I think you missed one small factor. You have to consider how many contesters are on the other end of each path. Figuring a great figure of merit to zd8 is good, but if it only produces one qso its n
Propagation predictions can come from VOACAP (ITSHF is the name of the package, and it's free from NTIA to download). There was an article about VOACAP (which is, by the way, based on the old IONCAP