Forecasting returns in the US market through fuzzy rule-based classification systems

The paper aims to investigate the forecasting ability of fuzzy rule-based classification systems (FRBCS) on future direction of the S&P500 index. To this end, we apply four FRBCS methods. Moreover, we compare both the forecasting accuracy and the interpretability of the results of FRBCS with the recently used machine learning techniques. Overall, among the two approaches, we prefer the FRBCS methods, since they allow a good balance between accuracy and interpretability, and provide sharper results than the machine learning techniques