A Neural Networks filtering mechanism for foreign exchange trading signals

Neural Networks have been successfully used in several financial applications. In the stock market and foreign exchange domains, Neural Networks have been used with considerable success to predict the future prices of stocks and currency pairs, their rate of return, risk analysis, and several other features that might be of benefit. In this paper, we present a methodology to filter the high-frequency signals of a rule-based foreign exchange trading strategy, through a neural network-based, intelligent selection mechanism. We then compare the results vs. a random selection mechanism and again vs. the overall signal pool, in terms of profit and correctness. We can clearly show that the neural network filtering approach yields a better performance than its random baseline.