Article ID: | iaor200525 |
Country: | Netherlands |
Volume: | 19 |
Issue: | 2 |
Start Page Number: | 241 |
End Page Number: | 256 |
Publication Date: | Apr 2003 |
Journal: | International Journal of Forecasting |
Authors: | Pollock Andrew C., nkal-Atay Dilek, Macaulay Alex, Thomson Mary E. |
Keywords: | forecasting: applications |
An experiment is reported which compares the judgmental forecasting performance of experts and novices using simulated currency series with differing trend strengths. Analyses of directional probability forecasts reveal: (1) significant effects of trend strength on all aspects of predictive performance being studied, with evidence for the hard–easy effect where overconfidence is exhibited on weak (i.e., more difficult) trends, while underconfidence is shown on strong (i.e., less difficult) trends; (2) lower performance of experts on relative accuracy and profitability measures, reflecting experts' resistance to strong trends; (3) better overall performance on negative trends; and (4) superior performance of composite forecasts. Possible explanations are offered for these results and future research directions are outlined.