Novel techniques to forecast daily currency exchange rates and a system that incorporates these techniques are presented. Forecastings are generated from automatically generated probabilistic decision rules, threshold values and global optimization techniques for multiple predicte learning. As extensive experimentations reveal, the system's prediction accuracy compares favourably with those of actual currency traders. It is also shown that the presented techniques outperform statistical techniques in this specific application domain.
Permanent URL for this record: https://lbezone.hkust.edu.hk/bib/b499842