So, judging by the numbers I had going into yesterday, I am not surprised that Bevin won. If I had to bet money on the race going into yesterday, I would probably have put my money on Conway without much confidence. That being said, the model nailed the Mississippi Gubernatorial race. There it overestimated the GOP by 1.1%. For a "blowout" race, that is about as accurate as you can get.
Going into the Louisiana run off later this month, I have Vitter with a 39% chance of winning. This is mainly fundamentals dragging him down because he has been polling in high double digits behind the Democrat here. Based upon the primary and historical trends, I would say Edwards would likely lose 58%-42%. That being said, those fundamentals are basically decent for an election where David Vitter is not the Republican candidate. Coupled with polling margins, which average ~15% unweighted and ~14% weighted, this projects a 52.6%-47.4% margin of victory for Edwards. Vitter still has a pretty good chance of winning this one, but things aren't looking good for him. Even with a large polling miss (which my model is projecting), Edwards is leading by a high enough margin that he could still win with a few points to spare.