A Novel Methodology for Credit Spread Prediction: Depth-Gated Recurrent Neural Network with Self-Attention Mechanism

This paper develops a depth-gated recurrent neural network (DGRNN) with self-attention mechanism (SAM) based on long-short-term memory (LSTM)\gated unit (GRU) \Just Another NETwork (JANET) to improve the accuracy of credit spread prediction. The empirical results U.S. bond market indicate that DGRNN model is more effective than traditional machine learning methods. Besides, we discovered Depth-JANET one gated performs better Depth-GRU and Depth-LSTM models units. Furthermore, comparative analyses reveal SAM significantly improves DGRNN’s prediction performance. show outperforms most other methods in