In the book Handbook on decision making: Trends and challenges in intelligent decision support systems. 3
Springer, Cham, Switzerland
Summary
The cryptocurrency market is an innovative alternative to the traditional currency markets that has increased the interest of investors, academics, and governments. In the last years, recent studies have tried to model the price of Bitcoin for investment and forecast purposes. Therefore, the objective of this study was to evaluate different crypto market variables through a quantile regression model and thus identify the best predictors for Bitcoin price forecasting by machine learning models. The main finding was that the Gaussian Process Regression models allowed the best performance metrics through the following predictors: high and low Bitcoin price, ask-sum, and Bitcoin price lagged. Likewise, the Bitcoin price was predicted for the next seven days, and it was observed a significant approximation by the confidence intervals of Gaussian Process Regression.
Keywords: Bitcoin price ·Forecasting ·Gaussian process regression ·Machine learning ·Quantile regression model
Editors: Zapata-Cortes, J.A.; et al.,
ISBN: 978-3-031-08245-0
DOI: 
https://doi.org/10.1007/978-3-031-08246-7_11
DOI of the book:
https://doi.org/10.1007/978-3-031-08246-7
Published: 2022
Citation:
A. Oviedo-Gómez, J.M. Candelo Viáfara, D-F. Manotas Duque, Bitcoin price forecasting through crypto market variables: quantile regression and machine learning approaches, in Handbook on decision making: Trends and challenges in intelligent decision support systems. 3. Editors Zapata-Cortes, J.A.; et al., . Ed. Springer. Cham, Switzerland, 2022.
IIT-22-380L
