Predictability of Financial Markets in ASEAN Countries using Machine Learning Techniques

Abstract

This paper develops several efficient machine learning models and compare their performance in forecasting the value and direction of stock prices and indices from the ASEAN countries. Although all models adequately forecast the stock indices ranging from 40% to 95% accuracy and outperform traditional regression models, ANN models outperform all other models. This study identifies several important variables as important predictors. Finally, this study concludes that the emerging economies of the ASEAN countries are indeed predictable with more than 95% accuracy.

Keywords: Machine Learning, ASEAN, Financial Markets, Support Vector Machines, Neural Networks, Random Forest, Gradient Boosting, Backpropagation, Multitask Learning

JEL Classification: C02, O33