We use really simple and transparent methods for creating forecasts for the gubernatorial and senatorial elections. Everything I do is outlined is this forthcoming paper. The method is unchanged from 2012, but the coefficients are updated with 2012 data. I consider three different types