In the past I've made the mistake of getting a little too fancy with my picks. I was prone to going for the "high upside" guys like Quincy Morgan, Onterrio Smith, Lee Suggs, Tyrone Calico, Adrian McPherson, and Matt Jones.
These days I take a more conservative approach. When evaluating rookie prospects and unproven players, I tend to look at draft position and production more than message board hype and "upside." I no longer reach for prospects with character issues or for "raw" prospects who have "huge upside." A guy like RB Chris Henry is a great example. He's a workout warrior who never produced. These guys never seem to pan out. Tyrone Calico and Quincy Morgan are out of football. Meanwhile modest, but productive athletes like Mark Clayton and Santonio Holmes have made a smooth transition to the NFL.
Also, I'm much more reluctant to assume that a given prospect will break out. If you look at the dynasty rankings of guys like DeAngelo Williams, Jay Cutler, Vince Young, Matt Leinart, DJ Hackett, Vincent Jackson, and Brandon Marshall, it's clear that the rankings don't necessarily lend adequate consideration to the risk side of the risk vs. reward equation. I have no doubt that many of these guys will eventually become stars, but some of their rankings leave no margin for error. That's a bad thing.
Never assume that a player will break out. Always consider the probability that he'll disappoint just like countless other "next big things" have in the past. There are dozens of guys like TJ Duckett, Michael Bennett, William Green, Tyrone Calico, Lee Suggs, and Quincy Morgan lingering in the FF graveyard.
I make exceptions in cases where the player is clearly a rare talent (i.e. Reggie Bush, Adrian Peterson, Calvin Johnson, Andre Johnson, Lee Evans).
FF isn't that complicated. If all you do is draft good players, you should end up with a good team. Don't overthink things. Just take good players and take them at a fair cost. Where you get into trouble is when you start overpaying for unknowns and bank on breakouts that may or may not happen.