To be honest, I didn't expect the gap between my state-by-state simulations and the national contracts to continue to this extent for this long.
Prior to this point, it was completely plausible that the national contract was taking into account the possibility of some external event that would convince many people across the nation to vote for McCain, causing the state-by-state results to move together in a way that my simulations don't model. At this point, that seems pretty unlikely.
So why the continuing discrepancy? Well, it's possible that people trading state-by-state contracts just have different opinions/information/motivation than the (more numerous) people trading the national contracts. But that suggests the possibility of an arbitrage opportunity (albeit not a risk-free opportunity). Let's assume for argument's sake that the national numbers and the state-by-state numbers are in fact consistent.
In that case, the only explanation I can think of is that the market truly believes that there's about a 9% chance that there's something that could turn the election McCain's way that they haven't taken into account when pricing the state-by-state contracts. That something could be a massive Bradley effect or some set of legal challenges, for example. (An assassination of Obama wouldn't account for it, since there's not much spread between the Obama-wins-president and Democrat-wins-president contracts.)
But my guess is that it's just that some people want to bet on McCain because they want him to win rather than because they think it's a good investment, and that those people are betting on the national contract because they care more about McCain winning overall than they care about the outcome in any particular state.
What are your thoughts?
By the way, I'll be posting with significantly higher than average frequency tomorrow. It should be interesting to see if any "insider traders" who know something about early returns wind up moving the state-by-state prices. Stay tuned!
Posts
