In December, Kwenta launched their token and started a Liquidity Mining (LM)/Bribing program on Velodrome Finance, one of the leading Automated Market Makers (AMMs) on Optimism. In late February, in addition to the LM on Velodrome, Kwenta decided to utilize Arrakis Finance's newest product: PALM, which actively market-makes liquidity on top of Uniswap V3.
In this article, we analyze several metrics on both DEXs to provide an objective and comprehensive comparison. We will explore the benefits and drawbacks of each alternative and evaluate their impact on Kwenta's token liquidity.
Kwenta spent 1,400 KWENTA tokens (around $523k) in bribes to incentivize 3rd-party liquidity in Velodrome. As a result of these bribes the pool at the time of writing has $2.43M of liquidity.
Kwenta committed 1,300 KWENTA and 16 ETH (around $500k) of Protocol-Owned Liquidity (POL) to Arrakis PALM, paying a 1% yearly management fee of 13 KWENTA and 0.16 ETH ($5k).
Despite Velodrome having 5.5x more TVL in its pool, both pools have offered a similar price impact on a $10k trade.
As Velodrome is a 0.05% fee pool, and Arrakis PALM is on the 1% fee tier on UniV3, Velodrome has facilitated more Volume.
By Arrakis PALM being on the 1% fee tier Kwenta has been able to generate enough trading fees (80 KWENTA and 25 ETH) to compensate for its active management costs, and even result in a $33k profit (at current valuations) since inception.
Arrakis has managed to beat impermanent loss (IL), and its vault is currently outperforming a hold strategy by 17%.
Velodrome is a Solidly fork that uses the constant product invariant (x*y=k) and is one of the most successful DEXs by volume on Optimism for volatile pairs. On top of that, the protocol uses a tokenomics system that is designed to attract liquidity into their markets by using built-in incentives (also known as bribes).
Arrakis is an autonomous and trustless market maker built on top of Uniswap V3. Arrakis helps protocols bootstrap deep liquidity by actively managing concentrated positions on Uniswap V3, similar to how market makers use central limit order books. Protocols can deposit any asset compositions into a PALM vault (e.g Kwenta deposited 98% in Kwenta and 2% in ETH), PALM then deploys capital into the market on both the buy and sell side, by leveraging the volatility of the market, PALM accumulates base assets first (bootstrapping phase) in order to reach an asset composition of 50/50. Then it actively makes the market with equal bid and sell orders.
PALM's bootstrapping mechanism enables protocols to provide liquidity in an unbalanced manner. In many cases, protocols may not have a surplus of base assets available for liquidity provision. PALM addresses this issue by placing LP positions close to the price and ejecting them to accumulate the base asset when the price moves. These price movements can be either price trends or mean reversions. The resulting unbalanced asset composition causes the protocol to have larger price impacts during the period of imbalance until the asset composition changes. For Kwenta, the bootstrapping phase lasted approximately 4 weeks. This case study is not including the bootstrapping phase as it is not comparative.
As per several community-approved proposals (KIP-46, KIP-49, KIP-50, and KIP-64), at the time of writing, Kwenta spent a total of 1,400 KWENTA tokens over the course of 25 weeks in Velodrome bribes. At the time of issuance, these bribes translate to a total of $523,185 spent in about 6 months.
Thanks to such a considerable amount of incentives (plus the extra rewards issued by Velodrome itself), Kwenta has been able to rent a historical maximum of roughly $4.4M of liquidity and about $2.43M at the time of writing. Therefore, Kwenta is at the time of writing renting $4.6 dollars of external liquidity per dollar spent on bribes, or paying 22 cents per dollar of liquidity rented.
In late February Kwenta decided to build a Protocol-Owned Liquidity position by using Arrakis PALM, depositing 16 ETH and 1,300 KWENTA. Kwenta pays a management fee of 1% of the Total Value Locked (TVL) as well as forfeiting 50% of the generated Uniswap trading fees.
Since the Arrakis vault had only been live for roughly 3 months and the management fees are paid quarterly, it is not possible to know the yearly amounts in advance. For the sake of this article, we present 2 different calculations. Still, readers can come up with their own conclusions by extrapolating their price expectations for ETH and KWENTA:
Fees based on the TVL at initialization: Kwenta initially committed 1300 KWENTA and 16 ETH, which in total is worth around $510k. Assuming that the USD value of the vault stays flat during the whole year, Kwenta would pay 12.9 KWENTA tokens and 0.16 ETH, which were worth $5,100.
Fees based on the peak TVL: After several months, due to the price of KWENTA appreciating and the ETH acquisition, the TVL increased up to $1M USD. Again, taking this value as a yearly reference, the management fees would be 8 KWENTA and 1.9 ETH, which were worth around $10,100.
As readers may have noticed, the management fees are orders of magnitude lower than bribes.
With these numbers in mind, we can already get a pretty accurate idea of the magnitude of the management fees regardless of further price appreciation. Therefore, by calculating the liquidity / yearly fees, we estimate that Kwenta will be paying $1 per year for 7$ to 14$ of managed liquidity, or 14 to 7 cents for each dollar of actively managed liquidity per year.
As with any constant product AMM, Velodrome’s execution price is determined by the total liquidity in the pool. Therefore, the more liquidity, the better execution users will get, and vice versa.
As we can see in the chart below, thanks to the substantial liquidity depth that the incentives managed to bootstrap, Kwenta users can perform decently sized swaps without incurring huge slippage.
Unlike the classic constant product AMM, Uniswap V3 is more sophisticated and closer to a CLOB model thanks to its concentrated liquidity feature. Due to this, skilled LPs are able to provide liquidity in concentrated ranges and increase the capital efficiency of their assets when compared to a constant product AMM, which is equivalent to a range that covers from 0 to infinity.
In our previous discussion on liquidity and costs, we observed that the Velodrome pool has nearly 6 times more liquidity than the Uniswap pool. However, thanks to PALM's effective liquidity management, both pools offer similar price impacts for $10k trades. In fact, the Uniswap pool even performs better in certain time periods.
To provide further context to the chart, an important consideration is that managing liquidity positions requires active management as the price moves through the ranges. As shown in the chart, there are occasional peaks in price impact. This behavior occurs because as the price approaches the end of an LP range, the price impact worsens. However, it's important to note that these peaks are short-lived since the ranges are automatically rebalanced by Arrakis PALM.
Finally, to further demonstrate the capital efficiency of PALM, we have multiplied the price impact by the TVL, through this normalizing the price impact chart based on the TVL of each pool. This normalization allows us to observe a significant insight: PALM utilizes its liquidity 10 times more efficiently compared to a constant product AMM such as Velodrome.
Taking into account that price impact is quite similar in both pools for trades of up to $10k, one would expect similar volumes in both pools. Nevertheless, the reality is far from that. As we can see in the following charts, Velodrome is consistently getting 5 to 7 times more volume than Uniswap, which is similar to the TVL imbalance between both markets.
This behaviour is most likely caused by the lower trading fees on Velodrome, i.e. 0.05% on Velodrome vs. 1% on Uniswap, which means that despite both markets offering similar price impact, the execution price when adjusting fees is consistently better on Velodrome.
When analyzing the size of the trades, we can see that the results seem to match the previous hypothesis, since the share of trades that go through Uniswap increases for bigger sizes.Looking at the volume traded per dollar of liquidity in each of the markets, we can measure the concentration of these metrics, and use its relationship as a proxy for capital efficiency.
Comparing both markets, we can see a historical average of around 0.25 volume/dollar (0.257 for Velodrome overall and 0.21 for Uniswap). Conversely, when we analyze the subset of Uniswap volume which goes through the PALM Vault, this average increases to an astonishing value of 1.075 - 4 times greater than Velodrome overall and 5 times greater than the entire Uniswap pool. This high concentration and the fact that the vault routes around 70% of the Uniswap volume despite only accounting for 20% of the TVL is another demonstration of capital efficiency.
As previously explained, despite both pools having similar price impacts for $10k trades, Velodrome routes 5-7x more volume than Uniswap, which is explained by the fact that Velodrome’s fee (0.05%) is 20 times cheaper than Uniswap’s (1%).
If one solely focuses on volume, it is easy to miss the whole picture. Instead, the following KPIs compare the fees generated by Velodrome and the Arrakis PALM vault.
The results are quite impressive. Since the inception of the vault, PALM has generated almost 5 times more fees than Velodrome.
On top of that, it is important to remember that, unlike rented liquidity, Kwenta owns the assets in the Arrakis vault and, therefore, is able to earn half of the trading fees. Thus, at the time of writing, the POL has earned 42 KWENTA and 13 ETH, which is worth around $37.5k USD.
These numbers showcase that the management fee paid to PALM is completely offset by the trading fees earned, and the protocol is even able to book some profits.
To analyze the previously presented stats and derive any conclusions from them, it is important to understand the implications of renting liquidity via bribes, as well as those of owning the liquidity.
On one hand, despite the bribes seeming expensive, it is important to note that by renting liquidity, the protocol isn’t incurring any liquidity provisioning risks, and is externalizing all of them to the LPs. By doing so, Kwenta doesn’t need to worry about Impermanent Loss (IL) or smart contract risks.
On the other hand, the protocol will always have to rely on third-party LPs to have liquid markets. This approach most likely forces the protocol to keep issuing LP incentives indefinitely, since otherwise it may not be attractive enough for users to provide liquidity.
Finally, it is also important to remember that in the Velodrome model, LPs don’t get the bribed KWENTA token. Instead, they get $VELO emissions, which means that the KWENTA incentives go to $veVELO users who voted for the Kwenta pool. The fact that LPs (probably aligned with the protocol) don’t get the bribed tokens, makes it more likely that $veVELO holders dump those incentives.
As mentioned above, Protocol Owned Liquidity (POL) is cheaper, but it also comes with some risks.
Firstly, the protocol needs to commit its assets to the liquidity pool. This means that those assets are subject to IL. To minimize the amount of capital needed and maximize its impact, PALM actively market-makes for the underlying assets. As previously demonstrated, PALM does a great job at maximizing capital efficiency, nevertheless, it is still worth assessing how the vault performs VS. holding the naked assets in the treasury.
Secondly, there is the aspect of opportunity cost, e.g. using the treasury funds in other ways to grow the treasury (depositing into lending markets, buying LSTs, staking, etc). Protocols can use PALM in the essence of treasury diversification though, such as Kwenta which used PALM for liquidity bootstrapping and market making. The Kwenta team could only commit 16 ETH and at the time of writing has around 35 ETH. It is difficult to quantify opportunity cost, but it is clear that the Kwenta team has seen a favorable performance.
After analyzing the most crucial factors, including the cost of liquidity, the performance of liquidity, the price impact, the volume facilitated, and the fee accrual, we can conclude that the POL route is the most sustainable and best-performing liquidity mechanism for Kwenta.
Arrakis enabled Kwenta to get a similar liquidity depth for $10k trades even though Velodrome has 5.5x more TVL in its pool. At the same time, to attract Velodromes liquidity, Kwenta spent $523k in bribes, whereas in Arrakis, Kwenta earned +$38k in fees (at current valuations) on its protocol-owned liquidity which results in a $33k (at current valuations) PnL since inception.
These benefits and the proven performance, make Arrakis PALM more attractive than renting liquidity especially because the protocol is able to self-sustain with substantially less capital and avoid the negative impacts from the reliance on 3rd-party LPs and introduction of sell pressure by veVELO holders selling their rewards.
Arrakis PALM is able to compete against a liquidity pool that has 5 to 7 times more TVL and still manages to provide the same level of price impact for a $10k trade, even though the PALM vault is active on a higher fee tier (20 times higher than Velodrome). Yet, the pool gets decent volume and ends up capturing almost 5 times more fees than the Velodrome one. Again, all of this is without sacrificing profits to IL as Kwenta`s Arrakis PALM vault is currently outperforming a hodl strategy by 17%.
To finalize this case study, it is important to mention that due to the wide adoption of constant function market makers, profitable liquidity provisioning has not been historically discussed. With concentrated liquidity and active management solutions such as PALM, the industry enters a new paradigm.
All data can be monitored here: Dune
Arrakis is a protocol that builds trustless market making infrastructure & strategies on Uniswap V3.