« The ABC of Automated Forecast Model Profile Selection » The work presents a step-by-step procedure for automated selection of appropriate (useful & credible) forecasting models, based on the observable profiles drawn from historic data. Analyzing core features of past demand, decision metrics are assessed, and then applied as gateways through a decision tree. An approach for filtering trend and seasonality from noise is introduced, which is designed as outlier-resistant. The practical implementation of the procedure, steps and metrics are also presented, aiming at its continuous operation, embedded into a forecast decision support system (FDSS).