Who will govern after 2014?
October 26th, 2011 at 12:30 pm by David FarrariPredict has just launched stocks on which party’s leader will be Prime Minister after the 2014 election – National or Labour?
These stocks do not assume the result of the 2011 election. If people say National after 2014, then that could either be a 3rd term for National or it could be Labour win in 2011 and then National come back in 2014.
Currently the stocks sit at 69% for National and 32% for Labour, so National are seen at this stage to be twice as likely to win in 2014, than Labour.
What will be fascinating to see if how these stocks change over time, and especially how the actual 2011 election results affects these 2014 stocks.
When you consider the cost of capital, the market seems quite optimistic that National will govern after the 2014 election. Say you have $69 to invest. If you stuck it in a term deposit for three years at 6% then in three years you would have $82. If you stick it in iPredict and National wins in 2014, you would have $100. So in reality the market is saying they think National has around a 82% chance of winning in 2014. That seems a bit over-priced for me, but I amsure the price will vary a lot over the next three years.
Related posts:
- iPredict Election Update #44: Nats could govern alone
- iPredict election stocks
- iPredict on 2011 outcome
- iPredict on Labour Leader and PM
October 26th, 2011 at 1:41 pm
I am afeared that with the demise of nearly all the minor parties this election Greens and Labour will unite to a GreenLabour Party under David Shearer and Russell Norman (Finance) in 2014. MMP will be strenghtened so the left never lose power.
Vote:And I am serious.
October 26th, 2011 at 1:54 pm
I’m no statistician so maybe I’m not getting your numbers, but wouldn’t your logic also lead to the following conclusion:
Say you have $32 to invest. If you stuck it in a term deposit for three years at 6% then in three years you would have $38. If you stick it in iPredict and Labour wins in 2014, you would have $100. So in reality the market is saying they think Labour has around a 38% chance of winning in 2014.
———————————————————
Doesn’t that line of thinking just inflate both numbers?
Vote:October 26th, 2011 at 2:06 pm
@ speters. Yes, you are right. Under DPF’s logic, National has an 82% chance of winning in 2014 and Labour has a 38% chance. Which, of course, makes no sense. The reason DPF’s logic does not hold in this case is that betting on a political market is not a strict financial investment. This is for two reasons: (1) a person betting might be wagering not on the outcome but on price movements in the interim. DPF has said he himself sometimes bets this way; and (2) People betting are sometimes more interested in the pure gratification of being prescient and awesome and less concerned with the few dollars. It is more accurate to read the percentages as they stand, not once filtered in the way DPF has presented them.
Vote:October 26th, 2011 at 2:16 pm
Also worth noting that so far there have been only 41 shares traded, probably between many fewer than 41 people. I’m not at all ready to believe this as a true revelation of market sentiment on an N of 41.
Vote:October 26th, 2011 at 3:33 pm
I’m not sure that we want to run time discounting that strictly. Yes, discounting will distort prices a bit. But if folks aren’t expecting to hold their contracts to expiration but rather are expecting to on-sell after prices adjust, then we’re not looking at a three year discounting window but rather a couple of months.
That said, you can wind up accumulating a really large position pretty easily and having a hard time moving it. I have WAY too much exposure to PM.National.
On PM.National.2014 and PM.Labour.2014, I moved about 20 units getting the prices from 50/50 to 60/40, not expecting to hold ’till 2014, but expecting rather to clear the position once prices hit 70/30 or so. And by the sounds of it, it’s time to put in my orders to do that. But the grading again beckons.
Vote:October 26th, 2011 at 4:21 pm
David’s analysis is correct where taking a position is a one shot deal, but the ability to close out a position having taken one softens the effect of the length of the contract. The ability to close out a position is how the conundrum pointed out by Rob Salmond is resolved: iPredict contracts cannot sensibly predict 120% of the outcomes! Because traders can buy today and not have to wait three years to realise their return, discount rates are less important. A trader only has to believe a contract is mispriced and that market sentiment will back him up should he buy or sell to effect a correction. Provided it does, he can profitably close out his position tomorrow or next week. So the ability to buy today and sell tomorrow, or short today and cover tomorrow, allows contract prices to be accurate even on long contracts and even when discount rates would not allow trading in one shot deals.
That said, a contract’s length plainly matters. We find that length of contract is a stronger driver of trading on a contract: for each day added to the length of a contract, 1.5 shares fewer per day will be traded at the margin (average = 807 shares traded/contract/day). Expected length does damp trading. Whether that lower trading translates to reduced accuracy is another matter, but it can’t help.
Even if trading on iPredict was a one shot deal, like at the TAB, we’d expect some trading to occur. Imagine two traders who agree Party X has 50% chance of winning the 2014 election, and each has discounts the future at 6% per annum. These folks will never trade: the most a buyer will be prepared to pay for their expected 50 cent payout in 3 years is 50/(1.06^3) = 41.98 cents. The most a seller will be prepared to accept is 50*(1.06^3) = 59.55 cents. So might trading occur at all on a 3 year stock were it a one shot deal? One reason is if there is a wide divergence of opinion. In the example above, two traders with 6% discount rates will trade if the seller thinks Party X is no more than 41.98% likely to win and buyer thinks Party X has no less than a 59.55% chance of winning. Traders with low discount rates will also find it easier to reach acceptable trades.
What does this say about the believability of the forecast? First, the fact that traders do not have to wait three years to realise earnings on a three year contract means David’s discounting analysis does not apply. 69 cents really can be understood to mean a 69% likelihood, though the measure may be noisier. Second, longer contracts do get less trading compared to shorter contracts. If this reflects less attention and a smaller pool of folks thinking about the question, then this may translate to lower accuracy that is additional to the instrinsic uncertainty of more distant events, but the effect seems weak: long term contracts do have a habit of looking about right in spite of the lower trading volumes they attract. We’ll be able to test accuracy against polling after the election and see who converged faster to the actual result. Third, liquidity constraints seem to exert influence on contracts priced near $0 and near $1 which have more than a couple of months to run. It is far cheaper to buy a share at 1 cent than to short it, and upwards bias on very low priced contracts with more than a couple of months to run has been observed on iPredict.
Vote:October 26th, 2011 at 5:00 pm
If you’ve got Matt and Eric and me all making the same argument about why this kind of time discounting doesn’t make much sense, DPF, then it might be time to stop.
Vote: