I'm in Love!!!!

with 538. I've been curious about how they are managing to agree with Intrade on their Electoral College map, and spent some time readthing through their methodology. It's very impressive and about as good as you can get with objective math. There are a lot of factors in figuring out things like the average error rate of pollsters (by state) and it seemed liek that would be a pretty subjective number. Nope.

My process is to look at the average miss for each pollster across each contest they polled, and compare it to the average miss of other pollsters in those same contest, after going through a more-complicated-than-it-needs-to-be iterative process.The results are below, split into groups for 'regional' and 'national' pollsters. (This distinction is arbitrary -- some pollsters like Insider Advantage and Quinnipiac straddle the line between being regional and national -- but helpful for presentation). 'Error' represents the average error for the particular pollster, as compared to the 'IAE', which is the iterated average error for other pollsters in those same contests.

It's pretty good. I'd like to say that it's what I'd do if I had the time and motivation, but that would be a lie. It's much better than what I could have done. While sites like Intrade and 538 aren't actually predictive (they give you a very clear picture of what the chances are at a particular moment, but that's very different than telling you what's actually going to happen), I really like the way 538 has tried to use objective math everywhere.

Well done!