I am unclear on the details of this proposal. Are you proposing that the market for 20xx resolves according to whether the market for 20xx+1 has a price above 50% at the end of 20xx+1? If so, suppose I apply this scheme for a market which resolves YES if the first six-sided die I roll in 2030 comes up "6". Wouldn't all the intermediate markets come up 0%, leaving me with no information about the actual probability of the die roll? Wouldn't there be incentive to manipulate the markets?
Or maybe you are proposing that the 20xx market resolves-as-percent to the 20xx+1 price in a year? But then taking the lightspeed example, wouldn't I just expect the price for 2020 to be 90%, the price for 2021 to be 91%, the price for 2022 to be 92%, all the way up to 100% at 2030?
It seems to me that this problem actually has no real solution, but that the cleanest way to ameliorate the bias problem is just to denominate the 2030 market in 2030-dated mana futures, or make it harder in other ways to get out of the market once you are in.
Thank you for the comment. I see that you've already discovered the more detailed follow up article I wrote, but in case others are reading, I'll provide it here ( https://abstraction.substack.com/p/iterative-markets-explained ). Meanwhile, I agree that iterative markets are prone to manipulation, so I think this is mostly trying to get the idea out there for others to think about potential ways to use it.
This sounds a lot like options markets that gradually move from "out of the money" to "in the money" with the sellers getting the benefit of the premium and the buyer getting the benefit of being long some event they think is much more likely than the market.
I like this general idea, Jonathan. It really is a serious problem for prediction markets.
But it's not obvious to me how closing and reopening markets every year changes the incentive for traders. Market transactions are already effectively bets on the future price of a security. I don't see how this proposal would prevent contracts from being priced inefficiently for extended periods. The FTL market should open at close to 100% in 2025. But why do traders have more incentive to make that happen when you iterate the market when the underlying contract doesn't close until 2030?
I've always thought the fundamental issue for prediction markets is liquidity. These markets need more money to chase marginal returns. I think many of the inefficiencies would go away if the Goldmans of the world had big prediction market portfolios.
Thanks Robert. I realize I didn't explain this well in the post, but the idea is for betting on the sentiment / directionality of the next linking market, so the current market resolves after the directionality of the next market in the chain is determined. I'll make a demonstration on manifold and post a link to it to make this more concrete.
Interesting. It would be really interesting to see how this would work in practice. It seems to me that betting on the directionality of a market might distort it in other ways. Do you know if this kind of market has been implemented anywhere?
I am unclear on the details of this proposal. Are you proposing that the market for 20xx resolves according to whether the market for 20xx+1 has a price above 50% at the end of 20xx+1? If so, suppose I apply this scheme for a market which resolves YES if the first six-sided die I roll in 2030 comes up "6". Wouldn't all the intermediate markets come up 0%, leaving me with no information about the actual probability of the die roll? Wouldn't there be incentive to manipulate the markets?
Or maybe you are proposing that the 20xx market resolves-as-percent to the 20xx+1 price in a year? But then taking the lightspeed example, wouldn't I just expect the price for 2020 to be 90%, the price for 2021 to be 91%, the price for 2022 to be 92%, all the way up to 100% at 2030?
It seems to me that this problem actually has no real solution, but that the cleanest way to ameliorate the bias problem is just to denominate the 2030 market in 2030-dated mana futures, or make it harder in other ways to get out of the market once you are in.
Thank you for the comment. I see that you've already discovered the more detailed follow up article I wrote, but in case others are reading, I'll provide it here ( https://abstraction.substack.com/p/iterative-markets-explained ). Meanwhile, I agree that iterative markets are prone to manipulation, so I think this is mostly trying to get the idea out there for others to think about potential ways to use it.
This sounds a lot like options markets that gradually move from "out of the money" to "in the money" with the sellers getting the benefit of the premium and the buyer getting the benefit of being long some event they think is much more likely than the market.
I like this general idea, Jonathan. It really is a serious problem for prediction markets.
But it's not obvious to me how closing and reopening markets every year changes the incentive for traders. Market transactions are already effectively bets on the future price of a security. I don't see how this proposal would prevent contracts from being priced inefficiently for extended periods. The FTL market should open at close to 100% in 2025. But why do traders have more incentive to make that happen when you iterate the market when the underlying contract doesn't close until 2030?
I've always thought the fundamental issue for prediction markets is liquidity. These markets need more money to chase marginal returns. I think many of the inefficiencies would go away if the Goldmans of the world had big prediction market portfolios.
Thanks Robert. I realize I didn't explain this well in the post, but the idea is for betting on the sentiment / directionality of the next linking market, so the current market resolves after the directionality of the next market in the chain is determined. I'll make a demonstration on manifold and post a link to it to make this more concrete.
Interesting. It would be really interesting to see how this would work in practice. It seems to me that betting on the directionality of a market might distort it in other ways. Do you know if this kind of market has been implemented anywhere?
The closest example I can think of is Scott Alexander's forecasting contest for this year asking which candidate for the Republican primary will be leading in prediction markets at the end of this year: https://manifold.markets/ACXBot/18-at-the-end-of-2023-will-predicti?r=Sm9uYXRoYW5NYW5u
I should also mention that someone can arbitrarily set the directionality threshold. For example "Will ABC market be below 5% on such and such date" e.g. https://manifold.markets/jack/when-will-we-know-whether-the-room?r=Sm9uYXRoYW5NYW5u