Prediction market

[Babylon]

Ok, I don't have one yet, My coding skills may or may not be up to actually making one.  But I thought a prediction market in bitcoins would be a good idea.  Because of their anonymous nature it would allow for a wider range of possible predictions than the currently available markets.  My thought is a market in which anyone can offer something to be predicted, with the site admins taking a cut for judging which way it has gone.  Things to be predicted could be submitted and then the judges would decide if they are able to judge that particular thing, and how much they would charge to do so, if they can't judge it it is rejected, if they can but the judge's fee is not met then the market is dissolved at the end and bids are returned.  If they can judge it and the judge fee is met that is subtracted and the rest divided among those with correct predictions based on their market share.

[FreeMoney]

I think multiple peer judges who don't know who each other are until afterwards and get paid for choosing the same way as the majority might be a good way to go.

[Babylon]

I agree.  The challenge then becomes how to decide the judging fee.  Obviously something like "Obama wins presidential election in 2012" is much easier to judge than "homelessness rate in Topeka Kansas doubles by dec 31st 2011" (I had trouble coming up with a really hard one, but I know they are out there) 

[llama]

I've wanted a predictions market since I first heard about bitcoin.  It is the perfect medium for a predictions market. 

As for judging fee, it should be pretty minimal.  Even for "hard" ones like "homelessness rate doubles", the criteria just needs to be very very specific.  For example:

"On July 20th 2011 The annually reported homelessness rate by detroit.gov will be equal to or above two times the number reported on July 20th 2010.  Should Detroit.gov not publish a homelessness rate on July 20th 2011, then the number reported by..."

As long as you give ridiculously detailed criteria, then disputable outcomes should be very very rare.

[Quantumplation]

Haha, Babylon:  That's very easy to judge, but different people will come up with different answers, and still consider it an "easy" choice. =P

[Babylon]

I've wanted a predictions market since I first heard about bitcoin.  It is the perfect medium for a predictions market. 

As for judging fee, it should be pretty minimal.  Even for "hard" ones like "homelessness rate doubles", the criteria just needs to be very very specific.  For example:

"On July 20th 2011 The annually reported homelessness rate by detroit.gov will be equal to or above two times the number reported on July 20th 2010.  Should Detroit.gov not publish a homelessness rate on July 20th 2011, then the number reported by..."

As long as you give ridiculously detailed criteria, then disputable outcomes should be very very rare.

That still requires judgement if the submission of predictions is open, which I think it should be.  You need someone to judge if a question is sufficiently precise.  This could, of course, be the job of the prediction market admin, who would hire on more admins as the market grew to require more judges.  I'm just not sure if that is the most elegant solution.

[FreeMoney]

I think it can be taken care of naturally.

When you propose a prediction you spell out all terms in detail, including the judging fee. When X judges have volunteered you randomly pick a subset of them and let the trading begin.

As a judge or bettor I'd be most comfortable with very clear terms. If judges only get paid when they agree with a majority then judges will prefer easy to judge terms because they are much more likely to get paid. It doesn't have to be a simple majority either. Maybe 3/4 or 7/10, this can be specified in the details.

I expect standard forms for terms would emerge and would be self reinforcing because it's easiest to predict how people will judge in established formats.

If someone gets this up, I promise to lose a lot of money predicting all sorts of things inaccurately.

[Babylon]

I think it can be taken care of naturally.

When you propose a prediction you spell out all terms in detail, including the judging fee. When X judges have volunteered you randomly pick a subset of them and let the trading begin.

As a judge or bettor I'd be most comfortable with very clear terms. If judges only get paid when they agree with a majority then judges will prefer easy to judge terms because they are much more likely to get paid. It doesn't have to be a simple majority either. Maybe 3/4 or 7/10, this can be specified in the details.

I expect standard forms for terms would emerge and would be self reinforcing because it's easiest to predict how people will judge in established formats.

If someone gets this up, I promise to lose a lot of money predicting all sorts of things inaccurately.

heh, win or lose the hosting site makes money because they take a small cut.  Like Mt Gox exchange.  Any of you coding sorts out there have an idea how this could be set up?  my coding skills are pretty minimal but I'd love to be involved in some way.

[FreeMoney]

Sure the site makes money off of hosting, but having a big spending loser will attract others, which will give the site more to rake. ;-)

Of course, my prediction is that my predictions are bad, so what's that worth?

[Babylon]

Sure the site makes money off of hosting, but having a big spending loser will attract others, which will give the site more to rake. ;-)

Of course, my prediction is that my predictions are bad, so what's that worth?

That'd be an interesting prediction for you to set up.

"User freemoney will lose more money than he makes on bitpredictions site between August 16th 2010 and August 31st 2010"

I am checking out open source prediction market software right now. 

[Quantumplation]

I'm not sure I understand the concepts at play here, but if you give me detailed specifications on what it should do, I could probably code it for you ( for a small bitcoin fee, of course =) ).

[Babylon]

ok, it's a futures market.  The things that users would be able to do would be suggest something to predict (such as "Will Barak Obama win the 2012 Presidential Election" or "Will there be more than 3" of rainfall on June 13th 2011 in the greater Seattle Area as reported by the Seattle Post Intelligencer") this puts the predictions into a "looking for judges" pool.  The person proposing the prediction also sets a judging fee, which will be subtracted from the total and divided equally among judges voting in the majority.  Once enough judges sign up the prediction is opened to the users who can then purchase either a yes or a no with a certain spread between them.  The price of yes and no's updates with each transaction and is based on how many users have chosen that answer (and how many shares in that answer they have purchased).  When the issue is resolved the total is divided among the users based on share ownership.  It would need to be secure, and it would probably be a good idea for users to be able to establish accounts which they may deposit bitcoins into, withdraw bitcoins from, use to purchase yesses or noes, and have bitcoins deposited into when a market closes.

From a quick look at Open Source possibilities for this it looks as if Serotonin Prediction Markets may be an already existing solution.  However I have not dug into just what it is and how it works, or how easy it would be to convert it to bitcoin.

I am, at the moment, poor in both cash and coin,  I do have access to some server space and might be able to cover coder fees, depending on how much they come to.  If I cannot I may also consider a shared enterprise if people are interested in joining as financial backers. (obviously the business model needs some work before it gets to that point)

[Quantumplation]

What would the judges do? 

[Babylon]

What would the judges do? 

Judge the results.  So let's say it is the day after election day and Obama has just won the 2012 election.  The judges would log in and click on "yes" and the users that had bought yes shares would be paid out.

This is an extremely easy prediction, since it is covered in international news and verified in strict ways.  The market would be able to handle both easy and hard judging situations however.  If someone set up a prediction for instance, that "The world will be a better place on December 31st 2012 than it is today (July 19th 2010)"  this is an incredibly subjective question.  However it can still be judged and what people are essentially speculating on is what the majority of selected judges will decide in regards to the question.

[Quantumplation]

Ah ok.

Proposed revision:  Once you open up the question to the public, they vote on the prediction, as well on a 1 to 10 scale of subjectivity.  This is then averaged, and determines the "percentage split" between how the funds are split up majority/minority somehow.  So an extremely subjective question, the judges aren't penalized for their answer as much, because it could have gone either way.

[Babylon]

makes sense.  0-4 might be a better scale, with 0 meaning 100% goes to the winners and winning judges, 1 meaning 10% to the minority, 2 meaning 20% etc.  (of course this would get averaged and could be given to users as a 1-5 scale)

[Bitcoiner]

This is actually a pretty interesting idea. I might start doing something like this in my spare time, just for fun

[FreeMoney]

I'm wary of sending much money to near strangers before anything has been done. But if the basics are built I'll buy in then and supply money to improve, market, maintain, etc.

[Anonymous]

Is this similar to the hollywood stock exchange?

http://www.hsx.com/

People predict how movies and stars will fare and can buy stocks based on that.

[Quantumplation]

Obviously I wouldn't request money until services had been rendered. =P  Also, I've got my plateful with 4 other projects, so it's low on the priority list, but if no ones stepped up before then, I'll take a crack at it.