Cross-Sectional Machine Learning Approach on Predicting Stock Return of LQ45 Index

There have been numerous stock-related publications, including analyses and forecasts. Diverse approaches and methods have been applied to studying stock price fluctuations. Recent research in many fields of study uses machine learning for forecasting. Data collection and analysis techniques have also affected prediction outcomes. Cross-sectional is a rarely employed technique, but it potentially yields enhanced prediction results compared to conventional time-series methods. Therefore, this paper uses cross-sectional machine learning to forecast the stock returns comprising the LQ45 index on the Indonesia Stock Exchange. Real-world applications of the prediction results, specifically portfolio construction, are used to evaluate the performance of both prediction models. In every test, cross-sectional-based machine learning demonstrated superior performance to time-series-based machine learning.