Generative AI in Financial Mathematics: Evaluating LLMs on a Bond-Focused Benchmark Dataset

Nemanja Vučićević1 Marina Svičević1 Aleksandar Milenković1 Miloš Pavković2 Lazar Stošić3,4 Jelica Protić5

  1. Faculty of Science, University of Kragujevac, Radoja Domanovića 12 34000 Kragujevac, Serbia marina.svicevic@pmf.kg.ac.rs (corresponding author), {nemanja.vucicevic, marina.svicevic, aleksandar.milenkovic}@pmf.kg.ac.rs
  2. Singidunum University, Danijelova 32 11000 Belgrade, Serbia mpavkovic@singidunum.ac.rs
  3. Faculty of Informatics and Computer Science, University Union—Nikola Tesla, Belgrade, Serbia
  4. Don State Technical University, Rostov-on-Don, Russian Federation lstosic@unt.edu.rs
  5. School of Electrical Engineering, University of Belgrade, Bulevar kralja Aleksandra 73 11000 Belgrade, Serbia jelica.protic@etf.bg.ac.rs

Abstract

This paper presents a comprehensive evaluation of the performance of large language models (LLMs) on a newly created set of 50 tasks in financial math ematics, 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 theoreti cal) 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, nu merical, 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 chal lenging 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.

Key words

Large Language Models,Financial Mathematics,Bond Valuation,Dataset Creation, Model Evaluation.

Digital Object Identifier (DOI)

https://doi.org/10.2298/CSIS251025050V