PRISM

Inspiration

We were inspired by the unusual trends of cryptocurrency markets that are normally difficult to predict and we wanted to try and come up with a trading strategy to understand how the cryptocurrency market works.

What it does

Our trading strategy is based on the mean reversion model where it analyzes data from Bitcoin (BTC) and Ethereum (ETH) and figures out the best method to trade it. We also utilized other cryptocurrencies, Ripple (XRP) and Solana (SOL), as market indicators to provide an overview of the crypto market conditions. The sentiment score is our alternative data and it identifies another reason why the crypto market is changing as it is likely based on people's reactions and emotions, which cause them to buy or trade the stocks.

How we built it

We utilized Databento to obtain the past data of the stocks for BTC, ETH, XRP, and SOL and we utilized Alternative.me's Crypto Fear & Greed Index for the alternative data. Tiger Data stored the data and analyzed it, and Snowflake took the analyzed BTC and ETH data and trained a machine learning model. Webull provided backtesting information and helped utilize XRP and SOL data for market indicators and the Crypto Fear & Greed Index for a sentiment score to make decisions regarding cryptocurrency trading more efficiently.

Challenges we ran into

One challenge we ran into was finding suitable data sources as although we found datasets for XRP and SOL, they were not available for a longer amount of time than we originally predicated so we had to adapt to the lack of data available. Additionally, another difficulty was trying the ML model as it took a long time for the model to understand which trends were important to look out for to buy or trade assets then.

Accomplishments that we're proud of

We are proud of coming up with a systematic trading strategy that attempts to tackle one of the most unpredictable markets, cryptocurrency. We are also proud of training an ML in a short amount of time.