An Artificial Intelligence approach to Shadow Rating

TL;DR

This paper evaluates deep learning methods, specifically neural networks with categorical embeddings, for predicting credit ratings of global corporations, comparing their performance to traditional machine learning models.

Contribution

It demonstrates that neural networks with categorical embeddings achieve adequate accuracy in credit rating prediction, offering a modern alternative to traditional models.

Findings

Neural networks with categorical embeddings outperform traditional models in accuracy.

Deep learning models show promise in credit rating prediction tasks.

The approach is effective across multiple rating classes.

Abstract

We analyse the effectiveness of modern deep learning techniques in predicting credit ratings over a universe of thousands of global corporate entities obligations when compared to most popular, traditional machine-learning approaches such as linear models and tree-based classifiers. Our results show a adequate accuracy over different rating classes when applying categorical embeddings to artificial neural networks (ANN) architectures.