Saturday, 23 July 2011

Probabilistic Peters

Recall that under New Zealand's MMP system, a party enters Parliament on winning a single electorate or on passing the 5% threshold on the popular vote.

So New Zealand First re-enters Parliament if they get more than 5%. Winston Peters is party leader and rather likely first on their party list. If they get 5%, they get around 6 MPs. So if Peters is in the top six on their list, he's in if NZ First gets 5%.

iPredict says NZ First is likely to get 5% of the vote: trading's hovered on or just under 5% for the last month.

But iPredict says Winston Peters has a 22% chance of re-entering Parliament.

Potential explanations:
  1. The Vote Share market is wrong: thin, flat payoff curve
  2. The Peters to Re-Enter market is wrong (unlikely: thicker, binary contract)
  3. Peters likely to step down in favour of Michael Laws or someone else, declining to pursue re-election
Full disclosure: I'm 144 shares long on Peters to re-enter Parliament, but bought entirely on price swings that seemed odd; I paid an average of $0.0653 for those contracts. I'm also 90 shares short on Vote.2011.NZF, but again I was just running against what seemed to be high prices at $0.0518.

Because of the chances of #3, above, I'm pretty wary of viewing the two markets as a potential arbitrage play. Anybody else have insights here?


  1. I made the same play, but everytime I short the vote share down to about 4.8% or so it bounces back up. Somebody is buying the vote share contracts at just under 5%. Is the market really that thin?

  2. There's no return to be had in getting NZ First down from 5% to 3% if you believed that to be the true price; you'd have to have way too much capital locked up for way too long for a <3% return.

  3. Potential explanation 4: People are not rational (economic beings).

    Potntial explanation 5: There is uncertainty about the future (a point Keynes made).

  4. There is no contradiction. It is perfectly possible that there could be a 22% chance of the NZ first vote being greater than 5% (Peters contract), but that the expectation of the NZ first vote is 5% (vote share contract). You just need a probability distribution with expectation greater than median, something like the log normal distribution would probably be appropriate in this case.

  5. @Andrew: You just need a few rational arbitrageurs to solve your first one, and your second applies across all contracts; they other contracts have reasonable prices.

    @Alfred: You then need a fattish long right tail above 5%, right? SO that if they break 5%, they're likely to hit 8 or somesuch? I just don't buy it.

  6. Is there a derivatives market in iPredict? :-)

  7. lets get real. New Zealand is in trouble.
    Our currency and our economy are bankrupt care of successive Nat and LAB Governments.
    Peters advocates that we place our New Zealand first, ahead of international politic.
    Is that radical or what.