AIP - 71: ANGL Market Making Proposal - Arrakis PALM


Deploy Arrakis PALM to conduct market-making on UniV3 for ANGLE.

Introduction to Arrakis Finance and Arrakis PALM

Arrakis Finance is Web3’s trustless market-making infrastructure protocol that enables running sophisticated algorithmic strategies on Uniswap V3.

Since launch, Arrakis has achieved

  • $1.7b in TVL at its peak (currently around $200m), across Ethereum, Optimism, Arbitrum and Polygon
  • Over 25% Uniswap V3 total TVL
  • More than 80 projects have their liquidity managed via Arrakis vaults

Arrakis PALM - Protocol Automated Liquidity Management is a novel liquidity bootstrapping mechanism that taps into the organic trading volume on UniV3. It is the first product built on top of the Arrakis infrastructure.

In essence, PALM helps protocols bootstrap their base asset inventory (ETH, USDC, etc.) and attain deep and sustainable liquidity. The major advantages of using PALM are:

  • Zero incentive: no LM incentive needed, liquidity bootstrapping is done solely via market making.
  • Low requirement in base asset: the initial liquidity can be made of mostly ANGLE, and PALM will progressively pull it towards 50% to create an equal buy/sell support.
  • High capital efficiency: by autonomously and actively managing concentrated liquidity, especially once a sufficient amount of base asset has been bootstrapped, PALM can further reduce the slippage for large trades even if with limited overall liquidity.
  • No biased price impact: no swaps involved, PALM conducts market making by setting up ranges / limit orders…
  • Minimize or outperform IL: PALM is able to keep the value of the deposit close to that of holding the initial assets, and often outperform it.
  • Trustless & transparency: Angle protocol retains the custody of the liquidity and can withdraw at all times. PALM only autonomously manages the liquidity but can never remove it. All executions of PALM can be monitored on-chain with full transparency.

Background & Motivation

Currently, most of ANGLE on-chain liquidity is on SushiSwap, which is incentivized with ANGLE and takes most of its DEX volume.

However, a mere $10k swap in the Sushi pool that has $667k liquidity can already cause a price impact of about 3.5%, which reflects the low capital efficiency of the underlying x*y=k type AMM.

In addition, the average volume in this pool over the past 30 days (or even further back) has been below $10k, with $22k being the highest daily volume. This means that Angle is over-incentivizing an underutilized pool, and the low volume may also be due to the low capital efficiency that drives the trades to off-chain venues.

With that in mind, in conjunction with some preliminary discussions between Arrakis and Angle teams, we believe that Angle community can greatly benefit from deploying a certain amount of POL into PALM on UniV3, to build up deep and sustainable ANGLE liquidity that is far more capital efficient and cost effective than how it is currently.

Once PALM proves to be able to take in the majority of the volume, Angle community will have the option to wind down and eventually stop the incentives for LPs on Sushi. Furthermore, other incentivized pairs can migrate to PALM to enjoy the same benefits.

Therefore, we, as Arrakis, propose to provide Angle with the full spectrum of market-making services on UniV3 with PALM, to reach the objectives mentioned above.


Phase 1 - Bootstrap ETH
Angle Treasury allocates $300k worth of liquidity into a vault managed by PALM, in a ratio of 80/20 in ANGLE/WETH. PALM uses this allocation to bootstrap WETH over time, with an initial target ratio of 50/50.

Phase 2 - Deepen the market further
Once the ratio of 50/50 is reached, the focus will be on further increasing the liquidity depth for ANGLE, to minimize and equalize the price impact on both buy and sell side.

During the time of deployment, the Angle community has full transparency of the execution and performance of PALM via a custom dashboard, as well as retains full custody of the liquidity in the vault, meaning that at any point in time, the Angle community can withdraw the fund from the vault or revoke the managing access from PALM. PALM can only conduct market making with the liquidity deposited in the vault, and will never be able to remove the fund.

For the services provided, Arrakis charges fees on two fronts:

  • Management fee: 1% AUM fee on a yearly basis
  • Performance fee: 50% of trading fees generated


For more information regarding Arrakis and Arrakis PALM, feel free to have a look at our docs and join our community. I’m also more than happy to respond to any comments here from the Angle community towards this proposal!




This PALM idea is quite interesting.

I have some questions about the proposal:

  1. Do we have examples of DAOs who used PALM already?
  2. Phase 1: Why not 100% angle (and 0% weth) to start from?
  3. Phase 1: how much time do we forsee for the phase ? (days? months?)
  4. Phase 1: why $300K? If some investors want to buy Angle, they may “eat” that $300K quickly if it is are current “discount” (angle = 0.03$)

Thank you in advance for answers.


All valid questions!

Currently there are about 30 projects using PALM, including Stargate, Kwenta, Pika, Across, etc. just to name a few. Here you can have a look at our case study for Kwenta about a comparison between PALM and LM via bribe, with real data analysis.

In order to make markets, you need assets to support both bids and asks. If we only had 100% in ANGL, it would mean that we could not capture any volume on the sell side. In addition, since most likely, PALM will be the dominant LP in Uni pool given the current state of the ANGL liquidity, if there were no base asset for PALM to deploy, then the price impact for selling ANGL would be huge.

This entirely depends on the volume and ANGL. The more volume it has, the faster we can complete phase 1. Also look at it this way, if there is no volume, then the protocol doesn’t actually need much liquidity. LPing should be a reactive process, meaning that you only deploy as much capital as the market demands. Once the volume catches up, things can go fast.

PALM does NOT deploy a token into UniV3 position below its current price.
PALM does NOT deploy its liquidity all at once.
This means that swappers will not get a discount, and if they want to accumulate the token, PALM will actually deploy ANGL progressively on the way up so that ANGL will be in a price discovery state.

Hi barbarossa.
Thanks for the proposal. Besides the ticker of ANGLE token has not been proofread during the proposal, I have some concerns about Arrakis’s PALM.
First at the bootstrap phase, there is a criticism about KWENTA case study when PALM has sold KWENTA to shift from $25K ETH to $400K ETH. (Source: Analysis: Velodrome Outperforms Arrakis PALM | by 0x_DanW | Jun, 2023 | Medium)

Second, PALM has effectively built a large sell wall for KWENTA by adding outsized liquidity right above Kwenta’s current active price range, suppressing any upward price action. In result, KWENTA has removed most of their PoL from Arrakis following their KIP-93 (Kwenta State Log)

In the context that ANGLE is our main incentive token, damping the ANGLE price to build PoL and suppressing any upward price action will damage greatly the liquidity depth of agEUR pools across the chains which will eventually hurt protocol revenue and veANGLE lockers’ benefit.

I hope could receive convincing responses.

1 Like

Hey peter,

Thanks for the questions, and really sorry about the wrong ticker in the title!

There are many false accusations in this article from Velodrome (not entirely their fault since they didn’t take time to understand how PALM works at the time), but I won’t go into all that. For this specific aspect. there are several points worth discussing:

  • The job of a market maker is to provide assets to the market so that people can buy or sell. The reason it’s called market “maker” instead of “seller” or “buyer”, is that it doesn’t directly buy or sell the assets, rather, it sets limit orders (aka “asks” and “bids”) and wait for any market participants (aka. “takers”) to fill that order if they are fine with the price. If takers are not happy with the price, then the limit orders stay unfilled, nothing changes. Another way to think about it, An AMM is just one type of market maker, and we wouldn’t call LPs the market sellers or buyers.

    “Selling” on the other hand would imply that we made swaps directly from the pool, which would have caused huge price impact and pool imbalance. So there is a misconception made in the article.

  • Another thing is that PALM makes market on both sides, which isn’t mentioned at all in the article. It is true that at the beginning, PALM had to deploy more liquidity on the KWENTA side, which is simply because Kwenta team provide WETH/KWENTA in 2/98, a highly skewed ratio. So there was just not a whole lot of WETH available to make the market. Bootstrapping is to pull that ratio to 50/50 first.

Same as explained previously, “outsized liquidity” is because we had mostly KWENTA. If the team had provided us with around 50/50 ratio, there would have been more or less equal support on both side right from the start.

In addition, the “large sell wall” “suppressing any upward price action” statement is both false and senseless.

It’s false because the price clearly went up during the bootstrapping phase of KWENTA. Otherwise, We wouldn’t have bootstrapped. PALM attempts to supply as much as the market demands, as opposed to having $10m idle liquidity from LM farmers but only for $50k volume, while all liquidity is under risk of IL. If the market eats up the KWENTA we deployed, it means there is more demand, then the price is pushed up and PALM may have to deploy more KWENTA to fulfil that demand.

It’s senseless because a market maker’s job is to make deep liquidity so that traders are willing to trade without having to endure too much price impact, and it is definitely NOT to pump or dump a token’s price. If pumping price is the goal, the project can simply remove all the liquidity but only some dust in the pool, a typical tactic memecoins are using.

This is false as well. This KIP has not passed by Kwenta DAO, and Kwenta removed their POL from PALM recently for Treasury rebalance and has already redeposited into a new vault (BeaconProxy | Address 0xbc97c6d1df0b1edaf7fe9b7d4589790fe14ec9f9 | Optimistic Etherscan)

So by now I hope it is clear that PALM does NOT “dampen” price as our competitors claimed. More importantly, having deep liquidity for ANGLE should give users more confidence when receiving it as incentives since they know that there is a liquid market for the token when they want to exit, than holding onto it becomes a less of a concern.

Thank you for your response. Truly appreciate the time and effort you’ve dedicated to this.

I’d like to address a couple of points regarding PALM’s functioning, as described in your documentation.

Firstly the project docs wrote, “projects can seed the initial liquidity in any ratio, e.g. 95:5 and PALM will progressively pull it towards 50:50”. This clearly implies that PALM will establish a significant buy wall just above the current token price. Consequently, any buying effort would be eaten up immediately and taken out into inventory to be left in the same liquidity state at the beginning (95:5). Any selling activity after that would continue to result in huge slippage due to the super thin liquidity available which also suppresses further farming effort of liquidity providers on agEUR pools.

Secondly, the crucial issue here is that PALM operates under a rather weighty assumption – that it is viable only if the token price goes up. This assumption forms the basis for PALM’s strategy of accumulating base assets to create a 50:50 pool by continue taking out buying asset (base asset) into inventory. In reality, this assumption has proven to be flawed, primarily due to the token emission schedule for pool incentives. As a result, PALM’s implementation will contribute to a continuous decline in the ANGLE price

Furthermore, you mentioned that KIP-93 has not been approved by Kwenta DAO, but you haven’t provided a DAO Snapshot as evidence. In one of your case studies, the Kwenta PoL pool on Arrakis achieved a $1 million TVL during a market upswing and price appreciation. However, the pool has now dwindled to $111K and their liquidity moved to Velodrome which contributes most trading volume on Optimism. This example serves as an indicator that PALM’s effectiveness for long-term liquidity governance is questionable.

I want to emphasize that I’m not involved with both KWENTA or Velodrome and this is ultimately ANGLE DAO’s decision, but I felt it was important to present real-world instances that highlight PALM’s limitations in sustaining effective ANGLE liquidity at least in bootstrapping phase.

Thank you again for your attention to these matters.

1 Like

“significant” is somewhat subjective, though what is objectively true is that PALM only deploys a portion of the deposited liquidity, e.g. 10%, depending on the volume demand, and the so-called “wall” is a portion of that 10%. So from our perspective, it cannot be “significant”, unless the total liquidity Angle deposits is millions worth.

Secondly, 95:5, or any ratio we are referring to, is the ratio between two assets of the total vault deposit, while it can be, and normally is different when we talk about the ratio between the two assets of the LP amount, which is the “10%” I mentioned previously. This ratio can be entirely different than 95:5, e.g. it can be 50:50, meaning that if the overall ratio is 95:5, then PALM just deployed the entire amount of the smaller asset.

The slippage decreases progressively as PALM bootstraps more WETH. It’s inevitable not to have bigger slippage at the beginning, because PALM only has 5% in WETH to LP with. With more WETH being bootstrapped, more of it will be deployed, hence the slippage is getting smaller and smaller. The is also we see in many of the vaults that went through bootstrapping phase.

This is probably the biggest misconception most people have about PALM, or market making in general. The beauty of PALM is that it can bootstrap even if the price goes down in general. What makes it possible is not the price trend, but VOLATILITY. The example in the doc demonstrates the case where PALM pockets a gain of WETH after a consecutive buys and sells, as long as the net outcome is positive. What this translates to is the local peaks while the price still goes down. To put it simply, PALM can “sell” the token at local highs as the price in general goes south. Sure, we can only bootstrap when price at least momentarily increases, but denying the validity of this assumption is effectively admitting that the token price will go down linearly, i.e. zero buys.

As far as I am aware of, this proposal was never put up for snapshot (btw, we were involved in the discussion of this proposal as well).

Furthermore, the Uni pools you are looking at are not managed by PALM. Kwenta just removed their liquidity from PALM for Treasury rebalance, and deposited some liquidity into a new PALM vault (the link I posted in my previous response), and it hasn’t even started LPing yet, because the setup is not complete.

Another thing I want to emphasize is that moving liquidity around to different protocols is a governance decision that can be due to many reasons, and high capital efficiency or low slippage may not be one. So I would really just look at the hard data if the focus is on improving liquidity situation.

Thanks for being critical! :handshake:

How PALM could bootstrap under such circumstances particularly when there are minimal buying activities to absorb based assets (ETH) within the context of scheduled liquidity mining? This approach appears to exacerbate the decline in ANGLE price.
As I mentioned, by deploying too skewed pool of 90%/10% or 80%/20% regardless of any pool size above the current ANGLE price, PALM effectively builds a buying wall where selling is impossible with reasonable slippage and buying has no effect positively effect to ANGLE price.
My primary concern also lies in the bootstrap phase, as PALM’s strategy to establish a 50:50 ratio from scratch displays a distinct flaw that put on heavy assumptions on price moves. I propose an alternative solution:
Halt PALM implementation and instead directly deploy a 50:50 liquidity pool from the beginning. This would allow Arrakis to effectively manage and rebalance the pool, providing support to both selling and buying activities, ultimately benefiting the Angle liquidity mining program.
With $300k liquidity in mind, Angle Treasury will establish the 50:50 ANGLE/WETH pool at reasonable ranges regarding of current ANGLE price. The fee for managing and rebalancing will be allocated following the current Arrakis fee implementation without 1% AUM fee yearly as well as 50% trading fee which will save a lot of liquidity cost for the protocol yearly. All trading fees will belong to liquidity providers (Angle PoL) to deepen further the pool.

I think your assumption here is that people may sell their incentives after having received them, and PALM is not going to be able to handle all that sell pressure with little amount of base asset.

A couple of things here:

  • If people keep selling their incentives while there is no sign of buying pressure, it may be time to rethink whether such LM scheme makes sense anymore

  • PALM will not be the only liquidity source for ANGLE. The current Sushi pools will run in parallel with PALM. The objective here is to first bootstrap WETH until PALM can single-handed manage all the buys and sells of ANGLE. During bootstrapping, it is expected that most sell pressure will be probably handled by the other pools, while PALM will take in the buys because of the lower slippage.

I guess what you meant is “sell wall” as in deploying ANGLE to the market. So same as what I explained before, at the beginning selling will mainly be handled by the other pools. But I don’t understand your statement of “buying has no effect positively effect to ANGLE price”. Once the limit order, or sell wall is being filled, it means that the price has gone up, then PALM sets new limit orders alongside the higher price, and then the process repeats. If the argument is that the sell wall can somewhat suppress the price, also as I explained previously, the easiest way to pump the price is to only have dust liquidity in the pool, which also means huge slippage and eventually it ends up a vicious circle where no serious size will ever trade it.

PALM does not have to bootstrap if a project deposits directly in 50/50, in which case PALM just directly conducts market making for deep liquidity. Bootstrapping is not a requirement, but a capability offered by PALM if a project doesn’t have or doesn’t want to spend much base asset on liquidity. If Angle protocol would like to allocate that much base asset, we have absolutely no problem with that and it only makes the job easier. However, the premier of this proposal is that Angle protocol would like to use PALM to bootstrap WETH instead of providing it in 50% right at the beginning. This way, the protocol can spend its base asset on developing its core product and have a healthy runway.

Having said that, if the consensus among Angle DAO is that the protocol should allocate WETH in 50% to PALM, then we are fine with that as well.

You quote in the wrong direction intentionally. I understand your purpose here. But as I clearly stated that Angle should not use PALM to bootstrap ANGLE in any ratio but a balanced Arrakis V2 pool with more active rebalancing - it’s simple, effective, and free (Fees - Arrakis). It required more capital commit needed from Angle treasury at first, but in the long term, it will benefit ANGLE liquidity by effectively deepening the pool with all trading fees. 1% fee yearly management and 50% trading fee is not worth for PALM that builds an unbalanced pool as buying limit order right above the current active price range, effectively suppressing any upward price action. That will frustrate agEUR liquidity providers who are participating Angle liquidity mining program which eventually hurt protocol revenue and agEUR scalability and veANGLE lockers who put trust in protocol growth. These issues should be really considered.

This is totally senseless. We’re talking about building ANGLE liquidity here not to criticize our liquidity mining for our main product -agEUR. Creating Protocol Own Liquidity doesn’t imply it’s cost-free, but rather it continually increases expenses for ANGLE sellers. Furthermore, the effectiveness of liquidity mining hasn’t ceased to have a positive impact in the space. This is an entirely different context.
I’d like to emphasize that the Angle Merkl solution is currently conducting a collaborative liquidity mining effort involving both Uniswap Foundation and Gaultnet. The outcomes of this endeavor look very promising. You can find more information in this tweet.

I misunderstood what you meant.

  • Arrakis V2 is the infrastructure for users to LP themselves in any creative way that Uni doesn’t offer natively.

  • PALM is a market making (or LPing if you will) strategy R&D’d by Arrakis inhouse and built on Arrakis V2. Its main purpose is to create deep liquidity in the most cost effective and capital efficient way. Bootstrapping is an additional capability if the project wants it, which is the case here according to Angle team.

If you want to use Arrakis V2 only, then you will have to come up with your own market making strategy for deep liquidity and it would have nothing to do with any marketing making service Arrakis provides.

And I will refrain from commenting on the price suppression argument as I have already explained enough. This is something I think you can easily verify based on evidence, e.g. KWENTA. If there were price suppression, we wouldn’t have bootstrapped.

On the Angle LP argument, that’s something Angle DAO needs to think about. You seem to think of it as a binary situation, i.e. price pumping facilitated by thinner liquidity vs. price stabilizes because of active market making. All I can say is that the former applies to the other direction as well, while the latter is exactly what market makers are supposed to do.

Absolutely no intention to throw shade at the approach Angle is taking on LM. All I said was if the community is so worried about LM recipients dumping ANGLE, which is the argument you are using regarding selling ANGLE, then it makes sense to rethink how to mitigate this fear.

And I don’t want to mix the liquidity situation for ANGLE with the ANGLE incentive situation for agEUR. All we want is to make deep market for ANGLE that helps both buyers and sellers. How Angle DAO internally handle it’s LM for other products or purposes is not the scope of this proposal.

I think we are running circles here, and kind of spamming the post a bit so that it may become unreadable for other community members. If you still have any question, please dm me directly and I’ll send you my Calendly so that we can hop on a call to discuss directly, and ideally with Angle team members as well if that’s helpful :handshake:

Hey Barborossa. Thanks for suggesting. Honestly, you’ve been answering all my concerns but it looks like not convinced enough to me and taking a call probably won’t change anything if you could write here in a more transparent manner in front of public eyes. The thing is PALM design has a clear flaw that has put heavy assumptions on token price. Base asset, in this case, WETH doesn’t come from thin air in order to build 50:50 from 80:20 without veANGLE locker expense, and ANGLE sellers. Then Angle should retain to choose PALM solution but consider other solutions like taking advantage of ve(3,3) Dexs or co-bribing to build a more sustainable and healthy liquidity solution for ANGLE in the long term.

I’m all for transparency. If you don’t feel comfortable to get on a call with me and potentially with ANGLE team members as well, I’m also fine with having a community call to discuss with you in the open, point by point. Having this long format of writing is not going to resolve any problem as its gets longer and longer and no one else can follow.

And yes, base asset doesn’t come from thin air. It comes from buyers. I don’t get the argument of “veANGLE locker expense”. As to “ANGLE sellers”, I also explained that the existing pools will stay in parallel as they are for the time being. So I don’t get your concern there either.

You seem to prefer bribe or LM approach, fair enough. The Kwenta case study has demonstrated some comparisons and I believe we at least made our case on that.

Anyways, if you are up for it, I’m looking forward to a verbal discussion with you in front of Angle community. For now, I don’t think more texts would be any productive. Thanks for the engagement :+1:

veANGLE lockers keep part of ANGLE out of circulation to maintain its liquidity available. If PALM effectively builds a wall that suppresses any upward price movement at the bootstrap phase, it implies that they will experience dilution through emissions from the Angle liquidity mining program. Consequently, they will not have the opportunity to achieve a point of breakeven in terms of their initial locking during the early stages of the protocol, as the ANGLE price upward has been suppressed.
In the scenario where PALM successfully establishes a 50/50 pool, with the consideration of a 50% trading fee (not accounting for the 1% management fee) might appear inefficient. This fee could instead be utilized to increase the depth of the pool similar to normal UniV3 pools, requiring minimal active management from Arrakis.
Anyway, this proposal will go through voting and the clarification about PALM mechanism might be considered regarding Angle stakeholder’s benefit.

I’m eating my own words right now, lol. But it’s too tempting.

You keep using the term “suppress”. So follow along your logic, which is that PALM builds a wall that is bad for the price, then let say the base “wall” is the default Uni V2 style AMM, and PALM somehow makes a taller wall during the bootstrapping phase, would you be happy if PALM makes a shorter wall which is definitely less suppressing to the price than the Uni V2 AMM style base wall, e.g. Sushi? I’m genuinely curious how you define the significance of the wall.

PALM will build a wall against ANGLE price moving up because it set up an unbalance liquidity distribution (eg 80/20 or 90/10) which acts as an effective buy limit order right above the current price. This order absorbs all based assets (eg WETH) from buying orders from the market to take into inventory which means PALM is depreciating ANGLE liquidity for selling activities. And buying effort from the market doesn’t appreciate the ANGLE price.
If the price of ANGLE keeps going lower due to incentive program emission, PALM never goes through its bootstrap phase because there is no base asset from buyers.
Finally, the reason I emphasize greatly ANGLE liquidity building plan is important to protocol success and growth since ANGLE is our main incentive token and interest distribution tool for all Angle stakeholders including agEUR liquidity providers, veANGLE lockers, and investors.
I’m sure it’s great to save some bucks by committing less capital to PALM at first but Angle protocol eventually will have to pay PALM in other ways in the bootstrap phase since based asset for building a 50/50 pool doesn’t come from thin air as I mentioned above, but from protocol revenue which indicates of the depression of agEUR market cap due to inefficiency of the incentive program, veANGLE lockers, and any potential buyback attempt from Angle. Moreover, Angle protocol will waste 50% trading fee and 1% of the management fee for an inefficient market-making solution, which could be used to deepen the ANGLE liquidity or act as protocol revenue. PALM might look fancy at first, but it doesn’t mean it’s worked for building liquidity for the long term. That’s why KWENTA has abandoned this solution.
I just want to emphasize again that ultimately it’s Angle DAO’s decision and I respect that. All my effort to promote the balance of all stakeholders benefit and nurture protocol growth for the long run as an Angle contributor. This is my last comment on this proposal.