Seems auspicious!

The site’s final forecast in 2016 gave Hillary a 71.4 percent chance of winning versus a 28.6 percent chance for Trump, which turned out to be good enough for 300+ electoral votes. Nate Silver often stresses that he gave Trump a better than one in four chance of victory four years ago because so many other forecasters put Clinton’s odds ludicrously high, in some cases at 99 percent despite how the polls tightened in swing states over the final two weeks. Those hubristic estimates gave data sites a bad name. Don’t blame me, Silver counters, reminding readers in his write-up of today’s new 2020 model that his model never counted Trump out.

Despite the similarity in the numbers, Trump’s 29 percent chance then is qualitatively different from his 29 percent chance now. Like I say, in November 2016 the polls narrowed at the end as undecideds began breaking for Trump, possibly encouraged by James Comey’s eleventh-hour revelation that he had reopened the Emailgate probe. FiveThirtyEight noticed the shift towards Trump in the numbers at the time and recalibrated accordingly. The 29 percent shot he enjoys right now isn’t really driven by the polls: Although the race has tightened a bit in the past few weeks, Biden still enjoys a national lead in the six- to seven-point range, which has the makings of a comfortable win. If the election were held today, I suspect Trump would have a less than one in 10 chance at victory.

But the election’s not being held today, is it? And in a year like 2020, when anything seems possible, we should allow for the chance that the country will endure another major shock or two, or 12, before November 3 that might reorient the race in Trump’s favor. Uncertainty in an uncertain age, especially over COVID, is one key component of his chances, says Silver. There’s also the fact that Americans are still able to pay their bills — for the moment — thanks to federal assistance, although now that that’s momentarily dried up, it’s another source of uncertainty. Trump’s also an incumbent, of course, which improves a candidate’s chances, and he polls better in swing states than he does nationally, which, per FiveThirtyEight, gives him about a 10 percent chance of once again losing the popular vote while winning the electoral college. (It’s hard to imagine any scenario in which Trump wins the popular vote, which makes me wonder how to square that with his 29 percent overall chance of winning.) “It’s way too soon to count Trump out,” Silver concludes, which is certainly true.

But it’s also too soon to count out a scenario in which Biden blows the roof off:

It’s important to remember that the uncertainty in our forecast runs in both directions. There’s the chance that Trump could come back — but there’s also the chance that things could get really out of hand for him. Our model thinks there’s a 19 percent chance that Biden will win Alaska, for example, and a 13 percent chance that he will win South Carolina. The model also gives Biden a 30 percent chance of a double-digit win in the popular vote, which would be the first time that happened since 1984…

Biden is in a reasonably strong position: Having a 70-ish percent chance of beating an incumbent in early August before any conventions or debates is far better than the position that most challengers find themselves in. And his chances will improve in our model if he maintains his current lead. But for the time being, the data does not justify substantially more confidence than that.

If you look at the model itself, you’ll find that there are very few scenarios in which Trump does much better than 300 electoral votes. If he wins a second term, odds are it’ll be by the skin of his teeth. Biden is different: There are various scenarios in which he cracks 400 electoral votes.

Interestingly, if you compare FiveThirtyEight’s 2016 model to the new one you’ll find that Hillary’s odds of winning over the course of the campaign surged higher than Biden’s ever have. She reached 89 percent after the Democratic convention, declined, then came all the way back to 88 percent in mid-October. Biden has never topped 79 percent despite his gaudy polling leads in June and July. The difference is that Hillary also had moments when her odds crashed, twice dipping below 55 percent and declining by more than 20 points over the second half of October. Biden, by contrast, has been rock-steady since June 1, never dipping below 69 percent. That’s a paradox of this election thus far: As insane as the uncertainty of day-to-day life has been, that uncertainty isn’t showing up in the election. “In fact,” says Silver, describing his model, “the uncertainty index points toward the overall uncertainty going into November being about average relative to past presidential campaigns.”

Imagine that. A pandemic, an economic collapse, anti-racism riots and protracted violence, and FiveThirtyEight estimates that this race is about as stable as any other presidential contest. That’s a testament to the extent to which voters’ opinions so far are a pure referendum on Trump, I think. And opinions about the president tend to be baked in, not something that people are still sussing out three and a half years into his term.

But maybe that’s about to change as the campaign grows more active and Biden becomes more visible? New numbers from the Democratic firm Change Research, via RCP:

Apart from Florida (which is a very notable exception), all of those numbers suggest a tighter race in key states than most pollsters saw a month ago. If I were Sleepy Joe, I wouldn’t feel at all comfortable about a four-point lead in any battleground given how pollsters overlooked the strength of Trump’s working-class support last time. It wouldn’t surprise me if Biden finally slips below that 69 percent floor he’s had in FiveThirtyEight’s model over the next week.

Exit question: Will he get a bounce from choosing Harris?