An update on the election night arb...

The big night looms...

As with the last time I blogged about this, not exactly tech focused, just a bit of fun ahead of the Big Night tomorrow (I plan on staying up to watch the show... Just need to find some cooked pork products in the deli section to keep me company).

I wrote about the arbitrage between the two leading prediction markets, the Iowa Electronic Market and Intrade back at the end of September. Interestingly at the time you could make a $14.45 arbitrage profit trading the two.

The evolution of an arbitrage

You can see this on the charts below. The one on the left show's the actual price offered on the two candidates by the two markets (Obama quotes are the upper two lines, Romney quotes the lower two). The one on the right show's the spread between the two offered by each market:

As you can see, through August until about the end of September when I wrote about it, a noticeable difference opened up between the two candidates. It then closed right up in October, partly I think because the difference had started to attract some mainstream coverage (e.g. this article which came out before I'd posted my piece - although I'd come up with much the same thinking on my own,  FT Alphaville and PBS).

On 15th October when I updated the analysis the spread can come right in and actually there was a slight cost to the trade (around 1% annualised, which makes sense as a trading cost for a relative illiquid market).

Now as we roll into election day the spread has gone right out again. Here is the updated analysis - in theory (if you can get your mail check into the University of Iowa fast enough) the trade now yields a whopping $44.68 (9%) one-day return:

There are a couple of reasons for this. One is a technical one - the Intrade $4.99 monthly fee is payable on the 1st of the month, so if you sign up now and then close your account before 1st Dec you don't have to pay it.

The second one is more of a bummer - technically IEM pays out based on the winner of the popular vote and Intrade on the winner of the electoral college. While Obama retains a thin but material lead in battleground states the national vote is closer, raising the prospect of a 2000-style split verdict where he wins the college by not the popular vote. In which case of course the strategy isn't as risk-free as first appears! I suspect the risk around this unlikely but hard to value event is the main reason why the spread has gone up.

What's the chance of that?

Of course this leads us onto a different trade. Effectively Mr Market is now saying there is a roughly 9% chance of a split vote occurring (making the simplifying assumption of zero trading cost - in reality its probably a tad higher once you strip out trading costs). For context NYTimes pundit Nate Silver assigns that outcome a 7.6% probability in his influential FiveThirtyEight blog. Although I would add the safety warning that Silver's analysis is derived by analysing past trends (albeit with adjustments) - one of the great bugbears (IMHO) of stock analysis.

Now what's the chance of that? I guess we'll find out in around thirty hours time...

Anyhow no earth-shattering conclusions here. Just a cute little episode which illustrates some of the mechanics - and pitfalls of arbitrage, as well providing an example of how the market can struggle to value illiquid, esoteric assets. Which is probably how we got into this whole credit crunch mess in the first place!

Right I'm off to the shops now to hunt down some pork pies!