We provide a trend prediction classification framework named the random sampling method (RSM) for cryptocurrency time series that are non-stationary. This framework is based on deep learning (DL). We compare the performance of our approach to two classical baseline methods in the case of the...