tendency for that model to be overconfident, including in the 2014 midterm elections when i believe mr. wanggave the democrats a 95% chance of holding the senate in september. and you know, i would think of it, if you want to see how different the models are, don't think of it in terms of percent. think of it in terms of trump's odds of winning. 99 to 1 versus 6 to 1. that's a much bigger gap than the difference between our models. >> everybody says that the senate is literally too close to call. you have seven seats, and so what's your best, again, what does your model say about this? >> exactly the same thing. if you look at the upshot model and our model, it's basically 50/50. maybe the slightest of edges to the democrats. we had more polls this afternoon, it might change a little bit. in my opinion, maybe six or seven seats are going to determine this. what's so interesting is you have seen the polls in the different state move in different directions. if you look at a state like pennsylvania, katie mcginty, the democratic candidate, moved up. if you look at missouri, you saw jason cando