How do I create a conditional asymmetric objective function for regression?

Question

I\'m trying to use a conditional asymmetric loss function with a regression model and am having issues. I want to penalize wrong way results but the direction flips depending on