Predicting political events: polling organizations vs. bookmakers

Factual Questions
21

Some players in the betting markets are going to be tuned into Nate Silver’s predictions. And Nate Silver might incorporate market predictions into his own analyses. Et cetera. Silver (and other pundits) may be exploring new territory in this bizarre election. Is there a clear answer to what market or pundit is the best predictor? Of course not. I got involved in this discussion only in response to wrong-headed comments that prediction markets are a particularly poor predictor.

One certainly sees today that stock markets and forex markets were caught totally off-guard. Yet there are plenty of SDMB threads saying, in effect, that almost no one can outperform the S&P 500 “predictor.”

But it’s absolutely silly to try to draw a conclusion form a single example! :smack: That would be like saying “Deere & Co. stock rose 5% 3 weeks ago; therefore the Efficient Market Hypothesis is false.”

I don’t know where you get this. I click your link and see Betfair offers to Lay 1000-to-1 on Christie with no takers.

(Ordinary bookies, rather than markets like Betfair, take action only one way. You won’t get “fair” odds on longshots from bookies — they need high “vigorish” on high payoff bets.)

22

I swear it had Cristie at 2% in the Predictwise column when I put it up yesterday morning.

Who has been saying “that prediction markets are a particularly poor predictor”??? No one that I’ve read. The position often taken is that they are superior predictors. And I am not convnced.

The stock markets caught totally off guard? What leads you to conclude that? They likely were ending up giving a slight lead to “remain” - based more on a good last poll than on the betting market assessment, and have had a fairly modest drop as a result as some risk was already baked in. S&P 500 currently off about 3%; FTSE 100 was down about the same. Not really exceptionally big drops.

Of course one election is not a comprehensive analysis. It is merely a point consistent or inconsistent with, supportive or not supportive of, a hypothesis: the hypothesis that betting markets are superior to polling data is not supported by this election. No more no less can be said.