Uniflow

Uniflow logo
GIF
Uniflow animated logo
Diagram showcasing the various technologies Uniflow utilizes and how they are interconnected

Inspiration

With a prior interest in crypto and defi, we were attracted to Uniswap V3's simple yet brilliant automated market maker. The white papers were tantalizing and we had several eureka moments when pouring over them. However, we realized that the concepts were beyond the reach of most casual users who would be interested in using Uniswap. Consequently, we decided to build an algorithm that allowed Uniswap users to take a more hands-on and less theoretical approach, while mitigating risk, to understanding the nuances of the marketplace so they would be better suited to make decisions that aligned with their financial goals.

What it does

This project is intended to help new Uniswap users understand the novel processes that the financial protocol (Uniswap) operates upon, specifically with regards to its automated market maker. Taking an input of a hypothetical liquidity mining position in a liquidity pool of the user's choice, our predictive model uses past transactions within that liquidity pool to project the performance of the specified liquidity mining position over time - thus allowing Uniswap users to make better informed decisions regarding which liquidity pools and what currencies and what quantities to invest in.

How we built it

We divided the complete task into four main subproblems: the simulation model and rest of the backend, an intuitive UI with a frontend that emulated Uniswap's, the graphic design, and - most importantly - successfully integrating these three elements together. Each of these tasks took the entirety of the contest window to complete to a degree we were satisfied with given the time constraints.

Challenges we ran into and accomplishments we're proud of

Connecting all the different libraries, frameworks, and languages we used was by far the biggest and most frequent challenge we faced. This included running Python and NumPy through AWS, calling AWS with React and Node.js, making GraphQL queries to Uniswap V3's API, among many other tasks. Of course, re-implementing many of the key features Uniswap runs on to better our simulation was another major hurdle and took several hours of debugging. We had to return to the drawing board countless times to ensure we were correctly emulating the automated market maker as closely as possible. Another difficult task was making our UI as easy to use as possible for users. Notably, this meant correcting the inputs since there are many constraints for what position a user may actually take in a liquidity pool. Ultimately, in spite of the many technical hurdles, we are proud of what we have accomplished and believe our product is ready to be released pending a few final touches.

What we learned

Every aspect of this project introduced us to new concepts, or new implementations of concepts we had picked up previously. While we had dealt with similar subtasks in the past, this was our first time building something of this scope from the ground-up.

Built With

Submitted to

Cal Hacks 8.0
- Winner Uniswap - DeFi Innovation

Created by

I developed the frontend using React. I closely worked with Elden and Toby to find the best way of integrating with the data provided by the backend. It was my first time using Node.js and the React framework, so it was initially intimidating. However, I was able to learn on the fly and it was extremely satisfying to see the page load up.

Jonathan Yue
I came up with the idea of the model for predicting liquidity mining yields over time. I closely read the Uniswap V3 whitepaper, unpacking the mathematical intricacies of the Uniswap V3 automated market maker in order to write a Python class to emulate a real Uniswap V3 liquidity pool. Using Bayesian methods, I designed a model to take in past trade data and generate likely future trading patterns, which would be used to simulate a liquidity miner's earnings over time in a given position. After finishing development of the model, I worked with Tobias in figuring out how to get the necessary data from GraphQL queries and assisted him in integrating the model with the AWS backend.

Elden Ren
I created the GraphQL and REST api requests that our frontend and backend used to pull data from each other and Uniswap. I developed the AWS backend that we used to run Elden's python model. I also worked with Jonathan on the React frontend and put some time into debugging the model with Elden. Configuring the backend to receive a response from a React program while running Python was challenging but satisfying when I got it to work.

Tobias Worledge
Nishant B

Updates

Tobias Worledge posted an update — Oct 28, 2021 04:18 AM EDT

We won the 1st place prize in Uniswap's DeFi Innovation category! See the closing ceremony video (timestamped for 21:54): https://youtu.be/3MDpnGUTwyE?t=1314

Log in or sign up for Devpost to join the conversation.

Nishant B started this project — Oct 24, 2021 11:31 AM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.