tl; dr
today i go over bdex, a tool i built that scrutinizes arbitrage cases on x*y=k AMMs on ethereum.
🎶 today’s mood
The following are the resources we are fetching (but feel free to add your favorite token and pool):
Token pairs:
- WETH/DAI
Exchanges:
-
Uniswap (0xa478c2975ab1ea89e8196811f51a7b7ade33eb11)
-
Sushiswap (0xc3d03e4f041fd4cd388c549ee2a29a9e5075882f)
-
Shebaswap (0x8faf958e36c6970497386118030e6297fff8d275)
-
Sakeswap (0x2ad95483ac838e2884563ad278e933fba96bc242)
-
Croswap (0x60a26d69263ef43e9a68964ba141263f19d71d51)
Let’s start by taking a look at what this tool does, and then I will share the source code with you.
Setting your environment
Add your Alchemy API key and endpoint to a file named .env:
cp .env_example .env vim .env
Create a virtual environment:
virtualenv venv source venv/bin/activate
Install dependencies:
make install_deps
Install the CLI:
make install
Running the CLI
You can run the CLI with:
bdex
Checking the latest block
We leverage Alchemy API endpoint eth_blockNumber_hex to get the latest block:
bdex -c
💡 The block number can be checked against ETHstat.
💡 We are crafting the checksum address string by hand without directly Keccak-256 hashing the methods and parameters.
Getting the token balance for an exchange
We leverage Alchemy API endpoint eth_call to retrieve the current token balance for a specific exchange:
bdex -b TOKEN EXCHANGE
Getting all token balances for all the exchanges
We loop over the previous method for a list of tokens and exchanges:
bdex -a
Getting all token balances for all exchanges, but now with web3.py
To be able to compare our results from the previous steps, we implemented an alternative way to fetch pair balances utilizing the Python web3 library:
bdex -w
💡 For this library, it's necessary to supply the contracts' ABI (in our case, for DAI and WETH, located at ./docs)
💡* A third option to verify token balances is through Etherscan tokenholdings dashboard.*
Getting trading prices for all the exchanges
To get the current price for WETH/DAI in all exchanges (e.g., as shown in the projects' dashboards), run:
bdex -p QUANTITY TOKEN1 TOKEN2
Quote for trading 1 WETH:
Quote for trading 10 WETH:
Quote for trading 100 WETH:
How the price is calculated
An AMM replaces the buy and sell orders in an order book market with a liquidity pool of two assets, both valued relative to each other.
As one asset is traded for the other, the relative prices of the two assets shift, and the new market rate for both is determined.
The constant product is:
token_a_pool_size * token_b_pool_size = constant_product
All the exchanges are forks from Uniswap V2, so they all use the same price formula for trading:
market_price_token1 = token2_balance / token1_balance
For example, in a pool with 2,000,000 DAI and 1,000 WETH, the constant product is 2,000,000,000 and the market price for WETH is $2,000.
Buy price (e.g., buying WETH in a WETH/DAI pool)
To find the buy price for a certain quantity, first, we calculate how much WETH needs to remain in balance to keep the constant product unchanged:
token1_balance_buy = constant_product / (token2_balance + quantity)
Then we calculate how much WETH goes out to keep this constant:
t1_amount_out_buy = token1_balance - token1_balance_buy
The buy price to reflect this ratio is:
buy_price = quantity / t1_amount_out_buy
Sell price (e.g., selling WETH in a WETH/DAI pool)
To find how much we can sell a certain quantity of WETH for DAI, first, we calculate the ratio of DAI in the new pool, as we add WETH:
token2_balance_buy = constant_product / (token1_balance + quantity)
We then calculate how much DAI will go out:
t2_amount_out_buy = token2_balance + token2_balance_buy
We calculate the DAI balance reflected with the income WETH:
token1_balance_sell = constant_product / (token2_balance - quantity)
And what's the proportion of WETH in the new balance:
t1_amount_in_sell = token1_balance + token1_balance_sell
We can now calculate the sell price to reflect the balance change, keeping the constant:
sell_price = t2_amount_out_buy / t1_amount_in_sell
Getting arbitrage
Run an algorithm to search for arbitrage in the supported exchanges for a certain buy quantity:
bdex -x QUANTITY
Arbitrage opportunities for 10 WETH:
Arbitrage opportunities for 1 WETH:
Arbitrage opportunities for 0.01 WETH:
Running arbitrage algorithm in a loop
To run the arbitrage algorithm for a certain amount of minutes:
bdex -r MIN
Results are saved into results/<arbitrage_TIMESTAMP>.txt.
Here is a sample of the results running this algorithm for 100 minutes for trading 1 WETH:
{'buy_exchange': 'SUSHISWAP', 'sell_exchange': 'UNISWAP', 'arbitrage': '7.01', 'buy_price': 3475.14, 'sell_price': 3482.15} {'buy_exchange': 'SUSHISWAP', 'sell_exchange': 'SHEBASWAP', 'arbitrage': '4.27', 'buy_price': 3475.14, 'sell_price': 3479.41} {'buy_exchange': 'SHEBASWAP', 'sell_exchange': 'UNISWAP', 'arbitrage': '2.06', 'buy_price': 3480.09, 'sell_price': 3482.15} {'buy_exchange': 'CROSWAP', 'sell_exchange': 'UNISWAP', 'arbitrage': '13.06', 'buy_price': 3469.09, 'sell_price': 3482.15} {'buy_exchange': 'CROSWAP', 'sell_exchange': 'SUSHISWAP', 'arbitrage': '5.79', 'buy_price': 3469.09, 'sell_price': 3474.88} {'buy_exchange': 'CROSWAP', 'sell_exchange': 'SHEBASWAP', 'arbitrage': '10.32', 'buy_price': 3469.09, 'sell_price': 3479.41} ...
The Source Code for the main loop
The arbitrage API class
The util class
