Flash Talk by Alessandro Buser - CEO, Dialectic

Very nice to be here. Um, Buenosirus, beautiful city. Uh, stable summit is always my favorite event of the conference cycle. Um, so yeah, I'm Alessandro, CEO of Dialectic. Now, um, I will quickly go over some thoughts about understanding D5 contagion risk.

Uh, this is of course a very hot topic right now. Um, and I just want to walk through how we think about risk, how we think about contagion risk, how we quantify risk. Um and I hope this is just helpful for everyone who's active in DeFi. So a few words about us. Uh we are onchain asset management firm.

Uh we focus purely on DeFi yield. Uh so we have different vehicles. Uh we started out mainly having private offchain funds and just now very recently we deployed onchain vaults. Uh so it's very exciting for me. Usually when I speak at stable summit I just talk about these private vehicles where most people can't invest.

uh but this is the first time where I actually can talk about things uh that are actually relevant for everyone because now we are also active on chain and we have these uh three different strategies on on the machina infrastructure. So um what we really focus on is getting risk adjusted return uh so we allocate to anything that is in DeFi that has a good risk profile um and that pays a good risk adjusted return. So I'll go a bit about how we think about risk. So for us it's really a multi-layered approach. Um so at the core of it we have the due diligence layer uh which is really important.

So this just means like really understanding deeply what we get into understanding all the risks of the protocols, understanding um how the protocols work, what the smart contracts are, what the economic primitives are that we work with and and so on. Um and then once we have quantified the risk and I'll go a bit into how we do that there's a layer of diversification uh which is about just like having many positions that have uncorrelated risks and that don't bring too much concentration. Um so a very important thing for us is kind of understanding the multi-layered um nature of DI. So in D5 we talk a lot about composability and a very important thing about composability is that everything is connected. So that means if there is a vulnerability at any point um along the stack, these risks can propagate quite quickly throughout the entire kind of stack of of the DeFi ecosystem.

So we have this this tool which basically allows us to break down individual positions into the underlying dependencies. So in this case we have like a pool on aerodyum and as you can see this is like USDC EUSD pool and as you can see like EUSD is itself composed by many different underlying assets that all have exposures to to different risks and so like understanding how these risks propagate and understanding how they affect the the top level position um is really important and I'll I'll just like talk about the two concrete examples from from from last week. So on the one side uh above we have what happened with the Moonwell market where basically as many of you know there was an exploit on balancer v2 um and then like a day later there was an exploit on moonwell and the reason why this exploit was possible is because the moonwell market um was using a chain link oracle uh which was feeding data out of the balancer pool and this is something that many people didn't realize that the moment that there was a vulnerability on balancer there was like a ripple effect to many other protocols in the space. Um because a lot of the of the lending markets they use oracles and some of these oracles feed data from balancer. So like understanding how these risks propagate and understanding how a vulnerability on one protocol can affect another protocol on a different chain um is like very important.

And then the other example is uh the one below here um in which we have a morpho vault uh that is lending to a token which itself is backed by a basket of positions. One of these positions was an oiler market. Uh the oiler market itself was lending to another protocol um the USD and then the USD was backed by stream. So when when stream collapsed, uh there was like this ripple effect. We're kind of going back step by step.

Um the risk just manifests kind of going up the chain. And so the way that we that we look at these things is that we kind of break down this individual chain of dependencies and we assign risk scores. So kind of when when I had the the slide before where there was like the due diligence um kind of layer, uh what that basically means for us is that we quantify the risk with a score from 0 to one. And so in this case, I mean, these are just like example numbers. Um, but you can see that like if we assign a risk score of like 95 out of 100 on stream, uh, because our due diligence realize that there is like a lot of issues with how that protocol is set up, um, then this kind of risk score propagates up the the chain.

And so kind of once we get to the end position which is something that we may consider to deploy into uh we can see how kind of the risk of this one position is affected by all the underlying dependencies. And then in order to like calculate this we have a risk propagation formula uh which basically um is just like a compounding formula of the different risk components. And this allows us to kind of quantify these risks of the individual positions based on on on all the underlying dependencies. Um and then kind of like once we have the risk scores uh we can go into the portfolio optimization part of it. And here what's really important to understand is what what we mean by risk adjusted return and position efficiency.

Um so like two examples of two different positions. So on top uh we have a morpho vault that in this case like let's say it pays 8% APR but it has a risk score of 75. So this would be like this morpho vault here um which has a risk score of 75 because it has all these underlying dependencies. So then we can calculate the efficiency of this position which in this case would be 10.7.

Um but then let's say that there's another morpho vault uh which has a risk score of 30 but only pays 5% APR. And you can see how the efficiency of the second uh morph vault is higher than the efficiency of the first morpho vault. So when we construct our portfolios both in the private funds but also now in in the public uh strategies of machina this is how we we approach the construction of the portfolio. Um so in conclusion um I mean D5 composability is a feature not a bug but it comes with a lot of extra complexity and it's really difficult to to kind of go dig deep down what all of this means and that there needs to be a lot of automation to to do this. Um and then with just like being able to calculate these risk scores, um that's the the only way in which you can create an efficient portfolio is by kind of understanding what the risks are, how much you're getting paid for these risks, and then just like picking positions that are efficient from a risk return perspective.

Um so this was uh a lot of uh condensed information. Um I'll be around if if you have any questions on this. Uh thank you.