I've exported and run models. My outcome for the models was ERA; I should have also modeled WHIP to isolate effects on SLG, for example. So, when I say that GB/FB has a potent effect on ERA and and pjfoster says it has no impact on HR, we can both be right if, for example, high GB/FB leads to a big increase in double plays.
While it is theoretically possible to imagine an effective ML pitcher with a VsL or VsR < 50, I've never seen one; I just don't think the engine creates guys with splits of 45 and pitches of 95, which is what it would take. Similarly, you're playing with fire at control < 50, although I've seen some guys in the high 40s with splits in the mid 80s who were OK.
Rivero is a very back end option; I sometimes use guys like him in a mop-up role when I can't afford 11 legitimately good pitchers, since he has high DUR/ST.