pelicanstuff

As long as enough of the random element is kept... Not every player who starts out in the game IRL is massively suited to their position - we just never hear about most of these. Solid caps on attributes for positions are not realistic, they have to be merely a guide or a case of probability.

It would be nice to have the newgens sometimes generated at more random times of the season as well.

The development curves doo need overhauling, I will agree with you on that.


(P.S. I think although my way isn't a 100% overhaul, it might be slightly easier to implement and cause less bug-fixing time.)

Sir_Liam

I think regens have gone backwards since FM2007.

In 7.0.2 I thought SI were getting a lot nearer to getting it right, but on FM2008 I see the same problems that 2007 had, plus much worse physical attributes.

Celtic_1967

Sorry, I should have been clearer. I'm not saying that the problem didn't exist before, just that the current system for creating regens and developing them is relatively new.

We're not butting heads. Rather we're having a discussion about an issue we both feel strongly about. We're approaching it in different ways but we both want the same thing.

How's about this for an idea:

Instead of manually coding a development system, why not create code that creates a series of templates when we initially set up the game.

Essentially - the system looks at the top 100 (or how ever many) players in each position (ordered by current ability) and then builds a template based on their current stats compared to a '10 year old' - a blank regen with 1's in all positions.

This template describes the development curves with a high standard of training in all attributes required to bring the player to the equivilant abilities at the equivilant age.

When a regen is created it is automatically assigned a development template from that list and then progresses according to that.

A method like this *should* keep regens on a par with the content of the database at the start of the game - at the expense of taking longer to generate the initial game in the first place. </BLOCKQUOTE>

I love this idea. If it can be implemented correctly I can see this issue being largely solved. I believe that this solution is likely to be acceptable (if not ideal) to everybody

AB-forever

IMO the bold bit is very important. DaveRH, I really like your suggestion but the random element should be there. And maybe more than 100 players should be used to base templates on as there are many, many different types of players.

carthy1

I have never got that far myself but can I ask a question ? ........
If you were to go by good and bad players as opposed to old and young players are the standards still shocking?

DaveRH

It depends on what you mean by the random element.

In my suggestion most potential player quirks would be dealt with by the quirks in the template itself (so the quirks of the player that template is based on), combined with the factors of training regimen (which is likely to be slightly different to the training the original player had) and training effectiveness in different skills.

That should be variable enough (note - not random - I believe that randomness is part of the problem in the system already) enough that you'd have a pretty hard time saying "this new player is definately based on player x from the database".

Remember - we aren't looking to create the perfect newgen system, just one that creates satisfactory players for the game.

I agree about the "more than 100 players" comment - it's a fairly arbitary number. The more players the better, but if it's being generated as users are generating their games then you'd want to draw a line between the number of players being examined and a satisfactory processing time.

Liniert

in my game (currently in the year 2036) there are 162 players worth over £10million.

current best player in the world is Walter Morel who plays for AC Milan and has 40 goals in 43 appearances for Argentina and his stats are truley world class.

i dont see the problem with regens, maybe its just my game

DaveRH

Some further thought on this - with regards to training:

Training points are earned on a RPG-style experience point system - so once a player hits a certain amount of points then he is automatically given a suitable amount to distribute across his skills (the amount being determined by the difference between his CA and PA, some of his key mental attributes (such as professionalism and determination) and the quality of the coaches/facilities).

Two seperate ladders are used for physical and mental/technical improvement - so the two sets of attributes can improve completely seperately (this would stop players who done no physical training getting stats increased in those fields and vice versa).

Players earn points through training sessions, playing non-competitive games and playing competitive games.

The player template allocates the order in which points are assigned to the attributes - but the template is also modified by what training the player has been assigned to.

As players age, their PA is reduced and appropiate penalties given to both the physical and mental/technical attributes. The template (again modified by training) determines which stats receive point losses.

arrogantio

They're not though, that's the point.

The CPU goes afetr crap high PA players with reckless abandon, while ignoring players who are actually performing, and are positionally better.

THAT is one of the main reasons the execution of the PA system is so flawed.

That's why I recommended (with NATURAL ABILITY) that no-one gets to see it, not even the scouts/CPU.

Someone being able to see a PA of 190 in a crap 20-year-old rotting in the reserves is just utterly unrealistic and has no bearing on real life.
Someone being able too see a 20-year-old as a big talent because he is performing really well in a good league and improving a lot is UTTERLY REALISTIC.

That contrast illustrates why the current system is so flawed. </BLOCKQUOTE>

An easier way of doing this would be to make 'Judging Potential' attribute generally less effective, both for scouts and for AI managers, especially in cases where the player doesn't play much.
</BLOCKQUOTE>True, although in all honesty the



Thanks for that

SI's regen system must already use some form of templating system. In some respects, the system already in place is superior to a "regen" system even loosely based on existing players like the one suggested by DaveRH

Players of a low standard aren't simply linear translations of world class players (e.g slower, less technical Messis and dumber, less athletic Ferdinands). Lower levels are full of imbalanced players who excel in one particaular area but have serious deficiencies in other aspects of their position. Higher levels do not - some players may have minor weaknesses such as defenders who occasionally lapse in concentration - but a player with most of the attributes to be a world class centre back but no aerial ability will be considered a fullback or DM.

At CA 50, there will be quite a few natural central defenders who have no ability whatsoever in the air (but are picked there because of other abilities e.g marking, anticipation that are unusually good at the level they play). At CA 150 there aren't: any such players that as youngsters (with CA50) show the potential for all round improvement will be moved (most likely to DM or fullback), which is most adequately represented by the regen system simply not producing high potential players with certain characteristics ill-suited to their position at all. Conversely at lower levels, strikers with finishing as one of their weaker stats, defenders who can't tackle etc aren't exactly unusual)

My suggestion is simply to tighten up what is presumably already there:

This would involve having some very tight starting attribute ranges for some attributes in some position templates, some of which might actually contradict other templates for the same position (e.g. "quick, raw winger", "slow technical winger", "dribbling winger" "hard-working winger", "all-round winger"). Some of the looser templates (e.g. "pure physical target man") would only be applicable for lower potential players

Others may be governed by a loose probability distribution for their position. In some cases, such as long throws, the existing distribution of attributes is far too loose - the vast majority of players ought to have values in the low double digits with the exceptions being mostly wide players, but that shouldn't stop a mediocre centre-half having the stat at 20 - and the tag "viewed as the next Dave Challinor . Some, such as finishing for wingers, would have a fairly broad range and even distribution.

Other attributes like determination and natural fitness could be calculated entirely independent of position but linked (via a probability distribution) to PA and the starting attributes a template returned (you will find some, but not many potential Premiership footballers with low natural fitness; hardly any hard-working, professional and influential players that lack determination, nobody that can cross and take free kicks well but be incapable of taking a decent corner) These last ones seem to be attributes FM08 is particularly wrong

I think there's a lot of merit conceptually of viewing the "templates" as what the finished article should look like: a player reaching their PA, whether that's a flawless 180, solid 120 or one-dimensional 70 with starting attributes at 15; CA ~55 scaled by subtracting from the different attributes at peak level the differing number of points those attributes would naturally expect to grow as part of a normal development curve.
Obviously the actual "finished article" may end up somewhat different depending on how they are trained. I believe this already occurs in a manner not dissimilar from the one in DaveRH's post immediately above.

arrogantio

if any moderator wants to remove the top quote (from another thread) and half-sentence my above post will probably make more sense...