We answer this research question by comparing six well-established machine learning models trained on 9 months of minutely bitcoin-related data against each other. The results show that the trained models indeed significantly outperform random classification .