Chronos-Codex2022

This project explores the process of cointegreatiom between cryptocurrency price price series with the goal of developing statistical arbitrage trading strategies.

Inspiration

We stumbled across and academic paper from Stanford in 2019 where they introduced a method to utilize machine learning in statistical arbitrage pair selection.

What it does

This project uses machine learning to create and train a machine learning model to create an optimized statistical arbitrage trading strategy to be utilzed by trading bots.

How we built it

To build the machine learning model we had to fetch all the historic data on multiple coins using python and the python-binance API in jupyter notebook. After that we used the gathered data to train a machine learning model to optimize trading returns.

Challenges we ran into

Most of the challanges we ran into were of a technical nature. It always had something to do with python and importing/installing packages. Data manipulation and the subsequent cleaning also took a big chunk of our time to perfect.

Accomplishments that we're proud of

We are very proud of our data analysis that managed to prove that our intuition is true, there are cointegration relationships between crypto currencies, mainly DOTUSDT and BTCUSDT.

What we learned

We gained a deeper understanding of automated crypto trading and utilizing machine learning to tune optimal trading algorithms.

What's next for Statistical arbitrage bot ETH/MATIC

To continue optimizing and fine tuning the strategy using machine learning to maximize trading returns. We would also like to deploy this strategy on a live network using a bot.

Demo video

(We had a few technical issues so we're putting the video here) https://drive.google.com/file/d/1lwRDj2iifUw4QhVQbpi2SHn2pnDZmpJM/view?usp=sharing