A shot of optimism after a day of eeyorish state polls. How does Rove arrive at this result when eight of the last nine Ohio surveys have O ahead? In two steps: (1) He clearly trusts the national data over the state data, and (2) he cites historical numbers showing that incumbents recently have tended to overperform their national polling on election day by only one percent or so. Obama’s tied with Romney in the RCP national average tonight at 47.4. If, per the historical data, O’s ceiling is therefore at 48 percent, then it follows that most everyone else will break for Romney and that his national advantage will carry him to narrow wins in the states he needs.

As for Ohio:

Adrian Gray, who oversaw the Bush 2004 voter-contact operation and is now a policy analyst for a New York investment firm, makes the point that as of Tuesday, 530,813 Ohio Democrats had voted early or had requested or cast an absentee ballot. That’s down 181,275 from four years ago. But 448,357 Ohio Republicans had voted early or had requested or cast an absentee ballot, up 75,858 from the last presidential election.

That 257,133-vote swing almost wipes out Mr. Obama’s 2008 Ohio victory margin of 262,224. Since most observers expect Republicans to win Election Day turnout, these early vote numbers point toward a Romney victory in Ohio. They are also evidence that Scott Jennings, my former White House colleague and now Romney Ohio campaign director, was accurate when he told me that the Buckeye GOP effort is larger than the massive Bush 2004 get-out-the-vote operation.

Democrats explain away those numbers by saying that they are turning out new young Ohio voters. But I asked Kelly Nallen, the America Crossroads data maven, about this. She points out that there are 12,612 GOP “millennials” (voters aged 18-29) who’ve voted early compared with 9,501 Democratic millennials.

Are Democrats bringing out episodic voters who might not otherwise turn out? Not according to Ms. Nallen. She says that about 90% of each party’s early voters so far had also voted in three of the past four Ohio elections.

In other words, the dam that O’s built among early voters simply isn’t tall enough to hold back the red tide next Tuesday. One caveat to Rove’s point about the national numbers, though: According to RCP, Obama’s either tied or ahead in seven of the last 10 national polls taken. Romney still leads in Rasmussen and Gallup, and in only three of those 10 does Obama reach 49 percent or higher, but things have evened out a bit after Romney’s post-Denver debate surge. Case in point is the new Fox News poll tonight, which has the race dead even at 46 after Romney led by a point in early October. The topline number is not so good for Mitt, but the fundamentals are:

Independents give the edge to Romney by seven percentage points (46-39 percent). That’s down from a 12-point advantage in early October…

Among the subgroup of most interested voters, those who are “extremely” interested in the election, Romney leads Obama by 53-42 percent…

Romney’s supporters continue to be more enthusiastic: 69 percent say it’s extremely important he win, while 59 percent of those backing Obama feel that way.

Romney also leads on this metric, which will hopefully influence a lot of undecideds next week:

How is Obama even when the numbers look that rosy for his opponent? Partly because the partisan split has moved from D+1 in the last poll taken in October to D+5 in this one, which strains credulity as a prospect for election day. This is why it’s so hard to make guesses based on the polls right now — even some of the ones that are in sync, like the national polls showing O inching into a tie, have obvious weaknesses that may make the results questionable.

Rove’s other prediction, incidentally: At least 279 electoral votes, which jibes with Romney political director Rich Beeson telling reporters today that he thinks the campaign can win a few other midwestern states besides Ohio. Exit question: Second look at Karl Rove?