LogFAQs > #934386211

xp1337 posted...

do you have any examples to point to of 538 being egregiously wrong?

Remember that in 2016 they were mocked for giving Trump odds as high as they did such that they were the outlier. A 30% chance hitting isn't some awful miss. That's more tame than flipping a coin and getting heads twice in a row. You had models back then giving Clinton 99% (and I'll own up to having believed them at the time)

In 2018 the likeliest probability for House seats was a 39 Dem pickup. They got 40.

If you're judging on whether his likeliest options are coming through I think his model speaks for itself there that it is the best in the business.

And LotM is right how you should be evaluating his model but even by your own criteria I think he passes?

I defended the 538 model in 2016. I thought it would be wrong in 2018, and it was relatively right. I am saying that if it hits a 1% odd and the result is to say well that only happens once in a hundred times, the model is still right though. Instead of that the model was just not properly figured out then that is pointless as a predictor.

---

Xbox Live User Name - Corrik
Currently playing: Spider-Man (PS4), Quantum Break (X1)