gzeiger is correct in saying that being in the same genus does not mean species will breed together or not. Taxonomic ranks (e.g., families, genera and even species) are entirely arbitrary, even if they are natural (monophyletic - that is, include all descendants of a single common ancestor). Such natural groups are identified by the exclusive sharing of derived characters (synapomorphies). Sharing of primitive characters (symplesiomoprhy) does not indicate a close relationship. For example, lungfishes and coelacanths are more closely related to land vertebrates (tetrapods) than to other fishes because they share a number of derived features with land vertebrates. The presence of scales and fins (both primitive features in bony fishes) in lungfishes, coelacanths and other bony fishes is irrelevant, and does not indicate a close relationship.
Thirty years ago, the late Don Rosen (former fish curator at the American Museum of Natura History) noted that various poeciliid fish species hybrized with each other, while others did not. He further noted that some of the closely related species did not hybridize, while more distantly related ones did. He concluded that the ability to reproduce was in fact a primitive characteristic, and not indicative of close relationship. This was a major blow for the prevailing "biological species concept" of the time - which argued that reproductive isolation was the primary indicator of a species.
Hope that helps a bit - though sorry about the jargon.
As for crossing New World and African cichlids, I can't imagine anything good will come of it. I suspect you will end up with a lot of dead breeders (I particularly worry about how such distantly related, aggressive fish will behave towards each other), and even if you are successful, some pretty ugly (and no doubt unhealthy) offspring.
Tony