As I already mentioned, while SFDC is something a computer would follow exactly, it is intentionally not coded that way to add a factor of risk to an otherwise very rewarding training type (super concentrated one position, 48 min per game training). Only managers with highest level of dedication and yes, luck, can train NT material.
We do not want everyone to use the same exact type of training that yields the best possible results every time. And yeah, there will always be people that will attempt a more risky approach and others that will go the safer route. The difference adds width to the game, and the ones that go the risky route have the right to complain and be unhappy when the risk doesn't pay off - but that doesn't mean we're going to change a crucial part of the game to suit them and reduce that risk.