Automated Stock Price Motion Prediction Using Technical Analysis Datasets and Machine Learning

Abstract

The purpose of this paper is to study, develop and evaluate a prototype stock motion prediction system using Technical Analysis theory and indicators, in which the process of forecasting is based on Machine Learning. The system has been developed following the cloud computing paradigm consisting of a backend application in Google’s API Hosting Cloud and using an Android front end using inspiring, contemporary styles of tools and libraries and harmonized with modern technology trends. The focus of the paper is on system implementation and process automation. Due to the increased difficulty level of the task, the traditionally heavyweight stocks of the bank sector of the Athens Stocks Exchange market have been used in system modeling, as the attempt to generalize the model for all stocks would make our effort impossible. A use case with the famous US S&P 500 index has been also tested. We conclude with a discussion on the optimization of the accuracy of such systems.