US Election Predictions

Forget my bet - I'll either be the winner of 3 bottles of whisky or this election is going to cost me a run to Massachusetts for a few choice bottles to hand over to my betting comrades.
By way of preface: I have absolutely no track record for making correct predictions - I was way off in 2008. 
I am looking at the numbers and I want to now hazard a series of election eve observations.
Here is what I know:
1. Obama is going to underperform his 2008 victory of 52.9-45.7 in the raw vote, and 365-173 in the electoral college.
2. Obama is not going to pick up any new states over his 2008 victory. He will lose some.
3. The Senate and Congress, having tilted either a bit or sharply Republican in 2010, will produce a pickup or two in the Senate for Republicans (indicating not a particularly strong improvement, given that more Dems are up this cycle), and an essentially steady situation in the Congress as compared to 2010.
4. The electorate is a bit less white than it was in 2008 (works in Obama's favor).
5. Both candidates suffer from an enthusiasm problem with their base. Hard to know if the enthusiasm gap is bigger for one or the other.
6. Only rarely and with no consistency has any candidate hit 50% in any national poll in the last 30 days.
7. There are 4 things we did not know after both national conventions which we now know - if you will, the 4 surprises (in chronological order):

• the 47% tape (bad for Romney)
• the Benghazi attack of 9/11 (bad for Obama)
• the Oct. 3 debate performance of Obama (bad for Obama; good for Romney)
• Hurricane Sandy (the October Surprise, if you will - good for Obama)

Next, there are things I do not know, which are hotly contested by the two camps:
1. The electorate will either be composed of Dems, Repubs, and Independents in a mix that resembles past election trends - or it won't.
2. Independents are breaking significantly away from Obama to Romney - or they aren't.
3. One side has already learned it will lose - or it hasn't.
Given all that, I can foresee - if everything breaks Romney's way (the poll models are wrong, independents defect from Obama in a big way) - a possible maximum victory in the electoral college of 304-234 - but in order to accomplish this, it would mean Obama loses 10 states he carried in 2008: Florida, Pennsylvania, Ohio, Virginia, North Carolina, Indiana, Wisconsin. Colorado, Iowa, & New Hampshire. That would be amazing.
Much more likely is a scenario where only some of those 10 states will switch. Let's look at the best result Obama can expect in 2012 - only 2 states switch: North Carolina & Indiana (and the one electoral vote from Nebraska). Then Obama wins 332-206. That would be amazing.
Between these two extremes, the problem is figuring out which of the other 8 states Obama will lose from his 2008 victory.
Let's subtract from Obama & add to Romney the 2 likeliest states, based on current polling: Florida and Colorado. In this scenario, Obama loses 4 states he carried in 2008. Then it is Obama 294-244. Maybe.
In order for Romney to win, he'll then have to pick off one more "big state" - Pennsylvania or Ohio & then one or two other, or else run the table with all the littler states. That latter scenario is unlikely. But this is the scenario in which Obama retains Ohio & Pennsylvania, loses every other of the remaining, and loses in a squeaker: Romney 273-Obama 265. Doubt it.
So, it looks like I may be buying whisky for my betting buddies. Obama has more paths to electoral victory than Romney. And it will mean that Obama will have pulled off a modern election miracle unheralded in 20th century American politics: with unemployment above 7.2%; consumer confidence at a low point (but slightly up), and with anemic GDP, a sitting incumbent gets reelected. Never happened before. Something on the order of an African-American junior Senator with less than 2 years experience in the Senate winning the presidency.
What I'll be looking for in the earlier part of the evening (for glimmers of hope for my bet):

On the East Coast:
Bucks County, PA
Northern VA
South FL (Miami-Dade, Broward, Palm Beach counties)
Southern NH

In the Midwest:
Cuyahoga county, OH
Milwaukee & Madison, WI (Dane, Waukesha, & Milwaukee counties)
Polk County, IA
Hennepin & Ramsey counties, MN

Basically, a significant Obama underperform in these places might mean a larger shift of states away from Obama than the "smart money" currently has it.
I don't think Obama will lose 4 states, and I don't think he will lose 10. Instead, it will be somewhere in the middle. One or two more states. But which ones?
Everyone is zeroed in on Ohio (which has gone with the winner every time since at least 1972), and rightly so. I am looking at Wisconsin (a typically Blue state) & Virginia (a Red state that went Blue once in 2008).

I don't have any confidence in the prediction I am going to make (and I admit this is only to make my bet work out - and it only works if the polling models are wrong, and if independents are abandoning Obama in droves):
Romney wins the raw vote 49.7-49.1 (meaning the winning margin is going to be less than a million votes); Romney wins electoral college: 285-253. Maybe. (Update: I just learned this prediction agrees with one put up by Glenn Beck on his radio show last week. I am quite ill.)(Even later update, Nov. 5: Karl Rove has come up with a slightly different map, same number. Now I am really nauseous.)
Obama loses 6 states (in order of certainty): Indiana, North Carolina, Florida, Colorado, Virginia, Ohio, Wisconsin.
Take out only Wisconsin, Romney still wins the electoral college. Take out only Ohio, and Obama wins the electoral college.
Hopefully, the networks won't fuck this up with mistaken calls. Could be a long night.
But far more likely, I'll be running to Massachusetts this week. The range I am seeing for Obama is anywhere between 234-332; Romney between 206-304. Looking at it that way, the middle would be a win for Obama: 283-255.

PS: my prediction for the national Jewish vote: Obama 72%-Romney 28% (underperforming 2008, but just by a bit)