Draft-to-Funding Markets: When Bets Can Influence Outcomes - Martin Štrobl

Transcript

Hey everyone, thanks for coming in the morning. Uh I know sometimes it's tough uh on a second day of a conference and um so let's talk about markets and about draft to funding which is just a name I made up uh but perhaps it fits. Basically, we are interested in whether the betting thing that's happening on the prediction markets, which I I don't think I need to tell you is a big thing, right? Uh so I'm skipping on the motivation. Uh and we are trying to investigate uh whether we can perhaps also use this mechanism for something that actually helps the outcomes that these things are built on top of.

Uh I'm a researcher at Logos and I've been uh dealing and looking into this on and off for the past year. So let me start with with a short example. Uh so suppose we have a fundraiser and uh there's a community that wants to build a park. Uh the goal is 10K, right? And we are at a stage where you know they have 7K. It's still 3K short. Uh it's one week before before uh before the end of this fundraiser. uh and we enable betting on on whether this target uh this goal will be hit. So we we have just two betters. We have Betty. She is she's convinced it's going to hit the goal. So she's betting 2K on on that.

And on on on uh on the other side we have uh Nick who who is uh betting 4K on the contrary. And and of course, you know, if Nick wins, he gets bet Betty's stake. Uh and and in vice versa, right? And you can see that already in the money that's in this system now, we because we now have the fund, you know, fundraiser and then we have the betting on top. Uh it already exceeds the goal that that that was put out as 10K, right? Um so the clock's ticking, right? And no one's showing up. Uh it looks like Nick is about to win. um because we are still 3K short, but Betty has an idea. Um and she thinks, well, if I

donate the the 3K myself actually, I'm going to, you know, hit we are going to hit the goal. My bet's going to be successful and I I will get Nick's 4K. So, it's not actually such a big deal bad deal, right? Because like I make one extra K out of this move. Uh so you know if you want to look at the napkin math uh it's the 1k that she's getting from Nick um because she she is spending the 3k on topping up the fundraiser to succeed.

So uh in the end the park gets built and she she is profitable uh on her bed uh and it helped the underlying cause right. So now now that was a very special setup. Now we need to realize what what happens in reality right in reality we would have uh on that side of the market where Betty was we would have a number of betties like suppose there's 50 right uh and of course they could still top up with the 3k but who pays for it? How do they coordinate? Who will be it? you know it's the optimal strategy always to free ride and and wait if someone does it for you right so they they need some sort of coordination to overcome this

free rider problem uh that exists there and so so essentially the basis of this research is the idea that uh what if we had a protocol that actually does this betty trick but it does it for like tens of betties right and it does it automatically uh such that the it provides the accounting that uh that's resulting uh as as if she if they coordinated. So the idea basically is that you know if we pull all the bets that's the yes bet and no bet with the with the donations and if we we have enough money in the system such that we we we are going beyond the goal uh then

uh we are able to execute a strategy where uh with the gap being 3K we can actually just take Nick's money uh from the speculation capital and put it towards uh towards the fundraiser And you know, we satisfy this fundraiser. Betty wins and he she receives her original stake, which is 2K plus whatever's remaining uh in the system, which is the untouched 1K of Nyx. So her profit is again 1K. So if you want to compare side by side, you you would have like the very same outcome and because because we just pull the money uh and we we just make sure that that the campaign

is successful. So as I was saying, you know, the protocol settles as if all the yes better coordinated. Uh it's an opinionated thing, right? uh in itself but it optimizes for getting the thing uh funded whereas not getting it funded. Uh so uh so now that was like a motivation for for the the broader claim that you know actually in special cases we we can use the bad liquidity for potential funding of the underlying cause.

Uh so we we are able to use speculative capital as productive capital. So now now being more formal let me just talk about this draft to funding one mechanism and it's called one because there's just one campaign. Uh I'll show you certain extensions which are perhaps more interesting but let's just focus on this one today mostly. Uh so you know formally we can have we have a campaign that that's you know a goal G w with with with a period t and then you have like a private benefit right if you are a donor for this cause if I donate uh on a kickstarter I am getting something for it as a perk so that's the be uh that

actually I'm getting as a private benefit and we also have a parameter beta which is not that important what I'm presenting today but that one is just in case you know I'm better and my uh and my uh my actually bet actual bet is drafted then I also might be able to get some benefit from that. So that just uh denominates the fraction that that that would be of that private benefit that I'm getting. Well, but you have three actions essentially. So you can donate, right? And then you get the private benefit B. You can also bet on on the yes part uh that it's getting funded uh and and bet on the no part that it's not getting uh funded. And of course uh you know as I was saying the

the yes bets uh their potential upside is coming from the no bets because they those are always allocated to the winner and it's irrevocable. So once you bet you cannot revoke your bet. Uh now just this simple design leads to three uh phases. So essentially you have the pre-lock phase where you know the capital is not able to reach the goal in the system. Uh then once we satisfy this inequality, we have enough capital in the system. Uh we already know that the goal is getting funded right and drafting is what what I call is the uh is taking the bet and putting it into

the funding. So we know that this is going to happen via drafting. Uh so we cannot bet because we already know that the yes part is winning. Uh but we can still have donations. So that's why in the post lock uh period you have donations and they actually crowd out the the drafting that was happening earlier that was assumed earlier because you know it can still be before the end of the uh of the tea period before the end of the campaign uh and it could be that we already know it's getting funded but you can always have more donations coming in.

So that that's the main design and now this is a bunch of things which kind of just formally uh represent that some payoffs. I I I don't think it makes sense to go into that. I'm just uh I'll just show it as uh uh as something for the perhaps the video recording. Um what is interesting here is that on failure uh the donors just get back the their own donation, right? uh the yes bettererss they're losing so their payoff is zero. Uh but what's interesting in the no part is that uh you are getting back your nos no votes plus a proportional stake of the of the

yes pool. So you are getting paid you know if it if it's failing uh and at the same time uh there is no risk for donating and if it fails you get your fruit donation back. So, but but there was a bit of a uh un unfair thing to provide you with the example earlier because we didn't really look at how how those two two two parts the Betty and Nicks how they end up in that situation where uh where they are in a betting market and they they feel like they should bet whether the goal is hit. Uh so we should probably think about like why do they land there in the first place? Uh and the idea is that the

system is designed in a way that you know donations are are there just for the sake of the benefit of course independent of of the betting part. Uh but nos and yes it's like a vicious circle right because they are each pool is a reward pool for the other one. So why should no show up if there are no yeses? Why should yeses show up if there are no nos? what do you bet you know without knowing what's your reward because you don't know who there's no bets on the other part yet and and the system uh has properties uh such that of course at the start of the comp campaign the nos are um it's very attractive to it's the most attractive to bet no right

because only after that you would probably get donations that are coming in you would get more capital flowing into the system and that makes uh not hitting the goal perhaps uh always like less likely, right? Uh the the opposite holds for yes. As you are closing on on the goal, probably it's more likely that you're going to hit it. So so so the the the the answer to why these these guys would actually be in this market is that they don't arrive at the same time. So it's most likely that the nos actually show up very early when there's more uncertainty and the nos the yeses show up once they know uh you know this is this is more attractive and possibly you

know towards the end of the campaign but there's an inherent dependency chain as I was saying because one is a reward for the other so uh we need to figure out how we can break this vicious circle and uh actually introduce some mechanism that that helps trigger it uh from donations and uh also what I'm calling a mission aligned capital uh mission aligned capital what I mean by that is that it's just a capital that wants to get the the underlying cause funded and it it you know it doesn't have a preference whether they do it through donations or through some betting activity uh but they are uh willing to commit the capital just to to produce

the outcome uh and I I'll talk about that uh in minute because it actually turns out that uh the way this system behaves is very much shaped by the the benefit the private benefit from donations. So if your donation doesn't actually provide anything to you uh and it's perhaps more like a public uh a public service or uh any sort of like a shared uh shared outcome that that you are willing to fund uh but perhaps you don't draw like a private benefit out of it. It's a different system, right?

Because in in that case if if there's no reward actually it makes more sense for this mission aligned capital the capital that wants the thing funded uh to bet on not getting it funded. uh and that's because it is he's still a a funer who who does this is still getting the possibility that you know if there it's like a conditional funding so he's getting the the option to uh if we are near the goal to actually close the gap and and have his money committed to the cause but also he is uh you know in in terms of expected value he is he is uh possibly motivated by the fact that he

is getting the yes uh yes rewards right potentially on failure. So on failure he's getting the the yes rewards uh on and on success he's happy because he he contributed to funding the underlying cause. Uh whereas you know if he just donated he would lose uh he would lose all his money on success which is just fine but he wouldn't get any any profits on on the failure of the campaign. Uh so there with a small small B uh you know there's like a like a middle ground uh example but but so let's focus on the one with the large B uh wi-i which is uh one where the donation is still still the

thing to do for mission aligned capital and that's because that's because they are getting this private benefit which is big enough uh so it's still preferred and the issue is that you know even if you have donations uh running in it doesn't trigger uh this this this sort of dependency chain. So what we would need to have in that case to have some text on donations uh in the failure case and in that case they they would they would flow to to the no betters uh so that you know there's some donation coming in uh and if I'm betting no and there's no counterparty yes yet I am still getting that K KD uh on the

failure uh failure scenario. Good. Uh so that that's like the the the basic outline of the mechanism. Uh now how can it be extended right? So we had one campaign uh with one threshold. So obviously you can think about multiple thresholds. So you can have multiple G's uh and then we would have a system where there's uh bets allocated to to these different different thresholds. Perhaps not all of those bats actually have like a counterparty on on like the yes and no symmetry, right? Uh but but this this version opens up like some redistribution that that is possible because like a losing bat on one

threshold can fund a winning bet on a different one. Uh and the the challenge is also the clearing mechanism. Uh so because you could have uh you know a like a drafting that converts into satisfying one one threshold the next threshold doesn't get satisfied uh but there could be enough liquidity and enough betting on the subsequent threshold that you know doing a coalition sort of like clearing would would actually like push it to the to this next threshold. So there's there's more like algorithmic uh questions around this. Uh also you can have like continuous funding. Uh so suppose there's like a no deadline right but

there's like a reserve of like a protocol their treasury or something they have some burn rate and then it's more like an insurance you could have uh betting on this reserve not hitting zero and and obviously no is betting that actually they go bankrupt, right? And then you would have some sort of tax that funds the the the yes bettererss and the no guys would be paying for that. And if if the event actually happens that that they go beyond zero uh that potentially you could use drafting to like sanitize this sort of shortage in a short term.

uh taking taking some of the bad liquidity and then uh you know the system survives and and you need to probably also like have some follow-up mechanism to to to perhaps over time if if it goes into profitable terms you know re re repay this this this depth towards towards the betters potentially well and what what's possibly most interesting for me as a subsequent research is you know having endogenous goals so this means that you know you would have more projects like say there's five projects and like there's uh there's funds allocating across those projects and you you could and my goal as a project could be just

you know being the second project if we rank them in terms of funding uh or getting some sort of share. So it's like a moving target, right? Because if my project is satisfying that perhaps there's there's your project which is uh which has enough bad liquidity we can draft money from such that it just just supersedes mine project uh and and then you know I could iterate this right and I could also like take my bed liquidity and again like go go beyond that perhaps until we exhaust that. So that it's a question also of like efficiency of drafting. Uh and the settlement is ultimately like a fixed point in in in

this way how we draft across these projects. Um and I have just a few minutes left. So I'm just going to rush over applications here. Um and I have just three showcase applications for different regimes of the B B right. Uh so uh greater action markets with zero threshold crowdf funing with a small one and large uh which is a token launchpad. So let me just start with the threshold crowdf funing and not spend much much time on this because that's pretty much the example I showed you initially, right? There's there's a goal that we need to hit and the beckers are getting some perks for contributing and what what this mechanism does is that if we are close to the threshold at the end

and there's enough betting liquidity, we are able to push through and and satisfy the threshold. So it it produces more funded campaigns than not. Um now let me talk about these action markets. So this is assuming that there's like a local trusted authority that asks one question like and it's like using a prediction market to to trigger some sort of outcome in the society right we usually see it with the Trump administration but we don't see it uh in the real world so much. Uh so actually there was like a a guy at at the at the Paris airport who brought a haird dryer to to like uh to actually

uh change the temperature uh on a sensor to to win on poly market. So now we are asking like how can we set up like a prediction market such that we are able to trigger the underlying action right. So here we have like an illegal dump on a street and we are asking whether it's getting cleared and there's some local authority that just posts this question. Uh and if I am willing to pay for this it actually means that I need to bet that it's not getting cleared right because then I'm creating the opportunity for someone else to show up and and I call him a doer uh and bet yes uh which is like an active citizen who does it

themselves. So he bets yes and he goes and cleans it. Uh and the local authority says it was cleaned and and the bets you know are routed to him that he's winning from the founder funer. uh and and you know what what can draft to draft to funding do here is that we can have like a fallback contractor which is which is like a quote for 2k and if if the authority uh realizes that the market has the size of 2k already it can just close the market and order the contractor to do it and everyone's satisfied because all the market in the all the money in the market was just willing to to pay for this cost. They just didn't know who to

pay for example. So this just summarizes what it is. Uh so there is a proportional speculation to the cost. You know by design there's no way for speculation market to outgrow out of proportion and it's kept by the cost of the action in this case. And lastly, I want to present the token launchpad application uh which I think is interesting because here it it maps very nicely to like a D5 uh space. So on on like the pump.fund launchpads for example you have the bonding curve launchpad right and there there's a threshold at which you graduate to like like a unis swap uh amm

and so so here how it maps is that the donors are actually buyers of the token on the bonding curve initially like like you would have normally uh and and then you would have betting on whether the token manages to to go beyond uh beyond the threshold and graduate within some time frame. So you can bet on that as as you would normally. Um but what happens if if it doesn't clear uh you know and if we have enough money in the system to actually reach the threshold uh then the yes bets would and then actually all the bets uh would just be used uh for swaps, right?

They would be used because we have the bonding curve. So we know exactly how much capital we need to reach the reach the price uh that associated with this threshold and then uh we would just use this money and buy the token uh and that would pump the price towards the threshold and we still have uh the money in the system. It's not like we spend the money. We just swap them into a different token. So now if you have drafted capital actually like the yes bettererss uh and if they get drafted it means that only the money that they put in the system uh are converted to a different denomination potentially because it was like on an earlier part of the bonding curve you know it's more

more attractive uh it it's profitable uh and and they could be getting also the no capital which is again uh you know partly could be drafted partly not uh and that would that would be uh also converted to tokens and hold hold the value. So to conclude I I hope I convinced you that there exists uh ways in which we can turn speculative capital into productive one and we can do that if we really focus on the underlying actions that this uh these these huge markets are built on and we we think about how we could leverage that uh and we ground that with uh with this sort of mechanism uh such that we don't grow out of

proportions and have a cap on this speculation. So thank you very much. Uh there's a much longer talk that you can watch on this. Uh and u I'm on X. I am working on a paper regarding this and that's going to be hopefully available very soon. Uh and I also urge you to look at what we are doing at Logos. We have a new builder program and uh a lot of uh to offer also to dev studios. So thank you very much.