My idea is when a player reach a certain age(let's say 30 years old), they get a random mentoring score that does not go up or down. You would never know how good a player is good in mentoring until they reach that age.
I don't think the idea is without merit, but it needs much more thought. Right now it sounds like creating yet another transfer market where teams pick up players over 30 to more or less ride the pine and somehow give a boost to the younger players. That's pointless. There would need to be a clear link between the "mentor player" in-game performance and their effect in training. And it needs to be fine-tuned such that it's strategically pretty much as wise to go without veteran as it is to spend money/playing time on them to boost the young ones; perhaps the presence of the "mentor player" would affect the training speeds based on their own skill set or something.
Thinking about my two previous sentence for only 15 seconds or so, I get the mental image of the young players, like vampires, sucking the skills from the veterans who become worse and worse and finally writhe in agony until they get sacked. Yes, I like this idea, but it needs more work. I agree with the comment about considering also the player's experience level.
Any change like this has the potential problem of making the game more complicated. That's a real issue. If it's too complicated, it kills interest and accessibility for new managers and those not willing to spend much time; if it's too simple or has a very limited effect, why introduce it at all. There needs to be room for a real choice, where there are no completely wrong alternatives, only outcomes that differ enough to make it a strategic option.