It's still early (using play money, only binary yes/no markets supported) but we're hoping that a really simple interface for individuals to ask questions will push the frontier on how people actually use these things. Let us know if you have any thoughts!
Thanks so much !! And Mantic is very cool. My few questions on prediction markets as you've done comes down to these below:
1. Liquidity - how many informed and independent traders do you need (as units and $) before its valuable? What makes it truly useful, like Nasdaq Vs a low liquidity bad price signal, like the OTC equity markets
2. Tractability - knowing what you're betting on is hard (Byrne's point above too). It's not as simple as putting a $ value as in his Oil example esp if it includes also a prob distribution. There's a long tail ecosystem behind each prediction after all.
3. Adversarial insider trading ,- tension between market maker and insiders w knowledge (Byrne's worry), though this shouldn't be an issue most cases imo
4. General - I also worry folks like me (and you) are overused to thinking in terms of bets, whereas most of the public hates this mode. Yet they're invaluable in fin markets because they don't see it as a pure bet. I find this a difficult psychological barrier.
These aren't showstoppers but are genuine wonders I have about the widespread use of markets, for instance in corporate or political settings!
Dec 18, 2021·edited Dec 18, 2021Liked by Rohit Krishnan
I'm glad you like Mantic, and those are great questions -- thanks for posing them!
Re liquidity & insider trading: we're using an automated market maker to provide liquidity, rather than the more traditional centralized order book approach. This draws on the crypto insight that AMMs function better in situations with less volume.
To geek out a bit more, our AMM is a dynamic parimutuel market -- a setup that doesn't require the presence of liquidity providers to take on risk. Roughly speaking, bettors contribute to either the YES or NO pools, and are paid out proportional to the size of and odds at time of their bet. The dynamic parimutuel system is pretty novel, but we think it solves a lot of the liquidity/risk issues in user-created markets!
---
Re tractability: If i'm understanding you correctly, you're saying that any market will have potential ambiguities around the wording vs the intent of the question?
We're hoping this is one of the things that user-created markets alleviates; when you buy into e.g. Rohit's market on "Will Strange Loop Canon receive more viewers in 2022 than 2021", you're actually betting on "*Will Rohit decide that* Strange Loop Canon received more viewers...". Issues like "how are viewers counted? do views on guest posts count?" are left in the hands of the market creator. Then, bettors will seek out markets that are decided by creators whose judgement they trust.
(This might be kind of a dodge; we don't actually know! But a cheap web prototype should teach us and the rest of the world a lot)
---
Re thinking in bets: yeah, I've definitely seen an aversion to "betting" among the public. Some (fairly smart!) friends I've talked to churn off as soon as I talk about betting, saying things like "I don't believe in gambling" or "I don't like taking on risks".
In the short term, we'll promote these markets to the rationalist/EA audience; longer term, I could envision different UIs that make more sense for the populace. One wild hypothetical: we could present a Reddit-like interface with up/downvotes on each question (allowing you to "vote" repeatedly) -- and behind the scenes dynamically readjust your entire portfolio of bets, using the Kelly criterion to optimize your returns? But yeah, how to frame prediction markets for the median web user is a super-interesting question!
Hey thanks for the detailed reply! Really appreciate it. The latter two points I get it - honestly it seems like the thing we'll just have to try and see. Which I endorse.
Re the first it's less of the mechanics (though the parimutuel market seems cool). It's the q of how many investors or how many dollars are needed for proper price prediction. Eg in OTC equity markets tickers trade for like a few thousand shares a day, which isn't enough for either narrow bid ask spreads or price discovery and brings vol.
Woah, this interview has got to be the most ambitious crossover event in (econ Substack) history!
Re: prediction markets being weirdly underused - this is something I'm super curious about. So much that we built https://mantic.markets/, a prediction market interface where anyone can pose their own questions! Here's an example: https://mantic.markets/ManticMarkets/will-100-people-have-signed-up-for
It's still early (using play money, only binary yes/no markets supported) but we're hoping that a really simple interface for individuals to ask questions will push the frontier on how people actually use these things. Let us know if you have any thoughts!
Thanks so much !! And Mantic is very cool. My few questions on prediction markets as you've done comes down to these below:
1. Liquidity - how many informed and independent traders do you need (as units and $) before its valuable? What makes it truly useful, like Nasdaq Vs a low liquidity bad price signal, like the OTC equity markets
2. Tractability - knowing what you're betting on is hard (Byrne's point above too). It's not as simple as putting a $ value as in his Oil example esp if it includes also a prob distribution. There's a long tail ecosystem behind each prediction after all.
3. Adversarial insider trading ,- tension between market maker and insiders w knowledge (Byrne's worry), though this shouldn't be an issue most cases imo
4. General - I also worry folks like me (and you) are overused to thinking in terms of bets, whereas most of the public hates this mode. Yet they're invaluable in fin markets because they don't see it as a pure bet. I find this a difficult psychological barrier.
These aren't showstoppers but are genuine wonders I have about the widespread use of markets, for instance in corporate or political settings!
I'm glad you like Mantic, and those are great questions -- thanks for posing them!
Re liquidity & insider trading: we're using an automated market maker to provide liquidity, rather than the more traditional centralized order book approach. This draws on the crypto insight that AMMs function better in situations with less volume.
To geek out a bit more, our AMM is a dynamic parimutuel market -- a setup that doesn't require the presence of liquidity providers to take on risk. Roughly speaking, bettors contribute to either the YES or NO pools, and are paid out proportional to the size of and odds at time of their bet. The dynamic parimutuel system is pretty novel, but we think it solves a lot of the liquidity/risk issues in user-created markets!
---
Re tractability: If i'm understanding you correctly, you're saying that any market will have potential ambiguities around the wording vs the intent of the question?
We're hoping this is one of the things that user-created markets alleviates; when you buy into e.g. Rohit's market on "Will Strange Loop Canon receive more viewers in 2022 than 2021", you're actually betting on "*Will Rohit decide that* Strange Loop Canon received more viewers...". Issues like "how are viewers counted? do views on guest posts count?" are left in the hands of the market creator. Then, bettors will seek out markets that are decided by creators whose judgement they trust.
(This might be kind of a dodge; we don't actually know! But a cheap web prototype should teach us and the rest of the world a lot)
---
Re thinking in bets: yeah, I've definitely seen an aversion to "betting" among the public. Some (fairly smart!) friends I've talked to churn off as soon as I talk about betting, saying things like "I don't believe in gambling" or "I don't like taking on risks".
In the short term, we'll promote these markets to the rationalist/EA audience; longer term, I could envision different UIs that make more sense for the populace. One wild hypothetical: we could present a Reddit-like interface with up/downvotes on each question (allowing you to "vote" repeatedly) -- and behind the scenes dynamically readjust your entire portfolio of bets, using the Kelly criterion to optimize your returns? But yeah, how to frame prediction markets for the median web user is a super-interesting question!
Hey thanks for the detailed reply! Really appreciate it. The latter two points I get it - honestly it seems like the thing we'll just have to try and see. Which I endorse.
Re the first it's less of the mechanics (though the parimutuel market seems cool). It's the q of how many investors or how many dollars are needed for proper price prediction. Eg in OTC equity markets tickers trade for like a few thousand shares a day, which isn't enough for either narrow bid ask spreads or price discovery and brings vol.