Discussion about this post

User's avatar
Seth K's avatar

Have you looked at 538’s model?

Nate’s (and many others’) issue with 538 isn’t that it might be wrong.

Their issue is that it’s clearly broken. Or at the very least, in the early stages of beta testing with lots of bugs that haven’t been worked out.

Its predictions make no sense, some of which are mathematically impossible.

And its assumptions make little sense, strongly weighting historical state-level polling but without accounting for how the demographics of states change over time.

Not to mention the fact that state-level polling tends to be sparse, especially historically and especially in non swing states.

538 might end up being right...but that doesn’t mean it’s good.

Expand full comment
Steve Dougherty's avatar

Would love to see your forecast when it comes out, specifically what you define as the "fundamentals". In my admittedly amateur experience, there's no agreed upon set of fundamentals nor the appropriate sliding scale to weight them. And I totally agree with your view on polls. I have spent countless hours telling people that polls in July are about as accurate as an Imperial Stormtrooper in Star Wars.

Expand full comment
2 more comments...

No posts