We wanted to create a tool to help navigate the volatile rise and fall of cryptocurrencies like Ethereum. Fluctuations in cryptocurrencies are heavily influenced by public opinion, which can be gauged by social media. We chose to analyze tweets because Twitter is a public and frequently used platform that people use to both express their attitudes and post news articles. Additionally, the 140 character limit per tweet makes it easier for a computer to understand sentiments.
What it does
Cryptonite takes tweets on Twitter, determines their sentiment and the general sentiment of the public, and uses that information to predict the price of Ethereum.
How we built it
Cryptonite is a Django site hosted on Heroku. To construct its functionality, we used machine learning models built with Microsoft Azure. For more detail, please see the page on the bottom of the site.
What's next for Cryptonite
In the future, Cryptonite is looking to incorporate new data and more sophisticated algorithms to make more informed decisions. Currently, Cryptonite uses the sentiment of tweets as a predictor of the price of Ethereum. Integrating the sentiment of other social media, such as Facebook and Reddit, would allow for an even more powerful predictor. We also want to incorporate the sentiment of news articles about Ethereum and cryptocurrencies as a whole, as they would add another element to Cryptonite that twitter may not have. Finally, in the future, we want to expand to other currencies using a similar algorithm and use more historical data to better understand long term trends.