iPredict

Temperatures update

2009-02-23T16:00:00.003+13:00

As I've noted a couple of times, the prices on TEMP.2009.HIGH stock are very high when compared to any reasonable forecasting model. In previous posts, I used a variety of methods to forecast the 2009 temperature anomaly based only on the time series. A model using only a nonlinear time trend (year and year squared) and lagged temperature variables explains about 84% of the variation in the time series. Based on that model, it looked like there was no more than a 5% chance that 2009 would be warm enough to beat 1998.

We now have the global average temperature anomaly for January. Let's add it to the model and see what comes up.

The first way I did it was to use my model to predict January temperatures. That model provides a much worse fit than the one predicting annual temperatures, but that's to be expected. The model's predicted January temperature is spot-on: the model predicted a temperature anomaly of 0.37 and that's what we got. We would have needed an actual January temperature much higher than predicted to start me worrying about my call on this contract. Instead, I'm right on the money.

The next approach is to add January temperatures in to help predict average annual temperatures. I updated the forecasting model I previously was using to have January temperatures as an explanatory variable. I then took the regression residuals and checked how many times in the data series we've had residuals large enough that, if we had that bad a model fit this year, 2009 would beat 1998. It's happened 6 times in the 154 year series. That's 3.8%. The best fit model says we have a 3.8% chance that 2009 is warmer than 1998. Last time, we had a 7/154 chance. So, the addition of the January data makes it LESS likely that we'll beat 1998. Why is that, given that January temps are right on target? The addition of January data to the model tightens up the precision of the estimates!

Long story short, the January data suggests that the chances of 2009 beating 1998 are even lower than I'd previously thought. A fair price on this contract is $0.038.

Full disclosure: I have a very large short position on this contract, based largely on analyses like those conducted and linked to here. I'm also slightly long on the TEMP.2009 contract as my model suggests 2009 will be warmer than 2008 (74% chance), just not warm enough to beat 1998. If you want to check my work and try it for yourself, get a copy of Stata. Get the data series and import it. Add the obvious column headings. Then do the following:

gen year2 = year^2
tsset year
reg temp year year2 L1.temp L4.temp jan
predict temp_hat
predict resid, residuals
sum resid if resid > (0.543-0.3867909)

The numbers are the 1998 anomaly and the predicted 2009 anomaly. The last line will then tell you the number of times that the model has been so far out that it generated a residual sufficiently large that, if encountered again, 2009 will be warmer than 1998.

3.8%. I don't know who or what is keeping the prices up at $0.18. Current price is five times higher than it ought to be based on fundamentals.

14 New Stocks - Interest, Anyone?

2009-01-28T17:55:00.002+13:00

14 news stocks are now underway, all relating to interest rates and what they'll be doing over the next six months.

For the first time, iPredict is pointing at Australia and asking you to consider what the Reserve Bank of Australia will do with its cash rate when the RBA board next meets on 3 February, next Tuesday. Yes, a short sharp set of stocks this one, but we expect some healthy trading.

In another first, iPredict is also asking you to forecast retail mortgage rates for home owners by the middle of this year. How low will variable rate mortgages be in June? Below 7%? Below 6% below 5.5%? This should be a stock of interest to home owners considering whether to break their agreements with their banks and moving to a lower interest rate.

Finally, we launch another set of OCR stocks, this time for the RBNZ's 12 March annoucement. Of course, we'll be keeping an eye on these stocks as tomorrow's OCR news comes through.

iPredict Forum is Underway

2009-01-23T17:51:00.004+13:00

iPredict's forum is underway. Here's the link. Well done Aaron and Simon (our developers) for pulling this together and fighting the monster that is IE6.

If you have any trouble, please tell us by posting into the bug reports section of the forum. If you're having so much trouble that you can't see the forum, well just post it here :-)

91 Unleaded Stocks Launched

2009-01-23T17:29:00.004+13:00

Today sees the launch of our first petrol price stocks, aimed at forecasting the price of 91 unleaded petrol at the pump.

Five binary stocks are designed to measure this, each asking will the price of 91 unleaded be within the following ranges:

91.FEB09.VLOW: price less and 132 cents
91.FEB09.LOW: price between 132 and 139.9 cents
91.FEB09.MID: price between 140 and 148 cents
91.FEB09.HIGH: price between 148.1 and 156 cents
91.FEB09.VHIGH: price above 156 cents

How do we measure price? We use the Ministry of Economic Development's Regular Petrol Price series, available from here. Although the Ministry does not make it clear on the page, this series is measuring the GST inclusive, all other taxes-inclusive price of 91 unleaded petrol in the Wellington region. I had to call the Ministry to figure that out.

We have also launched a new bundle called BUNDLE.91.FEB09 which combines all five stocks and sells for $1.

Climate stocks

2009-01-22T13:39:00.003+13:00

Since folks seem to be discounting at 100% the previous post on why the TEMP.2009.HIGH stock seems massively overvalued, I started worrying that traders out there know something I don't. So, I threw the data into STATA instead and ran a few prediction models with lag effects and a time trend. The best fit model predicts a 2009 temperature anomaly of 0.38821, which is higher than I previously was estimating. But, the probability of exceeding 1998 seems largely unaffected because my confidence bands are tighter. The standard deviation of the regression residuals is 0.100832. The residual tells you the difference between the actual observation and the estimate predicted by the model for any particular year. So, if the actual temperature anomaly turns out to be 0.546, the residual would be -0.15779: the prediction minus the observation gives you that as residual. How often do we observe residuals at least that large on the down side? 7 times in 154 observations: 4.5% of the time.

The simple model explains 84% of the variance in the time series using only lagged dependent variables and a time trend.

Want to replicate this? Get the dataset, add in a null 2009 observation, throw it into STATA, type:

gen year2 = year^2
reg deviation L1.deviation L4.deviation year year2
predict y_hat
predict resid, residuals

Then just take (y_hat - 0.546) for 2009, and see how often the residual is smaller than that value:

sum resid if resid < -0.15778

7 observations in 154.

I wonder what the folks buying at 0.22 know that I don't. No way anybody's using this for hedging against a hot year, not at these stakes....

* Note I kept the first and fourth lags after a general to specific reduction starting with 7 lags.

Request for Your Feedback on New Features

2009-01-21T17:09:00.007+13:00

Heaps of great feedback in earlier posts on new features ideas. This morning we compiled a list of new features that we've been bouncing around for a while. Part of the list we came up with is below, much of it is from ideas given to us by you. We've left out some administrative features ideas and ideas only for our internal prediction market customers.

Please do two things in the comments.

One, list your priorities starting with most preferred e.g. F,D,E,B,Q. Prioritise as many as you like, I'll compile your opinions on ideas for however many as you are willing to give.

Two, add new suggestions. I apologise in advance for all the features already submitted in the last few months not on the list - the list was prepared in a rush, please submit them again, I'll add them.

I will update the list with your new ideas. I may reverse the list's order so that newest ideas are put at the top.

For option D, what minimum, if any, is appropriate? $20?

Some of these may require clarification - if so please ask.

Here is what we have so far, in no particular order.

A. Forum

B. Continued API development

C. Confirm trade button disabled after one click

D. Option to display trader earnings by percentage (minimum deposit $20?) + make named appearance on list optional

E. Funds frozen following a withdrawal request + option to cancel request if not processed

F. User's cash, portfolio value and net worth plotted over time

G. User-added events/news on stock details, editable/removable/flaggable by other users via wiki, added items plotted on stock price chart

H. User stock board – lists all stocks in one place, current price/volume/day's high,low/lifetime high,low/high buy,low sell. Ajaxed and sortable.

I. Daily stock high/low prices added to stock details

J. Multiple stock trading system - allows buying and selling of multiple stocks at one time from single page – facilitates arbitraging

K. Stock details page widget which converts price to prediction for that stock taking into account any special rules (e.g. price floors) for that stock + remove probability from browse stock page and replace with "Avg. Cost" or "Rev/Stock Received At Sale" values if position taken, otherwise blank

L. Home page public notification – alerts to users arriving on the home page, warning of approaching downtime for maintenance, for example, or new stocks launch time

M. User alerts system - user can get emails/text messages sent if prices move outside defined bands

N. Basic look/feel site adjustments (colours, some page design changes)

O. Re-work trading interfaces - reduce steps, make clearer, ajax-enable

P. Re-work short selling to make clearer to user what funds have been transferred and why

Q. Post-trade suggested stocks "Users who bought this stock also traded …" displayed on the trade confirmed screen

R. Browse closed stocks page

S. Stocks vary between $0 and $max, where $max can be <> $1.

T. Connect forum and stock details pages - allow forum browsing (for posts relating to that stock only) and posting from stock details

U. Download (anonymous) raw trading data interface (currently available via API only)

V. Add Google news items to stock details pages using stock-specific search terms. Plot news item flags on stock price line.

W. One-click trading interface: buy and sell price can be clicked for a standard parcel size (eg 25 contracts) subject to there being at least the std number available at that price (if not the lesser number)

X. Either streamline browse stocks to load only selected categories or set up a browse stocks for mobile/PDA only

Y. Remove the division into 'page 1, page 2' of stocks in the lists on My Portfolio, e.g. the Watch list

Z. Show current bid and offer prices in the portfolio lists

Climate stocks

2009-01-19T15:28:00.007+13:00

At current prices, the markets are forecasting a 60% chance that 2009 temperatures will be above 2008 temperatures, with a 24% chance that 2009 temperatures will beat the 1998 high of a 0.546 degree anomaly. The 2008 anomaly was 0.324, so traders must reckon 60% of the temperature probability distribution lies above 0.324 and 24% lies above 0.546.

If you go to the underlying data series, which goes back to 1850, we find a mean change from year to year of close to zero (because of course it's anomaly relative to an average baseline) with a standard deviation of the year-on-year change of 0.115. So, in any given year, the next year's anomaly will fall within +/- 0.115 of the prior year's anomaly about 68% of the time. For the 2009 anomaly to be greater than 0.546 would require an increase equivalent to at least 1.93 standard deviations of the year-on-year change. A jump that large shouldn't happen more than about 2.5% of the time in a normally distributed series. You could try arguing that the variance of the series has increased, but the standard deviation of year on year changes over the last decade is almost identical to that of the whole 159 year series.

You might also well say that I ought to account for the warming trend. Let's try that. From 1980 through 2008, the average change in the anomaly has been 0.009483. We can add that to the 2008 temperature to get a prediction for 2009. The 2009 observation would need to be 1.83 SD above the expected temperature increase to beat the 1998 anomaly. We would expect that to happen 3.36% of the time.

I've included a graph with the underlying temperature series and bands showing changes 1.83 standard deviations above and below the prior year's observation. How often in the series' 158 year* history do we see the subsequent year's anomaly outside the 1.83SD band? 14 times. Or, 8.8% of the time, with 4.4% above the band and 4.4% below the band. That's a bit more than we'd expect from the standard normal distribution (6.7% of observations, 3.35% each side), but not a ton more. I've marked these with big red dots: 1863, 1865, 1877, 1879, 1890, 1916, 1930, 1954, 1957, 1964, 1974, 1977, 1997 and 1999. Those are years with temperatures that varied by more than 1.83 standard deviations of the normal year-on-year change from the prior year's observation. You'll probably need to click on the graph to enlarge it to see things properly.

I've been shorting this stock since it launched. I plan to continue shorting this stock. I'm not saying that 2009 can't be warmer than 1998, I just can't see how it's more than 20% likely to happen. I can't see that it's more than 10% likely to happen either. The analysis above should give an upper bound estimate of the likelihood of 2009 exceeding 1998: for a midpoint estimate, I'd not account for a warming trend and instead evaluate at the mean zero change. At what price would I start covering my shorts? Well, I think a fair price wouldn't be higher than $0.05, and I'm a bit risk averse.

Of course, feel free to follow the links provided in the stock description to get the underlying data and have your own play with it. I'm not doing anything high tech here.

*Yes, the series is 159 years long, but you can't look at the year-ahead for the most recent year. That's the one we're trading on!

Note: Post updated to provide analysis based on 1.83SD cutpoints (the upperbound case) rather than the midpoint case provided earlier.

iPredict Forum in Testing

2009-01-16T17:40:00.003+13:00

The iPredict forum went into testing today. We're aiming for a live site launch next week.

Newstalk ZB on Sunday

2009-01-16T17:38:00.003+13:00

I'll be on Newstalk ZB this Sunday morning at 9:20 talking to Garry Ward about iPredict.

Rare Tie in QV Stocks

2009-01-16T17:19:00.005+13:00

The QV stocks, asking where in the country residential house prices would fall most, closed today in a rare tie between two binary stocks.

According to QV, residential property prices fell by 8% in both Auckland and Christchurch in the year ended December 2008.

So, for the first time, we had to convert the binary stocks to indexed, then judge Auckland and Christchurch stocks at $0.5000 each. Wellington closed at $0.

These QV contracts are a nice example of what makes prediction markets interesting - at least from my perspective. That perspective is from somebody who doesn't have even the first clue about property values. Running a prediction market contract gives me, and anybody in a similarly ignorant position, to find out from people who actually know about residential property, the best possible view about what is most likely to happen.

These contracts were a first for another reason, which is that they were sponsored by our friends at SuburbWatch. Check them out - if you're interested in timing the market and getting in at the bottom, this site can help.

iPredict on Nightline

2009-01-09T16:10:00.002+13:00

TV3 have just visited our incredibly untidy office to film a piece for Nightline on our climate change contracts. Fingers crossed, we'll be on tonight's show at 10:40.

19 New Contracts Launched

2009-01-09T11:49:00.004+13:00

iPredict launches 19 new contracts this morning covering politics, the economy, and our first foray into science with two contracts on climate change.

The full set of contracts launched today are:

FIJI.PM: Frank Bainimarama to lose position as Interim Prime Minister of Fiji in 2009
ZIM.MUGABE: Robert Mugabe to lose position as President of Zimbabwe in 2009
MP.CLARK: Helen Clark to leave Parliament in 2009
MP.CULLEN: Michael Cullen to leave Parliament in 2009
MP.ANDERTON: Jim Anderton to announce in 2009 he will not stand in next election
LEAD.GOFF.09: Phil Goff to be replaced as Labour leader in 2009
DL.KING.09: Annette King to be replaced as Labour deputy leader in 2009
NAT.MAORI.09: Maori/National agreement to be terminated in 2009
NAT.ACT.09: ACT/National agreement to be terminated in 2009
UNEM.DEC08: What will the unemployment rate for the December 2008 quarter be?
UNEM.MAR09: What will the unemployment rate for the March 2009 quarter be?
UNEM.JUN09: What will the unemployment rate for the June 2009 quarter be?
UNEM.SEP09: What will the unemployment rate for the September 2009 quarter be?
OCR.09.SUB45: OCR to go below 4.5% in 2009
OCR.09.SUB40: OCR to go below 4.0% in 2009
OCR.09.SUB35: OCR to go below 3.5% in 2009
OCR.09.SUB30: OCR to go below 3.0% in 2009
TEMP.2009: Global temperatures to be warmer in 2009 than 2008
TEMP.2009.HIGH: Global temperatures in 2009 to be highest ever

All but four of the contracts are binary. This is the result of demand: over 95% of all trading to date on iPredict has been on binary stocks.

Probably the newest aspect to these stocks is the introduction of two questions on climate change. 2009 is shaping to be an interesting year for climate science. There are, as I understand it, two camps in the climate change community. One says that greenhouse gases is a major driver of changes in climate. The second says changes in the Sun's energy output is responsible. What makes 2009 interesting is that these drivers are expected to head in opposite directions - greenhouse gas concentrations will continue their inevitable march upwards, but the Sun's energy is expected to continue falling. So which driver will temperatures tend to follow in 09?

We have more contracts in the works on petrol prices, mortgage rates and the weather, and possibly commodity prices.

GM.BAILOUT closed

2009-01-09T11:47:00.002+13:00

GM.BAILOUT closed at $1 this morning. The US Treasury has loaned GM $4 billion and has committed to another loan of $5.4 billion on January 16.

Thanks for your patience on this contract. Onward and upwards!

Fourth Quarter Negative Growth? Ummm, Yes

2008-12-23T15:30:00.004+13:00

Well it seems New Zealand is in for a fourth consecutive quarter of negative growth in GDP, at least according to early trading on our new recession stock. Our RECESSION.DEC08 contract launched at 70 cents about 15 minutes ago, and has just shot past 90.

Earlier today, Statistics NZ announced growth of -0.4% in production-based GDP and -0.7% growth in expenditure-based GDP in the September quarter.

It is hard to find any good news for the New Zealand economy anywhere in today's numbers, and based on that 90 cents, its hard to see any good news coming soon.

Merry Christmas!

New Features Launched

2008-12-23T15:23:00.004+13:00

A big update to the site this week which saw had tons going on under the bonnet and some much-requested new features:

Remember me: checking this box when you next sign in means iPredict will remember you on this machine for two weeks. iPredict will un-remember you on a machine if you sign out and on all machines if you reset your password. If you don't click remember me, you'll be logged out after four hours of inactivity.

Edit active orders: from your my Portfolio page, you can now edit your active orders rather than cancel and re-set.

Active trader count: a live count of all users who have been active within the last 30 minutes.

iPredict API: iPredict's new API lets applications talk to iPredict without visiting the site. Join our API development community where you can learn about the iPredict API and contribute directly to its development.

New Resources page: this is an updated Web Tools page. You can now embed zoom-able flash charts as well as the traditional jpg-based charts, and there is a link to our API development community.

My Portfolio menu widget: This is a menu item which shows your portfolio value and cash balance, and when you hover over it expands to show the stocks in your portfolio. Clicking on this locks it open.

Our focus now is on getting some new stocks launched. Thanks for all the recent new stock ideas, keep them coming, we'll be using the holiday break to write contracts for launch in the new year.

We will be open for trading throughout the holiday period. Please report any issues via our Contact Us page. The next funds withdrawal run will occur on Tuesday 30 December and then Monday 5 January.

Have a great break.

GM.BAILOUT Contract Clarification 15/12/08 5:15pm

2008-12-15T17:08:00.006+13:00

iPredict makes the following clarification to the GM.BAILOUT contract:

Funds from the Federal Reserve System will be considered funds from the US Government.

This is for two reasons. First, the Federal Reserve is an independent agency of the US government, as is the CIA, the CFTC, and EPA and the Securities and Exchange Commission source

Second, the Federal Reserve's bailout of American International Group (AIG) in September and October 2008 is widely understood to be a bailout by the US government source, source, source, source

GM.BAILOUT Contract Clarification 10/12/08 11:30pm

2008-12-10T23:12:00.004+13:00

iPredict adds the following clarification to the GM.BAILOUT contract:

Funds, or a portion of funds, GM is ENTITLED to receive from the US Government before 1 January 2010 will be considered ‘approved’ and contribute to the US$5 billion minimum. The date GM draws, or intends to draw, upon those funds is irrelevant. Funds, or a portion of funds, which remain subject to conditions being met (whether those are in GM’s control or not) will not be considered approved until those conditions are met or certain to be met.

The basis for this decision is as follows. In the present contract, the relevant wording is:

[A] bail out will be deemed to have occurred if the US Government approves a transfer of at least US$5 billion to General Motors.

[O]nly transfers planned to be made before 1 January 2010 contribute to the US$5 billion minimum.

The contract specifies approval, not the actual movement of funds, as a condition for contributing to the US$5 billion minimum. This clarification is based on that position.

The purpose is to have all funds that GM is entitled to receive with certainty prior to 1 January 2010 count towards the $5 billion minimum, and to not count funds which depend on some non-trivial conditions being met.

Note also that the contract specifies that it “is unaffected by decisions made by his successor's administration.” Any announcement by a representative of the incoming Obama administration to cancel funds to GM approved during the term of the Bush Administration has no effect on this contract.

In response to the question:

If Congress approves a maximum loan of $7bn for the period through to 31 Mar but GM only draw down the $4bn they need this month by 1 Jan, is the "transfer" under this contract deemed to be $7bn or $4bn?

If GM is entitled to receive US$7 billion before 1 January 2010 and no part of that $7 billion is subject to conditions that may not be met (whether those are in GM’s control or not), then the whole US$7 billion will contribute to this contract’s $5 billion minimum. The date GM draws, or intends to draw, upon those funds is irrelevant.

GM.BAILOUT Contract Clarification 9/12/08 12:00am

2008-12-08T23:27:00.014+13:00

For the contract GM.BAILOUT, iPredict considers it important to clarify its position regarding any loan or loans made to General Motors under the Advanced Technology Vehicles Manufacturing Loan Program (ATVMLP). The GM.BAILOUT contract includes the condition that a transfer of at least US$5 billion to General Motors is required for the contract to be judged true and pay $1.

For the reasons outlined below, iPredict considers ATVMLP funds WERE NOT committed to GM by the US Government prior to the contract's launch on 14 November. Accordingly, any loan or loans made to GM under the ATVMLP WILL contribute to the $5 billion minimum in the GM.BAILOUT contract.

Some background on this decision is relevant.

On 30 September 2008 President Bush signed into law the Consolidated Security, Disaster Assistance, and Continuing Appropriations Act 2009 (Pub. L. No. 110-329; "Continuing Resolution 2009") source in which Congress set aside $7.5 billion as a guarantee on loans to eligible parties totalling no more than $25 billion. Loans will be funded by the Department of Treasury and administered by the Department Of Energy. source

Car makers must apply for loans under the ATVMLP, and to qualify for a loan a car maker must meet certain criteria. Criteria include being financially viable and having improved the fuel efficiency of their light duty vehicles since 2005. Applications for the first tranche of loans are due by December 31, 2008. Additional tranches will be due every 90 days thereafter as long as authority remains to make additional loans source, source, source.

Although the Continuing Resolution 2009 funded a loan guarantee, which on its own would disqualify this as a transfer to GM (clarification 4) unless GM defaults prior to the contract's close (clarification 7), loan principle will be funded by the Department of the Treasury, a department of the US Government.

iPredict has judged ATVMLP loan(s) to be included for the following reasons:

a) ATVMLP loans come with eligibility criteria and GM's eligibility under these criteria was uncertain at this contract's launch;

b) this contract's wording refers to "funds committed to GM". For the purposes of this contract iPredict considers no commitment was made to any one company on that date and that conditions required to satisfy "funds committed to GM" was not met; and

c) if GM is judged eligible for loans under ATVMLP then this will result in a new transfer to GM from the US Government.

ATVMLP loan(s) must be approved (but not necessarily paid) by the Department of Energy or any other US Government agency prior to this contract's close to contribute to the $5 billion minimum.

In the post "GM.BAILOUT Contract Definition Clarifications" on this blog dated 16 November, updated on 19 November, the author wrote: "Specifically, a decision to relax restrictions on a US$25 billion loan previously made to GM will NOT contribute to the US$5 billion minimum transfer total, because this results in no new transfers to GM." This comment was made assuming a loan had already been made to GM and at question was whether a change in this loan's purpose would affect the GM.BAILOUT contract. It continues to be the case that relaxation of conditions on committed to GM by the US Government prior to the contract's launch on 14 November will not contribute to the $5 billion minimum transfer total. However, as explained here it is iPredict's position that any loan made to GM under the ATVMLP was not committed to GM by the US Government prior to the contract's launch on 14 November.

New Stocks: OCR January 29

2008-12-04T11:42:00.005+13:00

Yesterday saw the launch of five new contracts asking what the Reserve Bank will do in its 29 January OCR announcement. Previously we have launched new OCR contracts after the previous one, but this time we went ahead of today's announcement to see what effect today's announcement would have on expectations for January 29.

The theory was that a big drop today would quell xpectations of another large drop next month. So did it?

The strongest response was for a 75 point drop on 29 January, this stock settled 11 cents higher than just before 9am today.

The next strongest ressponse was on the 100 point stock, this was about level with 75 points prior to the announcement but fell from 18 cents to 12 cents in the two hours after today's OCR announcement.

Other stocks were more sedate. Stock for a 50 point stock increased, but much more gradually than the 75 point stock, settling around 3 cents higher at this time. A 50 point drop is the market's pick for the 29 January OCR announcement.

Stock for a 25 point drop (not shown on the chart) bounced downwards but ended exactly where it started at about 9%.

You can, by the way, produce your own charts for any active or closed stock using our web tools page. The chart above was produced using this url call. The bottom of the web tools page lists the full set of arguments available to you - you'll need the end and count arguments to see closed stock data, here for example.

New Stocks: Ferry Strikes and Building Consents

2008-11-28T17:51:00.003+13:00

Two new sets of stocks tonight.

The first group asks about building consents. Earlier today, Statistics NZ announced residential building consents were at their lowest in October since 1992. What's in store for the next four months?

We have launched four stocks aimed at answering this question: BC.NOV08, BC.DEC08, BC.JAN09 and BC.FEB09. These contracts pay $0.01 for each 100 building consents for residential dwelling units granted.

We are using Statistics NZ's Building Consents issued report, usually issued towards the end of month following the period in question.

The other stock launched tonight, thanks to a suggestion by commenter (and trader?) Colonel, asks whether the Interislander will be hit by industrial action during the Christmas/New Year period. The Interislander is a division of Kiwirail which until July this year was owned by Toll NZ and now owned by the NZ government. Early trading is interesting - the contract opened at $0.10 and bounced all the way up to $0.28 before adjusting back to around $0.15. A contract to keep your eye on if you're planning a vacation on the other side of Cook Strait these holidays.

We're aiming to keep rolling out new contracts every few days for time leading up to Christmas. Please keep in mind we are obliged by our authorisation to limit the scope of stocks being offered to those political, business, economic, sociological, or demographic events, and we continue to welcome your ideas.

Matt Nolan Picks 75 Points on OCR

2008-11-28T06:31:00.003+13:00

Matt Nolan is picking a 75 basis point cut in the OCR on December 4, and with OCR.75.4DEC trading at around 4.5 cents, he's going to do very well if he's right.

Matt's reasoning:

All that has changed since October 23 is lower oil prices, a lower exchange rate, and the govt promising a larger fiscal stimulus. This is what the RBNZ said would limit any movement in the OCR in December:
“The reduction in domestic spending will be partly offset by the depreciation of the New Zealand dollar over the past few months, falling oil prices and the recent loosening of fiscal policy”

New Stocks on iPredict: Inflation, Exchange Rates

2008-11-26T19:37:00.004+13:00

Seven new stocks just launched on iPredict tonight:

NZ Exchange Rate - 3 Stocks

NZD.US.JAN09, NZD.US.FEB09, NZD.US.MAR09

Definition: What will the New Zealand dollar be worth against the
greenback in early January/February/March 2009?

Stock Type: Indexed

Initial Price (all three): $0.5480

Inflation - 4 Stocks

INFL.DEC08, INFL.MAR09, INFL.JUN09, INFL.SEP09

Definition: What will December 2008/March 2009/June 2009/September 2009 inflation be?

Stock Type: Indexed

Initial Prices: $0.50/$0.47/$0.44/$0.41

We'll be launching some slightly less serious contracts this Friday - stay tuned.

Your Feedback: What Stocks for iPredict?

2008-11-24T19:56:00.005+13:00

iPredict closed nine contracts this morning, and with the cupboard looking temporarily bare we thought it might be useful to let you know about contract suggestions before us, for a possible launch in the next few days.

In addition to core stocks for iPredict – political leadership and economics statistics, we have received many good and great ideas for forecasting. Here is a summary of requests, most of them, from users. What do you like? If you have more, please send them to us at stockidea@ipredict.co.nz

Our goal – as the market operator – is to run contracts which say something of consequence about the future. Traders' goals are, of course, different – the goal is to run contracts that satisfy both objectives.

Please note that iPredict is not authorised to run sports markets.

Update 28/11: iPredict is also not authorised to run entertainment contracts. We are authorised to run contracts that depend on "political, business, economic, sociological, or demographic events or circumstances".

Politics

National/Maori/Act/UnitedFuture longevity

Helen Clark appointments

Economics

Emissions Trading Scheme – what will carbon trade for?

Will oil be below US$40 a barrel before 1 Jan 2009?

Retail gas price to exceed NZ$5 on 1 Jan 2010, 1 Jan 2011, 1 Jan 2012

Mortgage rates

New Zealand visitor arrivals

Australian OCR-equivalent

Listed company profits

Fonterra payout

Bluff oyster prices

Social

Transmission Gully – construction to begin before 2011?

Cook Strait ferries on strike at Christmas/New year

Tolls implemented on Waterview Motorway

Science

Weather - rainfall totals?

SETI detects alien intelligence before 2010-2012

International

Mugabe to lose job in 2009

Other

The number 1 trader as reported on iPredicts rankings page at 5pm on 31 Dec 2008 will have a net worth at least $4,000.

About the iPredict Market Maker

2008-11-20T07:55:00.007+13:00

When you look at a stock on the iPredict public site, you may have noticed the regular nature of some buy and sell orders apparent in the order book, which you can see in the Advanced trading interface or on the home page of any stock (provided you are logged in). Many orders are for the exact same number of stocks. You may have even noticed regularity in the order prices.
That is the iPredict market maker (arrowed instances of this in the image above). The market maker seeds the market with offers to buy and sell, so that the first trader into the market after it opens has somebody to trade with immediately. The purpose of the market maker is to constantly seed the market with buy and sell orders ready for human traders to come in and fill. The effect of this seeding is to increase the liquidity of the market. Higher liquidity means better prediction.

The market maker is effectively another trader, albeit a hyperactive one. Each contract on your prediction market has its own market maker, and each contract's market maker is explicitly funded. You give the market maker a budget, and when that budget expires the market maker will stop making orders.

What rules does the market maker use to seed the market?

When a stock is created, the market maker is set up at the same time. The market is told how densely to seed the market. Offers to buy and sell are priced along an S-curve that varies between $0 and $1. Establishing the market requires the administrator to decide how steep the S-curve is, and how densely offers will be made along the S-curve.

Setting up the market in effect puts "notches" in the S-curve. These notches are the places where all offers to buy and sell and placed by the market maker for as long as its current settings persist. The administrator can change settings mid-steam – but as long as the settings remain unchanged these 'notches' are where all offers to buy and sell are made.
When the market is established, the administrator defines an equilibrium starting price. The market maker will then seed the market with offers to buy and sell. All notches below the equilibrium price are seeded with buy offers, and all notches above the equilibrium price are seeded with sell offers. All offers have the same number of stocks in them, generally we set this to 25 stocks.

The first trader into the market may decide to accept one of the orders on offer, or she could make her own offer for other traders to fill. Let's say she accepts a buy offer and sells to the market maker, and the entire order is filled, leaving a notch on the S-curve vacant.

The market then fills that empty notch with a new sell order for 25 stocks. The rule the market maker follows is to fill empty notches with new orders on the other side of the double auction. Offers by human traders occupying the same price as a notch are ignored by the market maker – the market maker will fill an empty notch of its own regardless of offers by other traders.

The market maker loses money on average – it will seed the market with offers without foresight and it will not react to news, unlike human traders. If the market maker runs out of money then it will stop seeding on the side of the market where it must make further outlays – that is, if money runs out then the market maker will stop making further buy offers on stock it is long on, and it will stop making further sell orders on stock it has a short position on. that can produce price instability in markets when other traders are not making buy and sell orders that are not immediately filled.

Our decisions on market maker settings – the steepness of the S-curve, and the number of stocks offered at each notch – is guided by two goals, one to produce a liquid market, and two to minimise the risk that the market maker runs out of funds.

Genesis Energy Looking For A Spot Trader

2008-11-17T15:18:00.005+13:00

If you've done well on iPredict, you might consider going pro - as it were - by joining Genesis as a trader on the electricity spot market. They're advertising a position right now.

I have to say I have not seen the spot market in action, but I'm told the principles are precisely those behind iPredict.

If you think trading for a living might be your thing, you can apply for a position as spot trader here:

http://careers.genesisenergy.co.nz/jobcentre/applycod.asp?job=464

Good luck!

GM.BAILOUT Contract Definition Clarifications Updated 19/11

2008-11-16T13:43:00.004+13:00

Four clarifications to our GM.BAILOUT contract:

1. This contract will pay $1 only if President Bush approves a bail out. The contract is unaffected by decisions made by his successor's administration.

2. $1 billion = $1,000,000,000

3. A decision by the US Government to take over some (or all) of GM's obligations to its creditors will be considered a transfer to GM. If the value of this transfer is unclear, iPredict will use the average of the maximum estimated values reported on cnn.com, news.bbc.co.uk and nytimes.com.

4. A loan guarantee by the US Government will not count as a transfer to GM.

These clarifications have been added to the contract definition on the site.

Clarification 19/11/08:

5. For the purposes of this contract, funds committed to GM by the US Government prior to the contract's launch on 14 November do not contribute to the $5 billion minimum transfer total.

6. Relaxation of conditions on funds committed to GM by the US Government prior to the contract's launch on 14 November will NOT contribute to the $5 billion minimum transfer total, because this results in no new transfers to GM. Specifically, a decision to relax restrictions on a US$25 billion loan previously made to GM will NOT contribute to the US$5 billion minimum transfer total, because this results in no new transfers to GM.

7. If GM defaults on a loan, and the date of that default is after the launch of this contract, and that loan was guaranteed by the US Government, then the value of the additional liability the US Government bears as a result of default WILL contribute to the $5 billion minimum transfer total, even if that guarantee was made prior to the launch of this contract. If the value of this transfer is unclear, iPredict will use the average of the maximum estimated values reported on cnn.com, news.bbc.co.uk and nytimes.com.

8. A bail out that is formally and fully approved by any means during the term of the Bush Administration will be considered for the purposes of this contract to have been approved by President Bush.

National Official Results Bundle

2008-11-13T08:14:00.003+13:00

We've just launched another official election results stock, OFFICIAL.NAT.OTH, which completes the set of all possible National seat outcomes from official election results and allows us to launch a bundle, BUNDLE.NATOFFICIAL.

This bundle contains OFFICIAL.NAT.NC, OFFICIAL.NAT.L1, OFFICIAL.NAT.L2, and OFFICIAL.NAT.OTH and covers every possible official election results outcome

PM Stock Closing

2008-11-09T16:51:00.004+13:00

We are very aware that the election result appears decisive and we are as eager as you to close the PM stocks and pay out its value.

However, it is important that they be closed once a result is beyond any doubt, and right now it is not.

There are two sources of uncertainty at this time. The first is the 220,000 special votes, which could remove as many as two seats from National in the next few days. Special votes do not tend to lift all boats equally, and have historically favoured the Green Party in particular. I believe special votes will be counted in the few few days. For an explanation of how National could lose up to two seats from special votes, see here.

The second source of uncertainty is coalition negotiations. It is possible a coalition partner will make outrageous demands that jeopardise coalition building.

It is likely these uncertainties will be resolved by the end of next week - although nothing is certain in MMP. Until then we will not close the PM stocks. Obviously, there is not enough uncertainty to prevent Helen Clark and Michael Cullen from stepping down - but the worst nightmare for a prediction market is to pay out on an event that does not in fact go on to happen.

If you wish close out your position ahead of the stock closing, you may do so at any time on the open market.

Thank you for your patience.

New Stocks on iPredict

2008-11-09T16:41:00.002+13:00

The election is over and iPredict has many new contracts. In the last 24 hours we have launched 17 stocks asking:

Who will replace Helen Clark as leader of Labour? Goff, Cunliffe, Mallard, Other
Who will replace Michael Cullen as deputy leader of Labour? Cunliffe, Street, Jones, King, Goff, Other
By how much will the Reserve Bank cut interest rates in their 4 December OCR announcement? 100 points, 75 points, 50 points, other.
Will the Electoral Finance Act be repealed by May?
Will the economy's negative growth continue in the September quarter?

We will also be launching new PM stocks asking who will win in 2011 in the next few days.

Election morning roundup

2008-11-08T08:49:00.004+13:00

Hi all,

Some interesting trading late last night the effects of which haven't yet entirely unwound. See the prior post for details. The upshot of these trades is that PM.National is lower than it was yesterday, PM.Labour is higher, and MP.Peters is higher. I rather suspect that a strong partisan came in, buying as much PM.Labour as possible against the order book.

Now, if we run the underlying prices through the spreadsheet, PM.National should still be up around 90%. One thing to keep in mind if you're using the spreadsheet is that it assumes independence between the Act-National-UF seat share and the probability of the Maori Party supporting that coalition. Obviously, if the former coalition looks very likely to get a majority on its own, then the likelihood of no agreement with the Maori Party goes up as the coalition doesn't need it. So, the spreadsheet understates the true value of PM.National. If you want to correct for that, replace the prices in Maori.Nat.Supp and .Abst with the prices that were dominant when the probability of a National win was more like 0.7: they then summed up to about 60%.

Many of you may now be facing the liquidity-constrained trader's dilemma: whether to freeze more margin and take more exposure in stocks to which you're already heavily exposed, figuring that the prices are out by about 10%, or whether to seed the order book with bids and asks far away from current prices in case another strong partisan comes into the market and wants to throw money away. Spend too much now, and you've no margin left to cover seeded orders later. Dilemmas!

As for me, I'd invested nowhere near $1000 in the market to start with, so I had room to deposit a couple hundred in cash balances (that's why I've been rising in the rankings lately, not due to any recent brilliant profit-taking) and have room to throw a fair bit more in should really tasty profit opportunities arise. I just really don't like my current exposure in PM.National, PM.Labour and MP.Peters.

Evening roundup

2008-11-07T23:27:00.002+13:00

The thing I like best about Matt's shiny new graphs is that the price spikes from heavily partisan trading become very apparent. Have a look at PM.Labour. At 10:46 tonight, somebody spent several hundred dollars (at a first guess) running the price up from $0.14 to about $0.30. It's since dropped back to $0.1978 but has a ways to go yet before getting back to its ex ante price. Market response to the spike was near instantaneous, with traders making the obvious arbitrage play shorting both National and Labour, and then buying back the bundle. Prices in PM.National dropped as consequence, but not as far as PM.Labour dropped.

About the same time as the PM.Labour price run-up, somebody had a similar go at MP.Peters: it ran up from $0.16 to $0.21 and now is down to $0.20.

I had a few asks sitting in the MP.Peters order book, knowing this had happened a few times with MP.Peters. I didn't expect it to happen in PM.Labour though.

Again, there were no price moves in the complementary markets that would have signaled anything other than a partisan trade.

Recommendation: seed the order book with bids and asks well outside of the current trading range to take advantage of folks who come in and want to spend several hundred dollars buying the stock of their preferred party and who want to do it in a hurry. There's some risk in doing this: if new information comes into the market suggesting that there really ought to be a shift in price level, you'll take a hammering. But if you're watching the market reasonably closely, you can guard somewhat against that.

Happy trading.

New Charts On iPredict

2008-11-07T21:15:00.005+13:00

iPredict has new charts on stock pages.

I love them.

But what do you think? Feedback, good or bad, is most welcome. We weill keep developing them over the coming weeks.

If you find any problems with them please report to us asap letting us know your browser via our contact page.

Update: IE7 users cannot see the charts and receive a message "Error retrieving XML data". We are working to resolve this. If it cannot be resolved before this evening we will roll back.

Update: Fixed - IE6 & 7 users should be able to see the dynamic charts now.

Morning update

2008-11-07T08:55:00.005+13:00

At current prices in the VS markets, and subbing in Ron Mark's chances in his electorate rather than Peters's chances (as either is sufficient for NZ First to return), PM.National is currently substantially underpriced. Only two unlikely potential Parliamentary configurations yield PM.National probabilities under 80%: if NZ First returns and the Maori Party supports Labour, PM.National would be at 75%; if ACT also doesn't return, that probability drops to 19%. The most likely scenario is:

ACT	4
National	60
UF	1
Maori	5
NZ First	0
Labour	41
Green	10
Progressive	1

The above gives us a 122 seat Parliament and a 65 seat Act-National-UF coalition governing handily, with or without Maori Party support.

Unless prices elsewhere in the system are out, and they don't look out to me, National's a strong buy at anything less than 90%. The spreadsheet's telling me 97.5% chance of a National coalition government, but I'm not confident enough in it to go above 90. Full disclosure: I'm current high bidder on PM.National and have substantial position long PM.National and short PM.Labour.

63 of you have downloaded the spreadsheet to play with: any thoughts on it?

Yesterday saw some interesting trading on MP.Peters: what looked like a couple of attempts to drive the price up, or a strong Peters partisan hastily trying to build position, or somebody with inside info about Ron Mark's chances of winning his seat. We now have a market on whether Ron Mark will take his seat, and that's saying 6.1% chance that he does. Is there really a 10% chance that NZ First beats 5%?

Addendum It's happened again. Somebody's run the MP.Peters price up from 0.16 to 0.20; it's now back down again to 0.1825. Why do I think one trade? Because it happened between quick refreshes of my browser window! If this were somebody trading on fundamentals, we'd see movements in Rimutaka.Mark or in Tauranga.Peters or in VOTE.NZFirst. I suppose it's possible that somebody's making this play figuring that the true standard deviation of the minor party stocks is more like 1.1 than 0.9, but just ramming the price up seems an odd way of going about things. Recommendation: if you think that the fair price of MP.Peters is below $0.18, throw some asks into the order book above $0.18 in case our Peters fan returns!

First iPredict Trader to $2,000

2008-11-06T16:24:00.006+13:00

iPredict's first trader to $2,000 is jmex, who crossed the line on closing of the US presidential stocks last night. Congraulations jmex!

There is a $1,000 limit per user per six months for contributions to iPredict - so jmex has (at least) doubled his or her money. Nice going!

Volatilities

2008-11-06T15:17:00.002+13:00

There's a 12 point difference in the predicted probability of a National-led coalition, using my spreadsheet, if we compare probabilities at high bid for National and low ask for Labour as compared to low ask for National and high bid for Labour. The price spread in the vote share market corresponds to a 12 percentage point difference in the probability of a National-led coalition (all other stocks evaluated at high bid).

If National gets 48.5% of the vote (high bid) and Labour gets 34.68 (high bid), we expect National + Act + UF = 63 in a 121-seat Parliament and an 85% probability of forming government.

If National gets 49.17% (low ask) and Labour gets 34.68, National + Act + UF = 64 in a 121-seat Parliament and a 95% probability of forming government.

If we evaluate both at the midpoints of the bid-ask spread, we get an 83% probability of a National coalition.

I wonder why moves in the VS market aren't matched by larger moves in the PM markets. Somebody who thinks National's true vote share is likely to be upwards of 49% would do much better by going long on PM markets than by correcting the VS markets.

Peters fan!

2008-11-06T14:12:00.004+13:00

In the last 5 minutes, MP.Peters jumped from about 17% to about 22%. Prices didn't move in VOTE.NZFirst or in Peters.Tauranga. So somebody knows something we don't (perhaps about Ron Mark's chances), or a Peters partisan just wanted to buy a lot of stake in it in a hurry. I'm shorting in response, but has anybody seen any decent polling on Ron Mark's chances?

Addendum
If Vote.National is right at 0.4917 and Vote.Act is right at $0.3467, then PM.National ought to be closer to $0.94 than $0.79. One side of the equation is wrong. Which one?

TV3 Leaders' Debate Analysis

2008-11-05T19:57:00.004+13:00

The TV3 leaders' debate on Monday night was telling for how little was revealed by the leaders that wasn't already known. Both leaders, and John key in particular, stayed on-message for much of the debate.

The most notable feature of the debate was the near-complete absence of moment in the price of PM.National. It was a flat line the whole way, but for a very minor upwards price movement about 10 minutes from the debate's end. Traders were sending a message: we have heard this before.

National's flat line was particularly striking given the rollercoaster ride that stock had been on a few hours earlier. In the morning, news was that Peter Dunne had been offered thousands of dollars from a businessman after agreeing to oppose international fishing restrictions. This news sent National's stock falling 4% in the space of an hour. However, the afternoon saw a strong recovery for National following John Key receiving endorsements from former rugby stars Michael Jones and Inga Tuigamala.

So the contrast between earlier price movements and the leaders' debate itself was striking.

PM.LABOUR did move during the debate, downwards in two steps and but recovered to where it started shortly after the debate's end. There was an initial downward's movement as the leaders spoke about trust at about 7:20pm, but the price quickly returned to where it was in the minutes following. This was followed by a larger downwards price adjustment of about 1% at about 7:35pm as the leaders discussed longevity and change. Traders responded negatively to the Prime Minister on this point.

However, this price movement did not stick either, PM.LABOUR returning right back to its initial price in the minutes following the debate.

With neither party's chances of leading the next government making any significant change end to end in the debate, this debate was a draw.

Prices in our VOTE markets

2008-11-05T09:03:00.003+13:00

David Farrar at Kiwiblog maintains a running time and size weighted average of public polls. Let's compare current polling figures with our prices (at current high bid, 9:15 am):

Party	Polls	iPredict
National	47.4	46
ACT	1.9	3.468
Maori	2.4	4
Labour	35.3	40.13
Green	7.7	7.968
Maori	2.4	4
NZ First	3	4.013

The PM markets suggest a 76.26% chance of National leading the next government.

If we run the current market prices through the spreadsheet, I get a 45.4% chance of a National-led government. But, if I use the poll numbers instead (and combine it with polling estimates of 4 Maori electorate seats rather than 6), I get 75.17 -- pretty close to the current trading price in PM.National. Now, it's somewhat dodgy putting polling numbers into the spreadsheet because the variances then don't make sense, but something's wrong here. As a first guess, I'd say Labour is 3-5 cents too high in the VS markets.

I can believe National being lower and ACT being higher than the polling numbers, netting out to about the same total. I can believe Labour being a bit lower than the polling numbers and the Greens a bit higher. I can't see Labour on 40. But if the folks trading here are happy with Labour on 40, then PM.National just isn't worth 76.

I'm going to short VOTE.LABOUR if prices haven't moved before the US election contract settles and I get my $132 in Obama payout.

Addendum Lots of volatility this afternoon in VOTE.NATIONAL and VOTE.LABOUR. I saw VOTE.NATIONAL as high as 52; it's since dropped back down to 46.67. If I had cash balances, I'd be seeding both sides of the order book!

iPredict Server Upgrade

2008-11-04T12:45:00.002+13:00

iPredict was slammed with traffic around 8:00pm last night as a result of our appearance on Tv3, resulting in one of our web servers crashing and affecting some users.

This was a RAM problem and occurred in spite of testing in March which we thought guaranteed performance on the night. We were wrong.

We have ordered more memory for our servers to be here in time for election night this Saturday and will continue testing in the meantime. iPredict staff will be working right through election night to keep the site running.

I'll keep you posted on progress ahead of the weekend.

Obama prices as markets near conclusion

2008-11-03T13:38:00.004+13:00

Two more days 'till uncertainty, such as it is, is resolved in the US Presidential markets.

Currently, Obama contracts are trading at $0.886 at the Iowa Election Market, $0.8895 at InTrade, $0.901 at BetFair, and $0.9137 at iPredict. Justin Wolfers over at the Freakonomics Blog notes that the favorite-longshot bias suggests that Obama's true chances are slightly higher than that. On the whole, prediction markets do a fantastic job of predicting outcomes. Draw a 45 degree line on a graph whose axes are "probability of event as expressed by prices" and "frequency with which event happens", and observations track the line pretty well. Except that prices of low-probability events tend to lie above the 45 degree line and those of high-probability events tend to lie below the 45 degree line. Longshots (McCain) tend to be overpriced while favorites (Obama) tend to be underpriced. So we have some reason to expect that Obama's true chances are somewhat higher than the prices in the various markets, and McCain's true chances are somewhat lower.

One proferred reason for this bias is that exchanges charge fees, so investing $9.50 in hopes of getting $10 doesn't make much sense if you're going to have to net fees out of the earnings. If that were the case, we'd expect Obama prices on BetFair and InTrade to be lower than those on either the IEM or iPredict as neither of the latter charge fees. But, the IEM has the lowest Obama prices going.

I'd expect Obama's chances to be much closer to 95% than 90%, and have started building position in that contract. Wolfers agrees in a Wall Street Journal column. I'd previously divested most of my position, seeing that iPredict's prices were a decent way above the prices in the other markets. Thinking more on favorite-longshot bias has me heading back in.

In other news, Foreign Policy has a nice article on prediction markets. Enjoy!

The Hitchhiker's Guide to PM Probabilities

2008-11-03T10:02:00.002+13:00

I've uploaded my handy excel spreadsheet that works out what the price of PM.National would be if the prices in the underlying markets are correct and if my bold simplifying assumptions aren't too bold. Though the spreadsheet contains much that is apocryphal, or at least wildly inaccurate, if the calculations show a big gap between the calculated price and the current market price, there's likely an arbitrage opportunity playing around somewhere in the prices. Either in the prices in the underlying markets, or in the PM markets themselves.

Be warned: the file is 24 MB, mostly because I cribbed somebody else's Sainte-Lague spreadsheet and because my Excel coding skills are padawan-level at best, I copied those sheets four times over to work out the seat shares in four relevant cases: Act returns (or doesn't) and NZ First returns (or doesn't).

Happy trading!

Calling for your ideas on an iPredict API

2008-11-02T19:09:00.004+13:00

iPredict is developing an API and we'd like to hear from you if its something you think you might use.

There's a lot of decisions that go into making an API. Our thinking at this point is that for a start a "read-only" API that gives access to raw trading data. You will be able to get price and trade volumes by date and time and by stock. Simple stuff - but useful for setting up alerts and collecting and alanysing data.

The next step will be whether and exactly how to add trading functions to the API. What are the things you'd like to be able to do without having to visit iPredict and use our interface? What do you need in an API to do it?

Please contact me if you'd like to add your ideas into the API development: matt at ipredict dot co dot nz.

Unofficial blog

2008-11-02T19:06:00.002+13:00

iPredict now has an unofficial blog:

The unofficial iPredict Blog

Weekend market ponderings

2008-11-01T11:45:00.006+13:00

Shouldn't Maori.Nat.Other trade at the same price as PM.Labour + PM.Other? Maori.Nat.Other pays off in the case that the government isn't headed by National; PM.Labour + PM.Other span the space of non-National heads of government. Those prices were out of step with each other for a good chunk of the day: an arbitrage worth keeping an eye on.

Current prices in other markets suggest the right price in the PM.National market is about $0.74. I've now set this up in a handy Excel sheet with a reasonable implementation of Sainte-Laguë to back out list seats from prices in the VOTE.XXX markets. It then puts in overhang and accounts for configurations of Parliament where each of ACT and New Zealand First can either succeed or fail in returning to Parliament.

Here's how things currently are looking in the most likely case:

ACT 4
NAT 58
UF 1
MAO 6
NZ1 -
LAB 43
GRN 10
PRG 1

In that case (where ACT returns to Parliament but NZ First does not), if the Maori Party sides with Labour, we have a 64% chance of a National-led coalition; that rises to 87% if Maori supports National or abstains.

I'm happy to upload the .xls file after the weekend if anyone's interested in playing around with it. It's currently 20MB though because of the S-L tables.

Unfortunately, I'm rather beginning to think that the PM markets are far more reliable than the underlying VS markets and that the better approach would be to work backwards from the PM markets to what prices would need to be in the other markets. As 13 different prices go into working out the PM price, it would more than a little difficult to figure out how much each of those would need to be tweaked if working the process in reverse.

And so is solved the mystery of the inconsistent prices

2008-10-31T22:01:00.002+13:00

Prices in the vote share market have moved since my last posting and only partially because of my own trading. At current VS market prices (in combination with expectations on Dunne and Anderton where we have no markets, and overhang from the Maori seats where we do have a market), we're looking at

ACT 4
NAT 58
UF 1
MAO 6
LAB 44
GRN 9
PRG 1

That makes for a 63 seat National-Act-UF coalition in a 123 seat Parliament. That's substantially more than a 60 seat Maori-Labour-Green-Progressive coalition. Let's continue to assume that Maori opposes National on supply and confidence. Are the prices in the PM markets now consistent with the prices in the VS markets?

Again, we need to start by working out the standard deviation in the VS markets. We'll start by correcting an error that I'd made previously which meant that my prior estimates overstated National's chances of becoming government: the Majority.National market pays off based on whether National gets a majority of seats, not of the popular vote. This matters with overhang. If we were looking at a 120 seat Parliament, current prices would imply a standard deviation of 2.19 seats for National; at 123 seats, it's 5.46.

For the minor parties, I'll again benchmark against NZ First: 1.23 in the vote share market for 1.48 in the seats market.

The standard error of the seat share for the National coalition is then 5.66 seats.

What proportion of the distribution (63,5.66) lies above 61.5? About 60.4% of the distribution: about a 60.4% chance that we have a National-led government under these conditions.

Does that mean that PM.National is overstated? Not necessarily! I'd earlier given reason 4: that the Maori Party might not oppose National. If Maori either supports National or abstains on confidence and supply, we need the proportion of the distribution (63,5.66) that lies above 58.5, not 61.5: about 78.6%. If Maori abstains or supports National, National has a 78.6% chance of forming government. We now have some markets on what the Maori Party might do post election: those markets suggest about a 40% chance that Maori abstain on confidence and supply for a National-led government and a 20% chance that they join National in coalition. So, roughly a 60% chance that the Maori party does not oppose National. So taking the Maori Party into account, what's the chance that we have a National-led government? 71%. The current price in the PM market!

But we haven't accounted for that there's a 25% chance that Peters returns to Parliament. If Peters returns, we have a rather different configuration of Parliament:

ACT 4
NAT 56
UF 1
MAO 6
LAB 42
GRN 9
PRG 1
NZ1 5

and 124 seats: the Maori Party now get 2 by overhang rather than 1. 63 are needed for a majority.

ACT+NAT+UF = 61
LAB+GRN+PRG+NZ1 = 57

If Maori support National or abstain, National forms government; otherwise, Labour does. In the first case, we need to know the proportion of the National coalition vote that exceeds 59; in the second, 63. 63.68% and 36.32.

So, to take NZ First into account, we have a 75% chance (no NZ First) that the 71% probability was right and a 25% (NZ First) chance of a 52.7% likelihood of a National government: about 66% overall chance of a National-led government.

Overall, it might be the case that PM.National is overpriced. But if it is, it isn't by much. Given the chain of assumptions built into the analysis, I'd be reluctant now to short PM.National at 71%.

Does this mean I had things horribly wrong in the prior posts? Well, I did note that if the chances of the Maori party supporting National were higher than the chances of Peters returning to Parliament, then PM.National could be close to right. And, it looks at current prices like the chances of the Maori Party supporting National are more than double the chances of Peters returning to Parliament.

I'll be putting in buy orders to cover my prior PM positions a few places back in the order book!

Our next government redux

2008-10-31T14:47:00.004+13:00

One of the advantages of being at Canterbury in October is that Sir Clive comes to visit. Sir Clive won the Nobel a couple of years back for his work in economic forecasting and economic statistics. So I asked him if I'd screwed up anything major in the post below. He said the overall technique should be reasonable for what we're trying to do here, but that I'd made a minor error: the correct probability of National forming the next government, at the current prices in the vote share market, is 60.6%, not 57%. I'd made the rather silly error of adding up the standard deviations rather than adding up the variances.

So we have four options.
1. The prices in the PM markets are wrong
2. The prices in the VS markets are wrong
3. The underlying distributions are highly skewed rather than symmetrical.
4. The Maori party is more likely to go with National than NZ First is likely to make it into Parliament.

We've eliminated 3a, which was the "Eric's completely screwed this up" option as I now have the Sir Clive seal of approval. And you're not going to do better than that in New Zealand currently.

I'm complementing my positions in the PM markets with the appropriate inverse positions in the VS markets as a quasi-arbitrage.

Addendum
Prices in the VS markets have moved such that they're now consistent with a 63 seat National-ACT-Dunne coalition to a 60-seat Maori-Labour-Green-Anderton block. This is going to be a lot closer to the probabilities implied in the PM market, but I need to work the numbers out. We've now opened markets on whether Maori will support National on confidence and supply. The most likely outcome, according to that market, is that Maori will abstain on supply and confidence, in which case PM.National is more likely: it's then a 63 - 54 situation with 6 abstentions. The market has it more than twice as likely that Maori either support National or abstain on confidence and supply than that Winston Peters returns to Parliament with a block of seats for a Labour coalition.

Our next government?

2008-10-30T09:10:00.005+13:00

The markets today are saying 70% National, 30% Labour as next PM. Let's have a look at the underlying numbers.

At high bid for each party, assuming that Peters does not return to Parliament, Hide takes Epsom, Dunne and Anderton return to Parliament and the Maori Party take 6 electorates, we're looking at a 123 seat Parliament arrayed as follows:

GRN: 9
PRG: 1
LAB: 45
MAO: 6
UF: 1
NAT: 57
ACT: 4

62 seats are needed for a majority. National + Act + UF are at 62; Labour + Green + Progressive + Maori are at 61.

The markets also are saying there's a 17% chance that National's true vote share will be above 50%. If National's price is normally distributed around the true mean, though, that means there's also a 17% chance that National's true vote share will be below 43.3%. Again, we can take the combination of the prices in VOTE.NATIONAL and MAJORITY.NAT to work out the standard deviation of VOTE.NATIONAL: it's currently at 3.5. Let's assume that the standard deviation of VOTE.LABOUR is the same. We'll benchmark minor party vote share standard deviation by that of NZ First, which we can work out from the combination of VOTE.NZFIRST, TAURANGA.PETERS and MP.PETERS. At current prices, that suggests that the minor party standard deviation is 1.15.

Ok. National's coalition's expected vote share is 50.65%, getting it 61 seats plus Dunne. While the vote share point estimate is 50.65, the standard deviation will be the sum of the standard deviations of the underlying components: 3.5 + 1.15 = 4.65. National's seat share is then expected to be 62, with standard deviation 5.58 seats (as 1% of the vote gets you 1.2 seats).

Labour's coalition's expected vote share is 49.37%, getting it 59 seats plus 1 Maori overhang plus 1 Anderton. The standard deviation will be (as a rough cut) the same as that for National's coalition because we don't need to worry as much about the variance of the Maori party vote if they're set anyway for overhang, so we're really only worried about variance in Green and Labour. So, 4.65. Labour's seat share is then expected to be 61, with standard deviation 5.58 seats.

Now, what's the probability that the realized seat share for the National coalition really lies above the realized seat share for the Labour coalition? A cheap way of checking this is just to check the probability that National lies at 61 seats or less. Our standard normal table tells us the chances of this are 42.86% as only 7.14% of the distribution lies between 62 and 61 seats.

So, National's chances of getting more seats than Labour look more like 57% than like 70%. And those chances get much worse in the 26% probability case that Peters returns to Parliament. I'm not going to bother working them out at this point though.

The market has something wrong here or my calculations are way out. It could be that the normality assumption is just wrong and that National has much more upwards risk than downwards risk. I've made tons of simplifying assumptions along the way, but I can't see that they'd push me that far out. Or, the prices are out somewhere: either Vote.Labour (and partners) is overpriced and Vote.National (and partners) is underpriced, or PM.Labour is underpriced and PM.National is overpriced. Either is rather plausible. There's not much candle in moving the VOTE.Party markets if they're out by a point or two. But similarly folks could just be underestimating the effects of overhang on the PM markets: National gets +1 with overhang but Labour gets +2, which matters in a tight race. Finally, Maori could decide to go with National. If you think that's reasonably likely, then the current prices might not be too bad.

Full disclosure: I'm currently short PM.National and long PM.Labour, with positions taken after having done a rough eyeball version of the above a few days back in the comments section of a prior post. I'm going to further short PM.National and further purchase PM.Labour as result.

Addendum Since posting, PM.National has gone up a couple points, PM.Labour has dropped a couple points, and NOBODY'S vote share has moved. How does the probability of one party or the other winning change when the underlying vote shares don't move? Are folks upweighting the likelihood of Maori going with National?

Do Scandals Help With Re-Election?

2008-10-29T08:06:00.003+13:00

During an interview with MP Winston Peters this morning, Radio NZ put it to him that the publicity generated by new developments in the Owen Glenn saga would assist in Mr Peters' re-election. Mr Peters replied by saying stories like this never help.

So what does iPredict say?

iPredict has been running a contract MP.PETERS asking whether Mr Peters will be returned to Parliament. This contract pays $1 if Mr Peters remains in Parliament by any means after the election, and movements in the price this contract trades for over the past 36 hours gives us an answer.

The answer, in this case, is no. Mr Peters' re-election chances are not helped by the publicity produced by these developments, as indicated by a decline in the price of MP.PETERS.

These developments have been accompanied by an increase in the chance of Mr Peters' sacking, up from 10% two days ago to 15% this morning - a significant increase starting from a small base but Mr Peters sacking or resignation from his position as Foreign Affairs minister before the election remains a one in six chance according to iPredict traders.

Peters prices now internally consistent?

2008-10-23T15:35:00.004+13:00

The introduction of PETERS.TAURANGA seems to have brought Peters prices closer into consistency with one another.

MP.PETERS: 0.3
VOTE.NZFIRST: 0.37
PETERS.TAURANGA: 0.18

Now,

MP.PETERS = probability (Vote.NZFirst>0.5) + Peters.Tauranga - [probability Vote.NZFirst >0.5 AND Peters.Tauranga]

We don't know the probability that Vote.NZ First > 0.5 and we don't know the joint probability of the last term. If we (obviously incorrectly) assume independence of the two terms, we can back out a biased measure of the probability that Vote.NZFirst > 0.5. Alternatively, we can take the limit case and say that Vote.NZFirst will definitely be above 0.5 if Peters wins Tauranga.

First, the independence case.

0.3 = p(V.NZ1>0.5) + 0.18 - p(V.NZ1>0.5)*0.18
0.12 = 0.82p(V.NZ1>0.5)
0.146 = p(V.NZ1 > 0.5)

Is it plausible that there's about an 18% chance that NZ First's vote share will be above 5%? Let's assume normality and look at our handy standard normal distribution. At a Z score of 1.05 we have the right mass in the upper tail. The implied standard deviation is then 0.13: 1.3 percentage points. That's still on the high side, with a 95% CI of 1% - 6.2%.

What happens if we take the limit case where NZ First's vote share definitely exceeds 5% in any state of the world in which Winston wins Tauranga? MP.Peters then reduces to being just the probability of NZ First getting 5% of the vote.

0.3 = p(V.NZ1>0.5) + 0.18 - 0.18
0.3 = p(V.NZ1>0.5)
Giving an implied standard deviation of 2.5% -- much too high.

If we make the obviously too-restrictive assumption of independence, we wind up with an upper-bound of plausibility estimate on the variance of VOTE.NZFIRST. If we relax that assumption, we move to far too high estimates of the variance (in my view). What would tighten things up? Either MP.Peters is still overvalued or PETERS.TAURANGA is undervalued or VOTE.NZFIRST is undervalued.

I'll consequently further short MP.Peters while going long on PETERS.TAURANGA and VOTE.NZFirst!

Update The latest polls have pushed up MP.Peters but haven't touched either Peters.Tauranga or VOTE.NZFirst. Very odd.

More thoughts on Peters prices

2008-10-20T16:48:00.003+13:00

Interesting discussion in the comments section on the prior post. Let's refine things a bit.

First, we can work out the standard deviation of the VOTE.NATIONAL price by looking at MAJ.NATIONAL. The point estimate price is $0.47 and the chances of their beating 50% are 26%. The Z score on the standard normal table for 24% (the area lying between the point estimate and $0.50 given that 50% lies above $0.47 and 26% lies above $0.50) is 0.645. Z = [x-mu / sigma], so 0.645 = [0.5 - 0.47]/sigma and sigma must then be 0.0465. So, we can be 95% sure that the actual National vote share will lie somewhere between 38% and 56%.

Now, there's no easy way of working from the standard error of VOTE.NATIONAL to that of VOTE.NZFIRST. A standard error of 4.65 percentage points is entirely wrong, but simply scaling it by the scale difference between the two markets also would be in error: we then could have arbitrarily-precise point estimates simply by increasing the payout per percentage point of the popular vote. Moreover, we should be less confident in the vote shares of the minor parties as compared to those of the major parties. I suppose that the best we can say is that the standard error of VOTE.NZFIRST should be higher than 0.465 percentage points: the scaled standard error for this market, or $0.0465.

Given that the standard deviation on the VOTE.NATIONAL market seems reasonably high, I'm less confident that the standard deviation on VOTE.NZFIRST is as tight as I was thinking initially. If the 95% CI on VOTE.NATIONAL runs from 38 to 56%, it's not that implausible that the 95% CI on VOTE.NZFIRST runs also over a pretty wide range.

For sake of argument, suppose that the standard error for VOTE.NZFIRST runs double the scaled value from VOTE.NATIONAL, so 0.093. If VOTE.NZFIRST is $0.37, then there's a 6% chance that VOTE.NZFIRST winds up above $0.50. If the standard error were triple that (scaled) from VOTE.NATIONAL, it would be a 15% chance. MP.Peters is trading at $0.39. What are Peters' chances in Tauranga given these prices? Somewhere between about 30-40%, depending on how you'd want to specify the covariance between VOTE.NZFIRST and Peters in Tauranga, and what estimate you'd like to use for the standard deviation of VOTE.NZFIRST.

I'm going to maintain my short position as I don't think it between 30-40% likely that Peters will win Tauranga.

Addendum I told you that it was overpriced! Tauranga.Peters opened at $0.24 and dropped within hours to $0.12.

Peters prices

2008-10-19T19:29:00.004+13:00

Does anyone else find it very odd that MP.PETERS is trading at about $0.385 while VOTE.NZFIRST is trading at $0.36?

Winston Peters has two ways to get back into Parliament. The first is by winning Tauranga; the second, by getting 5% of the vote. MP.PETERS consequently should then be trading at a price equal to the sum of the probability of Peters winning Tauranga and the mass of the probability distribution of VOTE.NZFIRST that's above $0.50. We don't have a market on whether Peters will win Tauranga but I'd be shorting that market on any initial price above 20%. If his chances in Tauranga are about 20%, that means that 18.5% of the probability distribution on VOTE.NZFIRST is above $0.50. For those with less stats, that means that, if the prior analysis is right, NZ First has an 18.5% chance of bettering 5%. Assuming normality (always risky with this subject), that means that the standard error of the price estimate on VOTE.NZFIRST is about $0.156 [corrected]. I find that a bit high. If the standard error is 0.156, the 95% confidence interval is from $0.06 to $0.667 [corrected].

I could be way out on Winston's chances in Tauranga; if they're much higher, that tightens up the variance on VOTE.NZFIRST. Or, I could be way out in my distributional assumptions. Or, we could really just have a very high variance estimate of the true vote share for NZ First. If none of those are the case, though, either MP.PETERS is overpriced or VOTE.NZFIRST is underpriced.

Full disclosure: I shorted MP.PETERS after thinking this through as I find it more likely that MP.PETERS is overpriced than that either VOTE.NZFIRST is underpriced or that my prior assumptions are seriously out.

Obama prices

2008-10-19T07:30:00.002+13:00

I'd noted previously that there have been pretty big price discrepancies in the Obama/McCain prices across different markets. Turns out now that the reason InTrade has been so low, relative to the others, is that a decent-sized institutional investor has been hedging on InTrade against the possibility of a McCain win by buying McCain shares. For more discussion, see here and here and here and here and here.

Now, why might somebody want to hedge against political risk? For the same reason that they might want to hedge against any kind of risk! So, you could imagine that a defence contractor might want to hedge against the possibility of an Obama win, and a biotech firm big on stem cell research might want to hedge against the possibility of a McCain win.

Hedgers will increase the return to informed traders, just like anybody else who's trading for non-informational reasons. But, if the hedger is big relative to the market, it can have a reasonable effect on prices for a very long time, as shown in the InTrade example. It's only in the last few days that InTrade prices have come up to match those here, on the IEM, and at BetFair.

As iPredict allows only $1000 deposited every 6 months, we're not a useful vehicle for hedging against political risk. So, we shouldn't see that kind of non-informational trading affecting our prices.

I note this morning that our prices on Obama are on par with those over at Iowa and higher than those at BetFair and InTrade.

Full disclosure: I'm long Obama, but sold off a good chunk of my position at $0.885.

National Wins Leaders' Debate

2008-10-15T10:14:00.007+13:00

National's chances of leading the next government increased by 5.1% last night according to iPredict traders, handing John Key a decisive victory over Helen Clark in the first leaders' debate.

Labour's re-election chances fell 4.4%.

By the end of the debate, Labour was a 30% likelihood of leading the next government, a significant increase from its position a week ago where it was tracking in the low twenties. However, this morning PM.LABOUR is trading down at 25 cents. Last week's gains have been largely undone.

As in share markets, price movements in prediction markets are caused by the arrival of new information and outcomes which are different from expectations. TVNZ commentators after the debate said that while the Prime Minister performed well, this was to expected, whereas Key in his first leaders' debate outperformed expectations.

Print media commentators generally agree, Audrey Young saying

She had by far the best campaign launch on Sunday but he wiped the floor on the debate.

John Armstrong:

While there was little to separate the pair in a pretty even contest, John Key has to be declared the winner of tonight’s debate…She was as rock solid as always, but predictable.

Fran O'Sullivan, however, gave the debate to Clark:

Clark scored best on the issue du jour - the international credit crisis. She has a post-election plan. Key doesn't.

Traders on iPredict are unequivocal: National's chances of an election victory increased sharply at Labour's expense, and on this measure victory in the first leaders' debate is National's.

Click on the chart for a larger view.

Who Will Win The Leaders' Debate?

2008-10-14T15:30:00.003+13:00

One of the great things about prediction markets is their real-time evaluation of news events. Price changes in PM.LABOUR or PM.NATIONAL in response to news is like running a real-time poll. And a prediction market can tell you the political value of that news - a statistic not available from any other source, certainly not traditional polling or even, arguably, The Worm.

In view of that, iPredict will be measuring the performance of leaders in tonight's debate in the way that matters most: who's election chances go up during the debate, and whose goes down? That is, ultimately, the measure of any politician's performance.

Following the lead from InTrade, we will use the following method for determining the debate winner: the largest increase in absolute likelihood of being elected government. The before and after values will be calculated as the average market price for PM.NATIONAL and PM.LABOUR in the thirty minutes before the debate commences and the thirty minutes after the debate finishes.

The leaders' debate is on TV One at 7:00pm tonight.

Labour's Seismic Shift

2008-10-14T11:03:00.007+13:00

Labour has seen a step change in their chances of leading the next government in the past four days, a point brought home in the chart (click for a larger view).

In the days following iPredict's launch in early September, Labour's chances of leading the next government hovered around the 25% mark. Towards the end of September, a downwards trend in this market began with Labour sinking to the low 20's by the end of last week.

And then there was the Morgan Poll on 10 October, which produced a surprisingly narrow gap of 3% between National and Labour. This was followed up later on Friday by the release of a TV3 poll also showing a narrowing gap between Labour and National.

This shocked the market and the result was a step change up for Labour by 6% - a significant change give Labour's relatively low starting point. And there is some indication of an additional upwards movements in Labour's re-election prospects since Friday, which may be a product of two major announcements, the first that the government will insure bank deposits, and the second being Labour's universal student allowances policy in kick in four years from now.

This upwards trend in Labour's chances of leading the next government has of course been mirrored by a decline in National's chances. National is now down to $0.69, not its lowest level since iPredict launched but close to it.

What is striking is how little other recent events have moved Labour's election prospects. The global financial meltdown, the release of National's tax policy last week, and the disclosure that the government is looking at 10 years of budget deficits did not move traders on iPredict - producing at most a gradual downwards movement in re-election chances.

The current momentum is with Labour, it will be interesting to see what National does reply and whether that response is enough to convince traders.

Morgan Poll Hits PM Stocks

2008-10-10T17:46:00.006+13:00

Take a look at the effect of tonight's shock Morgan Poll, out in the past hour, on our PM stocks.

The Morgan Poll is an opinion poll on vote share and its latest results show Labour closing the gap on National, and traders on iPredict have reacted vigorously to the news.

The PM.LABOUR contract, which closes at $1 if Labour leads the next government, skyrocketed all the way to 38 cents, up a full 15 cents from what has been a fairly long term average in the low to mid-twenties.

However, the spike did not last, and other traders appear to have come in and take advantage of some traders' willingness to pay up to 38 cents for PM.LABOUR contracts.

PM.NATIONAL also reacted but not as strongly and few trades were being done than on the Labour contract.

The present volatility in the PM stocks reflects traders trying to understand the new information that has unexpectedly arrived and the consequences for Labour and National. The question of who will lead the next government is complex in an MMP environment. Traders must not only consider vote share, but also the combinations of parties which could go into forming the next government.

What is clear is that this Morgan Poll is of a consequence that other major events this week - the shock news that the government is forecast to be in deficit for the next decade, the release of National's tax policy, and the unexpected turn for the worse in global finance markets - are not. None of these other events moved the PM stocks at all. For that, we need shock poll results.

It will be interesting to see where the market settles at as this new information is understood and incorporated.

Click on the chart for a bigger view.

Update 11/10: Labour's chances of leading the next government seems to have settled 5% up from their long term average in the low 20's. National's likelihood of leading the next government has seen a corresponding 5% fall. National's 50 point lead in the leadership stakes has been cut to 40.

Another poll out tomorrow, Sunday. Will be interesting to see whether it has a similar effect.

Contract Types on iPredict

2008-10-09T14:11:00.006+13:00

When I talk to people about iPredict, I think many come away with the idea that the only kind of contract on iPredict is all-or-nothing contracts that either pay $1 or $0.

There are three kinds of contracts that can be run on iPredict, two of which currently appear on iPredict, and one we'd like to develop and add to the site soon.

The contract types are:

# Binary: these are the yes/no questions that either pay $1 or $0. The interpretation of binary contract prices is as "likelihoods" e.g. $0.51 = 51% chance of being judged true.

# Indexed: these contracts have their payout tied to an observable variable such as inflation or vote share or sales revenue. For example, iPredict is running INFL.3Q08. This contract pays 1 cent for each 0.1% of inflation that is reported for New Zealand in the third quarter of 2008 (this contract will be closing soon). So, for example, an inflation rate of 5.1% would result in the contract closing at 51 cents. The interpretation of the contract price is the expected value of the variable being predicted e.g. a contract price of $0.51 means we expect inflation will be 5.1%.

These two contract types give us considerable flexibility. An important idea on designing a contract is having the contract's payout struct vary in a way that is compatible with the parameter being measured. One could obtain an expected value for inflation by runing 20 binary contracts asking whether inflation will be between 4.9-5.0%, 5.0-5.1%, etc. But a single indexed contract will do the same job and will almost certainly be more efficient (read: accurate). (running 20 binary contracts is useful to answer a different question: what is the distribution of beliefs about a future event).

# Conditional: these contracts pay out according to the outcome of one variable depending on some other variable being true. For example, you could define a contract that pay out 1 cent for each 1% of inflation reported in Q3 2009 given Labour leads the next government. This contract would payout $0 if National (or anybody else other than Labour) leads the next goverment.

Conditional contracts is the prediction market tool used for investigating policy questions. If we were to run another conditional contract alongside the example above, but instead conditional on National leading the next government, then an examination of the difference between the prices of the two conditional contracts - one conditional on Labour leading, the other conditional on National - can tell us what traders think about the effect of each party's policies on unemployment by this time next year.

The analysis is a bit more complicated than that, actually. The conditional market prices have to be weighted by the likelihoods that Labour or National will lead the next government. But bottom line, even with this mild complexity, we should expect some reasonable estimates to come out of these markets given enough liquidity.

But there's the rub with conditional markets. They are fine in theory, yet nobody has yet found a way to make them sufficiently attractive to traders to make those markets truly liquid. InTrade runs conditional markets, and removed all fees from trades on some of those markets in an effort to stimulate activity. It didn't work.

We at iPredict have some ideas on making conditional markets easy to understand, and we intend launching them in the coming weeks.

Cross-market prices

2008-10-07T16:35:00.007+13:00

Arbitrage opportunities continue to exist between iPredict and other US election markets. Current prices at midpoint of bid-ask spread:

Exchange	Obama	McCain
iPredict	0.7266	0.2572
InTrade	0.6860	0.3135
BetFair	0.7692	0.2247
Iowa	0.7750	0.2255

Note that Iowa event derivatives pay off based on which party gets the larger proportion of votes cast, not on specific individuals, and takes no account of electoral college machinations.

From prior posts, there's at least some reason to suspect that InTrade prices are on the low side for the Democrats. It's neat to see that iPredict falls roughly at the midpoint of the different exchanges.

So, what's limiting arbitrage? First, it's difficult to open an account on the Iowa market if you're not American. Here's their sign-up form. If you want to pay other than by a paper cheque drawn on a US bank, you need to email them to get details for a wire transfer. Second, Americans can't easily trade on BetFair or InTrade because of one crusading moralist from Arizona who can't stand that anybody else anywhere might ever enjoy himself.

But that doesn't explain why non-Americans aren't arbitraging across BetFair and InTrade.

I can't answer for anybody else, but I can say why I haven't started arbitraging between those two markets. First, BetFair's prices are reported as the amount you would earn for a $1 stake, not the amount you would have to pay for a derivative paying $1 if the event pans out your way. So you have to transform prices by 1/x to make the comparison. That isn't too big a barrier though. 1/x is pretty easy. More importantly, BetFair's commission structure is a bit tougher. They take 5% of your winnings. That starts eating into arbitrage profits pretty quickly. And, it's unclear to me whether you can cancel your positions by buying the opposite. In InTrade and in iPredict, if I think the odds of Obama winning are 74%, I can buy Obama whenever the price is less than that and sell those positions if the price gets higher. On BetFair, it looks like the system would treat those as separate positions. I could easily be wrong about that, and I'd love it if someone with more experience in BetFair could set me straight here. But I'm worried that, in BetFair, my position would be locked in until expiration, at which point I'd be subject to the 5% fee. The help files suggest that, if you have positions that cancel each other out, the system doesn't freeze margin on it. But that doesn't save you from having to pay commission on whichever side of the market pans out.

Any better explanations on why arbitrage isn't happening?

Update 9 October, 2 PM: Egads!!! Obama prices:
InTrade: 0.657
Iowa: 0.824
BetFair: 0.8197
iPredict: 0.765

There is a 17 cent spread between InTrade and Iowa. Spend $0.657 to buy Obama on InTrade, spend $0.176 to short Obama on Iowa, and you're guaranteed a dollar for a $0.833 investment: a guaranteed 20% return. Well, near guaranteed. If Obama wins the popular vote but loses the election, you lose both sides. If you go with BetFair instead of InTrade, you lose 5% as commission, but that still seems worth it at those arbitrage margins.

Update 2:21 PM: Obama's up to 70 cents on InTrade. Still a big gap though!

Robertson v Franks Tonight

2008-10-06T09:42:00.004+13:00

For those in town and looking for some insider information, as it were, on our Wellington Central contracts, it might be worth coming down to hear three candidates for Wellington Central share their thoughts:

The Victoria University Debating Society Election Debate 2008

The motion: That we need a centre-right government

Affirmative:

Stephen Franks - National candidate for Wellington Central
Christopher Finlayson MP - National List MP and Rongotai candidate
Stephen Whittington - champion Victoria student debater

Negative:

Grant Robertson - Labour candidate for Wellington Central
Sue Kedgley MP - Green candidate for Wellington Central
Polly Higbee - champion Victoria student debater

The chair is Sean Plunket.

Monday 6 October, 6.30pm - 8pm

Lecture Theatre One, Rutherford House, Victoria University of Wellington. Gold coin entry. Questions after the debate, then tea and coffee.

Don't forget your laptop and wireless card!

Grass Roots Data vs. Formal Data

2008-10-05T08:51:00.005+13:00

With turmoil in US financial markets, a question is what effect with this chaos can be expected to have across the economy.

Inkling blog reports on a recent seminar for market researchers where the following question came up:

[I]f I'm a market researcher at a luxury hotel or fancy car company, should I be quaking in my boots? And vice versa, if I'm a market researcher at a budget hotel company, should I be ramping up the marketing budget?

Using traditional methods, an empirical answer to that question is weeks away in the next report on the latest round of surveys. However:

[W]earing my prediction market hat I had a simple thought: why not augment this traditional methodology with a constant stream of questions posed in a prediction market? Grass roots data vs. formal data.

The solution:

Exploring this thought further, let's say we did work for a budget hotel company. We could first identify what the levers are that drive our business. Business travel patterns. Gas prices. Stock market performance. Competitive activity. Then we could recruit people who have insight in to these factors and ask them to participate in an ongoing prediction marketplace. These aren't necessarily just hotel goers, they could be economists, wall street traders, bankers, and professionals from the energy industry.

Prediction markets as a compliment to traditional survey-based empirical research. Not as radical as it seems.

On Short Selling

2008-10-04T11:42:00.002+13:00

From the Spectator, hat tip The Master.

7 Traders Is All You Need

2008-10-03T17:36:00.003+13:00

Prediction markets are not like surveys. You don't need 1,000 people on a prediction market to get something sensible.

In fact you need about seven traders. Seven. That's one reason why businesses are picking up prediction markets - you can learn so much with so little, and at close to zero risk.

Where did this magic number seven come from?

In the 1950s, Nobel Laureate Vernon Smith began running experimental markets in his "lab", which in this case meant his first year economics students. He made half the students buyers, half sellers, and gave buyers secret values (a demand curve) and sellers secret costs (a supply curve). And then he told them to go to it.

Buyers would announce bids, and sellers announce asking prices (sound familiar, iPredict traders?) and trade would occur when overlapping bids could be matched.

Although traders operated with very incomplete information, the market generally converged to prices and volumes consistent with the theoretical intersection of supply and demand curves, also known as a competitive equilibrium.

Smith's experiments were designed to find out the conditions in which markets start failing, and some of these tests asked how many traders does a market need to work? The answer was seven. Of course, more is always better - but only because that increases the available pool of information, not because markets require hundreds of traders to properly work.

Prediction markets really are fundamentally different to surveys, both in the number of people required and in the protections from various sources of bias.

Click here to listen to Vernon Smith talk about his experiments and to read more.

Electorate Boundaries on Google Maps

2008-10-02T08:56:00.003+13:00

This is cool. Electorate boundaries superimposed on Google Maps.

Two surprises for me. One, MMP electorates are big.

Two, many electorates are all at at sea. Check out Wellington Central to see what I mean. Going by the proportion of that electorate that is on land and the proportion at sea, whoever wins WC is going to be spending about 80% of their time administering Cook Strait.

Check it out.

Wellington Central contracts launch

2008-10-01T16:41:00.005+13:00

Wellington Central is a marquee electorate, and this election it is being contested by four high-profile candidates: Grant Robertson for Labour, Stephen Franks for National, Sue Kedgley for the Green Party, and Heather Roy for ACT.

Three contracts launched today on iPredict asking who will win Wellington Central: WC.FRANKS, WC.ROBERTSON, and WC.OTHER.

The timing of this launch was influenced by the appearance of these candidates tonight on TVNZ7's live show "Back Benches". It will be interesting to see contract price movements during and immediately after the show's airing at 9:10pm tonight. A real-time poll, if you like, of the candidate's performance on the night.

A Comment on the Close Dates statement

2008-09-30T11:07:00.003+13:00

The last word mentioned on every contract is this:

"This stock will be closed no later than 3 working days after the Close Date. The Close Date may be extended if the event outcome cannot be determined by that date."

The purpose of this statement is to give ourselves some lee-way if it isn't clear what an outcome is.

The reason for posting this item on the blog is that we did receive a couple of questions about what this statement means when it comes to a contract like WGTN.BUSSTRIKE. This contract asked whether an agreement between bus drivers and GO Wellington would be announced before midnight on Monday 29 September.

The effect of the statement that, "The Close Date may be extended if the event outcome cannot be determined by that date" is not to extend the WGTN.BUSSTRIKE contract. At midnight last night it was clear and unambiguous that no announcement has been made, and the contract could be closed immediately.

What we have in mind with a statement like this is the election. Let's say the negotiations to form the next government drag on past the close date on the PM contracts. These contracts should not close before those negotiations are completed, they are best extended.

If you can think of a wording that makes this clearer, please let us know in comments or just send us an email from the contact page on the site.

Your daily arbitrage

2008-09-30T09:50:00.008+13:00

Today's prices (9:55 am), at midpoint of bid-ask spread:

Exchange	Obama	McCain
iPredict	0.6312	0.3491
InTrade	0.6095	0.385
BetFair	0.6826	0.3053
Iowa	0.6840	0.319

Note that Iowa event derivatives pay off based on which party gets the larger proportion of votes cast, not on specific individuals, and takes no account of electoral college machinations.

So: buy Obama on InTrade, sell him in Iowa (if you can; BetFair if not), sell McCain on InTrade, buy him back on BetFair.

Sounds like I need to open a BetFair account....

Addendum: Figured out table formatting...
Addendum 2: as of 11:15, InTrade Obama price is up to 0.62; McCain down to 0.375.

Making money on manipulation or partisans

2008-09-26T16:43:00.003+12:00

Matt sends me the following graph of some interesting trading on 22 September. At 9:11 am, a single trader (he tells me) bought about 1200 shares of PM.National, running the price up from $0.7376 to $0.82. Traders came in almost immediately to drive the price back down. They started selling PM.National starting at 9:13. By 9:30, the price had been driven back down to $0.7678; at noon, the price was $0.7503. At 5:30 PM, the price was right back where it started at $0.7376.

Was this a manipulation attempt? I find it doubtful. First, there was no complementary selling on the Labour side. Over the time period, the price on the PM.Labour event derivative dropped about 2 cents and then came back up. Second, there was no sustained attempt to keep the price up. What seems most likely to me is that somebody with a very strong belief that National is set to win the election, and very little patience to put in limit orders, wanted to buy a lot of shares and wanted to buy them very quickly.

What I find remarkable is how quickly other traders responded to the anomalous price spike. Two minutes after the price spike, the first selling started. Within 20 minutes, selling by other traders pushed the price down by more than 60 percent of the initial rise. These markets are robust.

Like I said in my previous post, partisans of either party are profit opportunities for informed traders. Folks who had left limit sell orders in place for a good run up were able to sell PM.National to somebody desperate to buy at any price and then could buy back the stock after the price had dropped to reasonable levels. Folks who were watching the market were able to profit by pushing the price back to a level in line with fundamentals. I wish I'd been paying more attention to the prices at 9 o'clock Monday morning so I could have shared in the fun!

Obama Contracts, Arbitrage Opportunities, and Manipulation

2008-09-26T11:40:00.005+12:00

A few posts back, Matt introduced iPredict's new event derivatives on the American election. I noted in the comments that an arbitrage opportunity existed between prices on iPredict, where one could buy McCain for 40 cents or sell Obama for 60 cents, and InTrade, the Irish exchange, where one could sell McCain for 48 cents or buy Obama for 51.4 cents. I of course then went and did exactly that, taking a nice bit of arbitrage profit across the two markets.

Now, we shouldn't expect to see price differences across different exchanges: arbitrageurs, like me, should work to ensure that prices are the same. And, most of the time, we find that to be the case. Recently, though, it looks as though things have gotten a bit out of whack over at InTrade. Chris Masse, the world's leading, and most bombastic, prediction market blogger[1], provides the evidence. In short, Obama prices have been lower on Intrade than on a host of other exchanges, and it looks like somebody keeps going in to drive down Obama prices while simultaneously driving up Hillary prices. InTrade still has event derivatives on whether Hillary will become the next president.

A couple notes.

First, research by Robin Hanson and Ryan Oprea has shown that the presence of folks who try to manipulate these sorts of markets winds up improving markets rather than hurting them: it means that there's a stronger incentive for each of us informed traders to watch for opportunities to take money away from the manipulators. They're every bit as lucrative to informed traders as are the hard-core partisans who come into the market and just want to buy their preferred party's shares no matter the odds. I traded pretty heavily in the Iowa Election Stock Market back in 2000. As they didn't have an automated market maker (like we have), I often served as market maker, happily selling overpriced Democrat shares to folks coming in to the market who seemed to want to buy lots of Democrat shares, and same for Republicans. Great fun. In short, manipulators and partisans make the rest of us richer and the market more efficient.

Second, a potential manipulator on InTrade provides all of us with a profit opportunity if we're happy to arbitrage across markets. Arbitrage in general is hampered by that it's illegal for Americans to trade at InTrade or Betfair or iPredict. That makes the set of potential arbitrageurs smaller and keeps the field more open for us! All you have to do is open an account with InTrade and make the arbitrage play whenever prices differ across the two markets. Buy Obama at InTrade and sell Obama at iPredict, or vice versa for McCain, whenever the prices differ appropriately. Since folks started pointing out the arbitrage opportunities, the price difference has eroded considerably, but you can still buy Obama there for 57 cents and sell him here for 59 cents. That's probably not worth the candle, but it might be worth keeping an eye on prices in other markets in case of odd price spikes. Risk free profits await.[2]

[1]No, Chris, this does not make me a little Chris Masse fanboy.
[2]Yes, I know there's currency risk, but you can always hedge that elsewhere.

Robin Hanson on the history and future of prediction markets

2008-09-23T11:11:00.003+12:00

Robin Hanson was one of my professors at George Mason University, where I did my graduate work. He's also the God of Prediction Markets. Indeed, his theoretical work was essential to iPredict being able to offer an automated market maker to facilitate trades.

To hear the good word about the prospects for prediction markets, go listen to his lecture at videolectures.net. It's his contribution to the Pascal Lecture Series. He covers social and corporate applications of the technology. His example of how a firm could use prediction markets to figure out whether they should switch ad agencies is pretty neat! Enjoy!

Stock alerts launched

2008-09-22T19:09:00.005+12:00

In response to some strong opinions sent back from our newsletter on the value of alerts, our first stock alert email went out today announcing the immediate launch of contract BENSON.POPE, which asks whether David Benson-Pope will stand as a candidate in the upcoming election for any party other than the Labour Party (or as an independent), and the launch of US presidential contracts.

These presidential contracts are our first outwards looking stocks, and I had to make a quick call to our lawyer today to double check that running stocks pointing at events outside New Zealand is consistent with our Securities Commission authorisation.

It is.

However, to help protect against my arrest next time I'm at LAX - which has actually happened, I believe, to the CEO of Dublin-based InTrade - these contracts remind traders that iPredict is for NZ Residents only.

So, friends in the US, please do not even think about trading on our US Presidential stocks, at least until more sensible rules on online predictions markets trading come into effect in the US (which, by the way, may be in the works), and also before our terms and conditions are amended to allow you to.

How to make money when you see overpriced stock

2008-09-22T18:29:00.008+12:00

Picture this. You're on iPredict, you see a stock price for a future event that you are sure is not that likely. Doesn't this overpricing create an opportunity to make money?

Certainly it does, if your instincts are correct. Here's how.

Sell. Even if you do not own stock, you can sell it and make money the lower is the future price of that stock.

Selling stock you do not own is called short selling, and on iPredict short selling works exactly the same as if you were selling stock you do own.

Before I explain the mechanics of what is going on, the key thing to know about short selling is that your earnings from short selling are the exact mirror of the money you make when buying: when you buy a stock, you make money if the stock's price goes up. When you short sell a stock, you make money if the stock's price goes down.

How much money do you make from short selling?

When short selling, the money you earn per contract is the difference between the stock price when you short sell, and when you "cover" or pay or stock you shorted.

An example might help clarify. Let's say you see a stock that is trading at 80 cents for an event that you are sure will not come true. That is, you think this stock will close at $0, and is currently massively overpriced. To make money on this overpricing, you need to short sell this stock. Here's how.

Sign in, and click Trade on this stock. In the Standard or Advanced trade interfaces, tell iPredict you want to sell the stock. You can do this even if you don't own any of this stock - that is why this is called short selling. Say how much you are willing to sell this stock for, as you would if you actually owned this stock, and confirm your trade.

Congratulations, you have just sold stock you don't own, and you are making money every time the price of the stock falls. How?

What you have just done is a) borrowed stock and sold it for money to a buyer on iPredict, and b) committed to paying for the stock you just borrowed some time in the future at whatever the price is at the time you decide to pay.

Now think about it. You've just sold shares you don't own and got the money for them up front. That's your revenue and it's locked in. But your costs - the money to buy the shares you just borrowed and sold - will only be determined in the future when you purchase them on the open market to repay the lender. The lower that future price, the lower your costs, and the bigger the difference between the revenue you received up front and the cost of the stocks you sold.

In other words, when you short sell, you make when the price of stocks fall.

One final note: when you short sell, the iPredict takes $1 off you as a deposit to cover the worst possible outcome, which is that the stock closes at its maximum price of $1. When you cover, you get your dollar deposit back at the same time you cover the cost of the stock you purchased. That's why money comes out of your account when you short sell and comes into your account when you cover - without the deposit it would be the other way around.

Clear as mud?

Check out our tutorial page for a video and a written tutorial on short selling. Short selling is trivially simple to do on iPredict, even if the theory behind it takes a bit of getting used to.

Update: from Australia, a ban on short selling is credited with buoying a volatile market.

An Australian Securities and Investments Commission ban on short selling, designed to reduce volatility and improve liquidity in the local equity market, also buoyed the share market today and lifted shares in companies that had recently been targeted by hedge funds.

On its face, banning short selling seems a very odd way to raise liquidity. If I see stock that is overpriced and cannot sell it because I happen not to own any of it, I'm not likely to instead dive in with offers to buy. Limiting the set of traders who can move a stock's price down to those who happen to have taken long positions on shares before the crisis sounds like a recipe for inefficiency - an arbitrary introduction of downwards rigidity in stock prices does not sound helpful even if prices are volatile. Should we expect it to be helpful to prevent volatility from being expressed in trading if that volatility is grounded in genuine uncertainty?

Arbitrage and US elections

2008-09-22T18:03:00.005+12:00

Our US election stock launched a few minutes ago and traders were ready, with stocks in Barack Obama leaping 15 cents in the seconds after its 6pm start.

We launched three contracts: one in an Obama win, one in a McCain win, and a third contract in neither of those two winning the presidency. These three contracts cover every possible result of the 2008 election, and we will accordingly be offering to sell these three contracts in a bundle for exactly (and always) $1.

Bundles make it easy to exploit arbitrage opportunities on the open market. If the combined trading prices of Obama, McCain and Other contracts exceeds $1 on the open market, buy bundles and sell them for a certain profit.

You can also make certain money if the combined value of Obama, McCain and Other is less than $1. Let's say the sum of trading prices for Obama+McCain+Other is $0.97. You can make certain money by buying equal amounts of all three and then sitting on them. The combined payout from the three must be $1, since one and only one of the three contracts must close at $1. You've made a guaranteed $0.03 per contract.

A combined price of $0.97 implies at least one of the stocks in the bundle of three is underrpriced. If you think you can pick which one then your returns increase by buying and holding that underpriced stock until the market corrects itself or the stock closes. But this is also more risky than the certain return on offer by buying all three. As always, risk is traded off with reward.

Remember, there is zero risk when trading a complementary bundle of stocks for a combined price that varies from $1.00.

As I write, the combined price of Obama, McCain and Other is about $1.06. If you are quick enough, there is certain money to be made right now by shorting all three in equal quantity.

iPredict on Kiwiblog

2008-09-21T15:52:00.002+12:00

David Farrar writes a weekly election commentary for our newsletter, and his thoughts are posted on Kiwiblog along with some nice big charts.

iPredict 1.1.3

2008-09-17T17:23:00.004+12:00

Today sees the release of version 1.1.3 of iPredict, with some some changes in response to feedback from iPredict users.

Changes include:

Front page prices
Spread information for each contract on the Browse Stock page
Tweaks to embedded graphing to get lines in the right place and to make embedded links hyperlink back to iPredict
Some adjustments to the left menu to (we hope) make it easier to get to your portfolio page and to credit funds - and to get to this blog

One or two bug fixes are in there as well.

We've also been talking about getting new charts. Right now, site charts are - let's face it - not great, and we are looking around for some software that can make the some of the incredible trading - WINSTON.RESIGN this week and ANNOUNCE.SEPT12 last week in particular - come alive.

Upcoming improvements include cookies to keep you logged in when you go away and come back to iPredict, email and (ultimately) text alerts when prices change, and greater bidding options.

As always, we welcome your feedback on making iPredict better. And we are most certainly listening.

We aim to put out a new software release every week. Keep those ideas coming in.

New Stocks launch today: SEATS.MAORI and MAJORITY.NAT

2008-09-17T17:18:00.006+12:00

Today we launch two contracts to do with Parliamentary seats that will be allocated to two parties – the Maori Party and the National Party, but with each stock we have a different purpose in mind.

SEATS.MAORI is a stock aimed at counting the number of electorate seats the Maori Party will win. The contract pays 10 cents for each seat won by the Maori Party. As with all our stocks, pay out is capped at $1 per contract. The purpose of this question is to measure one source of overhang in the next Parliament – that is, the total number of seats it will have.

Overhang occurs when a party wins more electorate seats than its share of the party vote would entitle it to. The current Parliament has an overhang of one seat because in 2005 the Maori Party won four electorate seats when their party vote entitled them to three seats. SEATS.MAORI asks how many electorates – Maori or General – the Maori Party will win in 2008. In combination with another contract on iPredict, VOTE.MAORI, we will be able to estimate the overhang the Maori Party will produce.

Overhang is important - it can determine who leads the next government - and tactical voting can produce large distortions. One idea floated recently was for Maori roll voters to split their vote: give their electorate vote to the Maori Party but their party vote to Greens. If every Maori Party voter did that, and if the Maori Party won seven electorate seats, then an overhang of 7 would occur, meaning a Parliament of 127.

As David Farrar pointed out earlier this year, overhang could determine the election:

Let’s say National gets 51% of the vote and 62 MPs. They should get to be the Government under MMP - this is exactly what MMP is meant to guarantee. But by this strategy of deliberate vote splitting to ensure over-hang, then Labour, Greens and Maori Party could gain 65 MPs instead of the 58 they would have on the party vote only, and get to form a Government which only a minority of NZers voted for.

SEATS.MAORI is trading at $0.5597.

Our second contract is MAJORITY.NAT, which asks a simple question: will the National Party win an absolute majority of seats in the next Parliament. This is a yes/no question which asks whether this election will produce the first FPP-style government under MMP. This contract is trading now at $0.3300.

Will Winston go today?

2008-09-17T08:22:00.002+12:00

Just before 10pm last night, iPredict launched a single day contract PETERS.SEPT17 on whether Winston Peters would be sacked or would resign on this day, September 17, after the testimony of Brian Henry delivered yesterday. This contract will close after midnight tonight - or sooner - depending on today's events.

The Prime Minister has previously said developments in the Privileges Committee hearing yesterday would have to be 'devastating' for Mr Peters to be sacked.

Yesterday's trading on our PETERS.RESIGN contract, which pays $1 per contract if Mr Peters is sacked as Foreign Minister any time before the election, suggests the testimony was at the very least highly relevant to Mr Peters future. The price for PETERS.RESIGN fluctuated hugely throughout the day as Mr Henry delivered his testimony yesterday morning, varying between 36 cents and 78 cents before midday.

The Dominion Post this morning speculates today might be the day for Mr Peters:

Miss Clark could dismiss Mr Peters as soon as today after his lawyer, Brian Henry, told the privileges committee yesterday he spoke to him seconds before e-mailing bank account details to Mr Glenn for a $100,000 donation... Miss Clark was being briefed on Mr Henry's evidence by her deputy, Michael Cullen, who sits on the committee. His assessment of whether Mr Peters' version of events is credible will be crucial in her decision on whether he stays as a minister.

Absent news, we should expect to see what we saw last week when we ran our first single day stock, ANNOUNCE.SEPT12. This stock paid $1 if and only if the election date was announced that day. Prices for that contract were at around 30 cents at 8:30am i.e. 30 percent chance of the election being called, but by 11:00am, prices had dipped to just 14 cents.

At about that time, the Prime Minister's office announced an unscheduled press conference, and all hell broke loose: ANNOUNCE.SEPT12 had soared to 90 cents at 12:30 as the press conference was about to start, and closed about 99 cents when the Prime Minister confirmed the reason for the press conference was to announce the election date - and not to sack Mr Peters.

Will we see another unscheduled press conference called today?

Arbitrage opportunities

2008-09-16T10:51:00.004+12:00

The Prime Minister markets currently show a 26% chance of a Labour-led government and a 74% chance of a National-led government.

In the Party Vote markets, Labour and its most likely allies (Green, Maori, NZ First) sum up to 53% of the vote; National and Act together sum up to 51% of the vote. Yes, the sum of all bids is about $1.04 and there's then a risk-free arbitrage opportunity for anybody who wants to short the set of all party vote stocks. There is an inconsistency across the two markets though: from the Party Vote markets, it looks like we're headed for a Labour-led government, while that's very unlikely according to the Prime Minister markets.

I'm going to go and do a bit of work to take advantage of a bit of that arbitrage, but I'm curious why nobody else out there is working on this one. Either the sum of bids on the Labour side in the vote share market is too high, or Labour's chances in the PM market is too low, or a combination of both. Any other traders out there have any insights?

New Winston Peters contract

2008-08-06T11:41:00.002+12:00

On Monday, iPredict launched a new Winston Peters contract. The contract is binary and closes at $1 if Mr Peters resigns or is sacked from his position as Minister of Foreign Affairs.

Possibly relevant is that there is only one confidence vote remaining before the election. The Appropriation (2008/09 Estimates) Bill has its third reading this week. After that, the Government no longer requires NZ First's support, possibly clearing the way for decisive action.

The PETERS.RESIGN stock opened at $0.5000, was up initially, and is currently trading down at $0.4833.

Tip: if you think the stock is underpriced, buy it. However, if you think it is overpriced, then sell, even if you don't own any. This is called short selling, its as easy as buying stock, and you make money the lower the stock price goes.

Recession Contract Launched

2008-07-02T14:43:00.007+12:00

iPredict is pleased to launch a contract on the possibility of a New Zealand recession being declared for the June quarter.

On 27 June, Statistics NZ announced GDP growth for the March quarter of -0.3%.

If Statistics NZ reports another quarter of negative growth in June 2008, a recession - defined as two consecutive quarters of negatvie economic growth - will have occurred.

Our contract pays $1 per share if this occurs. We expect Statistics NZ to report this result in late September or early October.

If the March 2008 GDP growth figure is revised to a positive value before this contract closes, then this stock pays $0.

If you have any questions on the stock's definition, please ask us for clarification.

Important Fact #2: You can ask (almost) anything

2008-06-30T14:58:00.005+12:00

In case you have forgotten already, the first important fact about prediction markets is that they work.

The second important fact is this: you can ask just about any question with a prediction market.

And in light of Important Fact #1, you should expect a robust answer.

Fundamentally, there are two kinds of questions you can ask with a prediction market. First, you can use them to forecast outcomes. Second, you can use them to test policy effects.

Forecasting outcomes is the familiar application to prediction, and the most familar of those applications is forecasting elections. Forecasting future events means using two kinds of contracts.

Binary contracts: these contracts pay $1 if a defined event comes true, and $0 if it doesn't. Will Clark be Prime minister next year? Will Key? Will the share market be up by today's close? Will petrol cost $3 by Xmas? These are all yes/no questions that a binary contract can answer.
Indexed contracts: these contracts pay out depending on the outcome of some event that isn't yes/no. For example, what share of the party vote will National get at the election? Labour? By how much will the share market change today? What will the price of petrol on Xmas day be?

Binary contracts are actually just a special case of indexed stocks - its just that their indexing ties them two exactly two values of $1 and $0.

To test the effect of policy and decision making, a third kind of contract is required:

Conditional contracts: these contracts are inexed to a variable conditional on some other event, like a new policy being set, coming true.

For example, a contract that was indexed to pay out on the unemployment rate at 31 December conditional on a tax cut occurring some time this year could be used to test the effect of tax cuts on unemployment. The test would work as follows.

Two contracts would be run. The first contract would be indexed to unemployment, conditional on no tax cut. If a tax cut occurs, traders who have taken a position on this stock are refunded what they put in. Otherwise, this stock pays out an amount per share that is indexed to the unemployment rate.

The second contract would also be indexed to unemployment, but this time conditional on a tax cut occurring. Similarly, if no tax cut occurs, traders who have taken a position on this stock are refunded what they put in. Otherwise, this stock pays out an amount per share that is indexed to the unemployment rate.

The difference in the prices these stocks trade for is the prediction market's view on the effect of a tax cut on the unemployment rate.

This prediction relies on the same wisdom of crowds, bottom up prediction methodology that has worked so extraordinarily well when pointed at elections and within firms. iPredict, by the way, is aiming to roll out conditional markets later this year. Stay tuned!

Important fact #1: These things work

2008-06-30T14:37:00.011+12:00

Although fun and simple to use, there is one thing to know about prediction markets:

They work. Really, really well.

Prediction markets have been around for about twenty years, primarily in the US, long enough to build an impressive record in Presidential elections, as the following table shows.

A similarly impressive record in predicting elections is emerging from US companies which have in the last four years adopted prediction markets in large numbers.

Market beat official forecast 6/8 (Plott ‘00)
Eli Lily markets beat official 6/9 (Servan-Schreiber ’05)
Microsoft project markets beat managers (Proebsting ’05)

Here in New Zealand, a major company we are working with has seen their average forecasting error halve in the six months they have been using wisdom of crowds technology.

Six months ago when we started our pilot with them, they didn't want their name mentioned for fear of their market being seen as a game.

Now they won't let us mention their name for fear of ceding a competitive advantage!

If there is one thing about prediction markets to take away, it is that they really do deliver extraordinary insight.

Welcome to iPredict

2008-06-30T13:43:00.009+12:00

Welcome to iPredict, New Zealand's first real money prediction market.

We'll be using this page to post weekly commentary on political markets, announce the launch of new markets, the results of closing markets, and provide some pointers to recent developments in the academic literature on predictions markets, which, by the way, is taking off.

iPredict is close to go live. Right now nearly the entire site is accessible, with only add users and add/withdraw funds functions currently unavailable. These will be switched on when we are receive clearance to operate from the Securities Commission, and then we're underway.

In the meantime, please take a tour around the site, view a video tutorial, and perhaps have a look at some of the recommended literature on prediction markets.

We think prediction markets represent a paradigm shift - not only a new way to predict election outcomes, but a new way for businesses to forecast sales, gain market intelligence, even design new products, and a new way for all public and private organisations to test social and economic policy - and all without leaving the comfort of home.

It's hard to overstate this idea's potential.

Welcome on board.