Nowcasting Sectoral Revenue Growth Using Copula Enhanced Latent Spaces and Quarterly Filings

Date

2026-4-20

Author

Kavruk, Mehmet

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

84
views

0
downloads

Cite This

Traditional financial modeling faces challenges in managing the rise of high-dimensional, often unstructured data. This thesis proposes a robust nowcasting framework that moves beyond single indicator models by leveraging uncertainty‑aware generative networks. To address the common but unrealistic assumption of independence in latent spaces obtained by variational autoencoders, this research integrates copula‑based functions to model complex interdependencies between latent variables, resulting in more realistic representations of economic activity. Furthermore, the framework incorporates large language models (LLMs) to extract sentiment indices from unstructured official reports, bridging the gap between qualitative human analysis and quantitative mathematical modeling. The proposed model is evaluated by nowcasting revenue growth across five U.S. manufacturing-linked sectors, and by combining macroeconomic factor models with LLM‑derived sentiment, the results demonstrate significantly improved accuracy and robustness compared to traditional modeling approaches.

Subject Keywords

Sequential Representation Learning, Time-Varying Copula Dependence, Financial Text Sentiment

URI

https://hdl.handle.net/11511/119083

Collections

Graduate School of Applied Mathematics, Thesis