Financial time series forecasting is often plagued by a fundamental flaw: complex deep learning architectures are deployed against highly stochastic data, yielding impressive evaluation metrics that ...