Neural Network based Trading Signal Generation in Cypto-Currency Markets

Abstract

Machine learning algorithms are commonly used to automate stock market trading. Crypto-currencies are novel digital assets that attracted investors all over the world. In this paper, we developed a neural network estimator to generate trading signals. Unlike previous methods, our method uses more volatile and uses historical data divided five minutes intervals. First, our method uses technical indicators and optimizes their time periods then we developed artificial neural network (ANN) architectures to predict asset future directions. Classification and regression networks are developed and their results are compared. Our results are promising but need improvements in order to make more profitable trading.