The study of TAIEX Directional Forecasts for Human Judgmental Accuracy

• Main Authors: Yi-Wen Yang, 楊怡文
• Other Authors: Hsi-peng Lu
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/60199748477323251249

碩士 === 國立臺灣科技大學 === 管理學院MBA === 98 === ABSTRACT The process-performance paradox describes that highly knowledgeable people often fail to achieve highly accurate judgment. Much research work related to human judgmental forecasts had been done for the test of this paradox, and showed various results in different domains. However, most of the experimental forecasts were limited to perform a very short period with one-time or few time forecasts. The one-time forecasting result could lose its meaning or representativeness if the forecaster is purely lucky to hit the answer, causing a nil-discrimination in data analysis. This study conducted an empirical study in the case of Taiwan Stock Exchange Capitalization Weight Stock Index (TAIEX) directional probabilistic forecasts. To make the experimental results become more accurate so as to reflect practical phenomena, the experiment of this study was planned to collect a period of forecasting data for consecutive 60 business days. A web-based questionnaire was designed to improve the convenience for participants in providing their responses and achieve data gathering automatically. In this experiment, twenty-two voluntary forecasters are divided into three groups (3 security experts, 2 academic experts, and 17 non-professionals) together with a fair coin to make one-day-ahead prediction. The fair coin was used as a random tool to make probabilistic judgment for comparison purpose. Not only a comparison of accuracy in human judgment between professionals and non-professionals was provided, factors affecting the decision process, such as calibration, discrimination, bias, slope, and scat indicators to TAIEX forecast were also considered. The experimental results indicated that (1) on average, professionals’ calibration accuracy is superior to both academic professionals and non-professionals, though the predictive figures of some amateurs outperform professionals; (2) the bias of human judgment in the professionals is less than non-professionals, showing that investment experience provides a positive impact on prediction accuracy; (3) among the three groups, there is a tendency of over-confidence existing in the group of naïves; (4) security professionals have higher overall accuracy and well-calibration in TAIEX forecasts, but there is nil-discrimination regarding sample base rate prediction under the decision process.