Price Forecasting for Bitcoin: Linear Regression and SVM approaches

A long term analysis for the most renowned crypto asset (namely Bitcoin) is considered. Bitcoin has currently the largest market capitalization among the crypto assets, and in the last years has steadily attracted the attention of both private and institutional investors. Since 2008 Bitcoin price has always experienced high volatility, too, which earned it the title of risky asset in the literature of FinTech. Hence, practitioners have stressed the key role of robust models to reliably predicting its value, not to mention the relevance of a comparative analysis with safe haven assets as silver and gold. This paper focuses on a couple of approaches to predict long term Bitcoin price. Basically the first one relies on more standard regression and linear models. Our second proposal is conversely based on applying a Support Vector Machine (SVM), in the class of Machine Learning (ML) methods, which do not require any of the assumptions typically needed by solvers for standard regression problems. We remark that both the above proposals are inherently data-driven.