To be honest, I spend more time determining relative value rather than projections, because as you have said, it's too hard to be accurate.
You should look at injury history, look at last year's stats, look at college performance in the case of rookies, look for new coaches, coordinators, and offensive system.
Look at who was lost or picked up in free agency and the draft. A different QB affects RB and WR. A decimated OL hurts. A stifling defense helps an offense or makes a coach play ball-control, limiting the QB and WR, but helping a RB (did anyone hear me think BALTIMORE really loudly?)...
A lot of people look at schedule, but I think that's overrated. First, teams rise and fall too quickly. Once you have a roster of people, you can look for individual matchups, but you have no way of breaking down a season, game-by-game, for 200 fantasy players.
Sorry if I sound like a pessimist, but that's not even everything, as I am sure my peers will be about to tell you