H/T to Greg Mankiw for the analysis.

The GOP markets at Tradesport only have McCain and Giuiliani, so I don’t think that they are very meaningful. But Mankiw points out what is interesting about the Democrat ones:

[W]e can obtain for each candidate the conditional probability–the probability that the person will win the general election if nominated. Here are the results:Clinton 48.3
Edwards 57.3

Note the relative performance of Hillary Clinton and John Edwards. Although Clinton is more likely to end up President than Edwards is, Edwards is more likely to win the general election conditional on being nominated. At least that’s what the market says.

I think that jives with most people’s understandings. I also think, looking at some of the other stuff on there, that there’s some room for arbitrage…