Bitcoin is the first decentralized digital currency and is built on truly transformative technology. Some experts call bitcoin “the currency of the future” and, Twitter’s CEO stated his belief in its future potential: “The world ultimately will have a single currency, the internet will have a single currency. I personally believe that it will be bitcoin.”
At the same time, cryptocurrency is extremely volatile. This volatility can have high profit potential if smart investing strategies are taken. Unfortunately, due to their lack of indexes, cryptocurrencies are relatively unpredictable compared to traditional financial instruments. Many entities are interested in predicting cryptocurrency prices. It is especially important for government firms as they need to get ready for significant price movements and prepare consistent economic policy. The demand for cryptocurrency price prediction tools are high.
The Cryptorion team in UC Berkeley’s Data-X course set out to tackle this challenge. Using regression models, LSTM, and NLP sentiment analysis, the team aimed to develop a model that would inform investors when to buy, sell, or hold their cryptocurrency. The team’s future work will focus on gathering data from more exchanges, combining LSTM and Gradient Boosting for short-term prediction, adding sentiment analysis results to adjust the risk preference of the algorithm, and developing a user interface web application.