Building the Next-Gen Perps Engine on Solana | Tristan Frizza

We really have to buy into this vision of building a really good exchange and filling in that gap that I guess FTX left and doing it in a way that's obviously permissionless and then everything can be audited on chain and you don't run into these these same problems. You own your own funds.

Hey guys, Block Works Digital Asset Summit is going to be back in New York City March 24th through the 26th. We're going to have top speakers from leading asset managers, financial institutions, DeFi protocols, crypto companies, policy makers, you name it. Everybody all under one roof. Think Black Rock, the SEC, Tether, Robin Hood, Salana, and much, much more. Follow the link in today's show notes to get your tickets, and you can use code lightseed 200 for $200 off at checkout. Nothing said on Lightseed is a recommendation to buy or sell any investments or products. This podcast is for informational purposes only, and the views expressed by anyone on the show are solely their opinions, not financial advice or necessarily the views of Blockworks. Our hosts, guests, and the Blockworks team may hold positions in the companies, funds, or projects discussed.

Hey guys, welcome back to another episode of the Lightseed Podcast. Today I am joined by Tristan, the co-founder of Bullet. Perps have become a big talking point within the Salana space recently and of course Bullet is one of the teams building there. So, glad to have you on Tristan. How you doing today?

Yeah, doing really well and looking forward to being on the podcast again.

Awesome. Glad to have you on. I think spoken with several teams now building within this space or maybe even kind of building around this space. Prop AMMs are sort of in this arena as well that have become a big conversation. But maybe before we get directly into what bullet is and what you're building there, it'd be great to jump backwards a little bit and talk about your prior endeavors. So I know you were early relatively early to perps on Salana with Zeta markets previously which has since sort of pivoted into this bullet endeavor. Like I guess at what point was there a realization that you know Zeta the design or like per platforms natively on Salana at that point in time where it just wasn't going to cut it and you had to kind of shift in terms of the design or how to build that from the ground up?

Yeah, it's been quite a journey I would say. So we kind of I'd been in crypto since 2017 or just kind of trading on the side, you know, worked across different roles in kind of early stage startups, working in tech, you know, did a bunch of stuff in the AI space before it all got really hot. And then crypto kind of came back in like 2021 when I saw what was happening in DeFi and the promise of smart contracts. I thought it was very elegant or interesting what was being built, these kind of financial Legos that could all plug into each other with these codified rules that you can build financial products off the top of.

So really the kind of genesis of this whole thing was back in 2021 or even late 2020 when instead of socializing with my family, I was still spending the time learning Solidity in my room and just trying to get up to scratch with what the latest was. And I was quite impressed by you know unis swap this platform doing over a billion dollars in in daily volume that was 80 lines of code basically for its like swap contract. I thought this is very cool and I feel like it's a it's kind of an open field to to go and build your own stuff.

And so we started building essentially like a DEX protocol. At that point it was mostly just throwing ideas at a whiteboard and seeing what would stick. And one of my I guess friends at the time was a was an options trader. And so kind of he really brought me into that world and kind of explained a lot about how options trading worked. And that was where we thought we could fill a niche.

Like we looked at a bunch of different ideas. We looked at under collateralized lending and that was a very cool idea but very hard to I think make it work without some kind of reputation score or or other elements that weren't so DeFi native. And so we landed on options. We we kind of assume that the options market is just as big as the futures market in traditional finance. But in DeFi it was it was minuscule like pers were really dominant but options especially onchain options were basically non-existent.

There were handful of protocols that were kind of struggling there. They had like 5% spreads. You would have to trade them on Ethereum where the gas price was like 20 bucks just because the contract logic was was very expensive to do all these operations and people were trying to do central limit order books onchain offchain all these kind of different hacks. Some people were doing AMM kind of story that we've been seeing go over the last couple years on all these different designs and then yeah we did that for about two years had like decent success essentially I would say we were we were probably like the top volume options platform on Salana in those early two years.

You know we kind of came through one of the Salana hackathons ended up winning that and impressing people with what we could build in two or three weeks that was like a fun experience just sleeping on my on my friend's couch and and essentially just trying to build this thing out learning Anchor learning Salana development and being fortunate enough to meet people like Anatoli in the early days who kind of really inspired me to get back into crypto because I think I'd kind of lost the original vision when I when I first got in in 2017 seeing all these ICOs and other things kind of disillusioned me a bit.

But seeing legitimate builders and really smart people in the space I think wanted to get me back. And the focus on high performance blockchains and and things like that was was really attractive to me. Then obviously 2022, end of 2022, we're at Salana, FTX collapses the next day. It was a whole nightmare situation essentially to deal with liquidity and chains being down and and all kinds of issues. It was it was not a very fun time.

And at that point, we had kind of launched Per as a supplementary product to our options suite to try and get this like full derivative suite going. And at that point, yeah, we I think really believed in in PES was was kind of the way to go. Options was quite a hard sell, I think, to retail traders. It requires quite a bit of prior knowledge. There's a lot of different variables that you have to consider in terms of your risk management. And they're honestly just it requires kind of like a quant finance degree to really understand deeply and make money.

And so seeing where kind of crypto was at still in these early stages of adoption, we really wanted to to double down on PEPS. And so yeah, throughout that bare market period essentially, I think it was just us and maybe one or two other teams really building out PES on Salana. There was there was not too much happening there and like a lot of teams kind of coming and going. But the good thing was I think PES were kind of feasible to pull off. I think they're a really elegant product and they help consolidate liquidity down a lot.

So that's kind of what drove us to to kind of double down on that vision. And I remember very much sitting down, my team, my co-founders, and being like, we really have to like buy into this vision of building a really good exchange and and filling in that gap that I guess FTX left and doing it in a way that's obviously permissionless and that everything can be audited on chain and you don't run into these these same problems. You own your own funds.

And then back to your question, I guess, on like where did it click for us in terms of the infrastructure for Bullet. Well, we had basically been running Zeta for a couple of years at that point. Thankfully the the bull market came back. We got this surge in volume. You know, we were kind of through the bare market with, you know, just like a mere couple of hundred or a couple of thousand users. You know, it was nothing too exciting to be honest. And we just spent a lot of time on just honing our infrastructure and making things faster.

But then I guess at the time that like Jupiter and JO launched their tokens, there was this massive wealth effect. Everyone started pouring back into Salana, bridging crosschain. Salana became a whole thing and we got like a ton of traction which was which was very cool but it stressed all our infrastructure out to the max. You know we had our matching engine stock once or twice and I had to get up at 3:00 a.m. and and fix it and that was that was uh not not super fun.

But then yeah we kind of had good amount of success there over that year essentially 2024 we processed 15 billion worth of volume which was really nice to think this thing that we'd worked on for a couple of years was was doing real numbers you know with a with a B in front of them. And then there was a point essentially I think where Salana was kind of it wasn't down but it was essentially unusable for a good period of like a month. If you kind of remember when stake weighted quality of service was essentially introduced into Salana, this is a problem where people were just starting to spam validator nodes, it was basically sensible or or kind of like to get your transaction included, it was better to essentially spam all the nodes in the network with the hope that you get your one transaction in and everything else would essentially revert.

But this caused this massive DOS attack on the network. No one could get their packets in. I'm pretty sure like Jup was obviously trying to forward a lot of traffic and you know it was just kind of like a a really difficult situation where like smaller apps I think really who who didn't have these kind of priority RPC connections with a lot of stake kind of struggled and so yeah for for like a good month essentially could not land transactions reliably on Salana and as someone building essentially an exchange protocol where you can't get off an oracle update or you can't get off a liquidation and you're running like tens of millions of dollars of of risk or something like that.

It's it's really kind of keeps you up at night. You know, you just kind of wonder like if there's like a big move essentially like if we would have had 1010 at that point, I think that would have been really catastrophic just not being able to to use the chain. And so that kind of clicked for me is like we have to take matters somewhat into our own hand and and build something a bit more customized. And it was also very hard to just execute central limit order books on chain. there's this whole toxic taker problem that was starting to occur and become a big problem for us and many other platforms on Salana where people would come in they'd write these MBV bots essentially and they would basically focus on just sniping maker liquidity so we need this maker liquidity for people to come in and trade on the platform and if someone is coming there and essentially just trying to profit off that and extract it as much as possible I mean they're just kind of playing within I guess the economic framework there and just trying to make the max amount of money and it it just kind of shows I the the floor of the design of the protocol and and how things were working on the L1.

So all those factors combined essentially we decided you know let's take things out of our own hands. How can we build something that's a bit more deterministic that gives us a bit more control over what we do with dedicated block space and being able to make things fundamentally faster than we have before like 400 milliseconds was nice but now you have all these new platforms you know DYDXs vertex hyperlquid all coming out doing kind of 10 times you know faster transaction processing whether that be colllocated L1 and offchain order book all these kind of different models and it was very hard to compete with them for the same amount of liquidity for the same amount of volume when you have this kind of big rate limit in terms of the number of transactions that you can process.

And so that kind of led us to bullet and we can kind of talk about some of the innovations we made there. But yeah, we decided basically to build it from from the ground up as much as we could and think about what do traders want and solve the liquidity problem essentially from the ground up.

Gotcha. That was very thorough. answered all of my background questions on kind of how we went from uh sort of difficulties with with prior perplexes on Salana to now, you know, your decision to sort of build out something new. And I think what we're seeing generally across the board is that um you know, not to put put other teams above you, but just that the broad sort of like Salana ecosystem has recognized that we sort of need like this next gen of purpose exchanges on the platform. the the kind of the old um designs were not going to work at scale or kind of in the current you know microructure of the market at least on the chain.

Speaking to bullets specifically then I guess if we can get into that. I'd love if you could just kind of give an overview explain what that sort of general design looks like and also maybe speak to like I think in terms of the way that bullet is described it's you know you're referring to it as a network extension. Maybe you can talk to that point as well. I feel like network extensions were also maybe talked about quite a bit more like a year or so ago. Probably back during the time when some of that congestion and difficulty around landing transactions was like more of a you know prominent talking point. But since then it's it's maybe fallen out of the conversation a bit if you can talk to those pieces and and yeah give us give us the background on on bullet specifically.

Yeah, sounds good. I guess in terms of the terminology for network extension, I think this was coined by either Kyle or or Austin a little bit ago. And essentially what we're trying to do with that is we're trying to essentially offload execution from Salana L1. So still using it for consensus and ideally data availability like depending on the on the throughput essentially but also just like as a ground source of truth for for data and and proofs and other other such things.

But the execution itself you know the SVM actually computing these transactions you know we're quite limited in terms of the compute budget that you're given and then you have to fit into these discrete blocks obviously and I think that design is harder to make work with per some will argue obviously prop amms are a nice way around this there's also a lot of work being done on the kind of multiple concurrent proposes roadmap essentially to help with some of the censorship and and block reordering that's that's uh concerns that are happening there.

But the way we kind of approached it is if we can build a really specialized execution layer essentially and then still have the data posted down to a chain that has much bigger validator set and then anyone can kind of pull the data off. I think that's like a reasonable compromise and it kind of goes I know Ethereum Vitalic is kind of fing L2s these days but what we're essentially building is like a ZK rollup and you do have more centralized I guess block building but the beauty is you have this decentralized verification of blocks so you can compute these things process them build these blocks very very quickly and then you can go and you can post them to a larger network essentially where you can disseminate all this data and people can sample it and people can basically build their own conclusions and verify that these things have been computed correctly and you can do this very cheaply these days with with ZK proofs and everyone can kind of run a light client you could have that even on your phone you know you could verify the bridge yourself so that's that's kind of the nice way is it kind of keeps I guess the sequencing honest and that kind of trade-off essentially lets us really get around some of the consensus bottlenecks that I think make it very hard to do more mid to high frequency trading stuff on on Salana.

So one example which I think the ANZA guys wrote a really good blog on was like how to build the NASDAQ on chain and they talk about the kind of toxic taker or adverse selection essentially. And to give a concrete example of how that works and how it was working on Zeta and many other kind of fully onchain platforms was a market maker will post essentially set of quotes, a bid and an ask. You know, they might say Salana at $100, I'm willing to buy at 99.9 and sell it 10 100.1 essentially. And so that's the market they're willing to to quote.

And you know a lot of the time for example like a Binance or one of these centralized exchanges maybe even a hyperlquid you know you have price discovery tend to happen a bit more quickly there because you know I guess the the time scales quite fast. Binance can process orders within mere milliseconds essentially. So you know and you have a lot of people trading that and so typically price lags on chain. You might see, let's exaggerate the example and say Binance price jumps up to to 101 a full dollar and within that block essentially you could visualize it as everyone is racing to essentially get included into the block following that to update their prices. They know offchain the price has moved. It's it's kind of gone up and they know those quotes on chain are now kind of out of date.

So the ask that's at 100.1 is is now very very underpriced essentially and there's like a a big amount of profit, you know, 99 cents essentially to be to to be kind of sniped there in a sense. And so it's interesting. It kind of creates this you know first come first served dynamic where the maker essentially wants to cancel his quotes because he knows he's going to lose money on it. Meanwhile, the the taker, the bot that essentially sees this as free money that's on the table is going to come in and try and beat them to it and and kind of scoop up the profits.

The difficulty for the maker obviously is if you want to get included in a Salana block, you have to go through JTO or some kind of transaction landing service. You probably have to pay pretty high tip fees to get included. You know, same same slot or or kind of ahead of these guys. And so you're kind of in this interesting situation where either you consistently pay very high fees to always basically get out of the way of the the steamroller of these bots or you essentially just don't do anything and you update your quotes slowly and you just basically eat the loss.

And in either case, it creates this negative EV situation where the maker is basically posting liquidity and they're just like bleeding money over time. And unless you have extremely robust like retail flow, which you might have on Jupiter, you know, for example, as a as a proper mm plugging into that, it's just like a lot of toxic flow is hitting you and you're just losing money essentially.

And so there's a few ways to kind of get about that. One of them is obviously I think just speeding up the chain is is very much I think kind of helps especially with central limit order books. That's kind of what you see happening in traditional finance. It's what you see happening in centralized crypto finances like just building fast execution and and letting this kind of first come first-serve order book process really quickly I think is a good way to get around that.

And to speak to bullet specifically since we've engineered it from the ground up all in rust written this whole matching engine kind of as you can think of it almost as like the matching engine and the order book and the margin system are all basically enshrined into the chain like it's a first class citizen there's no virtualization that goes o over the top of it. This is basically baked into the chain and as a result it's it's very fast. So we're pretty proud that we can execute an order on our order book in basically 500 millisec 500 microsconds. half a millisecond which is incredibly fast.

If you were to compare that to a L1 block I guess it's 800 times faster or you can get 800 times more transactions in there which is really nice. So a lot of this stuff can happen on the more micro time scales. People can adjust move in and out of of kind of orders very quickly. So we we put a big focus on that. The other thing you can do is is potentially application specific sequencing. This was like a big focus of that article as well something that hyperlquid I guess pioneered was was doing this cancel prioritization.

So market makers actually get this affordance where they can come in and they basically get first look to be able to cancel their transactions. And this has the nice effect even though people will question the market fairness of it in the bootstrapping phases I guess of an exchange. It is really nice to have that benefit as a market maker. you're not competing purely on the latency of your connection and having to put all this, you know, crazy microwave towers and fiber optic cables, you know, to to your validator or something or other to get the the lowest possible connection speed.

You basically create a bit of an even playing ground where even a guy just running a Python script essentially who's not even colloccated within the same 50 or 100 milliseconds essentially can be competitive with Wintermute. And that's a really nice way to just bring in 50 or 100 of these kind of like hobbyist guys to to start market making. And and the more you layer on that liquidity, books get thicker, retail looks at this and they say, "Hey, I want to trade this." And you kind of get a bunch of retail flow. And then more of like, you know, the jump tradings of the world will come in because it's profitable for them.

So I think that's also a nice trade-off and and we have kind of a a system which is exploring that as well. So basically, yeah, building up all these different tools I think is is really nice. focusing on the speed, focusing on the ordering guarantees as well and just making it really suitable for providing deep onchain liquidity.

Gotcha. I think that makes a ton of sense. Definitely the I mean there was certainly sort of like an added cost of doing business for makers. So trying to alleviate for that problem makes a ton of sense. I guess in this particular design I know you guys use a couple of different sort of like infrastructure providers like I think Celestia for DA maybe even succinct for their their ZKVM like I don't know if you can maybe talk through some of the the thought process behind like incorporating with each of these like are there technical limitations and challenges to doing more of these things like you know natively on soul as well such that like it makes sense to go with like a Celestia or you know succinct for proving and kind of like the the reasoning and the decision- making behind that?

Yeah. Yeah. That's a really good question. I can talk about Celestia first up. So we had built over the period of many months essentially like a data availability adapter for for Salana and the the problem of data availability is like how can you take all the transaction data posted somewhere on chain that anyone permissionlessly can come in and essentially find this data you know make sure that it's not being censored because you know if it is censored you can't actually run the proofs and and you can't apply these things essentially and figure out what the what the end state is.

And so we need to kind of disseminate that data. We were trying to do it originally like purely on the L1 and we were just basically writing it to the like call data logs in in Salana messages and it was working like reasonably well. We kind of got it working but it was very hard to scale more than like a few hundred TPS essentially. And our benchmark really internally is like 5 to 10,000 TPS to be competitive with with other exchanges. And the thing is just like the data rate on Salana, we're just like hitting it with a lot of place order transactions.

And so we kind of like built that out. that was like harder to scale because we would have to be doing like double-digit percentages of you know Salana's network throughput and being able to land that much data I think is is very hard. Salana unfortunately as well is just like not very well optimized for storing kind of blob data or any kind of like binary data format. So it was it was not the most efficient it could be.

And then another thing is like Salana is inherently not really built around the concept of meralization. So creating these these proofs and these checkpoints almost along the way where you can kind of prove state from from different parts was was a lot harder on Salana. There was a way to do it but until basically a year ago they had uh removed it from I guess the Salana core codebase. It's this accounts delta hash that they essentially changed and that just made it very hard to build like clients like natively on Salana. Like some teams had been trying to do this for for some period of time and it just became like technically infeasible.

So that's when we had to evaluate Celestia. Celestia is basically just like purely a DA layer that they are their own L1. But they are really good for just hosting blob data and that's pretty much all they do is you just write blob data. And they just have like a very fast network and I forget the exact numbers of how much they're pushing through the network but they have a pretty aggressive road map. And we need from the NAP and math I think to get 5,000 TPS at our kind of transaction sizes it was something like 1.6 six megabytes per second and they're planning to scale to easily 10x that and and more over the next year, which is pretty cool. So, we're looking forward to just like pumping a lot of blob data through there.

Yeah, that integration is still kind of ongoing. We're going to have that live very very soon. And then on the proving side as well, this is kind of this is the magic of of how it works and how it is verifiable essentially because people will kind of say, hey, if you're you're running this kind of single sequencer network, like what's the point? house is better than a centralized exchange. But the cool thing is everything that runs in our exchange, you can basically generate a proof. Every single transaction, every single order that's matched, everything is going through what's called a ZKVM.

And the way you can understand a ZKVM is it's like writing Rust essentially, except you can basically spit out a proof of execution, which I think is like a very novel thing. And when I first heard about it, I was like, this is very cool that we can have provable computation essentially. And so that's definitely one of the the limitations that we have to work around. I think some of these ZKVMs have like quite limited memory essentially. So we always have to work within those constraints. There are certain crates that might not be supported essentially.

But the nice thing is that when we run bullet essentially everything that goes through there we can spit out a proof for and is like entirely provable. we can then spit those blocks out and all that data and then we can attach a proof with them and anyone can go through they have the data and they can verify the proof as well. So that's how we use those two providers.

Gotcha. That makes a ton of sense. And then maybe just one followup there and I don't know if you have clear insight to this quite yet but like a as a cost consideration as well. Like I know at least for Celestia I understand that the cost like per data kind of posted there is is very cheap. And there's been plenty of articles that kind of go down that rabbit hole. For on the proving side, I'm less familiar like proving this much data and making you know kind of that kind of throughput of of execution available via those proofs like is that costly or do you have like line of sight to what that would be as sort of like you know an operation cost for you guys?

Yeah, it's a really good question. We haven't run like extensive extensive benchmarks and we're kind of in the middle of doing this doing this exercise. There are a bunch of like nice ones essentially on you know basically you can prove Ethereum at the moment in real time or near real time which I think is quite impressive. Obviously the gas throughput that we're doing is is probably an order of magnitude or two greater than what we've seen for Ethereum. But the nice thing is with a lot of these proofs you can kind of condense or you can kind of compact a lot of the data together and to to kind of prove from block n to n plus one or n plus 100 it's actually the same thing and you can kind of amotize a lot of these costs and that's what some zk rollups like starkware from what I understand basically do So instead of proving every single block live as as it comes they might batch it into four or eight hour windows essentially prove a big block that saves on the costs.

The one downside is obviously the bridging times are what are affected here. Basically your your finality almost of the roll up that everyone can essentially see this this this proof is is delayed. But yeah I think it's it's kind of like a costbenefit trade-off there where we kind of want to land on that spectrum. The other really nice thing is ZK proving is or ZK tech in general is improving at such a rapid pace. I would say pretty much every year we're seeing something like an order of magnitude improvement. And proving costs I think on Ethereum for an Ethereum block are probably on the order of cents or or tens of cents essentially. Maybe that's per transaction. Sorry. But yeah, it's getting basically very cheap and on the order of you know some of these high throughput blockchains which I I think is really cool. And within that kind of range of cost, I think it becomes very affordable and feasible for consumers.

Gotcha. That makes a lot of sense, I guess. And if that sort of rate of improvement continues, then you know, we could go from cents per transaction to, you know, cents per block and then so on and so forth. So yeah, exactly. And it also scales like people are finding more ways to paralyze obviously with this AI boom going as well as massive investment into compute and GPUs and stuff like that. So that also I guess can all carry over.

Yep. makes a lot of sense. Maybe let's talk about something I've I've seen you write about is this unlquidatable basis trade, but that conversation kind of centers around the fact that bullet will have you know spot lending per all integrated into one engine. I guess can you talk about the benefits of that like sort of integrating all those pieces together and then to like some of the interesting sort of you know trading capabilities that maybe that enables and whether or not that's really like feasible at other junk you know locations today?

Yeah, this is a really good question and something I spent a lot of time building out. I think it's quite an important feature. We've already always cared quite a lot about capital efficiency even in the the options days. We kind of built our own very basic version of portfolio margin there where you could kind of net off some of your call options, put options. Like some of these things basically neutralize I guess some of the the risk curve and you can kind of look at things in a holistic view and and see where certain parts can actually discount margin from other positions.

In our case specifically, I would say our number one biggest piece of feedback or complaint on Zeta, other than being able to land trades like super reliably, was the fact that people wanted to use coins as their margin essentially. You see this on most centralized exchanges. You know, Binance has coin margin per. There are very small, I guess, select few exch like DeFi exchanges that also had enabled this this feature. drift being like a notable example on on Salana had had built this out and I think people trading across the two platforms really liked that feature and I think it was it was quite nice especially like thinking about the bull market right not a lot of people want to stay on sit on stable coins like USDC and if you're coming in to long Salana essentially imagine you have a big bag of spot Salana you come in there you would have to go and basically sell that down in the spot markets and then you'd have to deposit into the platform to to long it again so that was a bit of an inconvenient step But it is quite a difficult thing to build many platforms even hyperlquid until basically a week ago was uh was only USDC margined and taking that kind of feedback on when we started building bullet I kind of sat down with our engineers and I was like this has to be a first class feature and if we don't kind of ship it in the first version it's it's probably just going to get delayed indefinitely and and not shipped you really have to kind of think about this thing from from first principles and that was what really I guess mandated or kind of motivated the spot and lending components of the markets.

Like we're not trying to go out there and compete with Camino or you know some of these spot exchanges necessarily, but this is kind of a really nice addition that brings the the trifecta of per lending and and spot together. We kind of have a similar model to I guess how some of the centralized exchanges do it. So people can come in there, they can trade using multiple different assets. will whitelist a bunch of I guess like pretty liquid ones, a bunch of the majors, you know, sold BTC, ETH, GTO, sold a bunch of these liquid staked ones, you know, and a bit of a longer tail. And that's really nice because you can basically just take whatever's in your Phantom wallet, you can deposit it, you can use it as margin. I think that was like a a big draw on some of these centralized exchanges is you can just use the full basket of assets.

The other really nice thing is because you have the uh borrow lend market under the hood, you're actually getting the benefit of the underlying interest rates there. So you're kind of earning yield as you know money basically just sits in your account which is which is a really nice benefit there. So yeah, that was a big piece of feedback. We took many months essentially to think about how to build that in a in a scalable way and kind of delivered that I guess like early last year and have been continuing to to tune that and it's really cool to see Hyperlid moving in a similar direction shipping their own kind of portfolio margin which is kind of similar idea with with some extra kind of complexity in that and yeah I think it's it's kind of a really good way as like a TVL sync as well people can come in there's reason to kind of deposit a lot of collateral maybe on Zeta that was less reason because you just have idle stable coins sitting in your account, you're really incentivized to to trade a lot.

But we will have a few kind of really nice sources outside of just pure trading. You'll be able to put it in there, own the interest rates. We'll also have vaults and other things that people can deposit their collateral into that will basically go into market making the platform and we can hopefully make the the yields attractive there, you know, depending on how things go on the launch. But yeah, I think having stuff for active traders and for more passive guys, I think it's nice to balance the two.

Yeah, that that definitely resonates. I think what we've seen maybe generally speaking within finance apps but also within crypto and and the purpose base specifically is this almost like necess like the bare minimum of what a product or an exchange is like the the the requirements at the base level have continued to expand. It's like you you kind of need to be a multi-product platform from day zero now just because the the expectations are so high in terms of what's offered. And I think um your point about earning yield on like idle capital that's become a big one for kind of like financial products generally speaking right it's like the prediction markets are talking about you know earn some earn some yield on your idol trades or your idol USDC and um it's that's become a big one across