Interesting scoring performance in flexpepdock against random peptides

Dear all,

recently I have tried to dock some different peptides to binding pocket of protein kinase and see their performance in flexpepdock. In the meanwhile I also dock some random peptides, such as poly G (GGGGGSGG) and poly E peptide(EEEEESEE). Interestingly, the energy score of good peptides and random peptides didn't show much differences(from -650 to -690), sometimes the random peptides even showed the better energy score. In order to keep them in the pocket I have made some constraint. The orientation of peptides are similar. 

My question is, can rosetta flexpepdock distinguish peptides by reflecting to energy score? If yes, how should I do?

Thank you!