Forecast accuracy
Evaluate GPMI-AI by its retained historical record
Resolved forecasts, calibration and model performance are shown without hiding unsuccessful predictions or mixing unrelated horizons into one marketing score.
How quality is judged
Historical accountability can include calibration, error by horizon, regime performance, model-version comparisons and refusal history. Metrics appear only when enough resolved observations exist.
Research only. Outputs are probabilistic and do not guarantee future performance.