Buck, you are missing some crucial points. First, on the eve of the election, he predicted a 53 district swing, not 45. Second, he was very clear and specific that while he was predicting 53, that number had a very low confidence level, it was +/- 30 seats. Which means his actual prediction was around ~+21-80. Why such a broad prognostication? Only 25% of the house seats actually had polling data to go on. Mr Silver doesn't do polling, and he's not omniscient, but when he does have polling data to go on, his results have been very, very good.