Accuracy of judgmental forecasts: A comparison

This study compares the performance of judgmental and statistical forecasts, as well as the judgmental revision of statistically derived forecasts, for artificial time series with a number of controlled characteristics. Judgment is found to produce inferior and more biased forecasts overall. However, results suggest that judgmental revision may have a benefit for some low noise series, particularly when statistical techniques are applied ‘blindly’ to the time series.