This tool monitors all settlements that happen on-chain, focusing on EBBO testing. More concretely, we promise our users that they will get a price that is at least as good as what they can find on other DEXs. By parsing publicly available liquidity, we can monitor how often solver solutions with better surplus could have been provided.
The project consists of 3 main components:
- In the first component, for each auction, we look at all solutions that did NOT win and check whether any of them actually provided a better deal to an order that got executed by the winning solution. This currently is best used for analyzing historical data. This script utilizes the callData of all solutions.
The output consists of all orders that could have provided the user a better surplus with corresponding solver and absolute/relative values of the surplus difference. It also provides the error rate of each solver. The competition endpoint is used for this. To run this, set start_block, end_block or tx_hash in
test.py.
2. The second component is a spin-off from the first that serves as a live logging tool which gathers the most recent settlements on CoW Protocol, and checks them for EBBO.
3. The third component relies on on-chain data and the instance.json describing the auction. This component parses all settlements that happened on-chain, recovers the surplus that each order got, and for each executed order, calls Quasimodo and requests a solution that only executes that order. Then, it checks what surplus Quasimodo gives to that order and compares with what happened on-chain.
Clone Repository:
git clone https://github.com/cowprotocol/ebbo.git
Version python 3.12 is used.
Install the following libraries as dependencies.
pip install requests
pip install web3
pip install dune-client
pip install python-dotenv
To test either of the components, create a .env file in the root directory (ebbo) and set variables ETHERSCAN_KEY and INFURA_KEY as string values from Etherscan API Key in your profile and Infura once you have an account created one. You can also set NODE_URL instead of INFURA_KEY if you are using a different RPC provider. You will also need a DUNE_KEY for running the historical data test.
Sample .env structure:
INFURA_KEY = 'string key here'
ETHERSCAN_KEY = 'string key here'
DUNE_KEY = 'string key here'
To start the EBBO daemon, run the following from the ebbo directory:
python3 -m src.daemon
If you wish to run the EBBO tool over historical data, set the (start_block, end_block) or tx_hash in test.py and run the following from the ebbo directory:
python3 -m tests.e2e.test
You should be good to go now!