ALGORITHMIC TRADING STRATEGY DEVELOPMENT USING MACHINE LEARNING

Abstract

Algorithmic trading refers to using a computer application with some algorithms to identify and execute a trade, at a speed and frequency that is impossible for a human trader. As it is profitable, it is one of the applications which draw the most attention from researchers. In order to identify and execute a potential profitable trade, price prediction is the main challenge. In this research, the author will attempt to perform cryptocurrency price predictions with three artificial intelligence techniques – Recurrent Neural Networks (RNN), Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL). Machine learning
(ML) library tools such as Tensorflow and Keras will be used to develop and evaluate the models. This project also studies the factors that affect the performances of models built with these three techniques, and explores possible improvements to the models. The author has chosen mixed methods studies to integrate quantitative and qualitative data collection and analysis. It attempts to combine the best of both methodologies to integrate perspectives and create a rich picture. Author has interviewed some traders to gain the insights to the key areas in trading. As for the quantitative data collection and analysis, the selected cryptocurrency historical price data will be collected from Kaggle.