Here is my model for AI players development in career mode which I inspired off from NBA 2k lol so nevertheless here it goes.
There should be Players Initial skills which shall be randomize between overall skills of every regen players and it should be between 35 to 55 and then the Potential attribute which shall be randomize between 55 to 99 and the Regressed attribute which will be the lowest attribute a player will get regressed to close to his retirement age which shall be randomized between 25 to 50
So the regen player starts with initial skills and then there shall be development period which is in years again randomize between 3 years to 7 years and at the end of this period regen player will reach it's potential attribute then next come peak period in which the players remain in it's highest attributes which he attain at the end of development period again this period can be random between 3 to 7 years and then finally the regression period in which the players starts losing it's attributes to reach it's final lowest regressed attribute again this period can range from 3 to 7 years.
This way we will have truly unique regen players some will attain their peak slowly and remain in their peak so long and start losing the attribute so slowly till their retirement age potentially a career that will last 21 years or at the other hand we can players who reach their peak quickly but not remain in their peak and then start losing the attributes quickly so as to have potential of 9 years career and then we can have multiple combinations of players career in between because of randomize element.
Sounds great if get implemented by BAS.
That's generally how I'd like to see it. I'm coming from spending lots of time on Football Manager using the editor. Key things there were:
Current ability
Potential ability
So, a player might have a current skill level and then a potential skill level. For my skill spreadsheet I generated a total ability, which would be the current ability. If it was something in the game I'd easily add in a potential ability too because I base it on specific levels of ability, so you could estimate a range that player might reach.
For someone like Jimmy Anderson they might have, for example, 500 pts which I would then spread out across his bowling skills. If it were possible to give him a potential ability I would set it at the same amount, due to his age, therefore his skills would never increase.
Then you'd want a regression which brings the skill down. I'd imagine this could be generated for each player when you start a career mode. It could calculate a retirement date/age and a 'peak' age and the length of their peak. It could include separate dates for retiring from particular formats, as this could vary. So, after a player begins their decline they might retire from test cricket, then ODI cricket and then all cricket.
One thing I liked about Football Manager was the 'random' aspect of potential ability. In that, for younger players, you could give a 'range' rather than a fixed figure. So, a player might be as good as Jimmy Anderson or they might be as good as Chris Woakes. Equally, they might end up being an international or just a decent domestic bowler. It could differ for each career, so you would get quite varied 'worlds' for each player.
There were also reputation levels too which would determine factors like transfers, fees and probably international selection. Something like this would be quite good. I think you could implement something using skills and other factors. For example, you might get reputation boosts if you end up as a top scorer in a tournament or a small boost for each man of the match award or milestone. Ideally you'd have in a cricket game something that creates separate reputation elements for each format.
I don't know how it's done currently but it certainly seems like some aspects are there but not necessarily that deep. It may be more complex than it seems.