Deep learning approaches for forecasting the FOMC statement sentiment index

This paper explores the sentiment analysis of Federal Open Market Committee (FOMC) statements and their time series forecasting using both traditional and deep learning-based models. The models examined include ARIMA, Multilayer Perceptron (MLP), Generative Adversarial Networks (GANs), Long Short-Term Memory (LSTM) networks, Convolutional Neural Networks (CNNs), and Exponential Smoothing State Space Models (ETS). Overall, this paper demonstrates the power of deep learning models, particularly CNN and GAN, in effectively capturing sentiment and forecasting time series, offering valuable insights for economic and financial applications that rely on sentiment analysis of policy statements. © 2025 The Korean Statistical Society, and Korean International Statistical Society. All rights reserved.

Deep learning approaches for forecasting the FOMC statement sentiment index Academic Article in Scopus