Stop the presses: Data error!

Intrade has the contract on Obama winning California at 3 on Sep 11. But it was around 91 the day before and the day after. Changing that 3 to 91 has a huge effect on the covariance matrix.

Corrected results show Obama with a mean of 300EV (roughly the same as where he is leading acording to intrade markets, today). The markets are quite stable though. Here is a histogram of the standard deviation in returns per state market.



Alaska is the outlier, which went from leaning to Republican to Republican lock with the pick of Palin. The Alaska market is settling in on a price around 6 on the Obama contract now, and might have a much smaller standard deviation going forward.

The majority of the markets have a standard deviation of daily price movement under 4%. That is percent change in price (50 going to 52) not absolute change in price (50 going to 54).

So because of his lead and the stability of the markets, intrade state markets can really be interpreted as giving Obama a 90% chance of victory as of today.

Here's a chart of how many EV Obama has been implicitly predicted to win, each day for the last 90 days. Despite the pure Obama as president contract being close to 50, sometimes above sometimes below, the state markets have consistently been predicting him as the victor.




The trends are similar. But if you believe future volatility will be like past volatility, the Obama for President contract is selling at a huge discount. Obama-for-President traders may foresee much more volatility upcoming and be discounting the state results heavily. The recent introduction of options on these contracts should allow us to dig in deeper into what forward volatilities the market is implying.

Possibly, I am wrong in thinking of these contracts like normal assets and analyzing standard deviation of price. Because clearly markets with low prices have big percentage change standard deviations and markets with high prices have small percentage change standard deviations.
Ultimately, they seem similar to standard assets because you can buy the contract for 50 one day and sell it for 52 the next, and you have made a 4% profit.

But simulating off returns in terms of their percent standard deviations does introduce a bias for high prices to remain high and low prices to possibly creep up. I think this is the straw that will break the camel's back in terms of getting my lazy butt to analyze the 51 "Republican wins state x's presidential election" contracts in addition to the 51 "Democrat wins state x's presidential election" contracts.