My adventures on Intrade
For those who don’t follow the economics and politics literature obsessively, Intrade is a market in bets on various kinds of predictions, set up to follow the conventions of a share market. As I’ve discussed quite a few times in the past, the efficient financial markets hypothesis in its strong forms, implies that markets like this should give a better (more precisely, at least as good a) prediction of things like election outcomes than could be obtained from studying polls, pundit predictions and so on. I’ve been sceptical of this, on the basis of casual empiricism and some concerns about whether the empirical tests I’ve seen are biased in favor of the claim being tested.
One thing I haven’t done until now is to enter the actual market to see how it works. I finally signed up, and discovered a few items of interest. First, thanks (I assume) to US laws against online gambling, it’s quite difficult for Americans to participate in the market, which is, at least for legal purposes, based in Ireland. You can’t use a US credit or debit card, and my attempts at a wire transfer from my US bank account failed. Australia has no such restrictions.
Second, and relatedly, the market is quite thin. If the managers of Presidential campaigns cared what Intrade said, they could shift the markets a long way for a very modest outlay. For example, shares in Ron Paul, with a $10.00 payoff if we wins the Repub nomination, are currently trading at 0.27, implying a 2.7 per cent chance. But a Paul fan who wanted to raise his estimated chances could push them up to 0.40 for an outlay of $1000 (there are about 3000 shares for sale at prices between 0.27 and 0.40).
Third, there’s no margin trading, which means in particular, that you need a lot of collateral to go short on a long-odds candidate (at least if I have worked out the system right). Selling short costs $10 a share, less the current price, so if I wanted to sell short $100 worth of Paul shares at the current price (that is about 400 shares), I’d have to put up nearly $4000. I had an elegant Dutch book worked out, betting against Paul and Huntsman (zero chance, in my view) to finance a bet against my preferred dark horse whose odds were equal to the sum of the first two. But that didn’t it work, so I had to just put down my money. Over the fold, my trackside tip ….
Newt Gingrich: I bought my shares early this week, and am already ahead to the tune of $15.00. My reasoning is as follows. Even if he survives the harassment scandal (which looks less likely now than when I entered the market), Herman Cain isn’t a viable candidate. Perry and the other conservative candidates are already toast. So, Romney is the obvious favorite and is indeed odds-on to win.
But, as Eugene Robinson just pointed out, Romney isn’t as inevitable as centrist pundits think – on the contrary he’s quite “evitable”. There’s a large pool of conservative Repubs who would rather vote for (just about) anyone else*. Once Cain is out of the picture, Gingrich is the last choice left to them. And given their willingness to defend Cain, it will be that much easier to forgive Gingrich’s personal baggage, now long in the past.
A majority of Repubs will probably hold their noses and vote for Romney as long as he appears to be the only electable option. But if things turn up for Obama, or if the (entirely accurate) perception that Romney is the emptiest of empty suits, a haircut masquerading as a man, starts to take hold in the mainstream media narrative, that calculus might change. At that point, the conservative narrative might recast this election as a replay of 1964, a necessary defeat on the way to a greater victory.
There’s plenty of “ifs” in there, but I still think Gingrich was badly underpriced when I bought at 5.0 per cent, and still underpriced at 7.5 per cent. A quick look around the wingnut blogs (not an experience to be repeated often, so I won’t give links) suggests quite a few of them are thinking the same way.
Coming back to Intrade, what, if anything does all this prove. My guess is that, most of the time, betting markets display what’s called “semi-strong’ efficiency, that is, the market odds are about as good as you would get from a skilled analyst with access to published polls and the public predictions of other skilled analysts. By definition that has to be at least as good as any mechanical calculation based solely on poll numbers, so the markets should outperform the polls. On the other hand, the markets are thin and may be prone to panics or bubbles at times. And of course , if there were no polls, the markets and the pundits would be relying entirely on guesswork.
* The exceptions, for different reasons are Huntsman and Paul.