I've already modeled a structure of an OBP (insect proteins). I picked the structure  (pdb) with the lowest DOPE score, which is actually has the best stereochemical quality according to Procheck. After that, I used the script for "loop refining" in a positive range of values according to Anolea (showed as red bars). I was waiting a better stereochemical quality after the loop refinement, however, more bad contacts among residues and low allowed positions were obtained. My question is, What aspects must I take into consideration for loop refinement?, Is possible that the modeled structure do not need any loop refining?

Best regards

Herbert