MEASURING THE QUALITY OF MULTIPERIOD ROLLING COMPETING FORECASTS
MEASURING THE QUALITY OF MULTIPERIOD ROLLING COMPETING FORECASTS

Summary/Abstract: For multiperiod rolling forecasts for the same phenomena developed by many forecasters, a complex problem arises of the fair evaluation of the quality of these competing forecasts. The score for a single period should then be only a component of an overall assessment of the sequence of forecasts for multiple periods. Assuming the regularity in the preparation of forecasts and their regular revisions for a given period, we get several forecasts made with various advances. In the final evaluation of the whole sequence of forecasts one should take this issue into account, introducing weights for the forecasts with different horizons. The article discusses some aspects of measuring the quality of competing multiperiod rolling forecasts noting the above-mentioned problems and proposing some evaluation systems.

• Details
• Contents