Umm, no. What is the "model"? Crap in, crap out.
If you look at the RAW numbers, as told by Nate Silver himself, the numbers are:
1. Montana (99% chance of GOP win)
2. South Dakota (99%)
3. W Virginia (98%)
4. Arkansas (90%)
5. Louisiana (77&)
6. Colorado (75%)
7. Iowa (67%)
8. Alaska (67%)
9. N. Carolina (32%)
10. New Hampshire (17%)
Those are Nate's OWN admissions and numbers. Go, EXCLUDING some B.S. undisclosed algorithm/"model," looking to logic we see the six pick-ups necessary as being the six likeliest (i.e., numbers 1 through 6). Because the VEGAS FALLACY is why gamblers are poor and "the House" is rich, when we factor out the Vegas fallacy and we make all the events INTERDEPENENT, as logic dictates, we see a 99% x 99% x 98% x 90% x 77% x 75% = 49%.
By Nate's own numbers the GOP is, statistically and logically, NOT likely to win the Senate. The math is clear. He's "modeling" this to create viewership, drama, and added page views (i.e., revenue). It's a close election, but don't drink the Kool-Aid. Crunch HIS own numbers yourself and you'll see his "model" is defying logic and math.
The take away is "it's going to be close." If you want to get all "mathy" then the numbers are technically on the Democrats side. I can't with these crap-in, crap-out models. Just because you make a blog doesn't mean you're being true to math or logic. I can bake numbers too and hide it under a "model."