Prediction Markets = Collective Forecasting = Collective Intelligence That Predicts
You can follow all the replies to this entry through the comments feed.
on The anonymity of “Wiser Than The Crowd” will expire soon.
on The anonymity of “Wiser Than The Crowd” will expire soon. on The anonymity of “Wiser Than The Crowd” will expire soon.
on How confidential is your confidential data in the hands of online payment solutions? Is Moneybookers fully in tune with the law? on The OpenFund aims to help (Greek) entrepreneurs with fresh ideas to jumpstart their business. on How confidential is your confidential data in the hands of online payment solutions? Is Moneybookers fully in tune with the law? on The OpenFund aims to help (Greek) entrepreneurs with fresh ideas to jumpstart their business. on Behind Nigel Eccles (of HubDub fame), the remains of the West Pier in Brighton, England. on Behind Nigel Eccles (of HubDub fame), the remains of the West Pier in Brighton, England. on Behind Nigel Eccles (of HubDub fame), the remains of the West Pier in Brighton, England.
Yes, there is a fundamental problem that diminishes the intensity of the predictive power: It is called time.
The further away the expiration, the more things can happen that can change the outcome of the event.
Increased liquidity (achieved by paying interest on deposits) is unlikely to resolve the problem.
Remember, markets aggregate all available information. If until the event we expect to see a lot more additional information, then the predictive power will be necessarily diminished, no matter how much people trade.
Try to answer: What party will win the presidency of the US in 2052 (Dem, Rep, Other)? No matter what the interest that you pay, the market is unlikely to give you any non-random answer at this point.
Some long-term questions like those regarding climate and technology should be served reasonably well by interest payments and leverage.
A nice feature would be being able to post any collateral (t-bills, shares of ETFs, etc) in a prediction market. Yes that would introduce a large “basis risk” to the prediction, but remember, you think both the collateral and the prediction are good investments. You still have exposure to the prediction — especially if its payoff is leveraged with respect to the collateral.
Good point. I guess we should have some notion of “information flow speed” or something along these lines to capture that in some markets the necessary information arrives very close to the event, while in other markets the information flows much more slowly (like in the markets that you described).
In such markets, yes, you need the risk-free payment to encourage people to invest. Or you can correct for the lack of such payment and infer the “real” probabilities based on the price of the contracts that do not offer interest. (Should be rather easy to do the correction.)
Yes, there are two issues, attracting trade and correcting for discounted prices. Of course if the collateral isn’t uniform, clearing is very dangerous!
I was just wondering why the Intrade markets always consist of only one selection.
For instance, the 2008.GLOBALTEMP.TOP5 market.
Why not offer different selections in one market, as follows ;
- 1 to 5
- 6 to 30
- 31 to 100
- other
Is it because Intrade feel that by making markets inefficient, they can charge their customers more ?
Doesn’t make sense to me. Maybe they are just stupid. Maybe I am.
Same applies to financial markets, especially the financial markets.
“Yes, there is a fundamental problem that diminishes the intensity of the predictive power:”
- the exchanges don’t pay interest on deposits.
OK, let’s see how “fundamental” is this problem:
The risk free interest rate today ranges from 0.1% (Chase Savings) to 2.5% (for a pretty good money-market fund).
So, let’s see the effect on contract with expiration date 12-months ahead.
If the price of a contract is p in a market that pays interest (i.e., is objective), then the price in the current no-interest-paying markets would be p/1.025. In other words, a contract that should have been at 0.9 is trading at 0.87. A contract that should have been at 0.1, trades at 0.097.
For a contract with expiration 10-years from now, then we have to multiply the current price with 1.025^10 = 1.28. In that case the 0.90 contract trades at 70 and the 0.1 trades at 0.7028.
I can live with that level of bias.
And I kind of doubt that markets lack participation because the exchanges do not pay interest. Making the exchanges legal in the US, publicizing the legitimate uses, and removing the gambling “stigma” would be a much more important motivator to get people to participate.
I don’t believe that analysis is correct Panos.
With exchange betting the purse-holder is holding on to both parties monies. Both parties have to lock up money for a considerable amount of time. Hence, there is no incentive to get involved in long term bets.
When a market consists of more than one selection, liquidity can jump from one selection to another depending on where the price is at. The other selections will then adjust automatically as there will be traders who operate like bookmakers on the exchange.
I think the Intrade market approach is wrong, but happy to be corrected.
“A contract that should have been at 0.1, trades at 0.097.”
I don’t think this is right because the seller is the one tying up more funds . The price should be closer to .12, or 1 minus the discounted opposite contract. The prices at extremes will be biased towards the midpoint, this bias will fall off as you move away from the extremes, and at 50% there will be no bias but a wide bid/offer spread.
I just pointed that the risk-free rate today is so low, that the effect of interest rate is minimal. Only in multi-year contracts there is some significant effect.
Consider a contract trading at price p, and expiring in a year. The person who buys should believe that the outcome will occur with probability higher than p*(1+r), where r is the risk free rate (r=0.001 to 0.025 today). This ensures that the expected return is higher than the return gained by simply keeping the money in a safe account. Similarly the person that sells the contract expects that the event does not occur with probability higher than (1-p)*(1+r).
Essentially, the beliefs of the two parties about the outcome of the event should have difference more than 0.002-0.05 (for a contract that pays $1), for a contract that expires in a year. This is almost equivalent to having an implicit transaction cost that discourages people from trading in small probability changes. And this implicit transaction cost decreases over time as we approach expiration.
I admit that interest payment is important is in multi-year contracts in which the important information arrives slowly over time, and we do not expect to see all the relevant information released in the few months before the event. (See Jason’s comment above.)
Betfair are offering the 2012 US presidential election market. It’s a 4 year bet with no interest payment.
Above all, it seems betfair’s intention to hook customers for 4 years to the exchange because there is no way you can get out of that bet. In the heat of the battle, some people have fallen for that trap, one of betfair’s specialties. How does a gambling addict terminate a contract if he wishes to get out completely ?
And it’s not just interest payments. Who is to say betfair will still be around 4 years from now, and what happens if they go bust 3 years from now ?
The very moment betfair start offering value to their customers, any value, drinks are on me. I’ll come to New York and we’ll have a party.