I think you have to handicap the numbers *and* the game. Both are important, and you have to know how to balance the two.
For example, college basketball. Most teams have played 6-7 games, but they're mostly against cupcakes. If they shoot 58% against Navy and JMU, you need to determine how relevant that stat actually is. Just because they shoot 58% against Navy doesn't mean they're gonna shoot well against Duke! I don't think stats are very useful in college hoops until January/February.
Andrew Carnegie was considered the richest man in the world. Did he use formulas, stats, etc? Not really. He just had a feel for business. But if you take a business/finance class, you learn lots of formulas in order to determine good/bad investments. One of my teachers once said that these formulas simply try to quantify the decisions that Carnegie/Rockefeller/etc would make. But you still need the business-sense in order to apply these formulas to real-life situations!
As for halftime lines, look @ the Saints/Giants game. Saints were up 24-7 @ halftime. Given that the Giants were only an 8-point underdog for the game, wouldn't it follow that the Giants would decrease the 17-point deficit in the 2nd half? But if you were watching the game, you could probably guess that the 2nd half was gonna get UGLY. If you were watching the game, you would've observed that the Giants were simply over-matched. You can't cover Joe Horn with a 3rd-string CB!
So I guess you need to cap the math...you need to cap the game...and a little bit of luck never hurts!