Foreign Currency Exchange Market Prediction using Machine Learning Techniques

The importance of predicting foreign currency exchange rate is evident in both academic and business sectors, despite financial time series data being known to be chaotic, noisy and dynamic. Even though this problem has been vastly explored and studied since 1971, to the best of our knowledge, there is yet to be an effective systematic or theoretical framework for approaching the problem. Prior traditional approaches in predicting the future currency rate relied on technical and fundamental analysis, but over recent decades, prolific machine learning models capable of recognizing complex patterns, have been dominating the field with promising accuracy. Furthermore, researchers have demonstrated that hybridizing appropriate machine algorithms can improve the predictive ability, although their sustainability is still questionable. Of late, behavioral economist supposed that finance market is driven by sentiment of the market participants, thus, text mining has further enhanced the potential of machine learning techniques, whereby unstructured data for example news, forums and blogs can be used as features. These evaluations terminate by the place of insight recommendations for outlook research.