Open [source] minded
The students were tasked with building a tool that would leverage open-source technology to provide accurate price movement predictions for the volatile world of cryptocurrencies.
“As software engineers, I think a lot of us were really excited about this whole open-source idea, like building these public tools as a public good,” an Olin student describes as their motivation for tackling the project.
Open-source software is an umbrella term that refers to technology that is created for public use and can be further developed by individual contributors. Some notable examples of opensource software include the Apache HTTP Server, the language Python, the e-commerce platform osCommerce, and even the internet browser Mozilla Firefox.
The most well-known cryptocurrencies, Bitcoin and Ethereum, are open source, meaning the codebase is available for public inspection and utilization, which is key for the decentralized ethos of cryptocurrencies.
Currently, the internet - primarily social media - is rife with people promising outlandish gains from investing in cryptocurrencies, only for vulnerable investors to be scammed by bad actors. Some of these promoters include prominent Hollywood celebrities and digital influencers, which resulted in multiple well-known lawsuits earlier this year.1
The project brief sought to combat this phenomenon, with a specific focus on providing a more accurate alternative to current crypto technical analysis tools.
Technical analysis focuses on market action — specifically, volume and price –– to assess potential price movement scenarios.2 Investors perform technical analysis by analyzing stock charts and making educated predictions on how a stock’s price will move in the market. Right now, most of the crypto technical analysis tools fail to accurately represent risk.
The team’s proposed solution, “Coin-Test,” is described as an open-source Python package that generates synthetic data and distributional analysis of strategies. The tool intends to add a step toward a more balanced analysis. In other words, the students’ algorithm utilizes historical data to predict signals to buy and/or sell a particular cryptocurrency, while factoring in risk and specific trading strategies –– like buy high, sell low.
There are five steps to the Coin-Test process:
- Load historical data
- Generate synthetic datasets
- Define trading strategies
- Run backtests
- Analyze results
The results of the project show potential –– with the students and experts specifically highlighting the value Coin-Test has for visualizing data. It’s important to note though Coin-Test is still in development, so users should verify their results prior to relying on it exclusively for any investment choices. The current cryptocurrency market and asset class continues to be one that is emerging and experiences high levels of volatility, so any current or would-be investors should perform their own due diligence and exercise caution before making any investment decisions.
That said, the students deserve props for stepping into a field they weren’t familiar with – financial analysis– and bringing their knowledge of data and software engineering to create something they feel is promising.
Nikhil Murgai, Head of Data Science at FCAT says, “Coin-Test takes a rigorous approach to statistical back-testing of trading strategies. In a world where several online influencers peddle 'get rich quick' trading schemes, Coin-Test offers a way to test and validate these schemes before people put their hard-earned money into crypto trading.”
Dmitry Bisikalo, the FCAT VP of Data Science and the expert who provided feedback to the team echoed this statement saying: “Coin-Test offers novel ways for evaluating trading strategies.”
Coin-Test is open source and available on GitHub for continued development.
A special thank you to the Olin College student team, Gati Aher, Nathan Faber, Andrew Mascillaro, and Eamon Ito-Fisher.
Why FCAT Engages with Universities
Universities represent a way to challenge FCAT’s thinking and navigate the development of cutting-edge technology. In the spirit of innovation, FCAT collaborates with a range from smaller liberal arts colleges to large, global Tier-1 research universities. Some institutions that FCAT has worked with include Brandeis University, Smith College, Savannah College of Arts and Design, and North Carolina State University, to name a few.
FCAT’s University Research and Engagement Director, Chuck Collins summarizes FCAT’s approach to research: “We have a hypothesis that we’re exploring in each of our research areas, but ultimately, it's researching the unknowns of the unknown. To prove or disprove your hypothesis is equally valuable to us in the research space and university-industry collaborations. The purpose is to create and contribute net new knowledge to the research area or provide our teams with information that guides internal work. We’re exploring the future, so we need people – who are often in academia -- to envision the application of emerging technology or discover an approach that doesn’t currently exist.”
If you are interested in collaborating with FCAT, you can reach out here.