For an interesting discussion on NFL football related to prediction, please see On Any Given Sunday, NFL Favorites Win and Underdogs Lose. If things were as simple as the article suggests, finding an 85% model would be much less difficult than it's turning out to be. Upsets happen regularly and can happen with almost any game, and many games are close.
In fact, consider the following from just this weekend: IND over HOU by 7 points - CHI over KC by 1 point - CIN over SEA by 3 points - ATL over WAS by 6 points - TB over JAX by 7 points - CLE over BAL by 3 points - DEN over OAK by 6 points - NYG over SF by 3 points - PIT over SD by 4 points. (That's a full nine of the fourteen games played this weekend.) I would submit that any of these games could have gone either way. Unless I'm missing something, this doesn't look much like a stacked deck to me.
Furthermore, of course it's true that the NFL goes out of it's way to create close games. In fact, the draft is designed to equalize teams. Should it be otherwise? I think this just makes for good sports.
Readers might also be interested in How Is the NFL Schedule Created? from the Bleacher Report.