Bet any game where Silver's projected winning % (I converted to pt spread) differed with Vegas closer by more than 1. He tracked it in 2011 and it went 26-18: http://graphics8.nytimes.com/images/...1-popup-v2.jpg
I don't follow college bball that closely so I figured this was better than throwing darts. This is the sharpest publicly available info IMO, if you are a skeptic read here as to how he does it: http://fivethirtyeight.blogs.nytimes...n-c-a-a-picks/
Also, for the conversions of winning probability % to point spread, I used the same number (10.36) that he used in 2011. I have no clue whether that is standard or if it would differ from year to year.