In the last DeFi summer, we have experienced many DEX aggregators like 1inch, Matcha, ParaSwap, and etc., and they played an important role in our daily trading life for us to get a better price and less fee.
By the way, I always have an opinion that the value ceiling of tool type crypto assets is lower than infrastructure type crypto assets, because the value of tool type crypto assets relies on the evolution ability to prove the irreplaceability. It’s difficult. Also, they are mostly built on infrastructure type crypto assets, like trading aggregator to DEX, interest rate protocol to money market.
Finding a product which can grow up to an infrastructure is really difficult and more like to fail, thus researching on innovations of tool type crypto assets also has a lot of fun.
One year passed, the DEX aggregators evolve to a more powerful and comprehensive stage to figure out the MEV (Miner Extractable Value), gas and slippage problems addressing in the article Ethereum is a Dark Forest.
Elaboration by Real Trading Examples
As written in the Blog, ‘Archer Swap gives traders the absolute best possible price for large swaps on Uniswap and SushiSwap, without having to worry about front-running bots’. The realization of these features relies on Archer Relayer. Archer Relayer can cooperate with miners, search for the most valuable transaction for them, and let them submit the transaction to the Ethereum mainnet. Also, it can benefit traders, they just need to pay some tips to miners and Archer Relayer will help them bypass the public mempool to settle the transaction.
I spent 0.2 WETH swap to 1.418 AAVE. The procedure is, 0.2 WETH was firstly transferred to Archer Router Address, from which 0.0639 ETH was transferred to ArcherSwap TipJar contract and tipped to the Miner (UU pool). Then the Archer Router Address is routed to Uniswap. Thus I can bypass the public pool to accomplish this transaction, 0 gas fee but really high miner tips (automatically set).
On ArcherSwap UI, we can also choose to turn on/off the manually tip set, and input tip amount by yourself. But if the tip is not enough, the miner won’t be willing to package your transaction.
Also, the design that ETH must be sent to the ArcherSwap contract first may lead to some centralized problems. Whether the transaction succeeds or not highly depends on the quality of the contract. My friend Blanker has already found some bugs in their contracts as described in his Twitter which cause some untransferable tokens left in their contract.
CowSwap is backed by Gnosis Protocol V2 (GPv2) which is developed by Gnosis team and provides MEV protection. GPv2 optimizes for coincidence of wants (CoWs), which can be explained as “an economic phenomenon where two parties each hold an item the other wants, so they exchange these items directly.”
That is to say, CowSwap matches orders for traders off-chain first, and if it doesn’t find another demand transaction, it will submit the transaction on-chain to other DEXs.
CowSwap brings in ‘Solver’ conception to realize this function. Solvers are encouraged to compete against each other to deliver the best order settlement for traders in exchange for the reward of each batch. Users submit trade orders with a degree of flexibility as solvers need to find the most optimal way to settle them.
And since transactions can be settled off-chain, CowSwap doesn’t need external liquidity on-chain thus decreasing the trading cost. CowSwap will use a united price to settle all orders in the same batch, which is called batch auction mechanism.
Before confirming swap, you need to sign a message which contains sell/buy type, amount, expiry and other info, to allow CowSwap to take down your order off-chain. Then ‘Solvers’ begin to look for whether there is a CoWs for you or not.
The process is very like Ethereum meta-transaction, which was proposed by Austin in 2018. Gnosis team, who dives deep in Ethereum research, also launched a related relay service product and now applied it to CowSwap. In general, after you sign a message with your account, it will link to Gnosis relayer offchain, then relayer fowards the execution onchain.
You will find your transaction is recorded in the Gnosis Protocol Explorer. And click on the Transaction Hash, you can actually see the transaction details.
CoWs is found in this transaction, which is dealed off-chain and not routed to Uniswap. So, you won’t find this transaction in your address, it’s their contract who actually settles the transaction.
CoWs is not found in this transaction, Gnosis routed this transaction to Uniswap V2.
MistX is a project developed by Alchemist team, which works very similar like ArcherSwap. Also, a project named BackRunMe developed by bloXroute team runs the similar way. ArcherSwap is FlashBots compatible and MistX utilizes FlashBots directly, while BackRunMe is backed by bloXroute. The FlashBots, bloXroute and other similar tools are all utilized to submit transactions as a searcher to ETH mainnet.
To be honest, the UI of MistX looks like a total copy of ArcherSwap. But the logic of MistX exchange route improves a lot compared with that of ArcherSwap. Because it skips the first step that ETH must be sent to the ArcherSwap contract, which may lead to some centralized problems.
Although MistX and ArcherSwap can both automatically adjust the miner tips, the MistX performs better and tips more intelligently.
I spent 0.2 WETH swap to 1.43 AAVE. The procedure is, 0.00516 WETH was tipped to miner Address(Ethermine), 0.000271WETH was tipped to MistX address. Then the MistX Router Address is routed to Uniswap V2. Thus MistX publish transactions exclusively as a bundle in Flashbot’s private mempool bypassing the public mempool.
Comparison of the Features
I draw a sheet to take down the core features of the MEV protection DEXs for readers to better understand their similarities and differences.
Fee/Revenue Structure: Gasless?
The fee structure of these products is quite obscure. Maybe why they don’t want to make it clear is because Degens just care about the total trading efficiency.
But as I tried every product, I can give two general conclusions and some detailed explanations:
1. These projects generate revenues mostly from your trading capitals or miner tips.
2. There ain’t no such thing as a free lunch. Gasless is because the gas is compensated in other ways.
ArcherSwap, they don’t charge gas fee because it’s included in the miner tips. Miner tips can be adjusted but usually not friendly to micro-transactions. ArcherSwap is able to extract some fees as revenue from the miner tips but it looks like they don’t. Traders also need to pay Uniswap/SushiSwap trading fees.
CowSwap, claimed not to charge protocol fee in the testing period, now seems charge fee around 1–1.5% of the trading amount.
Again, compare the two transactions as an example.
If GPv2 finds CoWs, you just need to pay protocol fee, that is 0.0159aave. But if GPv2 does not find CoWs, I guess users need to pay Uniswap fee and protocol fee both, total is 0.005weth. That’s why fee accountings are different in the two transactions. (0.0159aave/1.4636aave=1.08%; 0.005weth/0.2weth=2.5% ).
Also, what is interesting is that in the first transaction, CowSwap pays 0.00498 eth as gas for you , so the net revenue is 0.0159aave-0.00498eth≈ -0.0028eth . And in the second transaction, CowSwap pays 0.02eth eth as gas for you, so the net revenue 0.005weth-0.02eth≈ -0.015 eth. CowSwap lost money in both transactions! It seems micro-transactions on CowSwap hurt both CowSwap and traders.
The fee/revenue structure of CowSwap is the most obscure because it can be settled off-chain. After my article published, CowSwap told me on Twitter that my above estimation was wrong.
As they written in the Twitter, I concluded in these elements:
1. Trading Fee: if a CoW happens, it’s settled off-chain, so no trading fee; if a CoW doesn’t happen, trading fee follows the DEX they route. The trading fee will be priced in in the price estimation that is shown to you in the UI.
2. Gas Fee: The Solver will estimate what will be the gas cost to settle your trade on chain and then charges you 40% of that, the 60% is subsidized by CowSwap.
3. Protocol Fee: Don’t charge now.
But I think relying on the estimation gas fee on-chain to decide how much should users pay is totally not transparent and not decentralized. I would rather they charge protocol fee directly.
MistX actually shares the tips with miners which can be seen in the transaction details. It charges around 5% of the total tips regardless of trading amount. (0.0002717 ETH/ 0.005435 ETH=5% in the above example). Traders also need to pay Uniswap/SushiSwap trading fees.
We must respect the innovation of these projects which let us see more possibilities in handling with MEV protection problems, and some large amount trading do have demands for such projects.
But all such projects are in the early stage and have a lot of things to upgrade. Sometimes, we normal traders need to care about the ‘invisible consumption’ when trading in these aggregators. And to be honest, large DEXs and low slippage set might be more efficient when the gas price is low.
Also, the tools like flashbots, bloXroute behind these products are those breathtaking infrastructures helping us safely pass through Ethereum dark forest, which have a vast application scenario in many DeFi products.