you're right, it would make a lot of sense to combine votes of experts -- or look for the common recommended bets. the real problem is that a site like DC doesn't get enough attention to bring in enough experts to make that worth while.
i'll tell you another idea that would make sense -- have the system observe which experts were correct after the first few games, and then weight their predictions more highly. might give you a quick way to pick out a good set of candidate expert ratings.