A Comparative Analysis of Prediction Markets and Their Resolution Mechanisms - Borut | IngiteMarket

Hi guys, I'm Boro. I'm from Kalmia. I'll be talking about uh some of things that are going on in the prediction markets or what we're building. So, but first uh what is a prediction market? So, if I ask you guys who knows what a prediction market is here, raise your hands.

Okay. So, so uh great. So we need I need a bit of introduction to prediction market. Uh so yeah it looks like we can use prediction market to predict the future right. Um it looks like so if you take a look at what the elections for the US elections the last one was so it predicted quite right you know the Trump won and this you can see that the it really predicted the future but if we take a look at some other thing so like the the pope uh prediction it was not so really really good so Robert Francis prest one and uh it was just one and the five percent uh in the prediction before the market result.

So uh what is a prediction market? It's actually a market where users trade on the future events. So the way it works, you buy shares for the future outcome events. Uh and this goes like this that you actually buy let's say uh shares for yes or shares for no. And the more people buy shares for gas, the higher the price will be and it's all colorated.

So at the end you have some of 100% and uh this is the way it goes. Uh you have like uh um example so like alltime bitcoin bitcoin high at the end of the June. So uh some people bought yes and it it went up. So uh at the end it always resolves. There's uh there's quite a question that I was uh when I talk with the people and they say you know is prediction market betting or not?

Uh and it's kind of gray area. So um I ask AI and the AI told me it's neither a share market, it's nor pure betting. So um it's neither of them. So yeah, if if you take a look at the if just at the markets, you can see that the shares are bought and sold uh their value fluctuates and if the trade sell before the market resolves uh the market concludes then it's actually a share market. But if we wait that the market resolves then the your share is not toward nothing.

So uh the share that you bet on and uh if this one was not resolved as as positive then uh it's actually worth nothing. So in this case participants are actually wagering the the future of outcome events. So uh so uh yeah the important thing that uh in in the prediction market is resolving the market. Uh so resolving the market means that we need to actually prove that the event really happened and uh in um so like uh as I said if Trump really won the the the US elections so we have in a centralized way there are quite a few centralized prediction markets and uh in a centralized way this is usually done by some committee of people but in the centralized way um we have some solution They're not like perfect. Um I exposed to to here uh you guys already how many of you guys know poly market?

Raise your hand. Some of them are. Um so yeah the the poly market uses the the social oracle and uh we have also one oracle the social oracle works works in not in a deterministic way but it works with the with the staking and then we have an oracle on zitgeist zitgeist is a also existing prediction market on on the uh polka dot chain and u there they use their own chain to actually get the events uh from from from the APIs or they scan something uh and so on. So uh if we go through through some existing markets what what exists uh I will start with just really simple one and the simple one is metaculous this is a really a centralized market they they u it's like a forum where people really um put their wager on some events and uh um you have a nice graphs of what will happen but they don't use any money Uh it's it's all like free more like academic style. Um and it's all centralized.

No, no decentralization here. Uh then we have uh predict it. It's it's actually universe uh type uh from Wellington universe. Uh they actually use real money USD. Um it's al also centralized.

It's more like on politics but they use their money actually to to incentivize uh some researchers in in in the university. Uh it's it's nonprofit. Uh then I mentioned already Zeitgeist. Zeitgeist is a is a prediction market on pola.chain.

Uh it uses its own chain for resolution. It's different. you have like uh um politics, some crypto things and so on. And uh they the the way that that they do the resolution is by using their own chain scanning uh scanning the actual um APIs and so on. Because it's on Pulca.

chain, it's not an EVM chain. They actually implemented everything on on substrate chain. And uh the last one, the the most known one is the Polygon market. uh Poly Market uh Poly Market actually is not fully it's kind of semideentralized uh because they use onchain um onchain resolution but the the uh the things that how how the the stock markets and the order book works is on offchain so they only settle on chain um so it's not fully decentralized uh so yeah Jud just an overview of what we we saw. So the the poly market is the only one that uses uh USDC and the ZU guys uses the SG token their own token.

Uh so uh and uh yeah we we kind of find out that we on the EVM chain for our prediction markets we need uh some more real reliable uh resolution. So what we used here we kind of uh been with with the flare guys. We used their uh FDC connector. Uh so the FTC connector is acts like a native oracle. Uh and this native oracle we can actually get the API results all the deterministic results uh on chain uh so without any any need for for users.

uh and what we decided here is to to have a consensus. So we need three APIs uh to say okay we agree that this this market resolves so like you know MBA results or something like that we need three three sources and at the end we still have some some things that if if the three sources won't won't uh have a consensus then we will say you know uh we will have like a DA people that will vote and then they will resolve the market. Okay. And uh yeah, we are in a we're kind of in AI area. So we we cannot miss the AI.

Uh a question for you guys. Which AI model do you think is the best for AI resolution? Raise your heart for Germany. Nobody. Open AI one.

Uh Mistral. Okay. Nobody. Great. Yeah.

the the real answer is I would say all of them or none of them u because you cannot again trust just one AI model. Uh you need consensus between them. So uh AI really brings us here the the option to to resolve without the need of people. Um and uh with the T trusted execution environment uh the solutions that we saw here uh one provided by Oasis and I think flare will also have uh this kind of resolutions soon uh we can actually do the AI resolution offchain uh do it in a trusted execution environment and then put the results on chain uh and uh the tees are really quite safe here and um provide really the safety that we need. Uh okay and at the end uh this is what exactly we're building.

We're building some new version of prediction market. Uh we want to be different from the a bit different from the poly market and the or the other markets that exists. Um so we currently are starting uh soon in July using the Flare FDC resolution and uh also a backup for on whitelisting on Dhisting. Uh in the near future AR resolution is coming and voting. Uh what's how different we going to be?

We're going to be different mainly in the area of uh of the markets that we want to have. So we will go with social media markets um which relate to social media events uh on LinkedIn, Twitter, IMDb, GitHub and so on. Uh we're targeting XRP holders and uh also a bit of Asian market. So uh launching soon with this it's already on on uh test on DevNet. So on test net yeah uh so at the end just a bit about us.

We are Calia. We're UA based uh uh team. We actually do projects for anybody who's a non- tech uh guy. So we provide you the the whole the whole I would say development cycle from from start to to the end. So if you have an idea, you don't have resources, uh come to us and we will provide you a startup.

Uh and um yeah so um yeah ignite market we are like uh launching soon we're happy to have somebody on board who's a non-dev so maybe some product guy u just if anybody's interested in this it's a I would say it's it's a big market so if you launch it properly uh I think there are quite a few uh gains we can get from this um and yeah this This is it um from me. Um thanks for listening and uh if there are any questions just uh shoot. Hey hey hey thank you for presentation. So previously you uh told us about like API resolution. Yes.

When like several operators uh going to the same API gathering results. Yes. And making some decision. But you know like let's don't take like tier one uh services but like tier two. And for example, I am as developer I can like uh write down like maybe some wrong info.

Yes. And for like six hours one day and like all operators will scan this API and get like the bad data for example. And for example I can somehow stim like effect on the results. Yes. in this case uh why the API resolution is better than like decentralized like Oracle resolution which for example poly market has thank you

um you ask why AI resolution is better

API API

API oh AI it doesn't matter because AI need to take somewhere somewhere info from

yeah yeah sure but uh I mean the the API resolution is better because you can actually if you go with the uh with the secure socket layer on the web you can and they have the certificates you can actually trust uh you can check the certificates and you can actually trust them uh that the information is correct and you have like three four five APIs and they need to consensus between them. So you really need to to hack a lot of this to to actually um you know kill the the consensus or have have own hack. Uh the the social um from the poly market the social resolution from the poly market is still like staking based. So if you have a lot of money you can actually you know hack it and it happened uh quite a few times in in the poly market solution. So uh it's still based on liquidity.

So in this this way it's better.

So so when we're talking about API or AI, it is all always about like gathering data from several sources, not one source.

Yeah. Yeah. Always at least three sources. At least three. Uh but would go for much more than three.

But yeah, at least three because you need three for consensus. You know, two are not enough. Thanks.