Prediction of cryptocurrency values using sentiment analysis of news and tweets

Virtual currencies (or cryptocurrencies) are increasingly becoming more competitive in the global market, atracting investors seeking for profit on the oscillation of this market. These investiments are oriented by a simple principle: buy the currencies when their market value is about to rise, and sell them when their market value is about to drop. Several sources of information can be used to support the decision making process-such as news or comments in social networks on the topic itself. Nonetheless, to deal with the such a huge amount of information presents itself as a big challenge. The goal of this work is to develop a model that predicts the movement of cryptocurrencies' prices based on the public perception about the currencies. Prediction models were derived from each source of information and from the combination of both sources. Results obtained up to 75% MDA using the model induced with XGBoost, from the combination of the two sources, being able to predict the results even during periods of oscilation.