I find the discussion of their model much more interesting than the results: http://fivethirtyeight.com/features/nate-silvers-ncaa-basket...