Meeting with Creativity for breakfast was the highlight of a busy weekend spent attached to screens monitoring events in Turkey. On my way to Firehook I passed a storefront with a live display of positions with prices on various political events. It was a prediction market, a powerful forecasting tool that I had little hope of seeing after Intrade was shut down. We talked a lot about science and the difficulties of obtaining knowledge. I saw Feynman and Kahneman in Creativity's book collection. I kept him over two hours yet we never got around to talking about running. At one point my phone chirped and we watched a Periscope feed of people climbing on an abandoned tank on a street in Turkey. I got a coupon from the prediction market and later opened an account, geting a $15 match as promised. Now I have $50 to bet and grow. It is an educational site, with limits on volume, but wins and losses are real. They collect 10% of all profit. I can today buy shares at 66¢ that will pay a dollar (less fee) once we are delivered from a Trump White House. I take some comfort whenever I hear reassurance that the lesser of two evils will prevail in November, and believe this is what will happen ... but when I have to put my money down my confidence drops. Maybe I should just wait a bit longer and see what happens at the convention. But then, if things become more clear, the price will change to reflect the new understanding. I see why prediction markets are so effective at weeding out baseless opinionators and encouraging the well-informed to speak up.
Well, I lied. CLINTON.USPREZ16 is sold out, so I can't buy shares in the national election. So I bought 20 shares of "No" in Will North Korea test a hydrogen bomb before the end of 2016? at 88¢. This seems like a safe win, but I stand to lose $17.60 if wrong and gain only $2.40 if right. It's also about a third of my funds out of play until the end of the year.
I don't know anything about the betting markets right now, but I can't imagine betting on Clinton pays too much at this point. NYT has her odds of winning at 76%, and I think that is staggeringly low. If I were selling bets, I'd put it at around 90%. 76% is a bet I would take, but still won't pay out much.
Would a meta-poll analysis sway your opinion on that? I'd like T to be 1:10 chance of winning, but his support is what it is (within some margin of polling error), assuming he doesn't crash the plane between now and November.
No, because my problem is with the polls. Polls early in races always show a tighter race than what will inevitably happen. At this point in 2012 I think Romney "held" a slight lead, but Obama's reelection was never in doubt for a moment. Why? Because Romney was out of touch and had a difficult time speaking to anyone who made less than $1,000,000 per year. Trump is so far beyond what constitutes a reasonable man that once Clinton's media machine starts in on him, he will crash and burn so hard that he might actually bow out of the race before election day. You think 47% was bad? Binders full of women was bad? Trump can't be that kind and decent on his best day. I think he's in danger of losing states like GA, LA, AL, MO, and anywhere else there are large but apathetic African American populations. I wouldn't at all be surprised if he lost TX, to be honest (assuming that Dems can register and convince to cast ballots large numbers of Latinos). So, no, 538 isn't impressing me. And they've been wrong about Trump (in the opposite direction) for 12 months now. Trump is already dead in the water.
I have seen evidence that you are a better Bayesian than I, but if I understand correctly this means that if you had to bet, you would be indifferent given the choice of betting for Clinton, paying 10-to-1 against, and betting for Trump, paying 1-to-10. I think I would go for Trump at those odds.I'd put it at around 90%
Looks like the prediction markets are catching up...
Fifteen days ago I bought 70 shares of NO in Will national polling give Gary Johnson at least 8% in a 4-way race on August 31? I paid an average of 25¢ for the shares and could sell them now for about 90¢ each, as the authority puts his numbers at 7.8%. I was encouraged by superforecaster Bryan Caplan who predicts that Johnson will win no more than 5% of the popular vote. Caplan says his technique is to 'Step back, calm down, look at the numbers, and target thinkers who say, "This time it's different."'
Yes, to make it an attractive bet for me (on Trump), it would have to be 10:1. I would bet on Clinton at 1:3 and think it was a safe bet, if not a little boring for the low payout. Obviously betting at 1:10 is way more boring, but in the second part I was speaking as if I was a handicapper. Sorry for the confusion.
I just realized that superforecaster Bryan Caplan bet against Clinton nine months ago.