Author Topic: Simple Binary Prediction Market Discussion  (Read 24570 times)

0 Members and 1 Guest are viewing this topic.

Offline vegolino

  • Sr. Member
  • ****
  • Posts: 450
  • Reality is Information
    • View Profile
Bytemaster,
I'm very glad you brought up this topic.  As it shows that you are considering adding a prediction market functionality to UIA. 

I think you know where I am going with this.

I earnestly encourage you to take a serious second-look at Truthcoin.  I truly feel Truthcoin will be a win-win for Bitshares.   Putting aside your issues of it for now, from my perspective, Bitshares already has a white-listing function to freeze and retrieve funds from accounts.  I think this works in total concert with Truthcoin's SVD process, where resolving markets requires profit-based incentives to punish bad voters and withdrawing their Truthcoins.  I'm not a developer by any means, but from a conceptual standpoint, your already half-way there.  The SVD rules just need to be built in.  Paul (who wrote the whitepaper if you remember) points out that his idea way simpler to implement than all Bitcoin 2.0 projects out there.  And there is already some codebase from Augur.net and Truthcoin floating out there.  It would save your team time.

However, I understand you will probably not want to do this. This could delay the 1.0 for many many months.  I don't know if this could take 80 man-hours.  Unless your planning this prediction market for a 6 month 2.0 release.

I'll note again, as I have on other forums, that Ethereum and a Bitcoin fork all right now have developments in Decentralized Prediction Market that are seeing some sort of release by August (caveat could be later because of project mgmt).  But ultimately I think not making a decision here may result in Bitshares to once again fall behind the curve.  If we were to get on this first, alongside our marketing campaigns we stand to gain tremendously.
  +5%

Offline bytemaster

It comes down to time value of money.  If the winner has 95 at stake and the loser 5 then the cost to the loser is much less by waiting.   You can change the ratio by shifting the 0 point to .5, in this case the 50/50 odds would be .75.   

As much as liars are irrational, so are those who seek justice.   Let the two of them fight it out for a fee while rational individuals settle. 
For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.

Offline Bitcoinfan

  • Sr. Member
  • ****
  • Posts: 240
    • View Profile
Bytemaster,
I'm very glad you brought up this topic.  As it shows that you are considering adding a prediction market functionality to UIA. 

I think you know where I am going with this.

I earnestly encourage you to take a serious second-look at Truthcoin.  I truly feel Truthcoin will be a win-win for Bitshares.   Putting aside your issues of it for now, from my perspective, Bitshares already has a white-listing function to freeze and retrieve funds from accounts.  I think this works in total concert with Truthcoin's SVD process, where resolving markets requires profit-based incentives to punish bad voters and withdrawing their Truthcoins.  I'm not a developer by any means, but from a conceptual standpoint, your already half-way there.  The SVD rules just need to be built in.  Paul (who wrote the whitepaper if you remember) points out that his idea way simpler to implement than all Bitcoin 2.0 projects out there.  And there is already some codebase from Augur.net and Truthcoin floating out there.  It would save your team time.

However, I understand you will probably not want to do this. This could delay the 1.0 for many many months.  I don't know if this could take 80 man-hours.  Unless your planning this prediction market for a 6 month 2.0 release.

I'll note again, as I have on other forums, that Ethereum and a Bitcoin fork all right now have developments in Decentralized Prediction Market that are seeing some sort of release by August (caveat could be later because of project mgmt).  But ultimately I think not making a decision here may result in Bitshares to once again fall behind the curve.  If we were to get on this first, alongside our marketing campaigns we stand to gain tremendously.   



« Last Edit: February 04, 2015, 03:01:37 pm by Bitcoinfan »

Offline Bitcoinfan

  • Sr. Member
  • ****
  • Posts: 240
    • View Profile
You've embraced the notion that every should act like adults and be rational.  But markets can be irrational for a loooooooong time-- thus the requirement for a resolution.

See you, based on the behavioral finance field, the Nobel Prize winner Daniel Kahneman wrote in his book Thinking Fast and Slow, the pain of losing money is (-$100), called loss aversion, is always greater than is the joy that the winner has from gaining money (+$100). 

http://en.wikipedia.org/wiki/Loss_aversion

Therefore this creates a asymmetric scenario.  The losers will always have greater leverage than the winners, because they will do everything they can to avoid that loss because it hurts so much more.  Think of it like the kid who plays board games with the adults and whines at the very end when he loses. 

For the child, it hurts so much more.  The adults don't understand how much it hurts.  But it pains the kid to the core.  The kid is playing another game than the adults.  He's playing the I have more information game (regardless of the facts of the game) because only he knows who long he can stall the game.  It could be for 5 minutes or maybe he didnt have his breakfast and is unusually more irritated and is planning to stall for the whole day. He has greater power to manipulate the adults.  The adults have no idea how to value that because this is the neighbors child and they don't have any history of his behavior. 

The adults don't want to put up with the child because they don't have the time to, and the reward of the game doesn't justify for them the value of making the child lose.  So they let him win.  But in the end the game is never played again because the adults have other opportunity costs and wouldn't want to bargain with this game in the future because its already set a precedent and somebody else might turn like the kid (maybe another adult).  Why not? The precedent has already been set before. 

I'll end my example with a question.  What if liars come in and buy cheaply into a market with the expectation of stalling longer than the honest bettor?  What if the unfairness of stalling has turned away all honest people?

« Last Edit: February 04, 2015, 02:31:49 pm by Bitcoinfan »

Offline Bitcoinfan

  • Sr. Member
  • ****
  • Posts: 240
    • View Profile
This was also briefly discussed on the truth coin forums. 

http://forum.truthcoin.info/index.php/topic,149.0.html

Markets need to have a guarantee of resolve.  Otherwise if there was no guarantee, what economic incentives would that give to the "honest bettors"  to come in and align the markets so that it is justified.  Not to mention that both sides of the bet could never agree to a sum to end the pm, because the losing bettor will always has greater value if he holds out.  By playing the waiting game He has more to gain from disgruntling the other side into agreeing at any price.  So just from waiting he will stand to gain more than nothing.  This internal market conflict in this transaction  happens irregardless of what actually happened on the real world.   If your telling the loser they can holdout and be free of face economic consequences but you may actually gain, then they will do it every time, frustrating the honest bettors.  As a result honest bettors will never want to enter.  The cost of resolving would be too much.   There is no causal link between the real world and the prediction market because economic incentives were not guaranteed to correspond with actual outcomes,  therefore people have no trust to bet according to their actual beliefs. 


Sent from my iPhone using Tapatalk

Offline Markus

  • Sr. Member
  • ****
  • Posts: 366
    • View Profile
I believe without any crutches (like the price feed for BitAssets) this might lead to very asymmetric average settlements.

Wins will settle much closer to 1 than losses to 0.

If I buy in at 0.05 after the prediction is already lost I can gain 100% waiting for somebody to buy at 0.1. This leverage is not possible for won predictions trading at 0.95, so I wouldn't bother with those.

Offline Empirical1.2

  • Hero Member
  • *****
  • Posts: 1366
    • View Profile
Could there be some people unable to find any settlement due to people who lose their wallets/forget about their trade/arrested/die/other?

If so, I forget how prediction markets work but maybe after 30 days the system can force settle at 1 or 0 depending on if the majority of trades were settled above 0.8 or below 0.2 in the 7 days after the event finished.

This should get everyone to settle at very close to 1 or 0 right after the event and it is very difficult to game as there's a three week period in which people could seek delegate intervention which will probably never be needed.

You could also charge a fee to losers who waited 30 days to encourage early settlement at very close to 1 or 0 too.
« Last Edit: February 04, 2015, 12:32:05 pm by Empirical1.2 »
If you want to take the island burn the boats


Offline arhag

  • Hero Member
  • *****
  • Posts: 1214
    • View Profile
    • My posts on Steem
  • BitShares: arhag
  • GitHub: arhag
From bingo to virtual mining pools to prediction markets, you are always looking for a way to add demand for BitShares (which is great!). But of those three, prediction markets are one that I am actually interested in.

Now, as for this specific proposal, I am not so sure about the game theory (I guess I need to think through it more). But can you tell me what makes this different that you think it is a workable model when you argued against (what I think is) a similar game with NashX?
« Last Edit: February 04, 2015, 10:08:29 am by arhag »

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
feels like an elegant solution .. +5%!

Offline bytemaster

Define a user created asset that can be issued 1:1 by anyone putting up collateral in a specific bitasset.
Define the prediction criteria in the description of the asset.
Allow collateral to be freed by burning the issued asset.
That is all.

There is no need for a judge to settle the bet.

Now how does the game theory operate?   The issued asset will trade based upon the markets belief that it someone will buy it back.  Market participants will buy it back because they have collateral at stake.    If it wasn't tied to a prediction criteria then the asset would sell at a price of 0.5 because that is the Nash Equilibrium price at which both parties break even. 

Once you tie a prediction to it the Nash Equilibrium changes because the parties have a sense of "justice".  If there are only two people and they are "honest bettors" then they would simply settle at the honest outcome.   

If one of the players is a "sore looser" then they may demand 0.1 for a "false" outcome or only offer 0.9 for a "true" outcome.    The ability of a "sore looser" to hold out depends upon the value the winner places on justice.  If the winner is "hard core" he will simply hold out until the looser caves and agrees to settle honestly.   If the winner is more practical he will offer the looser a financial incentive to settle at a reasonable price.   

There is financial incentive for the looser to settle at any price above 0 or just slightly below 1, but no incentive for them to settle for nothing.   

There is financial incentive for a winner to settle at less than full price so they can exit the market. 

If there were only two players and both place a bet with dishonest intentions then they are in an "all or nothing" choice.  If they settle then they break even, if they don't then they both lose.   

If the market never closes and people are allowed to enter the Nash Equilibrium at any time then you end up with a market where people are betting on the price of justice. 

Say an election ends and the winner is known.   The best settlement offered by any loser is 60/40 which isn't particularly appealing to someone who won and was expecting something closer to 95/5.   In this case the best settlement offered by any winner is 80/20.   At this point in time speculators can bet on how long the sore looser will hold out and what price he will ultimately settle for.    A speculator will take the 80/20 settlement and hold out for a 90/10 settlement. 

Over time the market will discover the long-term price at which losers would rather get nothing than something and winners would rather get something than nothing.    Once the market establishes this price then we can calibrate our prediction markets.   Say the "long-term" settlement is 90/10, then a prediction value of .9 == 100% true and a prediction value of .1 == 100% false and everything can be scaled accordingly.   

Because long-term settlement risk is being priced in from the start all markets will initially track in direction even if they don't track in magnitude until enough data is gathered to allow proper calibration.

 

For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.