Manifesto: The Path to $100m Daily Volume on Chainflip
![Manifesto: The Path to $100m Daily Volume on Chainflip](/content/images/size/w2000/2024/06/Path-to--100m-in-daily-volume.png)
Consider this blog post a bit of a strategy update. We’re six months into having a live protocol, and even less with it de-restricted and truly in the wild. We always knew that when we launched the protocol, we would quickly figure out what works and what doesn't. Our north star metric at Chainflip Labs to measure the success of the protocol is volume. We've had some good days, some slow ones too, but overall pretty consistent upward growth in volume since the swapping protocol launched.
Firstly, I wish to speak on a key thesis that we launched with, which is that if the Chainflip swapping product is competitive, nice to use, and easy enough to integrate, then we don’t have to do much to achieve organic volume. This is the “build it and they will come” philosophy, which has its drawbacks of course, but it does allow you to see (especially early on) the truth about a given product. Many criticise this approach, saying that we aren’t “promoting” the protocol or some other nonsense. My counter to that is that we are choosing to not poison the data we are collecting about the product with bullshit schemes to temporarily boost the metrics for temporary financial benefit. Instead, we are seeing what induces consistent demand for Chainflip, and those successes do a much better job at promoting the usage of the protocol than fakery.
All of the volume that you see on Chainflip is entirely organic. We do not have any form of wash trading, taker incentives, or internal swaps that can distort the statistics. Thanks to some great early partnerships that we unlocked, we have been able to create a user base and leverage others who have used the product only because it solves a real problem for them. In that respect, the Chainflip protocol has been successful, regularly seeing several million dollars being traded every day on a protocol that currently only supports five assets.
As impressive as that is, it now seems as though we've hit a local maximum of the daily volume we can expect with the protocol in its current state. Realising this would likely be the case over the preceding weeks, we've been adjusting and updating our development and ecosystem roadmap to achieve a relatively new goal that we’ve mentioned here and there: $100m volume days by the end of this year.
I watch the protocol swaps every day. I study what is driving our volume, what kind of swaps are being done, where they come from, what the LPs do, how they react, how competitive we are, how we recover from bad situations, user behaviour on other platforms, and a dozen other factors which might help me understand where we are succeeding and where we can improve.
With this information, I have to help shape the development and ecosystem roadmap to apply our limited resources in the most efficient possible way, prioritising the features and initiatives which, according to these hypotheses, should make the biggest impact over a given period of time. This is a constant game of tradeoffs, and you never know what impact any single initiative is going to have until you do it, but by discussing with the team and understanding the other factors that go into deciding what to do and when, we end up with a roadmap.
In this post I will lay out some of the key strategic insights we’ve been able to learn from the first months of being live, and what we plan to do with these ideas over the next six months in order to achieve the goal of $100 million dollar daily volumes through the Chainflip Protocol.
Insight #1 - The JIT AMM and pricing model - Great for small swaps, but hitting fundamental limits
The JIT AMM combines both the best of a v3 AMM with that of an intent-based system. We have been very pleased to see that with just $10m in liquidity rolling around the protocol, we have been filling big swaps daily at prices that outcompete other protocols. However, we have now had time to observe and understand the limitations of the liquidity model.
Of course, there are still lots of little tweaks the protocol and LPs can and will do to slightly improve pricing here and there (JIT quoter, tighter spreads, split bids, faster blocks, etc), but we need to talk about a more fundamental issue with the model.
I will get to the *amount* of liquidity in a minute - but for anyone thinking we just don’t have enough liquidity to compete for big swaps - you’re not entirely wrong, but having more liquidity doesn’t mean better pricing on big swaps, and therefore doesn’t mean more volume (in isolation at least, more on that later). I used to think that more liquidity would always lead to better pricing, but I have realised some flaws in this reasoning. To explain this, let’s explore the dynamics we’ve seen in the JIT AMM model in the wild.
The theory behind the JIT AMM states that liquidity providers would bid on every swap to the maximum extent all the time. In practice, that turns out to not be true, particularly for larger swaps.
For one, LPs are all bidding on swap using the same liquidity sources, and so often end up bidding very close to each other for any given swap. Paradoxically, having more LPs/Solvers makes this problem worse (ref below), as then they have to factor in the issues that arise when using the same liquidity sources to place bids - the act of two parties potentially trying to fill the swap using the same liquidity increases risk - and therefore forces both LPs/Solvers to increase their spreads, where if just one honest LP bid on the order, then there would be no risk in offering the best price.
This isn’t a problem when the swaps are small (below $50k generally), as there is little risk the liquidity can’t be matched. But instantaneous liquidity on the global markets isn’t as reliable for larger volumes.
I’ve spoken with the LPs and they say that they just can’t quote better and remain as profitable when swaps reach a certain size in Chainflip right now, due to a few seconds of price risk inherited from the block time and the depth available on global markets. There simply isn’t enough reliable instantaneous liquidity out there to provide a near-zero slippage quote when you start getting into the $300k+ territory. That’s why if the only change to the situation was that the JIT AMM had $100m in liquidity available to it today, there wouldn’t be a game-changing improvement in pricing on big swaps - we are already getting close to maxing out the swap sizes under the current model while remaining competitive.
However, that isn’t the end of the story. We all intuitively know that if we market sell $250k of something, we’d probably get a worse price than if we’d placed five $50k market sell orders over a period of time, even over 15 seconds, rather than all at once.
This is one of the great insights from Streaming Swaps (THORChain) and DCA swaps (Jupiter). Even if you can match instantaneous prices for swaps up to a large volume (as is the case with Chainflip), it still will be beaten where that trade is spread out, even by a tiny bit.
So therefore, in order to really leverage the JIT AMM to its full potential, we need to give LPs more time to fill swaps and make it less risky to do so. This is why we have prioritised adding a DCA/Streaming swaps feature to the protocol to give LPs greater flexibility in filling orders and requiring less instantaneously available liquidity on the markets to fill swaps with near-zero slippage.
We’ve seen that on most popular DEXes, the bulk of volume comes from large orders, and increasing the time LPs have to fill those orders is really the only way to achieve that under the current liquidity model.
The JIT AMM is still very important here: no other cross-chain DEX market model will allow LPs to fill as much volume per minute with minimal slippage. If you have a more efficient underlying market model, then you don’t need to split up a $1m swap into 50 pieces to achieve near-zero slippage - you could probably do it in 10, and over a much shorter period of time.
All in all, DCA Swaps and better pricing have a huge impact on the protocol’s “best case” competitiveness, but then there’s then there’s the topic of reliability.
reference: https://arxiv.org/pdf/2311.18164 (The Paradox of Just-in-Time Liquidity in Decentralized Exchanges: More Providers Can Lead to Less Liquidity)
Insight #2 - Reliability is More Important Than We Realised, and Guarantees Matter
If I’m gonna swap $100k of anything (let’s say SOL to USDC as an example), will I choose:
- The protocol that is faster and has better pricing, or
- The bridge/protocol/CEX that may be marginally slower and offer a slightly worse price, but does actually guarantee that my order will not be executed if the actual real execution price isn't as good?
This is speculation, but it’s likely that when traders are comparing routes, and see that Chainflip offers no meaningful price protection, they go elsewhere, and that is hurting volumes. I wish it were possible to be able to measure to what extent that is true, but it isn’t. In any case, we’ve heard from feedback and know intuitively that this is an issue.
It’d be nice to be able to hand-wave that away and say “well, you know, 99% of swaps are executed with good pricing”, but the reality isn’t as good as we’d hoped.
This is compounded by the fact that we regularly see instances of LPs running out of liquidity on an asset, offering a bad price to protect the liquidity, and then some unlucky soul eating as much as 4% slippage because someone else has been hammering the same side of the pool all day.
We have also seen instances where users have cleaned out the LPs entirely and triggered complete swapping halt until LPs have rebalanced and can fill the order - something which usually takes over 2 hours to execute (and to this point, has been quite a manual process). This global 5% slippage protection is a good backstop for some truly egregious outcomes, but it’s not a meaningful protection. 5% is a lot, and doesn’t inspire confidence.
All it takes is for one big trader to send multiple large swaps over the course of a couple of hours, and we are in a bad spot, and can’t offer subsequent traders any guarantee that we will be able to offer them a good price in the time we specified.
There’s a few interrelated issues here:
- No price guarantees
- Rebalancing takes too long
- Limited LP collateral on hand on busy days
One of the main factors that has separated the strong volume days from the weak volume days is balanced and available liquidity. If you have a lot of swaps going in both directions for each pool, the pools stay balanced, and LPs can keep quoting aggressively. As soon as you start to move out of balance, one side of the books becomes weak, and with rebalancing taking a while, a general shortage of liquidity is triggered, which then reduces the competitiveness of the protocol as a whole, hurting volumes.
Side insight: This balancing issue is one to address if we want to have consistent volumes, although we can probably see higher overall volumes if we just focus on the best pricing we can for each given route individually at any given time, rather than trying to average out the liquidity across all the pools. LPs already exhibit this behaviour naturally, but because arbitrage doesn’t really happen on Chainflip, it’s down to the LPs to rebalance their inventories, rather than them losing money on arbitrageurs taking advantage on over-quoting on assets they are low on, as is the case in many other DEX environments. Instead, particularly when competing for user flow through integrators, we can let other protocols take routes we can’t fill and instead compete on the ones we can, which will rebalance the situation naturally.
The main feature which will address these issues in the works is Fill or Kill. Working hand in hand with DCA swaps, they are set to greatly increase the level of control users have when making Chainflip swaps. Adding meaningful price protections does several things:
- Insulates users against liquidity shortages
- Promotes consistent strong quotes amongst LPs
- Increases user confidence and therefore willingness to make large swaps
Fill or Kill, which requires users to provide a refund address and a price limit, is not a silver bullet. The underlying liquidity issues are still there, but they will be less damaging, and with demand for larger swaps higher given DCA swaps and price protections, greater liquidity reserves are naturally incentivised.
However, rebalancing is still too slow. If swaps fail and are refunded, it’s painful for the user, and they may not be willing to come back later and try again. To avoid missing out on these users, we’ve got to somehow make rebalancing faster.
There are several factors that determine the speed of rebalancing, but they are:
- LP response time
- CEX withdrawal time
- Chainflip Deposit time
With JIT Strategies, currently the largest LP on the protocol, having implemented an auto-rebalancer, the response time from them is now close to zero. When an imbalance is detected, the system automatically starts to move money around. The CEX withdrawal time remains the most problematic component of this, with manual approvals often required from the CEX providers. This can vary dramatically but it is a challenge to overcome. Then finally, LP deposits on Chainflip add another 5 - 30 mins to the process.
Boost is a feature which could cut down rebalancing times dramatically. With less restrictive liquidity sources organised, deposits and withdrawals could be automatically triggered within seconds, and the deposit boosted by Boost LPs to allow the LP to quote with a balanced inventory within 30 seconds. This is the holy grail of rebalancing systems, but will still take some time on the LP’s end to set up operationally with the exchanges, and unfortunately outside the control of Chainflip Labs.
However, in this way, Boost contributes to the protocol’s chances of winning big swaps, and not just by offering a convenient speed-up option for swapper in a hurry. It allows LPs to rebalance on the fly and thereby keep liquidity available consistently throughout the day. LPs will be able to fill swap orders with deposits that haven’t yet hit the system through boosted LP deposits. Boost is rolling out very soon (starting with just BTC), but will likely take a bit of time before the full benefits of the feature are truly utilised by LPs.
To wrap up this point, we should talk more about the underlying liquidity. Unlike pricing in the JIT AMM model, having more liquidity does make a big difference when it comes to reliability. Obviously, the more liquidity you have on hand, the more you have on hand to use before rebalancing is required. We have launched a Liquidity Incentive Program to encourage LPs to bring on more capital to enable this, though the program has not yet resulted in any meaningful amounts of capital coming in just yet. I wrote a separate report on this, but the long story short is that more work is needed to give LPs enough reason and enough data to meaningfully compete in this sphere.
As mentioned in the previous section, having more LPs isn’t necessarily an answer to this reliability question, but having more liquidity is. Currently, we see about $10m floating around the protocol, which could be deployed more efficiently. However, with DCA swaps the greater concern is going to be the total amount of capital available to prevent these rebalancing lockups. Until DCA is in, I don’t think our current liquidity levels are our greatest limitation. Once it is in, I don’t think much will be needed to attract more collateral into the system - the financial incentive to commit capital will simply be there if tens of millions in new swap flow is coming through the protocol every day and not enough capital is there to fill it. This isn’t a chicken and egg problem - this is a clear market structuring issue that can’t be fully addressed without Fill or Kill and DCA swaps, which when both in, will naturally cause more capital to be deployed in the Chainflip protocol to capture the opportunities they will create.
At the moment the cost-benefit ratio for LPs to commit more just isn’t there without further incentives. That changes when the natural benefit jumps drastically.
A quick pause…
So - to recap - DCA swaps allow us to compete for big swaps, FoK guarantees users won’t get screwed when they swap, Boost helps address underlying rebalancing issues, Liquidity Incentives help encourage more collateral to work with, along with more flow generated through the aforementioned features. They should all work in unison to make Chainflip the most efficient route for large swaps between the assets we support.
However, there’s two more major topics to cover to really cap off this strategy update. I know we’re already 2,000+ words deep - but there’s a lot to discuss here.
Insight #3 - Integrators *Are* the Distribution Strategy, And They Care About Chain Support.
While we have some great integrator support already in place and more lined up for the future, we’ve made several learnings about what they care about and how we can make life for them easier, and most importantly, what makes Chainflip attractive to new integrators.
Integrations remain the number one pillar in our growth strategy. I want to stress this point. Our go-to-market strategy hinges on other teams, products, and protocols leveraging Chainflip to serve their existing users. We can promote the hell out of our own frontend until the cows come home, but that is a slow burn process compared to just tapping into the billions of dollars in daily volumes traded through existing products.
Assuming the underlying market structure improvements are addressed as outlined above, the key factor in driving up volumes to new levels is to secure new integrations and boost the existing ones. We are also aware of several exciting “sub-integrations” (that is, wallets and services that will/are already using our partners to offer cross-chain/swaps to their user bases) in the works that could put Chainflip in the running to compete for some serious swap flow. We are not in a position to disclose what we have heard about this topic, but we need to be ready for them in any case.
Side Insight: We’re introducing a new feature to 1.4 that was also requested by some of these partners. It’s called Affiliate Brokers, and it allows brokers to set up their channels to allow “sub-integrators' ' to be paid a share of the broker fee automatically. This is great, as it means that if something like SwapKit (by THORSwap) is used by a third party, SwapKit can split the broker fee between themselves and that external partner. It makes managing the overhead of these indirect relationships much simpler. Again, this really helps the growth strategy by enabling partners to do a lot of the outreach and integration work for us.
We’ve been really satisfied with the level of usage of the frontend we offer at Chainflip Labs, and feel confident that as long as we can maintain a quality user experience and an increased feature set over time, the userbase of the frontend will continue to grow along with other venues. Part of the user acquisition strategy for the frontend is to show Chainflip as a route on other venues, which may attract those users to check out the native frontend.
So what have we learned about achieving and maintaining these lucrative integrations? The outcome of the discussions we’ve had with current and future partners is that chain support is the number one attractive feature of Chainflip. Bitcoin, having made up over half of our existing volume, is currently the “killer app” of Chainflip, but there are other routes that need to be added in order to truly leverage this.
I’ve spoken in the past before about how Arbitrum, despite being a popular L2 in its own right, serves Chainflip primarily as a strategic listing to allow for cheap multi-protocol hops. A BTC to Base USDC swap is suddenly cheap and possible, for example, by using Chainflip to swap the BTC to Arbitrum USDC, and then calling Across protocol to bridge the USDC over to Base or any other widely supported L2 cheaply, quickly, and efficiently. Doing the same using Ethereum USDC is prohibitively expensive with the cross-chain messaging calls supported by the protocol incurring large gas fees.
On launch, I’m not expecting large volumes to/from Arbitrum. I think it’s more likely that this is a listing that will have a compounding effect over time as more integrators leverage the listing to craft elegant routes for users. But for many, without Arbitrum, Chainflip remains too limited in its support of popular L2s to be considered useful for anything other than simple BTC swaps.
That said, the one future chain listing that gets our partners really excited is Solana. It is currently very underserved outside of its own chain ecosystem. The existing bridges are clunky, inefficient, and often entail wrapping/unwrapping in multiple steps. The pricing on these swaps also leaves a lot to be desired considering the extremely good liquidity available on centralised exchanges.
Chainflip is also uniquely good at abstracting the underlying mechanics away from external developers and handling a lot of interactions through the SDK. In conversations we’ve had with our partners, we are pretty excited about the potential of Solana to elevate the Chainflip protocol above and beyond the current levels of volume. It would be a truly unique offering in the space, and I’m very pleased to report that progress on Solana has accelerated quite a lot since the Arbitrum update in 1.4 has been finalised and we have had more space to work on it. Still, there have been a number of unique technical challenges to solve in order to achieve this level of support, but by the same token, that also makes us confident that it will be relatively difficult to replicate in other protocols, and thus will become a major unique selling point for Chainflip in the eyes of external integrators.
Adding new chains does more than just introduce new markets. It also boosts the potential of the existing pairs on Chainflip. SOL support will likely also boost BTC volume, for example, as it opens up a whole market of users looking to swap between the two assets that didn’t previously exist.
The pace of chain support has not been as fast as we would have liked, but all of the new chains we are adding have unique architectures that require extensive bespoke engineering to safely include in the Chainflip protocol and network. That’s not a bad thing - it means we are offering a unique product in the marketplace, and it is this specialisation that is especially appealing to integrators. Why add 15 L2s when they are already extensively covered by competitive and effective protocols? We should focus on the protocol’s strengths rather than trying to match the performance of half a dozen bridges that can already move stuff around the EVM ecosystem efficiently.
Armed with this knowledge, we can make informed decisions about what to integrate next once our pipeline is cleared - but already the two new chains being launched are the two key chains that can bring Chainflip in front of hundreds of thousands of potential swappers - and as a consequence, bring that sweet volume.
So please don’t ask me what comes next after Solana. I don’t know yet, and if I had formed an opinion solid enough to share this early, you should slap me. Seeing as we have no capacity to do research or planning for new chains at this moment in time, we would be making assumptions about the future market for chains before we really needed to, which would create unnecessary lock-in and cloud judgment regarding the next steps. We have plenty of ideas in various stages of development, but we will tackle that as a discrete research subject soon.
Insight #4 - Illicit Flows Threaten Consistent Liquidity
Like any other decentralised and permissionless DEX, bridge, or protocol, there are limits on what can be done about illicit flows running through the protocol. We had implemented address screening on the front-end before launch, but we found that it wasn’t as effective as we had hoped. As a result, tainted assets made their way into the protocol, and eventually triggered a flag on the LP’s centralised exchange accounts, causing liquidity to dry up while this was sorted out earlier in May.
We quickly moved to harden our countermeasures when these flows were detected, and since then, we aren't aware of anyone attempting to move these types of funds through the protocol. Still, it did make us acutely aware that if this isn't taken seriously by all Chainflip Brokers, then LPs may not be able to consistently service the protocol. It created unanticipated operational overhead and meant that we had a couple of days where swaps effectively came to a halt.
Thus, we need to do more work to keep this threat at bay, and ensure that LPs can operate without the risk that their funds/accounts will be restricted. For now, things are in a better place, but we’ll keep our eye on this issue so that it doesn’t block us from walking the path to $100m days.
A Summary of Initiatives, Timelines, and Impact
Let’s list all the features/initiatives I have mentioned in this post. It is totally unclear which of these features will have the most impact on volumes, but it is more likely that the gradual combination of the features compound upon one another to create a synergistic effect on volumes rather than any one feature being game-changing in and of itself.
- DCA Swaps
- Fill or Kill
- Solana
- Arbitrum
- Boost
- Affiliate Brokers
- JIT Quoting
All of which can be seen in the Roadmap: https://chainflip.io/roadmap
Talking about timelines, it’s a constant juggling act to figure out what to work on at any given time. Context switching sucks for engineering and slows everything down. Lessons are quickly learned and tweaks need to be made to existing features, and no doubt the new ones listed here when and as they arrive.
So I can’t give you firm timelines on the complete rollout of any one of these features, but I can say quite confidently that all of these will be at least partially implemented and live by Autumn, starting in two weeks when 1.4 hits the network with Boost and Arbitrum.
As they are launched, it will take time for integrators to respond to these changes, and for LPs to develop their systems to respond more effectively to the changing market structure and the new inflows triggered by these changes.
That’s why the goal isn’t $100m by then. I’m being realistic and saying that *yes* - these changes are big and will set Chainflip up to be *capable* of such volumes, but when those volumes actually happen is likely towards the end of the year rather than just as they are released.
And of course, everything I’ve said in this post is a hypothesis. Being the complex protocol that it is, and with very limited data to measure the success of any one feature in isolation, I can speculate on what will make the biggest difference, but I am not going to sit here and claim that one is going to be more critical than the other.
Overall, the strategy I’ve laid out in this post should make it clear that I believe that the combined effect of all of these new improvements will be greater than the sum of their parts. If this is true, we should see compounding growth some time *after* each improvement is made, rather than immediate and significant jumps in volume after any given release. Let’s see if I’m right.
I’ve been wrong before about a few design choices, or made some incorrect assumptions that have led us to do stuff that didn’t move the needle. This whole post is a demonstration of that - we have to respond to the information in front of us - and rest assured, I am constantly monitoring the situation and updating my mental model of the system in service of our broader goals, aided by the help and insights from my awesome colleagues and the wider Chainflip community.
All that said - if we can achieve this goal, then the buy-burn mechanism of the protocol should eliminate any overhanging concern about the economic viability of the network. If we can keep the $100m daily average volume figure consistent over 2025, one would estimate that somewhere in the region of $20m to $30m in fees will be used to wipe huge wads of FLIP off the face of the earth next year. Much less is needed to make the protocol sustainable.
So thanks for coming to my TED talk. I hope this has brought you closer to the decision-making behind the roadmap and has helped spur some thinking about ways to help.
The path to $100m is before us. Now let’s walk down it and see where it really goes.