whereas in FM19 my squad contained only 1/16 (though another in this set was also signed by me, just at a previous club).

One adjacent issue I've noticed in exploring this is how many wonderkids have a CA equal to their PA despite being very young. In other words, they seem incredibly promising, but have somehow already reached their full potential at the very start of their careers. I understand some players peak early, but the sheer volume here seems unrealistic: In 2024 in FM19, 4/11of the top CA 18-20 year-olds have somehow already reached that top PA rating. Does this truly mean that even if played and trained in optimal conditions these wonderkids will never actually improve at all..? See below: 

[enigmatic]

14 hours ago, Weston said:

Hello again! Almost exactly one year later I've decided to compare FM19 to FM18, adding up how many 18-20 year-old players fall into certain CA/PA categories both at the start of my save and at the end (note this is not a perfect comparison as the data set is 5 seasons on FM18 and 6 on FM19).

CA - 2018 / 2024

200-191 - 0 / 0
190-181 - 0 / 1
180-171 - 1 / 5
170-161 - 0 / 6
160-151 - 1 / 26
150-141 - 7 / 77
140-131 - 46 / 155

PA - 2018 / 2024

200-191 - 2 / 7
190-181 - 1 / 10
180-171 - 13 / 58
170-161 - 44 / 63

Again it seems pretty clear that the newgens are overpowered compared to the real-life players that populate the starting database. In fact, it seems that the top tier of newgens in FM19 are not have drastically stronger CAs than their real-life counterparts, but also have been boosted even further compared to the previous edition of the game. It'll be interesting to see if SI can get this balance a little better in FM20...

Interestingly, I did not own nearly as large a percentage of these top players this edition, so perhaps scouting has changed in a way that balances things out in a way. In FM18 I had 5/6 top CA 18-20 year-olds, whereas in FM19 my squad contained only 1/16 (though another in this set was also signed by me, just at a previous club).

One adjacent issue I've noticed in exploring this is how many wonderkids have a CA equal to their PA despite being very young. In other words, they seem incredibly promising, but have somehow already reached their full potential at the very start of their careers. I understand some players peak early, but the sheer volume here seems unrealistic: In 2024 in FM19, 4/11of the top CA 18-20 year-olds have somehow already reached that top PA rating. Does this truly mean that even if played and trained in optimal conditions these wonderkids will never actually improve at all..? See below: 

That's quite interesting, although my first instinct is to wonder whether there are actually more 18-20 year olds in your database, which would skew the numbers hitting 130 plus even more. I'd also be interested to see the geographic distribution in relation to the leagues you have selected (creating some Argentine wonderkids in a European focused save is obviously less distorting than the Premier League becoming full of 20 year old academy products)

The higher PA skew is there as always and is obviously deliberate, but the CA comparison between the two games does rather look like the current development model is too fast at the high end - players should generally need quite a bit of time or first team football to get beyond 120 or so PA and especially to hit a potential above that level.

Edited August 18, 2019 by enigmatic

[Gross_Ballon]

On 12/08/2018 at 16:31, Weston said:

That's what I'm saying, I think the newgens may be too good TOO SOON.

I don't know if me looking so much is a problem in itself considering it's perfectly fine for me to have a focus on young talent, many of these players were already on the radar of other top clubs when I discovered them, and there are also many other even better newgens that other top clubs picked up before I ever discovered them.

I for one would love help or guidance in finding one of these lads ! 

[RBKalle]

Currently managing Italy (non-playable, non viewable nation throughout the save), in season 2029.

10 years into the game, my 23-men squad has a staggering 14 newgens in it. Out of the original db, only Donnarumma, Tonali, Chiesa, Kean and either Cutrone or Pellegri are good enough to be starters, whereas all the other youngsters from the original db are either already retired, have stagnated or have simply been surpassed by a couple of way-too-good 24yo newgens.

If you think about it, how reasonable or likely is it that in ten years 2/3 of the current "New Generation" will have been ousted by new kids we don't even know about? Doesn't Italy have enough good/potentially good players born from 1998-2002? I hardly think so, therefore the newgens are simply much better or start from a much higher point and thus "overtake" prospect who have been misused by the AI....

It'd be as if a current NT wouldn't field ANYONE born before 1988 except maybe a keeper or an experienced CB. They either have found an unprecedented Golden Generation, or there's something wrong with the callups...

BTW, some of those younger players have already 20+ caps, so they're not recent additions (under my enlightened management... If anything, I've dug up some 28yo who had been sidelined to play a bunch of mediocre newgens, likely with high rep or PPA).

And if that happens at a NT level, it's even more prominent at a club level where money talks and high-CA/PPA/Rep players don't go unnoticed... Which also explains why you can build a competitive team with free agents after the second season.

Edited August 18, 2019 by RBKalle

[CFuller]

20 minutes ago, RBKalle said:

Currently managing Italy (non-playable, non viewable nation throughout the save), in season 2029.

10 years into the game, my 23-men squad has a staggering 14 newgens in it. Out of the original db, only Donnarumma, Tonali, Chiesa, Kean and either Cutrone or Pellegri are good enough to be starters, whereas all the other youngsters from the original db are either already retired, have stagnated or have simply been surpassed by a couple of way-too-good 24yo newgens.

If you think about it, how reasonable or likely is it that in ten years 2/3 of the current "New Generation" will have been ousted by new kids we don't even know about?

 It'd be as if a current NT wouldn't field ANYONE born before 1988 except maybe a keeper or an experienced CB. They either have found an unprecedented Golden Generation, or there's something wrong with the callups...

Have you looked at the real-life England squad? Only one player born pre-1988 has been called up in the last 12 months, and that one player is Tom Heaton. Only three other 80s babies - Fabian Delph, Adam Lallana and Alex McCarthy - have received call-ups in that time.

In fact, of the 23 players England selected for the Nations League Finals, 12 are currently aged 25 or under*. That's the equivalent of having 12 'newgens' in a 2029 national team on FM19.

And let's use Germany as another example. Manuel Neuer and Marco Reus are still in the squad, but everyone else who's featured in the last 12 months - bar Nils Petersen - is in their 20s. Their last Euro 2020 qualifying squad included 13 players aged 25 or under.

I'd say that your scenario is not ridiculous at all.

 

* Just to clarify, I'm not including Harry Kane, who was 25 at the time of the Nations League Finals but has since turned 26.

Edited August 18, 2019 by CFuller

[RBKalle]

I see your point, but often real-life NT seem to keep around their veterans a tad too long, basically shafting the following generation which end up becoming "old" (think 26-28) but still woefully inexperienced. Therefore it's easier to call up players in their early 20s so they can "grow into" the NT setup and still be useful for years to come.

It happened to Italy with the 2006 WC heroes, who "ruined it" for the 1984-1989 generation. It happened to England, who relied too much on Gerrard & co and only recently has learnt to let aging stars go before they become a liability.

In FM terms though, I can't really accept that the only players in the 28-32 bracket who are still viable starters/alternatives for Italy are those who were already good enough for Serie A football in the opening season.
That means that NO PLAYER from the original db with good or top-potential developed into a worldclass player.

BTW: I checked how many of the England players in the 2019 Nations League squad were already in FM10 (database from 2009):

Walker (122/145), Rose (105/-8),  Maguire (35/-5!!!), Lingard (50/-6), Henderson (110/-8), Kane (68/-9), Butland (56/-8), Keane (45/-6), Delph (125/-9), Wilson (52/-5), Heaton (111/133)

So it's 11 players out of 23 present in the game, with the likes of Sterling and Barkley probably barely missing the db (I guess in recent years promising 15-16yo players'd have made it into the db, likely with a high negative PA).
Regardless of some hilariously low CA/PA, they existed in FM10 and could have become NT players in FM10's own version of 2019... In my FM19's take on 2029, no Italian player who wasn't at a Serie A level at the start of the game developed into a NT player and newgens took over rather quickly and decisively.

[Weston]

19 hours ago, Weston said:

Interestingly, I did not own nearly as large a percentage of these top players this edition, so perhaps scouting has changed in a way that balances things out in a way. In FM18 I had 5/6 top CA 18-20 year-olds, whereas in FM19 my squad contained only 1/16 (though another in this set was also signed by me, just at a previous club).

Having thought about this more I think it could also be because in the first experiment I had been at my current top club for multiple seasons and thus had more time with top scouts and more transfer windows through which to find and sign these wonderkids, whereas in the second experiment I only had one year. And again, in general I recognize this is all anecdotal and a small data set.

19 hours ago, Weston said:

One adjacent issue I've noticed in exploring this is how many wonderkids have a CA equal to their PA despite being very young. In other words, they seem incredibly promising, but have somehow already reached their full potential at the very start of their careers. I understand some players peak early, but the sheer volume here seems unrealistic: In 2024 in FM19, 4/11of the top CA 18-20 year-olds have somehow already reached that top PA rating. Does this truly mean that even if played and trained in optimal conditions these wonderkids will never actually improve at all..? See below: 

Can anyone explain this to me? Am I misunderstanding something, or is this a bug?

[Weston]

5 hours ago, enigmatic said:

That's quite interesting, although my first instinct is to wonder whether there are actually more 18-20 year olds in your database, which would skew the numbers hitting 130 plus even more. I'd also be interested to see the geographic distribution in relation to the leagues you have selected (creating some Argentine wonderkids in a European focused save is obviously less distorting than the Premier League becoming full of 20 year old academy products)

The higher PA skew is there as always and is obviously deliberate, but the CA comparison between the two games does rather look like the current development model is too fast at the high end - players should generally need quite a bit of time or first team football to get beyond 120 or so PA and especially to hit a potential above that level.

In 2018 there are 29,738 18-20 year-olds in the game, compared to 30,006 in 2024. Below are the comparisons of what leagues I had loaded in 2018 and 2024 - it's more or less the same but I deleted some of the lower leagues once I'd climbed from Serie D to Serie A as they'd become less relevant and the save file was growing quite large. The difference is pretty small, all things considered, but yes it does seem like there are more overall even with slightly less leagues downloaded (though with Serie D, for example, being an extra add-on I got from someone on this forum, I'm not sure it added too many players to begin with). Still, I don't think under 300 extra players could account for the drastic increase in top-quality players across the different CA amounts.

2018:

2024:

[Weston]

2 hours ago, Gross_Ballon said:

I for one would love help or guidance in finding one of these lads ! 

Hire scouts with high PA judging and send them on assignments specifically for players with a max age of 20! Have scouts report on youth tournaments. Specifically request they watch players that win youth awards and/or feature prominently. Search for young players who are making lots of appearances. Generally scout nations with promising youth like Brazil, Argentina, Belgium, Netherlands, etc. If you put time into it you'll find them!

[Weston]

2 hours ago, RBKalle said:

Currently managing Italy (non-playable, non viewable nation throughout the save), in season 2029.

10 years into the game, my 23-men squad has a staggering 14 newgens in it. Out of the original db, only Donnarumma, Tonali, Chiesa, Kean and either Cutrone or Pellegri are good enough to be starters, whereas all the other youngsters from the original db are either already retired, have stagnated or have simply been surpassed by a couple of way-too-good 24yo newgens.

If you think about it, how reasonable or likely is it that in ten years 2/3 of the current "New Generation" will have been ousted by new kids we don't even know about? Doesn't Italy have enough good/potentially good players born from 1998-2002? I hardly think so, therefore the newgens are simply much better or start from a much higher point and thus "overtake" prospect who have been misused by the AI....

It'd be as if a current NT wouldn't field ANYONE born before 1988 except maybe a keeper or an experienced CB. They either have found an unprecedented Golden Generation, or there's something wrong with the callups...

BTW, some of those younger players have already 20+ caps, so they're not recent additions (under my enlightened management... If anything, I've dug up some 28yo who had been sidelined to play a bunch of mediocre newgens, likely with high rep or PPA).

And if that happens at a NT level, it's even more prominent at a club level where money talks and high-CA/PPA/Rep players don't go unnoticed... Which also explains why you can build a competitive team with free agents after the second season.

Right. This is what I'm saying. It seems pretty evident that the starting database of real players is by far weaker than any following generation of newgens. I think this may be in part because the scouts don't want to overestimate real life players because it is a bit silly when everyone has worldbeaters in FM that grow up to be mediocre. But it seems far more likely that the largest reason behind this skew is that it is marketable to focus on the excitement of finding wonderkids, and so they make too many of them to aid in their discovery. Not to sound cynical, but maybe they should just be honest about it and offer "Easy," "Normal," and "Hard" difficult levels of newgen databases based on the frequency of such wonderkids. Yes, you could argue that Italy are in a bit of a slump this generation and a future one could far surpass it just as previous ones have, but I don't think that's what's going on here...

[Weston]

2 hours ago, CFuller said:

Have you looked at the real-life England squad? Only one player born pre-1988 has been called up in the last 12 months, and that one player is Tom Heaton. Only three other 80s babies - Fabian Delph, Adam Lallana and Alex McCarthy - have received call-ups in that time.

In fact, of the 23 players England selected for the Nations League Finals, 12 are currently aged 25 or under*. That's the equivalent of having 12 'newgens' in a 2029 national team on FM19.

And let's use Germany as another example. Manuel Neuer and Marco Reus are still in the squad, but everyone else who's featured in the last 12 months - bar Nils Petersen - is in their 20s. Their last Euro 2020 qualifying squad included 13 players aged 25 or under.

I'd say that your scenario is not ridiculous at all.

 

* Just to clarify, I'm not including Harry Kane, who was 25 at the time of the Nations League Finals but has since turned 26.

All good points, but you have to consider that England and Germany play youth more regularly than Italy do as a matter of culture. Also, the year following a World Cup (especially a disastrous one - or lack of one! - as is the case for Germany and Italy, respectively) would naturally see far more youth being tested as part of a natural regrowth period.

[ClemB]

Il y a 6 heures, RBKalle a dit :

Currently managing Italy (non-playable, non viewable nation throughout the save), in season 2029.

10 years into the game, my 23-men squad has a staggering 14 newgens in it. Out of the original db, only Donnarumma, Tonali, Chiesa, Kean and either Cutrone or Pellegri are good enough to be starters, whereas all the other youngsters from the original db are either already retired, have stagnated or have simply been surpassed by a couple of way-too-good 24yo newgens.

If you think about it, how reasonable or likely is it that in ten years 2/3 of the current "New Generation" will have been ousted by new kids we don't even know about? Doesn't Italy have enough good/potentially good players born from 1998-2002? I hardly think so, therefore the newgens are simply much better or start from a much higher point and thus "overtake" prospect who have been misused by the AI....

It'd be as if a current NT wouldn't field ANYONE born before 1988 except maybe a keeper or an experienced CB. They either have found an unprecedented Golden Generation, or there's something wrong with the callups...

BTW, some of those younger players have already 20+ caps, so they're not recent additions (under my enlightened management... If anything, I've dug up some 28yo who had been sidelined to play a bunch of mediocre newgens, likely with high rep or PPA).

And if that happens at a NT level, it's even more prominent at a club level where money talks and high-CA/PPA/Rep players don't go unnoticed... Which also explains why you can build a competitive team with free agents after the second season.

I agree with this and outside regens being too overpowered, the other explanation I found is that SI have been really reluctant the last years to attribute great PA to unconfirmed/unknown youngsters. Most youth players of top division teams are woeful and it's very hard to make something of them... I believe SI chose the easy/safe strategy to rate good only youngsters with already first team experience or high media exposure to avoid producing new "Maxim Tsigalko" and "Cherno Samba" who are bad publicity for the realism of the game.

 

This could also explain why so many op regens are generated, to fill this "gap".

I hope i've been clear enough:)