Because you are talking apples and oranges aka starters and relievers.
True. Just an example that two pitchers can be close, but if their ERAs (or some other stats) are far different the market might not see it that way
First thing you do is evaluate your starters using up to date information, to decide how much better the starter who is going for the favored team is than the one pitching for the dog (if at all better).
Exactly. And if the pitchers are close, yet the lines are set so heavily favoring one, look to bet the other side. Penny is a perfect example. With the exception of a couple starts in April, he has been pretty much the same pitcher every single time he has stepped on the mound, whether he has given up four runs or zero runs. If someone fails to see that, it is not the fault of the stats
I really don't see how there is a disagreement other than the insistence that if we take the wrong path the stats could somehow be at fault rather than us