Selecting the best pose/cluster after docking

Hi...

I have a query.....After running ZDOCK for the docking of 2 proteins, how do I select the best pose or the best cluster after obtaining the results for the top 2000 poses ?

All this while, I was using a possible faulty technique of observing the cluster with the maximum number of representative poses in the top 50 poses and considering such cluster as the "best cluster".

Is this a right technique ? If not, how should I evaluate the poses and the clusters ?