This prediction assumes Romney lead in the popular vote from 0-2%. That would mean a swing of at least 7.5% compared to the previous election, but more likely around 8.5%.
The swing would be reflected mostly in the swing states, and to some degree in Obama states, less in Romney states.

North Carolina is a very inflexible state with very few persuadables/undecideds so with a swing of 8% it should go to Romney with a margin of 5-8% (accounting for decreased white turnout).

Florida has demographics which are particularly difficult for Obama, such as a large proportion of older people, Cuban origin Hispanics and Jews which are more national security conscious. However, other groups of Hispanics are on the rise. A uniform swing would show a Romney victory by at least 5, but due to demographic trends negative for Romney my prediction is Romney by 2%.

Virginia will see a reduced turnout in Obama areas by at least 10% and increased turnout in McCain areas by the same margin, as early voting indicates. Based on the data available it is logical to predict a Romney uniform 8% swing which will give him the state by 1-2%.

Colorado early voting numbers are good for Romney showing a 3% advantage for Republicans. The Republican margin is larger than the 2004 early voting margin. But Hispanics are more numerous in Colorado and they tend to register as Independents giving Obama an edge with Independents. Colorado will be very close and I predict a Romney victory by 1%, maybe even less.

New Hampshire will have a swing toward Romney mirroring the national swing, but will also have two additional elements: additional Romney advantage due to home state status and additional Obama disadvantage due to his incumbency. Final verdict: Romney by 2%

Wisconsin has a very powerful and organized Republican Party led by Governer Scott Walker, a veteran of many election and recall campaigns in the previous two years. Their turnout machine is effective and the Republican base is further energized by selection of Paul Ryan as VP on Romney's ticket. Past elections in a polarized environment which depended on turnout gave close results. This one will not be an exception. Romney by 1-2%.

Early voting yields an advantage to Obama in Ohio and the advantage is approximately 100000 votes. If the Independents break in Obama's direction as some polls indicate, the advantage will be even greater. Let's assume the advantage is 130000 before the Election Day. That would require Romney to at least match or even surpass the results accomplished by Bush in 2004. It is possible but unlikely. Obama by 1-2%.

Iowa early votes place Iowa squarely in Obama column. Obama by at least 2%.

Nevada early votes leave little room for questioning the outcome. Obama by at least 4%.