XFinBench: Benchmarking LLMs in Complex Financial Problem Solving and Reasoning

TL;DR

XFinBench is a comprehensive benchmark with 4,235 examples designed to evaluate large language models' ability to solve complex, knowledge-intensive financial problems involving multimodal data, revealing current limitations and areas for improvement.

Contribution

The paper introduces XFinBench, a novel benchmark for assessing LLMs in complex financial reasoning, and provides extensive experimental analysis on 18 models highlighting their strengths and weaknesses.

Findings

O1 is the best text-only model with 67.3% accuracy.

Models lag behind human experts by 54.8% in accuracy.

Knowledge augmentation improves small open-source models' performance.

Abstract

Solving financial problems demands complex reasoning, multimodal data processing, and a broad technical understanding, presenting unique challenges for current large language models (LLMs). We introduce XFinBench, a novel benchmark with 4,235 examples designed to evaluate LLM's ability in solving complex, knowledge-intensive financial problems across diverse graduate-level finance topics with multi-modal context. We identify five core capabilities of LLMs using XFinBench, i.e, terminology understanding, temporal reasoning, future forecasting, scenario planning, and numerical modelling. Upon XFinBench, we conduct extensive experiments on 18 leading models. The result shows that o1 is the best-performing text-only model with an overall accuracy of 67.3%, but still lags significantly behind human experts with 12.5%, especially in temporal reasoning and scenario planning capabilities. We further construct a knowledge bank with 3,032 finance terms for knowledge augmentation analysis, and find that relevant knowledge to the question only brings consistent accuracy improvements to small open-source model. Additionally, our error analysis reveals that rounding errors during calculation and blindness to position and intersection of curves in the image are two primary issues leading to model's poor performance in calculating and visual-context questions, respectively. Code and dataset are accessible via GitHub: https://github.com/Zhihan72/XFinBench.