Now... It's pretty clear why these are randomized and hidden, and I don't plan on taking that away.... not entirely. What I'm going to suggest should help give players a better idea of what their character is, rather than letting them stumble around in the dark.
So here's the idea:
For every stat with two substats, there will be adjective pairs. The adjective that is selected to be shown upon the @stats screen will be determined by the sub-stat that is greater than the other. In the case of equal values, it will either select a default or pick one at random. The adjective selected reflects the nature of the character.
For example, the sub-stats in agility might have a pair like:
Precise (CRD > QCK) : Nimble (QCK > CRD)