Forecasting Model Selection Using Intermediate Classification: Application to MonarchFx Corporation

•A new forecasting model selection based on intermediate classification is proposed.•In the proposed algorithm, train and validation sets are used to train classifiers.•Comparative out-of-sample performance is evaluated using trained classifiers.•The results compared favorably to traditional out-of-sample performance models.•The new selection method has been adopted and applied to data provided by MonarchFx.