CNN  — 

The polls look awful for President Donald Trump. He’s down by about 10 points nationally and in almost all of the swing states to Democratic presidential nominee Joe Biden. Trump’s first debate performance was widely panned, and he needs a lot to go right the rest of the way.

Chances are Trump will lose and could lose by a wide margin.

Yet, with more than three weeks to go until Election Day, it’s important to note that Trump still has a non-trivial chance of winning.

I’ve written about this many times. A few weeks ago, I stated “Biden is ahead, but the race has been and will likely remain within the margin of error until Election Day.” That’s still the case today.

View Trump and Biden head-to-head polling

Why?

It’s not because Trump stands much of a chance in the popular vote. With a little less than a month to go, the national polling aggregate has differed from the actual result by an average of a little more than four points since 1936.

Even if you were to construct a 95% confidence interval for how the polls at this point have differed from the result, it’s about +/- 12 points. Biden’s advantage is just inside of that. It suggests there is only a roughly 1-in-20 chance that Trump wins the popular vote. In other words, it’s something that could occur, though is improbable.

Instead, Trump still has a decent chance because of the possibility of a popular vote/Electoral College split. We don’t know the true extent to which Trump has a better shot in the Electoral College than he does in the popular vote, but we know it exists.

One baseline is assuming Trump does about three points better in the state that determines the Electoral College winner than he does in the popular vote. That’s what occurred in 2016. This means in reality that Trump needs things to shift seven points in his direction nationally to win the Electoral College given he’s down 10 points nationally right now.

(A look at the state level polling generally confirms this rough estimate.)

It’s far from impossible that Trump closes the margin with Biden by seven points. It’s an event that occurs about 1-in-7 times, if historical trends hold.

Indeed, you can see that the chance Trump wins is more than negligible in statistical models such as those created by Jack Kersting, FiveThirtyEight and the Economist. They all point in a similar direction.

Trump has roughly a 1-in-7 to 1-in-11 chance of pulling off the victory in November, according to all these different models.

This may not seem like a lot, but it’s not nothing.

Pick up a simple six sided die that you probably have somewhere around your house. Trump’s chance of winning is nearly as good as you throwing the die and hitting a six on your first roll of the die.

To use an example closer to my heart, the Buffalo Bills had about a 1-in-6 chance of making the playoffs heading into the final weekend of the 2017 NFL season. By winning against Miami and Baltimore losing to Cincinnati, the Bills ended up in the playoffs for the first start since 1999.

In fact, we’ve already seen a 1-in-6 possibility become reality this presidential election season. Back in the primaries, Biden seemed like he was done for after losing the first three contests. He had about a 1-in-6 shot of winning the plurality of delegates.

Of course, we all know what happened after that. Biden won the South Carolina primary convincingly and was off and running to the nomination.

In the real world, the seemingly unlikely things happen all of the time.

The bottom line is that the most likely outcome is Biden winning and probably with some ease. But don’t be shocked though if Trump pulls another rabbit out of his hat.