In the days and months before the mid-term elections, I barely spent any time reading polls. Instead, I closely followed the betting market at tradesports.com/intrade.com, trusting it to aggregate the relevant information into a small list of prices easily interpretable as probabilities. I think this approach paid off. There were three very close Senate races last night: Virginia, Montana, and Missouri. The Republicans only needed to hold one to keep the Senate. In the first few hours after the polls closed in the East, the contract paying $100 if the GOP keeps control of the Senate was trading at around $85. In hindsight, this 85% probability made perfect sense. Given that there were three almost perfect toss-ups, one would expect a price around seven eighths (87.5%) of a dollar.

As the returns came in, things kept looking worse and worse for the GOP incumbents in the above three races. Over the course of half an hour, the price fell from 85 to 45. Ninety minutes later, it was in the low twenties. (Right now, it's still at 6%, waiting on the final count & recount from Virginia.) I got a kick out of watching the returns come in and instantly seeing the market's interpretation. It was far more efficient than scouring the web, TV, and radio for the comments of the precinct experts.

Alas, the markets don't tell me everything I need to know about politics. Things like "Congress will be more protectionist" (HT: Eric) are hard to directly translate into betting contracts. (Though there could be (but aren't, as far as I know) contracts on Trade Promotion Authority, the Doha round, increased tariffs against China...) And then there are exit polls...