I say make what you know and like.  If you play puzzle games, make a puzzle game.  If you play fighting games, make a fighting game.  If you like a genre of game, then you probably know what makes it good (at least for you).
As a side note, I used to be a shareware computer game programmer.  My first and biggest effort was an RPG that I did as a senior in high school.  I played the gold box SSI D&D games, so it was heavily influenced by that.  Anyway, crude as it was, it was one of the first (if not THE first) RPGs for Windows 95, and sold a few hundred copies.  This was in the early WWW days, self-published, and I didn't have a way to accept credit cards, so I consider it a moderate success.  It gave my storywriter and I plenty of play money during our first couple of years of college before tapering off.