Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Insilico Terminal Podcast Episode 7 - Extended Founder Ruslan
53 Plays1 month ago
In this episode, Ruslan, founder of Extended, shares his journey from Revolut to building a fully self-custodial exchange. He explains the vision behind Extended’s unified margin and multi-asset collateral system, the choice to build on Starknet, and the challenges of trustless architecture. The discussion dives into liquidation logic, exchange stability during volatile markets, and what sets Extended apart - closing with thoughts on sustainable growth, token plans, and the future of decentralized trading.
00:00 Intro, Ruslan’s background, founding Extended, and vision for a self-custody exchange
06:11 Leaving Revolut to build independently, benefits of a small agile team
13:09 What sets Extended apart - unified margin, multi-asset collateral, and long-term roadmap
18:42 Why they chose Starknet, trustless design, and Ethereum ecosystem integration
24:59 Deep dive into multi-asset margin mechanics, liquidation logic, and risk handling
33:46 Why others don’t build this yet, technical complexity, and future competition
39:18 Exchange stability, recent crash lessons, and how Extended handled downtime
45:56 Sustainable growth, token plans, long-term goals, and closing remarks
Recommended
Transcript
Introduction and Guest Overview
00:00:14
Speaker
welcome to another episode of the in silico terminal podcast today we have a bit of a different guest with us since we have mostly had traders on so far and interviewed them.
00:00:28
Speaker
Today we have the founder of an exchange. And I just want to make a little disclaimer at the beginning that this is not an official announcement for integration or anything or collaboration or endorsement or whatever.
What is Extended?
00:00:41
Speaker
it's just, some I think It's a good opportunity to also switch up the guest list a bit and not only talk to traders, but also to people that are building stuff in crypto and in the trading world around us.
00:00:54
Speaker
And um I'm also not not getting this is not a paid deal. I'm not paid. I do have extended points. I'm going to admit that so might be a little bit biased, but we still Just going to have a good conversation here, I think.
00:01:10
Speaker
um Yeah, so why why don't you tell us a bit about who you are, about yourself, about what you're building? Why don't we just start with that? Yeah, sure, sure. Hi, Sokja, and thanks a lot for having me.
00:01:22
Speaker
right And it's great to hear that you're active on Extended. Um, yeah. So as for the intro, basically, my name is Ruslan. I'm the founder of extended with extended. We have a very simple and clear vision, right? We want to build a fully fledged self custody exchange with have a unified margin across, uh, perhaps spot, uh, integrated lending market.
00:01:46
Speaker
And that's basically our midterm calls towards, towards which we are, uh, working. Right. Mm-hmm.
Origin Story and Vision
00:01:55
Speaker
and um i'm actually curious what what made you what made you start extended what is like your your goal with it why are you passionate about the project sure sure look uh we've started extended in early 2023 right basically back then me as well as most of the team members were at revelant right where was the head of crypto operations
00:02:21
Speaker
And the back then in late 2022, right, we chatted to a lot of partners in this space regarding various potential collaborations, just exploring opportunities.
00:02:34
Speaker
And we also chatted to ah FTX back then and essentially It was it was quite quite weird because we had multiple calls with them. They were super positive, right? Super bullish on everything they are doing. And then all of a sudden they collapsed, right? Which was completely unexpected.
00:02:55
Speaker
And the we also got a bit of the weird feeling, right? When you just know the people, right? You talk to them and then the week after, right? Everything completely changes.
00:03:06
Speaker
And we essentially, we had a very simple a idea back then that basically you can build a very similar product, right? However, all the user funds will be self-custodied.
00:03:18
Speaker
The operator of the exchange would not have access to the user funds, wouldn't be able to impact the tracing logic. All the tracing logic would be validated on chain.
00:03:30
Speaker
And essentially, like I just got super interested in in that,
Crypto Culture and Development
00:03:35
Speaker
right? And they saw a massive opportunity because back then, effectively, there was no hyperliquid, right? The only perp taxes were DYDX and GMX. And already back then, it was clear that the user experience theory is suboptimal, right? A lot of things can be done better. The product portfolio can be expanded.
00:03:56
Speaker
And yes, essentially we started building towards that that idea. Right. And, um, that's, that's how we, that's, uh, basically how we came up with, uh, with the idea of extended, right. That's for why it makes me passionate.
00:04:10
Speaker
Look, ah One, I think crypto overall is a super fast moving space, right? Where a lot of the things change on the
Independence from Revolut
00:04:20
Speaker
fly. A lot of opportunities exist.
00:04:22
Speaker
That also presents a challenge to the team, right? And it's super interesting to work on solving that challenge. The other thing is that the feedback loop in crypto is insane quick, right? So you release a feature, you basically understand whether it works or not.
00:04:39
Speaker
immediately essentially right because if the if you have active users of the platform you basically hear the feedback the same thing right you get concerns about your product both as well as the positive things right some like effectively almost at the same time as you release new things and this is uh this is great to be honest right because
Team Dynamics and Benefits
00:05:00
Speaker
it allows you to iterate on the product super fast And if you are receptive to the feedback, basically your community helps you build that product, great which is which is obviously great. like Honestly, I don't think there is any other industry like that where the users and the builders can be so close together.
00:05:18
Speaker
right Obviously, this is like builders builders need to want that. right They need to be open to the community. They need to be talking to them, et cetera, et cetera. And this is what we are trying to trying to to do as much as we can.
00:05:31
Speaker
And this helps us improve the product a lot, right? As well as like, it's when you actually talk to the people who use your product, it gives you additional motivation, right? It gives you additional motivation and drive to to to do better, right? to um try to effectively become the best product in the in the space, right? And the space is super competitive, right? Burp Dex is ah probably one of the most competitive markets overall.
00:06:02
Speaker
And that's just a great challenge to take on, right? I don't think I can think of any any other topic or business which would be as exciting or as fast moving right or effectively as interesting to build in as versus what we're doing now yeah yeah okay i can i can see that like i definitely agree about that obviously i think crypto is uh probably
Comparison with Other Exchanges
00:06:30
Speaker
the the most interesting place to be which is uh the reason why i'm here why probably most of the listeners are here
00:06:37
Speaker
I'm wondering, ah i don't know how much you can ah we can say about that, but um since you said like that you and the many parts of the team were like at Revolut before, is there a reason why um you decided to like... like is Is there still a connection or is there a reason why you decided to like split off and like do stuff by yourself? like Why is it not called Revolut Dex, for example? Why did you like make it its own product?
00:07:06
Speaker
I mean, we are we are an independent team, right? So essentially our connection to Revolut is as follows. Basically, most of the most of extended employees and the team are former employees at Revolut. That's one.
00:07:21
Speaker
Secondly, we have ah quite a lot of existing, both existing and former C-level employees of Revolut being investors in Extended, right?
00:07:36
Speaker
As well as we are talking to the Revolut to the rebel team on dar on various topics, right? But that's basically the connection, right? So Extended is a separate is a separate team, right? That's why we're not called the Revolut Dex.
Unique Features and Future Goals
00:07:54
Speaker
I see. Yeah, was so I was just wondering if ah since you did like the crypto um operations at Revolut? Why maybe they, i don't know if I can like speak about that, but but maybe why they didn't decide to like do something like that themselves?
00:08:13
Speaker
Look, um probably the best way to answer that is that dear overall they are super ambitious, right? And their roadmap, I'm sure back when I was there as well as now,
00:08:30
Speaker
right is quite aggressive and they want to do a lot of things. But by definition, they will always be moving slightly slower than the essentially the rest of the market and especially than DeFi. Yeah.
00:08:46
Speaker
Just because one, it's a huge company, right? And secondly, it's like they are regulated, they are regulated bank, right? Which has its own consequences on the way the things move. and And to me, it was clear basically back then that both the base at which you can move as well as essentially the opportunities which you can explore are much bigger and wider versus doing it at Revolut directly.
00:09:22
Speaker
I see. I see. Yeah, that actually makes a lot of sense. um I'm not um'm also not trying to be like too critical critical or anything about that. i' Actually, i'm i'm a huge, I'm like a daily user of Revolut. I use it for my ah my every like payments and whatever. I'm i'm not really like buying or selling assets on there or stuff, but for for payments and everything, it's a really, really nice product. So,
00:09:47
Speaker
like go Look, it's an amazing product. It's an amazing company, yeah red which is super bullish on crypto, super active in crypto right to the extent they can be. They have a very ambitious roadmap.
00:10:01
Speaker
And that like I'm quite confident they will execute on that. It will take some time, but they will. right and yeah How big is the the extended team?
00:10:12
Speaker
How many team members do you Look, we're super small, right? So we have 10 people in the team. Essentially, we have four backend engineers, three frontend engineers, one person on the product slash all the operations, including support, and one person on the growth and go to market.
00:10:35
Speaker
Then myself, I'm basically doing trying to do everything, right? Growth, operations, product, whatever requires my attention, I'm doing that.
00:10:46
Speaker
ah But we are relatively small team, right? And like we're thinking about expanding it, but there are no specific plans because And I think you understand me. it's much easier to It's much easier to move with the smaller team.
00:11:03
Speaker
Right? Mm-hmm. You can do things faster. You can decide on the you can decide you can take decision decisions faster. You can iterate quicker. Right? So there is a lot of benefit of remaining a small team.
00:11:16
Speaker
Right? As well as we've obviously seen a lot of examples when small teams can build a huge products. Right? Mm-hmm.
00:11:27
Speaker
No, I definitely agree. I mean, we're also not like the the biggest team and I think Hyperliquid is probably like the greatest recent example. I think they're also only having like 10 people or whatever on the the core team. 100%. Yeah, it really doesn't take that many to to build something great. And also, like I've talked a lot about this with Dennis recently.
00:11:49
Speaker
i think if you have like a ah small team where everyone is like really very very like close to each other and dedicated and all of that stuff, it like makes communication and everything a lot easier. And as team scale, you need
Technical Architecture Explained
00:12:02
Speaker
to like implement a lot more of, ah I guess, these process-oriented things just so that everyone is like on the same page and all of that stuff. So I definitely see the the advantages of ah of having a smaller smaller team. And I mean, 10 people is also not that little.
00:12:21
Speaker
There is also like the results of the indirect benefit. is If you have a big team, right, you are by definition working on multiple small products in parallel.
00:12:33
Speaker
Right. Yeah. And effectively, people who are responsible for those internal projects or internal products, they can sometimes think Effectively, there can be a conflict of interest between ah couple of smaller products within the same company, right?
00:12:52
Speaker
ah Just for the purpose of releasing faster, right? Or achieving own targets. Whereas the small team, ah basically everyone works on a couple of prioritized features, right? And all the interests are aligned and hence you don't really agree require complex ah processes, right? Everything can be managed yeah quite simply and remain efficient essentially.
00:13:21
Speaker
So what would you say is like the main distinguishing parts of Extended? I think right now is actually a really good time to do to to have this conversation because um or maybe, I don't know, maybe the meta died down already after the latest Nuke.
00:13:35
Speaker
But before that, last month we had the perp DEX meta where every new thing like popped up. We had like obviously Hyperliquid in the last year, which was really successful. And then Lighter got a lot more attention. And there's so like many other perp DEXs as well. And Xsendid is also part of that category. like What distinguishes it from the others? and how do you how do you plan to like take market share and then make it succeed in the face of the competition?
00:14:06
Speaker
Yeah, sure. Look, we believe that in the long run, the only way to get sustainable product usage is product differentiation, right? So your product has to be unique in a certain way, and it has to cater to the needs of the users and solve some problems, right?
00:14:23
Speaker
Essentially, our stance on that is that product-wise, the key differentiator is everything around the logic of the unified margin, right? And basically, we see it as a multi-step process for us.
00:14:36
Speaker
So this month we are planning to release the tokenization of the vault shares, which in the simple words would allow you to use the funds locked in the vault as the collateral in the exchange.
00:14:50
Speaker
That will be the step one. Then until the end of the year, we are planning to release the cross asset collateral margin. meaning that users will be able to post non-stable coin assets as the collateral, such as WPTC, LPTC, RAPState Thief, USD, and use that as the collateral to trade the USDC set of pairs.
00:15:15
Speaker
That also implies that basically all the USDC deposits into the exchange will become yield-belling, right? Just because if there is person who deposited non-stable CoinAsset as the collateral and then he has a negative unrealized DNA. and He's basically borrowing funds and paying interest to all the USDC depositors.
00:15:37
Speaker
That's the plan until the end of the year. right Then in Q1, we are planning to release the spot markets. And once we have the spot markets, basically we'll have unified margin across spot perps and the integrated lending market.
00:15:56
Speaker
Longer term, we have a plan to essentially both decentralize the operator as well as build our own EBM compatible chain on top of StartNet. The key differentiator of that chain will also essentially be centered around the unified margin logic.
00:16:16
Speaker
We've margin of the exchange being available on our network for the users. So effectively, if you as a user, let's say, posted one BTC, one WBTC as the collateral and extended exchange, right? You started trading burps.
00:16:34
Speaker
Let's say you have a positive unrealized BNN, your total like which is, I don't know, like 150K, for example, right? That equity will be tokenized as an ERC20 token, which will be exposed to our EVM compatible chain.
00:16:48
Speaker
right So as a user, you will basically have the same margin available both on the exchange and on the EVM compatible network, which would effectively, again, reinforce the efficiency of the margin usage.
00:17:05
Speaker
And that's that's our longer term idea. Then validators of the EVM-compatible network will also be basically running the operator of the exchange, right? Meaning that our matching, liquidation, and all the currently centralized logic will be ah dis distributed across the validators of the EVM-compatible network.
00:17:32
Speaker
Does that mean it would be something like a layer three, maybe? I think, is Starknet a layer two of Ethereum? It is, right? I mean, Starknet is a layer two, right? Here, in terms of our chain, like I wouldn't position it as a layer two or layer three, right? It would rather, like, basically it would be something closer to to our own L1, right? Because we also accept the deposits directly from the other chains.
00:18:02
Speaker
um But essentially, I think the simple way to think about it is if you draw an analogy with Hyperliquid, right? They have HyperCore and HyperVM. So our HyperCore will the remain StartNet, right? Where effectively all the user funds will be secured.
00:18:18
Speaker
And our version of HyperVM will be like essentially our own EVM compatible network. with margin of the exchange being available the for the usage.
00:18:34
Speaker
I see. Can you speak a bit more about um why you chose StarkNet? Because um I think that is not really most obvious choice, I guess. It's not really the most popular chain. I personally don't know of anything else that is going on on the chain, to be honest.
00:18:51
Speaker
So why why did you go with that? Yeah, sure. Look, it's it's quite a frequent question and the answer to that is pretty straightforward, right?
00:19:02
Speaker
Effectively, from the get-go, we did not want to control the settlement layer, right? So we wanted the user funds to be stored and secured by an independent public chain, right? And we wanted that chain to be part of the broader Ethereum ecosystem, right?
00:19:20
Speaker
And if you have those two requirements, right, basically there is no other choice besides StarkNet. which also full fulfills like the requirements on the throughput that we have as an exchange. Right.
00:19:38
Speaker
So for example, i think I think we're settling somewhere in between 500 to a million, 500,000 to a million transactions per day, right?
00:19:48
Speaker
On chain. And essentially, StarkNet is probably the only a chain that can handle that throughput at the required cost.
00:20:01
Speaker
As well as, like, besides that, I think security-wise, StartNet is one of the best choices because they are the only CK-enabled chain with the decentralized sequencer, right?
00:20:15
Speaker
However, weather at the same time, what's important to note is that our users, they don't really need to interact with StartNet.
Why Choose StarkNet?
00:20:22
Speaker
right Yes, their funds are secured by StartNet. right However, we support effectively all the EVM wallets.
00:20:30
Speaker
We support deposits and withdrawals from and to major EVM chains. So essentially, as a user, you don't need to interact with StartNet if you don't want to. Obviously, we also support like Archant.
00:20:45
Speaker
Argent as the main StartNet wallet, right? But for the rest of the EVM users, they effectively can use us as any other any other EVM application, right?
00:20:57
Speaker
Yeah.
00:21:00
Speaker
Yeah, I think that's ah that's really good and that's very very useful because um I've never really, I think I've tried to play around with SarkNet before, maybe before they did their airdrop, but I think the user experience was so annoying to me that I just like gave up at some point. So it's like very good that you can just ah use the normal wallets and normal EVM experience.
00:21:21
Speaker
i I wanted to ask why why were like ah the two requirements you mentioned at the beginning of not controlling the settlement layer and the being close to the Ethereum ecosystem, why is that important to you?
00:21:36
Speaker
Look, because effectively we want to build the trustless product where users don't need to rely on the operator of the exchange, basically on us, right? For the security of their own funds.
00:21:49
Speaker
And I think if you have that requirement in mind, that's effectively the best choice to proceed with the public chain, which is secure and decentralized and sufficiently decentralized, right?
00:22:04
Speaker
um Yeah, so basically that's that's the that's the core Okay, that makes sense. um I would like to ask more about the multi-asset margin system um because as you said, that is like one of your key
Multi-Asset Margin System
00:22:21
Speaker
differentiating features. If you could explain a bit more on how that will work and be architectured, because I think it's quite ah a complex thing to do, especially on chain. You have to do like a lot of hedging and make sure everything is secure.
00:22:36
Speaker
If could speak a bit more on that. Yeah, sure. like i think we can start basically from the basics, right? From from the basics of the on-chain logic that we have. So essentially every user of Extended has his own, let's say vault, right?
00:22:54
Speaker
in our In our smart contract. And every vault has its own positions, right? And has its own equity, right? Effectively, the two main parameters of every user's account right is the equity of the account and the maintenance margin requirement of the account.
00:23:13
Speaker
As long as equity remains above maintenance margin, the account is healthy. At the point if equity drops below the maintenance margin requirement, the account is subject to liquidation. equity belows If equity drops below zero, the account becomes subject to ADL.
00:23:30
Speaker
And by the way, we cannot liquidate or ADL users in case those two requirements are not met on chain. and Now, currently, equity of the account is basically the USDC collateral plus the value of the synthetic positions, so perpetual positions that the user has.
00:23:55
Speaker
Maintenance margin is just the it's just effectively the margin requirement of the position right multiplied by position size multiplied by yeah mark price.
00:24:05
Speaker
and When we add spot assets, right, as the collateral, that will basically mean that those spot assets will contribute to the equity of the account, right? no And essentially, as long as you have a price of the spot asset,
00:24:25
Speaker
and And you can also apply a discount factor to a certain asset, right? Basically, nothing really changes from the on-chain logic, right? So the pure on-chain logic remains relatively simple, right? But you add essentially one more parameter, which is the size of the non-stable coin asset as supposed as the collateral multiplied by its in index price, multiplied by the discount factor if if we're to apply.
00:24:55
Speaker
and So
Challenges of Trustless Systems
00:24:56
Speaker
the on-chain logic remains relatively straightforward and it doesn't really change a lot. The core principles that remain the same. right Now, the complexity with the cross asset collateral will be the liquidation of those spot assets.
00:25:11
Speaker
Yeah. Because imagine you as a user, you have deposited, let's say again, like one WBTC, right? And let's say that we apply, I a 5% discount factor to a BTC.
00:25:25
Speaker
So one BTC contributes 95% of its value to to the equity of the account. Now, if you become subject to liquidation, right, as the first step of the liquidation process, we would need to sell off, sell off, like we'll first try to close synthetic positions. If that doesn't work, we need to sell off your spot asset, right? Mm-hmm.
00:25:48
Speaker
In the beginning, until we have our own order book based spot, right we'll be executing liquidations of spot assets through the RFQ system.
00:25:59
Speaker
where effectively all the market makers who are currently integrating the extended will have an option to pick up the liquidations of the spot assets. Practically, it will work relatively simply. right like Let's say you know that you need to liquidate one BTC and the worst price at which you can liquidate it right is, I don't know, like 100,000.
00:26:20
Speaker
right Effectively, you're submitting a request for quote RFQ to all the market makers who are active on the exchange. right And then in response, the market makers give you the price at which they are ready to take on that asset.
00:26:36
Speaker
And you basically pick the best price which they offer right and execute the trade for them. Essentially, the trade in that that example would be just the fact that we transfer them one WPTC and they transfer us USDC collateral, right which is basically credit to the position of the liquidated user.
00:27:02
Speaker
right Once we have the spot-based order books, right all the liquidations will go through them. But until then, all the spot liquidations will go through the RFQ system.
00:27:15
Speaker
I see. And why why would the synthetic per position, for example, i think you mentioned that earlier, why why would that not be able to be liquidated? like um No, it can be it can be liquidated. right Essentially, if you look in the unified margin logic, right you essentially have various types of risk exposure, right and yeah you choose which ones to close first.
00:27:43
Speaker
right Essentially in the example, which we had where the user has one WBTC as the collateral, and let's say five open synthetic positions will start liquidating synthetic positions.
00:27:57
Speaker
Right. But imagine that as the result of the liquidation of synthetic positions, the account equities, for example, below zero. Right. Yeah. Then we would need to sell off the spot asset because we know that the spot asset goes with the 5% discount.
00:28:15
Speaker
yeah And we know that if we sell the spot asset i know at 1% discount, right the user's equity will still go up because we actually apply the 5% discount.
00:28:26
Speaker
So liquidating spot in that example right would essentially bring account health account back to the healthy state.
00:28:36
Speaker
I see. Yeah, that makes sense. I get it. What are the challenges that you that you encounter during this process? like you It's not really life yet. So what what is say um challenging you in that process? Or why is no one else building something like this?
00:28:58
Speaker
Do you have any ideas about that? What's challenging? It's a good question, right? i like Basically, it was able I was able to describe the entire logic, I think, in five or seven minutes, right? So the logic itself is not that complex, right?
00:29:17
Speaker
Yeah.
00:29:20
Speaker
And I think, like, honestly, I think the key challenge was to design it in a way that can be executed in steps, right?
00:29:34
Speaker
As well as simplify the design so that it can be done quickly, right? So two specific examples, right? One is if we didn't think of the RFQ, if we did not think of the RFQ system for the liquidation of the spot assets, we would need to build first the order book based spot, right?
00:29:59
Speaker
which is not that complex in itself, right? However, bootstrapping liquidity for spot is quite complex and much more complex than for perks because you don't have leverage, right? So that was the first challenge which we essentially solved with the RFQ-based system, right?
00:30:16
Speaker
Now, the second challenge is, as I've explained earlier, imagine if you have let's say, one WBTC and you have negative unrealized P&L in USDC, you need to pay an interest rate.
00:30:30
Speaker
and So the question is, how do you actually define that interest rate? And do you have multiple approaches. right You can build a lending market like Aver, or you can choose Morphos model, where essentially interest rate curve can also shift right upwards and downwards. so And those are relatively like the Morpho's model is a relatively complex product. Right.
00:30:55
Speaker
And you also need to make sure because like I will probably not go into the super details of that, but yeah imagine your unrealized, your negative unrealized P&L is not a fixed number.
00:31:11
Speaker
Right. So unlike any other borrow, unlike in other any lending markets, the amount which you borrow fluctuates with the change of the mark price.
00:31:23
Speaker
Right? So there is quite a lot of complexities in how you calculate the accrued interest payment. Because let's say you have 10 synthetic positions, right? Altogether, they have the negative unrealized P&L of 11,000, right?
00:31:42
Speaker
A second later, all the mark prices update. Your negative unrealized P&L is, I don't know, minus 11,152, right? yeah And you need to find a way to be accruing the interest payment correctly taking into account the changes in the market price, right?
00:32:04
Speaker
As well as the value of your spot asset also defines how much you actually borrow, right? So basically the entire interest payment, interest accrual and interest payment logic is quite complex, right? and and we just found ways to make it simpler and to make it work, right?
00:32:26
Speaker
Then the third example would be is imagine you have a spot asset as the collateral, which you cannot liquidate. Right? you No one wants to buy it.
00:32:37
Speaker
Right? You need to have an ADL of borrows and ADL of spot. it And it all has to fulfill the on-chain requirements of us remaining a trustless exchange in the sense that we cannot liquidate or ADL users without those transactions being valid on chain right yeah so there are quite a lot of complexities on how you can design that system right i wouldn't say that they are i mean they're obviously solvable because we've solved them right but they require quite a lot of really thorough system uh thinking right because perps in itself is already a relatively complex system right once you start adding
00:33:25
Speaker
non-stable coin assets as the collateral, right? With all the liquidation logic, interest payment, the crude logic, ADL, that quickly becomes a very complex system, right? And you just need to make sure that it will all work even in the worst case scenario so that there's never bad depth in the system.
Resilience of Trustless Systems
00:33:48
Speaker
That's one. And second, you need to make sure that it all works in terms of the but's say throughput and settlement requirements. Because imagine like we could do interest payments real time, but then that means that every millisecond would need to accrue an interest payment. right That would basically 10x the amount of transactions that that we need to settle.
00:34:11
Speaker
And basically there are a lot of nuances how in how the it's being built. right ah But we're quite confident in the design that we have and we believe we can deliver it relatively quickly.
00:34:23
Speaker
Now, in terms of the question on why others don't do it, right? my My personal opinion, right, is that in the end state, all the perp taxes should and probably will support the cross asset collateral margin law, yeah right?
00:34:40
Speaker
Now, practically why they don't do it is I guess, one depends on the architecture, Because there is a huge benefit of us building on top of StarkNet is that essentially all the... like StarkNet is a general purpose chain, right? So we can build whatever business logic we want in Cairo, right? to Basically, in general in general language, right?
00:35:11
Speaker
For other perp DEXs, for example, running their own and no ck ZK logic, right or dear they would need to effectively build it at a much lower level right of all the ZK circuits, etc., etc., which is a much bigger effort.
00:35:31
Speaker
right So I guess the short ah answer is that I believe everyone will get there at some point. Right. But I think that probably will be one of the first ones to get there or even the first one.
00:35:46
Speaker
I see.
00:35:49
Speaker
That sounds good. I mean, I hope i hope you do for for your sake or it's just ah in general would be a but good product to have in crypto. Do you think um making it because it's like you're doing it in a trustless manner that you're more vulnerable to attacks from um people that like try to abuse the system as well because you can't just like roll back stuff or or like send relax exchanges to it in in the background and and make users hold that way.
00:36:23
Speaker
Look, and by the way, sorry if my previous answer was too long or too boring. Right? I think oh I have a lot of technical details. No, no, it it definitely helped me to to understand it better. I don't think it's bad to have a bit more of a technical understanding. This is a podcast after all, so have time.
00:36:42
Speaker
Look, as for the question of whether it makes us more vulnerable look i think if it's built correctly it makes us more resilient right and i will give you a couple of examples right so even now for example for all the triggers of liquidations we're not relying on extended order book prices we're relying on the external mark price right so essentially you cannot, like, even if you manipulate extended order books, that that would not have any downstream impact on the liquidation logic.
00:37:19
Speaker
And the same would apply to the cross asset collateral margin, right? So essentially, in order to manipulate, for example, liquidations on extended, you would have to change the price of the asset externally, which obviously requires much more capital and much more effort than doing it on one is exchange, right?
00:37:38
Speaker
That's one.
00:37:40
Speaker
Secondly is let's imagine the worst case, right? Let's imagine that there is a transaction that extended cannot settle on chain because its own chain logic doesn't accept it.
00:37:54
Speaker
Right. And we cannot do anything with that. We cannot roll back the state or whatever. Right. That basically means that the state of the exchange in terms of collateral balances, in terms of equities, etc., etc., is just frozen at some point in time, right?
00:38:15
Speaker
And it doesn't move ahead. Yes, that's that's very bad for us as the exchanger as the operator of the exchange, right? But at the same time, all the user funds ah remain intact at the state at which the exchange was effectively frozen.
00:38:32
Speaker
right And basically, users can access those funds without relying on us as the operator of the exchange. that's the That's the second point. right And the third point is, essentially, assuming that all the on-chain validations are designed correctly, and right any fraudulent or invalid transaction which simply would not be accepted because it wouldn't pass all the on-chain validations.
00:39:06
Speaker
If it passes the on-chain validations, then it means that the transaction has been done in accordance to the rules of the on-chain logic, right? And hence, it's not fraudulent.
00:39:19
Speaker
So that's the way we think about it. So in the nutshell, we think it makes us more resilient and the user wants to be more secure.
Handling Market Crashes
00:39:27
Speaker
i see. That makes sense, yeah.
00:39:30
Speaker
um While we are why we are on the on the topic of risks, um I think it could be a good chance to talk about what happened, i guess, last Friday with the huge event that basically affected the whole market and then wiped out a ton of of leverage and everything across almost every every single exchange.
00:39:54
Speaker
um To be honest, i'm I'm not too aware. I think you have had some issues with that. from some stuff I saw. I didn't really look into it that much, but I guess almost every exchange probably had like some form of issues at that moment. So maybe you could expand a bit on what went down at extended and how you dealt with it and what happened basically. Yeah, sure.
00:40:15
Speaker
Look, um I think we definitely should have done better and we actually could have done better, right? so essentially extended, went down for an hour right after the crash, right? And it wasn't related to the load that exchange experienced. So we went through the, essentially the crash itself and we went through the highest load quite well, right?
00:40:44
Speaker
But then we experienced a bug in the logic of partial liquidations, right? where basically partial liquidations were rejected on chain to our on-chain logic and on-chain restrictions.
00:41:01
Speaker
And we had to update that logic on the fly in order to restart the exchange. So we had to update the logic of the centralized part. right Now, that specific bug could have happened essentially at any point in time.
00:41:19
Speaker
how whether during the... Effectively, it was related to two things, right? It was related, one, to how we... Because the on-chain logic is simple.
00:41:30
Speaker
Basically tells that every liquidation trade needs to improve the health of the account, right?
00:41:37
Speaker
And the bug was related to the fact that if you have a partial execution of liquidation trade, against the empty order books, which effectively happened during the crash where a lot of market makers just pulled the liquidity out.
00:41:54
Speaker
We could not settle a couple of trades and that bug could have happened at any point in time. Right. The good thing it was easy to fix, but ah obviously it's during the, um,
00:42:08
Speaker
crash, right? We had the the market crash. We had like 6,000, I think, liquidation trades. And the chance of facing that specific bug was just much higher, right? And then the chance materialized, right?
00:42:22
Speaker
That's basically what happened. We were able to fix it relatively quickly, ah restart the exchange in an hour. Obviously, there were users who were but impacted by our downtime and so basically our policy is super simple, right? Like we're essentially reimbursing all the direct losses, right? Related to the downtime of the exchange itself.
00:42:48
Speaker
I see. Do you have like an an insurance fund or something or anything like that? Or how do you how do you go about that stuff? Yeah, look, ah for now, we've just compensated the funds from basically the funds of the exchange, right?
00:43:05
Speaker
Because the vault on the extent that operates as the insurance fund, right? that Obviously, we do not want to impact the vault depositors because of our own mistake.
00:43:16
Speaker
But the other thing that, like basically a good learning right, of the downtime and of all the compensations that we are have already paid and will be paying is that we also need to start building up the insurance fund, right?
00:43:33
Speaker
And we are yet defining essentially the, let's say, revenue stream, right, which will be donating funds to that insurance fund.
00:43:47
Speaker
That makes sense. ah um I guess, I don't know if it if it makes sense to talk about a token or anything like that
Future Tokenomics and Governance
00:43:56
Speaker
already. so so I mean, I guess you will have and ah the token eventually. do you have any... How are you going to work with the tokenomics and stuff like that? oh Do you have to plan out yet? Is it like public info or anything like that? Yeah, look.
00:44:13
Speaker
Part of it is public, right? But I think at this stage, I will just share what I can, right? Essentially, the TGE is planned for H1 next year, right? That's one.
00:44:25
Speaker
And then, essentially, once we decentralize the exchange and effectively the exchange operator becomes distributed, right?
00:44:36
Speaker
ah the revenues of the exchange will effectively go to the validators, right, of the operator of the exchange as well as of the EVM compatible network. So the tokenomics is a pretty pretty straightforward here.
00:44:56
Speaker
The revenues, the way they will go, right? they can go both through the buybacks, right? Through other mechanics that is yet to be decided on, right? But the principle will remain very simple, right?
00:45:11
Speaker
So it's basically like... That value goes to the token holders, basically. Yeah, okay. Okay, okay. um I was also wondering why... um Because one thing that I like about your product is that it feels very organic to me or very non... non non-shield and stuff like that. Like you don't really pay people for advertising or anything, or at least that's all it is to me.
Focus on Sustainable Growth
00:45:36
Speaker
um And also like you, apart from from the point system, you don't really have any any incentives like that other exchanges maybe use. Like you don't have zero fees or anything like that, I think.
00:45:50
Speaker
So like why Why did you choose to go with that approach instead of giving out more incentives or like paying market makers, users or KOLs or anything like that to to promote the the product and grow that way?
00:46:06
Speaker
I mean, the short answer right is that the business has to be sustainable. right Without the business being sustainable, right you cannot really grow in the long run right as well as you cannot support the healthy token economics, right?
00:46:25
Speaker
And that's one, right? Obviously, like there will be a lot of counter arguments that with zero fees for takers, right? You can still earn money on the on the flow, right?
00:46:43
Speaker
Which I would say is partially true, right? However, in my opinion, that works much better for the retail apps where essentially the users don't see the order book, right? They just get the price at which they execute the trade, right?
00:47:02
Speaker
And effectively the fees built into the spread, right? So I think like if we have an example of Robinhood, right? Essentially, I think in their public documentation, they are saying that there is an 80 pips spread, right? Yeah.
00:47:21
Speaker
Which is built into every trade, right? When you run an order book based exchange, this is not something you can really do, right? No, that's a bit insane. Yes. So even if you make money on the, even if you make money essentially on reselling the flow or charging the takers, which is basically the same thing, right?
00:47:44
Speaker
One, I think in the long run, your order books will be a bit less competitive, right? Because makers would need to have a way to make money and the overall execution cost for the traders would be comparable as with paying a direct taker fee, right?
00:48:07
Speaker
Just because the economics for the market makers would essentially need to remain more or less the same right? Otherwise they would not, otherwise they wouldn't join. and So we believe that purely from the taker side, essentially the overall execution cost remains more or less the same, right?
00:48:29
Speaker
While for the exchange, it shrinks the opportunities, right? In terms of building a sustainable business model and building is sustainable token in the long run,
Beliefs in Business Models
00:48:40
Speaker
right? And obviously, right? Like you can succeed with the zero fees approach. ah But effectively, we think that with the sustainable revenue model, which is super transparent, clear and understandable right by all the stakeholders in the ecosystem, in the longer run, you can achieve a much, let's say, healthier health a business it as well as the healthier token economics.
00:49:16
Speaker
Yeah, I agree. I think that is also probably the the better approach. I mean, it will it would probably like play out in the long run. We don't really know what would happen, but it just seems a bit more more sustainable to me as well.
00:49:30
Speaker
I'm curious, do you do you personally like trade yourself? Have you ever really traded actively or are you just more of a builder? Yeah, look, like I traded a bit, right? And I traded more in the past.
00:49:46
Speaker
ah just because I had more time. Right. ah But I would say I'm still more of the more of the builder. Right. Like if I see an opportunity and if I have time, I do take it.
00:49:56
Speaker
Right. But like this is not something I'm actively actively looking for. Right. Yeah.
00:50:06
Speaker
As and just to add on the previous topic of the business models. Right. Yeah. I guess like I guess there will be a lot of different opinions in the space about this. right and We actually are yet to see which business model will will succeed.
00:50:27
Speaker
right But I think as the builder, you just need to take a bet on something right and the essentially continue pushing that. right We have a rationale on why we believe our approach is better.
00:50:38
Speaker
yeah right other Others will have counter arguments to that, and that's totally fine. In the end, right the market is sufficiently big and diverse for multiple different business models to succeed.
Building Extended: Challenges and Hiring
00:50:55
Speaker
believe in something, as they say. Yes. It makes a lot of sense here. um What have been, apart from like the the last like liquidation cascade and the bug that happened there, what have been like the biggest challenges along the way while you were building the product?
00:51:15
Speaker
Maybe not not even part as far as bugs or anything or like breaking stuff or just maybe just also acquiring users or liquid liquidity or anything like that. What would you say? What's like the the most difficult part so far?
00:51:26
Speaker
yes i would say that the most difficult part is probably hiring the team right because in the end if you have the right people and the right team you can build almost anything in this world right so it always boils down to people right and hiring is always is the hardest part right for me i actually personally realized the benefit that i have only after i left prevalent right but essentially that benefit is that you worked with
00:51:59
Speaker
tons of people and you essentially have access to engineers, designers, product owners, operations managers, whomever, right? Just because you have access to the people with whom you've worked in the past.
00:52:14
Speaker
Right. And I guess that's one of the biggest benefits of essentially me leading the Revolut crypto operations in the past is just because I have access to those people and I also saw how hiring processes work, right? In big companies, how do they select people, how they hire, et cetera, et cetera.
00:52:32
Speaker
And I would say that it was one of the advantages which we had early on is because we had access to people who could hire effectively, right? That's one.
00:52:44
Speaker
The second thing, like bootstrapping liquidity growth for users is obviously super complex as well. And yeah those things are interdependent, right? You need liquidity to bring users. You need the users and the taker flow for the liquidity to come, right?
00:53:03
Speaker
And basically the approach that we've chosen, we still follow is that we're just trying to grow both organically simultaneously. And I would say it works. Right. So both the usage of the exchange as well as the liquidity on the exchange has improved significantly right over the last three, four months.
00:53:23
Speaker
Right. And that's the only way, in my opinion, how you can solve it organically sustainably. Right. We'll just continue pushing that. Like obviously there are there are a lot of nuances on how you can do it.
00:53:37
Speaker
Right. How we can approach growth. liquidity, etc. Right. And their own like complexities, decisions you need to take, etc. But hiring is still by far the biggest challenge. What like what exactly are you looking for when you hire someone? or what do Do you maybe have any advice to to other people that you've learned while hiring people like what the what to look out for? Maybe just like a characteristic or something in general?
00:54:05
Speaker
Yes, I think that I sort of followed the Revolut's approach where effectively the, like they only had two criteria to hire people, right? You need to be smart and you need to be hardworking.
00:54:19
Speaker
Like your past experience, right? Yeah. And we've, we basically followed the same, the same principle. I mean, of course, like if you're hiring engineers, right? They need to have a past experience. Right. But like in the nutshell, the principle remains the same, right?
00:54:35
Speaker
You have to be smart. You have to be hardworking. You like need to be committed to the product you to the product, both the product and the team right you're working on.
00:54:47
Speaker
And for me, those criteria suffice.
Current Focus on Stability
00:54:49
Speaker
So I never really looked for people with past experience in crypto. right like We always prioritize smart and hardworking people.
00:55:04
Speaker
It sounds like a good question to Yeah, like if you're smart and if you're hardworking, like effectively you can understand almost everything, right? Yeah, that's true. And you gather the experience you That's very true. um I think my last question for now is,
00:55:24
Speaker
Well, you look at extended right now, like currently in the state it is, what what would you say is like the biggest weakness or like the biggest area of improvement that you're focusing on right now, apart from like the the things that you're building for the roadmap for the future?
00:55:40
Speaker
Is there anything, any area where where you would say like, this is something that we're maybe not as good at right now, but we're still like trying to increase our performance a bit.
00:55:51
Speaker
given that we went down during the last crash rate, so the stability is definitely one of the things which has to be essentially fully solved. phrase So regardless of the market events, the exchange always remains stable and available to the users.
00:56:11
Speaker
Obviously, we're trying to do our best to solve that and essentially to have, let's say, 100% downtime aside from the scheduled maintenance.
Conclusion and Reflections
00:56:24
Speaker
right So I would say that's the biggest the biggest other priority.
00:56:30
Speaker
Yeah, i think that's also probably like one of the most important important things in general, like just exchange uptime. I think in many... um actually... Since I haven't looked too much into into what happened last week, the only thing I really saw is I think Lighter went down for a couple of hours.
00:56:44
Speaker
And um um umm I'm not exactly sure. I've heard like they they said something like, yeah, they didn't have enough hardware or whatever. So it went down, which is kind of a really not that great thing to hear as a user because... ah you even, i mean, for whatever reason, but you you don't want the exchange to go down when you need it the most. Like that is just very important for you to happen. It's incredibly frustrating. You're going to lose money and stuff and you just need that to to happen. So I think that's really a good thing to to focus on.
00:57:18
Speaker
Totally agree. Totally agree.
00:57:22
Speaker
Is there, I think I guess this that's like it for me, from ah the questions that I wanted to ask, the topics that I wanted to talk about. Is there anything else that you would like to say ah at the end that you would like to to announce to shield, to shelter publicize, or to tell people, to give them some wisdom them or anything like that?
00:57:44
Speaker
i think the questions you've asked were really deep and really well thought through, right? Thank you. for for all the questions I really enjoyed the i really enjoyed the the conversation.
00:57:58
Speaker
Yeah, me too. Yeah, I mean, like there is nothing there is nothing super specific that I want to add to that. Basically, i think we've covered all the most important topics, right?
00:58:12
Speaker
Everything is good from my side, at least now. Great. I'm glad to hear that. I thank you very much for for coming on. I also think this was a very good discussion. This was, I guess, a bit more technical than the usual stuff. So people are still tune in and like listen to it. But I think it was definitely very interesting and I learned a lot about about the exchange and um I hope I can do also more technical stuff in the future, even though I'm also not like the most technical person myself. So I'm very happy that the
00:58:43
Speaker
that you like the the quality of my questions and um yeah, I guess ah just thank you very much for for coming on. Thank you for your time and thank you a lot. Thanks lot for advising me.
00:58:57
Speaker
Of course. Bye everyone.