iPredict and Minor Parties
September 7th, 2011 at 2:00 pm by David FarrarI am a fan of iPredict. I trust it because I think the fact that people have to put money up to back their beliefs, means that the wisdom of the crowds often wins out. If you think a market is wrongly priced, then you can make some money buying or selling some stock.
However there is a flaw in the way a few stocks operate. Not the binary stocks (x will happen by z date) but the party vote stocks. These are used to predict the a party’s vote percentage at the election.
VOTE.2011.NAT is currently at 47.5c. It will pay out 1c per 1% party vote, so it is predicting a party vote of 47.5%. Now let us say I genuinely think National will get 50% and I have $100 to spend. I buy $100 of shares at 47.5c, or 210 shares. And if I am right and National gets 50% I get 50c per share so get $105 back. Now as an investor a 5% return on capital doesn’t get me very excited, so you may have few investors in that market.
Now let’s say I am the National Party Campaign Chairman, Steven Joyce. Let’s say I don’t want National’s share price to be 47.5c as that may not be enough. Let’s say Steven decide to buy up VOTE.2011.NAT shares to 55c so there is a story about how the market thinks National will get 55%. Steven has a bit of money so spends $1,000 buying 1818 shares at 55c. Whenever the price drops below 55c he buys shares until he has spent his $1,000. Now if National really only gets 47.5% then Steven loses 7.5c a share or $136. So a 14% drop on his money.
Now if other investors out there think National is over-priced then they can move the price below 55c by selling the share, which is equivalent to buying the reverse. What it means is they pay 45c for their “reverse” share and when the stock closes they get $1 less the party vote share or $1 – 47.5c or 52.5c. So they could sell to Steven his 1818 shares for $818 and if National gets 47.5% will make $954. That is $136 profit (made off Steven) and is a return on capital of 17%.
So for a major party like National the relative rate of returns are not far off – 14% for the person buyign at 55c and 17% for the person selling at 55c.
But now let us try this with say NZ First. You are the NZ First Campaign Chair and you want NZ First to show at over 5%. You have $1,000. You can buy 20,000 shares at 5c. Now if NZ First actually get 4% then you have only lost $200.
Now what say you are an investor who thinks the correct share price is 4c. In theory you can make money by selling at 5c and buying at 4c when it closes. But you have to effectively buy the reverse share at 95c and sell at 96c.
Now to buy 20,000 shares at 95c will cost you $19,000. And the payout will be $19,200. You will get a miserly 1.1% return on your money. Would you want to lock up $19,000 for a 1.1% return? No.
So what is the moral of the story? When it comes to the minor party stocks, it is possible for supporters to push the share price up at almost no cost to them, but at a great cost for others to push it down. This only applies to the party vote stocks, not to all the binary stocks.
This is why NZ First often has a stock price over 5c indicating Winston will be back, while the binary stock of “Will Winston get back into Parliament is at 22c, indicating only a 22% chance.
Likewise I suggest the stock showing United Future will get 2.6% is also the result of party activists pushing it higher. The same for ACT at 4.0%.
As we get very very close to the election, this problem may sort itself out. When there are only a few days to go, then even a 1% return may be decent enough for people to push the stock prices down to what the true level would be.
iPredict can also probably avoid this issue in future by having the cost of buying minor party stock at 10c/% point instead of 1c/% point.
September 7th, 2011 at 2:12 pm
The fact that someone who runs a polling company genuinely believes that poll results are a useful tool for shaping public opinion is very disturbing.
Vote:September 7th, 2011 at 2:15 pm
That sounds like a fairly good bet to me considering the amount of money involved.
Vote:September 7th, 2011 at 2:30 pm
I would love to see another stock open on whether each of the minor parties crosses the 5% threshold. Binary contract. The flat payoff curve problems apparent in the Vote Share markets wouldn’t be found in these ones.
Vote:September 7th, 2011 at 2:45 pm
You are a “fun” of iPredict ? You certainly have fun with it !
Vote:September 7th, 2011 at 2:46 pm
I’ve had my first look at it today. Fun is all it’s good for, it’s just a trading game, far too easy to manipulate with small change to be any value as a political poll.
Vote:September 7th, 2011 at 3:00 pm
Pete George, that’s pretty a flippant write-off of iPredict! If you did your homework you would know that iPredict and other prediction markets have been shown to be much more accurate than political polls.
You would also know that manipulating stock prices (say, pushing Labour up to 60%) provides increased incentive for other traders to adjust the market prices back to more rational levels, so any manipulation is short-lived and pointless.
Vote:September 7th, 2011 at 3:10 pm
Luke – they may be accurate by polling day, but this far out, particularly with low traded stocks, they are far too easily pushed. That makes the weekly report very questionable. I virtually doubled a party % with $20, if I managed to time that for the weekly report snapshot it would be a nonsense result, and someone could distort the opposite way too.
Savvy traders must have a field day with political poll pushers.
It’s not just iPredict, all political polls this far out mean little apart from large party trends. Otherwise they are story fodder – that may end up influencing voters.
Vote:September 7th, 2011 at 6:10 pm
“The weekly political snapshot is taken at a random time each week to avoid market manipulation by political parties or activists.”
https://www.ipredict.co.nz/app.php?do=blog&page=%3Fp%3D901
Vote:September 8th, 2011 at 9:50 am
The iPredict value is being higher than expected voter turnout doesn’t do anything for any party save one. They are either clear of the 5% or have a seat.
Then again it is high-tech stuff…
Vote:September 8th, 2011 at 10:04 am
Comparing current iPredict % to the DPF weighted averages shows quite a difference on some:
National 54.2 – 47.5
Act 1.6 – 4.1
NZF 2.6 – 4.7
The smaller the party the more the disparity:
A Harawira led party – 13.4%
Vote:Conservative – 8.8%
Citizen – 4.2%