A Study on the Predictive Power of Leading Economic Indicators for Stock Market Returns

Abstract

This research investigates the predictive power of leading economic indicators (LEIs) for forecasting stock market returns using advanced machine learning techniques. A Long Short-Term Memory (LSTM) neural network, known for its ability to capture temporal dependencies, is employed as the predictive model. The feature selection process emphasizes macroeconomic variables such as GDP growth rates, unemployment rates, and consumer sentiment indices, which are highly correlated with stock market movements. Data preprocessing includes normalization to ensure the comparability of variables with varying scales. For classification, a binary framework is utilized to predict whether the market will experience a positive or negative return in the subsequent period. The results demonstrate that combining LEIs with deep learning methods enhances the accuracy of stock market predictions, providing insights for both investors and policymakers.