Prototyping a financial decision support system

This paper explores, by means of a laboratory study, the ability of a financial DSS of known theoretical merit to improve decision-making for currency and interest rate forecasting where typically the task is to forecast direction of movement and to assess confidence. The study finds that while the decision-making is improved in comparison with a control, most of the potential value of the DSS is lost. In addition, no improvement in confidence assessment is observed. Both the DSS users and the control group exhibit very poor calibration with confidence assessments uncorrelated with accuracy.