Rena or arbitrage?

National's chances of winning the 2011 election peaked around 95% just prior to the Rena crash. Since then, they've dropped to 92%. Matthew Hooton's chalked some of this up to Rena. I wonder if that's it though.

For the last couple of weeks, CentreBet has had National at 1.15: a $1 bet on National paid out at $1.15, which is about $0.87 in iPredict terms. At the same time, National was trading around 95 cents. And I started arbitraging a bit - clearing some of my large long-National short-Labour position at iPredict and putting some big orders in the book to do more of the same while throwing a few hundred onto National at CentreBet. Apparently a few other folks started reckoning National a cheap deal at CentreBet: HomePaddock points to some big money hitting CentreBet on National: they note one $10k bet - far far more than I'd played.

So what's the price at CentreBet now? $1.09. Or, $0.917 in iPredict terms. What's the price at iPredict? $0.9264, but with only about $20 of action in the book between the current price and the CentreBet price. So it's barely worth making the arbitrage play given iPredict's transaction fee for price-takers.

Arbitrage between iPredict and CentreBet can easily account for the entire price drop in National at iPredict.

The price in the Vote Share market has dropped too, though, and this is one that can't be directly arbitraged. National's plummeted there from about 50% just prior to Rena to about 46% now; Labour's risen from 28 to 30; the Greens have risen from just under 10% to just over 11%. The other parties have been somewhat stable: ACT around 3.6%, NZ First around 4.5%, United Future on 1.9%, the Maori Party on 1.25%, and all other parties (includes Mana, Conservatives, and anything else) at 2.2%.

But I'm not entirely confident in the prices in the vote share market. A couple reasons why:

• The price on NZ First is really spikey: somebody keeps coming in and boosting the price up above 5%, but the price on a contract paying $1 if Winston Peters enters Parliament has been below 18% for weeks, barring a few oddball spikes. If NZ First's likelihood of crossing the 5% threshold were accurately reflected in price moves in the vote share market, then we'd see a lot more action around the 50% line in the Peters contract. But we haven't. Why? Presumably because the trader trying to inflate the price of the Peters contract can afford to sustain manipulation on a penny stock where folks taking the other side of the trade would have to lock up tons of liquidity for trivial gains for another month but can't do it where traders taking the other side can earn a reasonable return by smacking him back. Any arbitrage bot working to keep the sum of contract prices at the $1 mark will trade down the big contracts in response to goofing around on the penny stocks.
• If National really were likely to earn 46% of the vote and Labour plus Green were really at 41%, and ACT were only 58.5% likely to return to Parliament, and UF were only 79% likely to return to Parliament, I'm not sure that National would be trading up around 92 cents. But I'll need to update the spreadsheet I'd worked out for the 2008 election and have a play; a lot will depend on how much overhang is generated in the Maori seats this time 'round. Absent substantial overhang, wasted vote from the minors likely doesn't tip the balance between National and Labour/Green.