Predicting the closing price of cryptocurrencies

Abstract

Current research shows that stock market price, collected as a type of time-series data, could be forecasted by machine learning. The pricing data of cryptocurrency could also be used to conduct time-series prediction by leveraging different models, such as Long Short-Term Memory, Bayesian regression, GLM/Random Forest. This paper compares some of the machine learning methods used in predicting the price of cryptocurrencies by illustrating the nature of cryptocurrency, data availability, model used, results associated, and challenges.