kennec

and another 3-4months before 3-4 patches have sorted the bugs in the new versions new extra functions that messed up old reliable functions

Nicki

Where this problem really hurts is in the american Major League Soccer, where the rules force you to have some development players because of the maxsimum of senior contracts and the wage cap, after 2 years the game just gets annoying, because there maybe 2 usable players per year overall, the rest are completely crap, and I mean crap, who can use a player with 6 in pace??? (for example). and you cannot get so many youngsters from better countries where the youngsters are a little better because of the limit of internationals. If you play with Toronto FC it gets even worse, because there are very few canadian regens and those who comes are mostly very bad and Toronto FC have even more limits than the American clubs. I have even read that other people starts off by using all their draft picks for players and then play 2 years and then jump on to another league because of this.

kennec

the key problem with this is.

as a human manager we know we need certain players at the differnt positions. we look at their speed and jumping and so on.

the computer doesnt do it that way,the AI just look at ca/pa/rep and set them out. this means a human player wich acts like a normal manager would will stokpile the rare players with special skills.

a team with some pace on the wings and a jumping striker will scoreloads against small defenders in the ai team.

Same in defence where u get togheter a team of defenders with good jumping so all 4 can attack on corners and defend on AI corners.

ai teams cant compete with this and it doesnt matter hes centre backs are superb at throw ins and crossing the ball, couse they never use those skills.

isuckatfm

Database size is irrelevant. It's about consistency in the statistical spread of players for a given sized database and as it expands within the game world.

So if in my starting game I have x% of under 21 players with a CA of Y or more then that number should remain reasonably close to ensure the quality of the players remain the same. There should be no arguments about 'whose to say in 5 years time there won't be less quality under 21 players' because for that to hold true requires the opposite should hold true. Yet no one has ever posted here any evidence of CA %s increasing hence we have a one way change in CA values for young players as the database evolves i.e. regens are worse than the real life players at the start of the game.

So for example:-

7 July 2007:-
Total Number of Players: 12411
Total number of players 21 and under: 6794 (54.74%)
Total Number of players 21 and under with CA >120: 311
Percentage of players 21 and under with CA >120: 4.58% (311/6794)

26 May 2014:-
Total number of players: 12673
Total number of players 21 and under: 6143(48.47%)
Total Number of players 21 and under with CA >120: 118
Percentage of players 21 and under with CA >120: 1.92% (118/6143)

Now that could be skewed due to age ranges so I'll look at 20 year olds only and examine percentages:-

7 July 2007:-
Total 20 year olds: 888
CA 130 to 160: 61 = 6.87% (61/888)
CA >100: 283 = 31.87% (of all 20 year olds, 283/888)
CA 130 to 160 as % of CA > 100 = (61/283) = 21.55%

26 May 2014:-
Total 20 year olds: 951
CA 130 to 160: 10 = 1.05% (10/951) more than a 6 fold drop
CA >100: 141 = 14.83% (of all 20 year olds, 141/951) halved
CA 130 to 160 as % of CA > 100 = (10/141) = 7.09% more than 3 fold drop

So we have the same issue with my small game as exists with John Shaft's larger game.
If someone can show numbers like these that show a steady value (either increasing, or slightly lower) then fair enough. Posting a few screenshots of great regens in your game is not convincing for me. But until then I believe this is an issue brought on by two things:-

1. Slow, generic development
2. Low starting CAs

Mentioned many times. The last time there was a response from SI was in a thread a while back where they said they were aware of certain things but also didn't consider some aspects to be a problem. The message though was that player development, attribute assignment etc. is not very stable in that a single change to the code can have massive knock on effects. So refining the model is a slow, painstaking iterative process and I'll be surprised if it's fixed for 09.