@article{ART003228287},
author={Eun Hong Park and Ji Hoon Park and Ha Young Kim},
title={Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={7},
pages={41-51}
TY - JOUR
AU - Eun Hong Park
AU - Ji Hoon Park
AU - Ha Young Kim
TI - Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 7
PB - The Korean Society Of Computer And Information
SP - 41
EP - 51
SN - 1598-849X
AB - In this paper, we propose an approach to evaluate the capability of Multimodal Large Language Models (MLLMs) to recognize and classify visual patterns in stock charts containing real-world visual noise. We apply three prompting strategies—Self-Consistency (SC), In-Context Learning (ICL), and Chain-of-Thought (CoT)—to assess their impact on classification accuracy, and introduce visual preprocessing using the Segment Anything Model (SAM) to analyze the effect of noise reduction.
Through experiments on multiple MLLMs, we compare the individual and combined effects of these strategies. The results show that the SC prompting strategy and SAM preprocessing significantly enhance performance. This study serves as an early empirical investigation into the practical applicability of MLLMs for visual financial analysis, providing quantitative insights into how prompt design and visual preprocessing influence model performance.
KW - Multimodal;Financial Chart;Prompt Engineering;Self-Consistency;Chain-of-Thought;In-Context Learning
DO -
UR -
ER -
Eun Hong Park, Ji Hoon Park and Ha Young Kim. (2025). Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts. Journal of The Korea Society of Computer and Information, 30(7), 41-51.
Eun Hong Park, Ji Hoon Park and Ha Young Kim. 2025, "Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts", Journal of The Korea Society of Computer and Information, vol.30, no.7 pp.41-51.
Eun Hong Park, Ji Hoon Park, Ha Young Kim "Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts" Journal of The Korea Society of Computer and Information 30.7 pp.41-51 (2025) : 41.
Eun Hong Park, Ji Hoon Park, Ha Young Kim. Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts. 2025; 30(7), 41-51.
Eun Hong Park, Ji Hoon Park and Ha Young Kim. "Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts" Journal of The Korea Society of Computer and Information 30, no.7 (2025) : 41-51.
Eun Hong Park; Ji Hoon Park; Ha Young Kim. Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts. Journal of The Korea Society of Computer and Information, 30(7), 41-51.
Eun Hong Park; Ji Hoon Park; Ha Young Kim. Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts. Journal of The Korea Society of Computer and Information. 2025; 30(7) 41-51.
Eun Hong Park, Ji Hoon Park, Ha Young Kim. Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts. 2025; 30(7), 41-51.
Eun Hong Park, Ji Hoon Park and Ha Young Kim. "Can MLLMs See the Market? Evaluating Pattern Reasoning in Real-World Financial Charts" Journal of The Korea Society of Computer and Information 30, no.7 (2025) : 41-51.
