Generative AI in financial mathematics: Evaluating LLMs on a bond-focused benchmark dataset
Computer Science and Information Systems · 2026 · (00) · 50-50
DOI: 10.2298/csis251025050v ↗Kurz gefasst
Fachartikel in der Zeitschrift Computer Science and Information Systems, 2026. Verfasst von Nemanja Vucicevic und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Ministry of Education, Science and Technological Development of the Republic of Serbia.
Abstract (Originalsprache)
This paper presents a comprehensive evaluation of the performance of large language models (LLMs) on a newly created set of 50 tasks in financial mathematics, with a focus on bond valuation. Models o3, Claude Opus 4, DeepSeek-R1, and Gemini were tested on three types of tasks (practical, real-life, and theoretical) and three difficulty levels (easy, medium, and hard). Model o3 achieved the highest average accuracy (80.9%), while DeepSeek-R1 showed the weakest results (70.6%). The detailed analysis included error classification (methodological, numerical, conceptual), measuring solution time as well as the application of ANOVA and post-hoc tests, which confirmed statistically significant differences between the models. The results indicate that more complex and theoretical tasks remain challenging for all models. The proposed methodology and the set of created tasks provide a basis for further research and for the development of specialized LLM systems for financial analysis and educational purposes.
Förderung
- Ministry of Education, Science and Technological Development of the Republic of Serbia
Zitieren
APA
Vucicevic, N., Svicevic, M., Milenkovic, A., Pavkovic, M., Stosic, L., & Protic, J. (2026). Generative AI in financial mathematics: Evaluating LLMs on a bond-focused benchmark dataset. Computer Science and Information Systems, 50-50. https://doi.org/10.2298/csis251025050v
Vancouver
Vucicevic N, Svicevic M, Milenkovic A, Pavkovic M, Stosic L, Protic J. Generative AI in financial mathematics: Evaluating LLMs on a bond-focused benchmark dataset. Computer Science and Information Systems. 2026:50-50. doi:10.2298/csis251025050v
BibTeX
@article{Vucicevic2026,
author = {Vucicevic, Nemanja and Svicevic, Marina and Milenkovic, Aleksandar and Pavkovic, Milos and Stosic, Lazar and Protic, Jelica},
title = {Generative AI in financial mathematics: Evaluating LLMs on a bond-focused benchmark dataset},
journal = {Computer Science and Information Systems},
year = {2026},
number = {00},
pages = {50--50},
doi = {10.2298/csis251025050v}
}
EUVIMED