I notice that the Oscar prediction season has started. I can understand why this is of no interest to many sane people, but I quite enjoy it. And I may enjoy it even more this year (though that will depend to a certain extent on the movies…) This is because last year, as I blogged, the excellent Kris Tapley told his podcast sparring-partner Anne Thompson that “There’s no way to Nate Silver this kind of thing” — and this year Nate Silver plans to Nate Silver not just this kind of thing, but the thing itself.
His track record is held by some to suggest that he won’t do very well. But it seems to me that the way to measure his predictions is not against the outcome per se, but against other people making predictions, such as those pooled together at the Gurus o’ Gold site. Last year the statistical model put together by Ben Zauzmer did better than half the gurus and not as well as the other half, though this was because he felt there was insufficient data to call some of the races: on the races he called, Ben did as well as one of the better gurus. I suspect that, with more experience, more resources and quite a strong incentive to shine Nate Silver may do better than Ben.
So my metaprediction is that, if Silver chooses to predict all of the races, or a large majority of them, he will beat most of the gurus, but not all of them; the best of the gurus will do better. My further prediction is that if he keeps it up over five years, no single human predictor will beat him continuously.
And while I am at it, I predict that the predictinator will predict that Gravity will win the special effects oscar — and that it will be right.
1 Comment so far
Leave a comment