This is interesting. They have been operating iPredict since 2008, but apparently got a “no action” letter from the Commodity Futures Trading Commission in the US to allow US participants in the market (as long as they limited the markets in the same way Iowa University does for the IEM) 4 months ago.
The market isn’t particularly efficient. For example, if you bought “No” on all the presidential candidates to win, it would cost $16.16, but would be worth at least $17 for a 5% gain. Of course, after paying the 10% fee on profits and 5% withdrawal fee you would be left with a loss, which is why this opportunity still exists.
The market isn’t particularly efficient. For example, if you bought “No” on all the presidential candidates to win, it would cost $16.16, but would be worth at least $17 for a 5% gain. Of course, after paying the 10% fee on profits and 5% withdrawal fee you would be left with a loss, which is why this opportunity still exists.
Does this affect the accuracy of the market? Serious question; I do not understand the nitty-gritty economics very well.
Does this affect the accuracy of the market? Serious question; I do not understand the nitty-gritty economics very well.
Yes. Think of the transaction costs as an upper bound on error. Any wrong beliefs that are off by more than the transaction costs are “free money” to correct, and thus people will spend time and effort looking for and correcting those errors, but any wrong beliefs that are off by less than the transaction costs cost money to correct, and so won’t be corrected.
For example, suppose there were a 6% difference between reality (here, the union of mutually exclusive and exhaustive options) and the market. Then the 10% profit fee would drop that down to 5.4%, and then the 5% withdrawal fee would drop that down to 0.4%, and there would be an opportunity to make money—until the price had shifted so the difference between reality and the market was 5%.
(If you already have money in the system, and thus the 5% withdrawal fee is a sunk cost, any mispricing is worth correcting. Putting money into the system is what the withdrawal fee disincentivizes, and so only mispricings more than 5% will attract new money.)
[Edit]And, of course, the fact that the contracts are not instant adds further costs. Yes, you could buy up the contracts and get a certain 0.4% gain (in the previous example), but if that’s a certain 0.4% gain six months from now that may actually represent a negative real return, and thus the actual error upper bound is even higher.
It directly affects your error estimates. If by “accuracy” you mean “forecasting capability” then inefficiency is a symptom of underlying problems (e.g. too few market participants) which do affect accuracy.
This is interesting. They have been operating iPredict since 2008, but apparently got a “no action” letter from the Commodity Futures Trading Commission in the US to allow US participants in the market (as long as they limited the markets in the same way Iowa University does for the IEM) 4 months ago.
The market isn’t particularly efficient. For example, if you bought “No” on all the presidential candidates to win, it would cost $16.16, but would be worth at least $17 for a 5% gain. Of course, after paying the 10% fee on profits and 5% withdrawal fee you would be left with a loss, which is why this opportunity still exists.
It’s not really arbitrage if you lose money.
Does this affect the accuracy of the market? Serious question; I do not understand the nitty-gritty economics very well.
Yes. Think of the transaction costs as an upper bound on error. Any wrong beliefs that are off by more than the transaction costs are “free money” to correct, and thus people will spend time and effort looking for and correcting those errors, but any wrong beliefs that are off by less than the transaction costs cost money to correct, and so won’t be corrected.
For example, suppose there were a 6% difference between reality (here, the union of mutually exclusive and exhaustive options) and the market. Then the 10% profit fee would drop that down to 5.4%, and then the 5% withdrawal fee would drop that down to 0.4%, and there would be an opportunity to make money—until the price had shifted so the difference between reality and the market was 5%.
(If you already have money in the system, and thus the 5% withdrawal fee is a sunk cost, any mispricing is worth correcting. Putting money into the system is what the withdrawal fee disincentivizes, and so only mispricings more than 5% will attract new money.)
[Edit]And, of course, the fact that the contracts are not instant adds further costs. Yes, you could buy up the contracts and get a certain 0.4% gain (in the previous example), but if that’s a certain 0.4% gain six months from now that may actually represent a negative real return, and thus the actual error upper bound is even higher.
It directly affects your error estimates. If by “accuracy” you mean “forecasting capability” then inefficiency is a symptom of underlying problems (e.g. too few market participants) which do affect accuracy.