A comprehensive analysis of the use of deep learning models for forecasting the cross-section of stock returns

Cem Öztürk
2020
A comprehensive analysis of the use of deep learning models for forecasting the cross-section of stock returns CemÖztürk Munich, 08.05.2020 Supervision: Prof. Abstract This thesis focuses on recent developments in neural networks and their predictive power for forecasting the cross-section of stock returns. Neural networks with architectures ranging from shallow to deep are used with different hyperparameter configurations. Company characteristics are considered as the feature set and
more » ... ip between features are explained. On the other hand, this work also challenges the recent conclusions about the applicability of neural networks in the financial domain with counter arguments and reasons are supported with the works from machine learning research.
doi:10.5282/ubm/epub.74082 fatcat:ua7zqb3mpfb2dn4j2dd3lpcmyq