I’m not sure that this is in fact the final map, as there may still be state polls trickling in this evening. But here’s how things stand on the eve of the election, after 5 p.m. ET. Let me remind you: The RCP state averages called the winner of 49 out of 50 states correctly in 2012, missing only Florida. This map has it … Clinton 272, Trump 266. Flip any battleground state, even a small one like New Hampshire, and Trump wins.

Trump leads in Florida by a minuscule 0.2 percent as I write this; in New Hampshire, it’s Clinton by 0.6 percent, and that’s helped along by the dubious 11-point lead she had in the WMUR poll of the state that was released this morning. The long and short of this is that if literally one more poll of NH were to drop showing Trump ahead by a few points, the average would shift and that state would probably also favor him very narrowly. And in that case, he’d be at … 270 electoral votes. He’s that close to being a perilously slight favorite to win the election per RCP’s polling averages. And note that RCP’s miss on Florida in 2012 involved a larger margin than either FL or NH are experiencing now. That year, Romney finished 1.5 points ahead in the Florida average and lost the state on Election Day by less than a point. Trump would be president with a more modest miss than that in New Hampshire this year.

That said, there’s a question mark on this map. Observers of Nevada’s early voting like Jon Ralston swear that Dems have piled up enough ballots there to put the state all but out of reach for Trump, notwithstanding his lead in the polls. If that’s true then flipping NH on the map above doesn’t win it for him after all. We end up with Clinton winning 274/264. And it’s not obvious which remaining blue state Trump could flip to tilt the election back to him. He’s already won Florida, Ohio, North Carolina, and New Hampshire in this scenario. He’s close-ish in Michigan and Colorado, but they’re not as much of a coin flip as FL and NC are. Unless Ralston’s wrong about Nevada, even one of the rosiest maps for Trump still requires him to pull off one more major upset somewhere. And this assumes, naturally, that you take RCP’s polling average as definitive rather than another site’s. Each election modeler chooses to include and exclude certain polls for their own reasons. Some, like FiveThirtyEight, choose to weight polls according to how accurate the pollster’s been in the past so that some polls count more than others. FiveThirtyEight currently has Clinton ahead by more than three points in New Hampshire’s polling average, making that state more durable for her; that being so, Trump is stuck at 266 even with Florida and Nevada in his column. The Upshot also has Clinton by three in New Hampshire and gives Clinton a better than 80 percent chance of winning the election.

Still, you can see why Nate Silver was so defensive this past weekend when the left started whining at him that he’s lowballing Hillary’s chances of winning at 69 percent. With Trump even with Clinton in linchpin states like Florida and North Carolina and uncomfortably close in firewall states like Michigan and Colorado, this is, in fact, a close election. If you had to bet, you’d bet on Clinton because she’s led for most of the race, because Latino early turnout is encouraging for her, because there’s a chance of a “hidden Hillary vote” just as there’s a chance of a hidden Trump one, and because she’s expected to have a major ground-game advantage. But none of that matters if the white working class turns out in high enough numbers for Trump — and no one will have any idea if they will until the votes are counted tomorrow night.