Code Monkey home page Code Monkey logo

arb-bot's Introduction

Arbitrage Bot

JS trading bot that uses custom Solidity contracts, in conjunction with DEX contracts, to execute arbitrage on any EVM compatible blockchain.

To potential employers reviewing my work, please reach out if you wish to see a private fork of this repo (currently in progress) with more complexity and DEX integration, along with more in-depth tests.

Background

Check out https://ethereum.org/en/developers/docs/mev/ for context on this project.

Technologies

Javascript/Node.js, Solidity, Hardhat, Ethers.js, Waffle.

Design

Simple Strategy Overview

The first-pass strategy implemented is only a simple example with two hardcoded ERC20 tokens. Note that profitable strategies may require more complexity.

Planning

Consider two types of tokens, one in which we're "arbing for profits", and one intermediary token we're "arbing against". We'll also use two DEXs in this example, one to swap from our "arb for" tokens to "arb against" tokens, and one to swap back from "arb against" to "arb for" tokens. The hope is to end up with more tokens than you've started with, taking advantage of price discrepancies, after accounting for gas and flash loan fees.

First, we get the token reserves from the second ("sell") DEX, then choose a portion of token1 ("arb against") reserves from the relevant liquidity pool on that DEX. This portion will be the theoretical amount of token1 to obtain from the first ("buy") DEX, right now this is portion is 1/2. Next, compute the minimum amount of token0 ("arb for") it'll take to obtain our set amount of token1 from the "buy" DEX. Lastly, we compute the maximum amount of token0 we can obtain from selling our set amount of token1 on the "sell" DEX.

If the amount of token0 that would be gained in this process exceeds gas fees in ETH (and potential flash loan fees), the theoretical trade would be profitable.

Execution

If the planning stage suggests a profitable trade is possible, a flash loan will be used to borrow the relevant amount of token0 planned above. The planned DEX swaps will execute within the context of the custom arbitrage contract. Once finished, borrowed funds (+fee) will automatically return to the flash loan provider, and relevant gains will be transfered to the deployer of the contract.

Anatomy of bot components

src/bot.js contains only a main function, which calls initialSetup and loops until the user kills the process, continuously waiting for events.

The bot relies on some core functions that follow, contained in src/botComponents.js:

  • initialSetup()
  • getPriceDifferencePercent()
  • determineDirection()
  • determineProfitability()
  • executeTrade()

initialSetup instantiates all relevant contract objects, instantiates an ethers signer object, and subscribes to on-chain DEX events.

When a swap event occurs, getPriceDifferencePercent() is first called. This function will query the current price of relevant tokens on both Uniswap & Sushiswap, and return the priceDifference as a percentage.

Then determineDirection() will determine the order of exchanges to execute token swaps. This function will return an array, routerPath. The array contains Uniswap & Sushiswap's router contracts in a specified order. If no array is returned, this means the priceDifference returned earlier is not higher than the PRICE_DIFFERENCE threshold defined in the .env file (by absolute value).

If routerPath is non-null, then determineProfitability() determines whether there is a potential arbitrage or not, and returns a boolean indicating this decision. This is the "planning stage", in that more complex logic and/or optimization could be contained here.

If determineProfitability() returns true, executeTrade() is called, where we make our call to the custom arbitrage contract to perform an arb trade. If the function returns false, no trade is executed and the bot resumes monitoring. Throughout this entire process, reports are logged.

Tests

Each .js file in Tests serves a unique purpose, and allowed for (pseudo) test driven development. Note that tests are not super thorough yet, and really only verify that critical functions are generally working in the way we want them to.

All tests fork the Ethereum network via Alchemy API, specified by a block number in the hardhat configuration file. They then execute a JSON RPC to the local hardhat provider to impersonate a specific ethereum account. From there, we have a lot of freedom to test arbitrary scenarios.

LocalPriceManipulationTests.js: Tests the module that impersonates a whale with enough relevant ERC20 tokens to manipulate the price of a token pair on a DEX contract already deployed to our local test network. The manipulation of price by dumping a large amount of tokens is tested and verified. Note, this functionality is only used to create arbitrage opportunities within a local testing environment.

ArbitrageTests.js: Verifies on-chain functionality for the arbitrage contract, and how it interacts with various deployed contracts.

BotTests.js: Tests critical funcitons within the Javascript bot. These determine profitability, monitor prices, token reserves, etc.

Setup

Create an .env file

Before running any scripts, you'll want to create a .env file with the following values (see .env.example):

  • ALCHEMY_API_KEY=""
  • ARB_FOR="0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2" (By default we are using WETH)
  • ARB_AGAINST="0x95aD61b0a150d79219dCF64E1E6Cc01f0B64C4cE" (By default we are using SHIB)
  • ACCOUNT="" (Account to recieve profit/execute arbitrage contract)
  • PRICE_DIFFERENCE=0.50 (Difference in price between Uniswap & Sushiswap, default is 0.50%)
  • UNITS=0 (Only used for price reporting)
  • GAS_LIMIT=600000 (Currently a hardcoded value, may need to adjust during testing)
  • GAS_PRICE=0.0093 (Currently a hardcoded value, may need to adjust during testing)

About config.json

Inside the config.json file, under the PROJECT_SETTINGS object, there are 2 keys that hold a boolean value:

  • isLocal
  • shouldExecuteTrade

isLocal: Whether bot should monitor a local hardhat netowork for arb opportunities. If false, the bot will monitor mainnet.

shouldExecuteTrade: Whether the bot should execute a trade on the custom contract if an arb opportunity is found. This is helpful if you want to monitor mainnet for arb opportunities, but don' t yet have a contract deployed.

Local Testing

  1. Install Node.js if needed.
  2. npm install should install needed dependencies to the node_modules folder. Confirm with npx hardhat compile.
  3. You're able to run tests against an ephemeral local network using npx hardhat test.
  4. To spin up a persistent local network forked off mainnet, first create an https://www.alchemy.com/ account, and copy the api key to .env.
  5. Next, run npx hardhat node.
  6. In a separate terminal, you can run scripts against this local network using hardhat CLI, example: npx hardhat run script.js --network localhost.
  7. If desired to run a script against an ephemeral network, leave out --network localhost.
  8. Run the bot with npx hardhat run bot.js --network localhost.

arb-bot's People

Contributors

shaspitz avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.