Biden, by contrast, has a healthy margin. Four days before the election, he’s leading Trump in the FiveThirtyEight poll average in all seven states. He’s ahead by 5 percentage points in Pennsylvania and by 8 to 9 in Wisconsin and Michigan. That’s well above Clinton, who didn’t have a five-point lead in any of those states. Suppose what happened in 2016 happens again: In the final days—due to late deciders, polling error, or the emergence of “shy” Trump voters—Trump gains as much ground against Biden as he did against Clinton. That’s 3.8 percentage points in Michigan, 4.1 in Pennsylvania, and 5 in Wisconsin. Guess what? Biden still wins all three.

But suppose Biden loses one. In that case, he’s likely to make it up from the bonus states. In Arizona, Florida, and Georgia, Trump’s margin against Clinton on the Friday before the election, as measured by the FiveThirtyEight “now-cast,” almost exactly matched the final result. No wave of “shy” voters materialized; no polls were suddenly “unskewed.” In those states, Biden now leads by 3.1 percent, 2.2 percent, and 1.7 percent, respectively. If he wins any of them, that’s insurance against a loss in the core states.

Biden isn’t just leading in more states. Across the board, his vote share is higher than Clinton’s was. On the Friday before the election, Clinton was averaging 48.8 percent in the core states and 46.3 percent in the bonus states. Biden is averaging 51.2 percent in the core states and 48.8 percent in the bonus states. To beat Clinton in Michigan, Pennsylvania, and Wisconsin, all Trump had to do was pick up undecided voters. That wouldn’t suffice against Biden, since he’s above 50 in each of those states.