Machine Learning Algorithm Selection for Forecasting Behavior of Global Institutional Investors

Recently Son et al. proposed early warning system (EWS) monitoring the behaviors of global institutional investors (GII) against their possible massive pullout from the local emerging stock market. They used machine learning algorithm for lag l classifier to forecast the behavior of GII. The main aim of this article is to implement various machine learning algorithms in constructing the EWS and to compare their performances to select the proper one. Our results address various important issues for machine learning forecasting problem. In particular, a proper machine learning algorithm will be recommended for both long term and short term forecasting. This is empirically studied for the Korean stock market.