Prediction Markets

I’m a big fan of online prediction markets — futures or bets on various useful propositions, particularly political ones. I particularly like Intrade, which pays this blog for accounts opened via click-throughs on its display here, but which is also IMO by far the best designed commercial prediction exchange. But there are other such exchanges and markets and I’m sure that their numbers will only grow with time.

Here’s a primer on prediction markets from Chris Masse. Chris has also produced a 2004 year-end evaluation of particular prediction markets that’s worth reading.

Manipulated Markets?

I’m becoming less confident about the accuracy of the online futures markets in predicting the outcome of the presidential race. Intrade’s Bush re-election market sold off into the debate, rallied during and sold off after. This behavior seems odd, and is similar to what the same market did during the previous presidential debate and the vice-presidential debate. (I didn’t follow the first presidential debate.)

It’s occurred to me that partisans might be trying to manipulate the markets (Glenn Reynolds and Tom Smith have similar ideas, as I suspect do other observers). While I have no evidence for this possibility, consider it on a cost/benefit basis. Political futures markets aren’t so deep that a trader, or consortium of traders, willing to spend a few hundred grand couldn’t bias them against Bush for days or weeks at a time. Given the increasing credibility of these markets, and given the high cost of conventional advertising, selling a big lot of Bush re-election futures might be a relatively inexpensive way to help Kerry. (Of course the bill would come due eventually, if Bush won, but in that case the inevitable market pop might not occur until shortly before the election, in which case money invested in holding down Bush’s numbers would have been well spent.)

I think that these markets are a great tool, and that they will eventually equal or surpass conventional opinion-polling in the degree to which they become accepted by the media and public for political prediction. But markets aren’t perfect. Someone told me that James Cramer on CNBC suggested that Democratic hedge-fund managers are buying oil futures in an attempt to boost oil prices and hurt Bush politically. I am skeptical about Cramer’s idea: oil has been rallying for a variety of reasons, it’s a huge market and manipulation of huge markets tends to be prohibitively costly for prospective manipulators. But if Cramer is saying it, it means other traders are thinking it. And if they think it about oil, what about the relatively illiquid elections markets? It seems likely that politically engaged traders have considered, if not actually attempted, trying to push those markets their way. I don’t know whether they have succeeded. Perhaps this would be a fruitful area for systematic research.

OTOH, it could be that the election really will be as close as the online markets suggest it will be. We’ll know soon enough.

(Previous related posts: here and here)

UPDATE: Lex reminds me that the election odds shown by online bookmakers have roughly paralleled the odds shown by Intrade and the Iowa Markets. His observation supports the close-election hypothesis.

UPDATE2: Don Luskin suspects manipulation.

Markets Aren’t Nice

Intrade’s contract on Bush’s reelection shortly ago experienced a sudden (and heart-stopping for Bush supporters) price drop from around 62/64 to around 50. When I first noticed it I thought some very bad news must have come out. However, there was no news on Drudge, and a glance at the contract’s intraday price chart showed a spike down, suggesting that someone had come in with a large sale in a thin lunch-time market. The price was already recovering when I started paying attention, and soon returned to around 60 where it is now. It looks like the seller either made a mistake, selected a bad time to initiate or liquidate a position, or was trying to set off stops into a resting bid — IOW, typical short-term market behavior with no long-term implication. Sure scared the crap out of me, though.

UPDATE: EconoPundit also wants to know what happened.

UPDATE2: As of around 6:00 PM CST the market has recovered to its pre-spike level, with size bids and offers around 62/64.

UPDATE3: EconoPundit attributes the spike down to a big seller trying (unsuccessfully) to punish those who bet on Bush.