I actually learnt opa tracing ( I just used zz tracing)
I haven’t actually done an opa solve yet lemme do one and see if it works
.I guess I should learn an actual opa tracing method
To elaborate: counting good (or bad) wings tells you how many edges are in the correct (or wrong) orbit, but doesn't tell you anything about the parity of the pieces that belong in each orbit.You can't just count good/bad wings; that won't tell you anything about parity.
This is kind of analogous to square-1 CSP tracing: first you figure out how the edges (/corners) split into two groups of four (this is easy because you just look at whether it's white or yellow), and then you still have to check the permutation parity within each group! If the second step weren't needed, (smart/fast) people would not be doing it.