High frequency Bitcoin price prediction: statistical learning approach

Abstract

While the next day price financial forecasting has been the most popular scenario in the literature, high frequency forecasting has received considerably less attention. In this paper, we consider the problem of predicting the Bitcoin price in the context of high frequency trading. To this end, we compare the performance of the elastic net (Enet) regression against the standard ordinary least squares (OLS) linear regression. The results are surprising in that OLS is more accurate than the more sophisticated Enet. We also identify the driving covariates under each model.