Visualizing and Forecasting the Cryptocurrency Ecosystem

By Shawn Anderson, Ka Hang Jacky Lok, Vijayavimohitha Sridhar

Cryptocurrencies are programmable digital assets which have perceived value due to designed scarcity. The first cryptocurrency, Bitcoin, created in 2009, currently has a market capital of more than 136 billion dollars. In the past 10 years, the price of Bitcoin has grown 80,000 times. No one could have imagined this upsurge in cryptocurrency value. Some news has defined cryptocurrency as a fraud or bubble, while others think that it can revolutionize the way people interact with money. For their project, Shawn Anderson, Ka Hang Jacky Lok, and Vijayavimohitha Sridhar collected data from, GitHub, Twitter, and Wikipedia to produce visualizations of the cryptocurrency ecosystem. These provide insights into how the value of cryptocurrencies might be affected by GitHub features, tweets, and the news. The team also uses deep learning techniques to forecast cryptocurrency prices based on historical price data, achieving an impressive 58% accuracy rate for binary next-day prediction (i.e. predicting whether prices go up or down the next day). The possibility of achieving 58% accuracy by chance on the same number of validation examples used is less than 1 in 100 million, and the implications for potential long-term profits in the cryptocurrency market are significant.

Project report | Project poster | Video