In this research paper, “Deep Learning and the Cross-Section of Expected Returns,” authored by Marcial Messmer, the exploration delves into the realm of deep learning, an actively researched domain within machine learning.
The study, conducted using deep feedforward neural networks (DFN), focuses on predicting the cross-section of stock returns in the United States based on a comprehensive set of 68 firm characteristics (FC).
Employing a meticulous network optimization strategy, the research reveals that DFN long-short portfolios exhibit the potential to generate compelling risk-adjusted returns when compared to a linear benchmark. This outcome underscores the significance of non-linear relationships among firm characteristics and expected returns. Importantly, the robustness of these results is demonstrated across variations in size, weighting schemes, and portfolio cutoff points.
Furthermore, the study sheds light on the key drivers of return predictions, emphasizing the pivotal role played by price-related firm characteristics, specifically, short-term reversal and twelve-month momentum.
Notably, the majority of firm characteristics have a minor impact on the variation of these predictions. The findings contribute to our understanding of the intricate dynamics between deep learning, firm characteristics, and the cross-section of expected returns in the financial landscape.
Abstract Of Paper
Deep learning is an active area of research in machine learning. I train deep feedforward neural networks (DFN) based on a set of 68 firm characteristics (FC) to predict the US cross-section of stock returns. After applying a network optimization strategy, I find that DFN long-short portfolios can generate attractive risk-adjusted returns compared to a linear benchmark. These findings underscore the importance of non-linear relationships among FC and expected returns. The results are robust to size, weighting schemes and portfolio cutoff points. Moreover, I show that price related FC, namely, short-term reversal and the twelve-months momentum, are among the main drivers of the return predictions. The majority of FC play a minor role in the variation of these predictions.
Original paper – Download PDF
Here you can download the PDF and original paper of Deep Learning and the Cross-Section of Expected Returns.
(An option to download will come shortly)
Author
Marcial Messmer
mu Capital Management; University of St. Gallen
Conclusion
In this research, Marcial Messmer explores the potential of deep learning, specifically deep feedforward neural networks (DFN), to predict the cross-section of expected returns in the U.S. stock market using a wide set of 68 firm characteristics (FC). The study’s systematic network optimization approach demonstrates that DFN long-short portfolios can yield strong risk-adjusted returns when compared to a linear benchmark.
This highlights the significance of non-linear relationships between firm characteristics and stock returns. Notably, the results remain robust across different factors, such as firm size, weighting methods, and portfolio cutoff points.
Moreover, the study identifies price-related firm characteristics, like short-term reversal and twelve-month momentum, as key drivers of return predictions, while most other FC play a minor role in explaining return variation.
FAQ:
– What is the main focus of the research paper “Deep Learning and the Cross-Section of Expected Returns” by Marcial Messmer?
The research paper explores the application of deep feedforward neural networks (DFN) trained with 68 firm characteristics to predict the cross-section of expected returns in the U.S. stock market. It delves into the potential impact of deep learning in understanding these returns.
– What are the key findings of the study regarding the use of deep feedforward neural networks for stock return predictions?
The research reveals that DFN long-short portfolios can generate appealing risk-adjusted returns when compared to a linear benchmark. This emphasizes the importance of non-linear relationships between firm characteristics and expected returns. The results also remain robust across various factors, such as firm size, weighting methods, and portfolio cutoff points.
– Which firm characteristics play a significant role in driving return predictions, according to the research?
The study identifies price-related firm characteristics, specifically short-term reversal and twelve-month momentum, as among the main drivers of return predictions. Most other firm characteristics play a minor role in explaining return variation in the U.S. stock market.
