TY - GEN
T1 - FinNLP-FNP-LLMFinLegal @ COLING 2025 Shared Task
T2 - Joint Workshop of the 9th Financial Technology and Natural Language Processing, FinNLP 2025, the 6th Financial Narrative Processing, FNP 2025, and the 1st Workshop on Large Language Models for Finance and Legal, LLMFinLegal 2025, co-located with the 31st International Conference on Computational Linguistics, COLING 2025
AU - Yu, Yangyang
AU - Li, Haohang
AU - Cao, Yupeng
AU - Wang, Keyi
AU - Deng, Zhiyang
AU - Yao, Zhiyuan
AU - Jiang, Yuechen
AU - Li, Dong
AU - Weng, Ruey Ling
AU - Suchow, Jordan W.
N1 - Publisher Copyright:
© 2025 Association for Computational Linguistics.
PY - 2025
Y1 - 2025
N2 - Despite the growing potential of large language model (LLM)-based agent frameworks in stock trading, their applicability to comprehensive analysis and multi-asset financial trading, particularly in cryptocurrency markets, remains underexplored. To bridge this gap, we introduce the Agent-Based Single Cryptocurrency Trading Challenge, a shared financial task featured at the COLING 2025 FinNLP-FNP-LLMFinLegal workshop. This challenge focuses on two prominent cryptocurrencies: Bitcoin and Ethereum. In this paper, we present an overview of the task and associated datasets, summarize the methodologies employed by participants, and evaluate their experimental results. Our findings highlight the effectiveness of LLMs in addressing the unique challenges of cryptocurrency trading, offering valuable insights into their capabilities and limitations in this domain. To the best of our knowledge, this challenge is among the first to systematically assess LLM-based agents in cryptocurrency trading. We conclude by providing detailed observations and actionable takeaways to guide future research and development in this emerging area.
AB - Despite the growing potential of large language model (LLM)-based agent frameworks in stock trading, their applicability to comprehensive analysis and multi-asset financial trading, particularly in cryptocurrency markets, remains underexplored. To bridge this gap, we introduce the Agent-Based Single Cryptocurrency Trading Challenge, a shared financial task featured at the COLING 2025 FinNLP-FNP-LLMFinLegal workshop. This challenge focuses on two prominent cryptocurrencies: Bitcoin and Ethereum. In this paper, we present an overview of the task and associated datasets, summarize the methodologies employed by participants, and evaluate their experimental results. Our findings highlight the effectiveness of LLMs in addressing the unique challenges of cryptocurrency trading, offering valuable insights into their capabilities and limitations in this domain. To the best of our knowledge, this challenge is among the first to systematically assess LLM-based agents in cryptocurrency trading. We conclude by providing detailed observations and actionable takeaways to guide future research and development in this emerging area.
UR - http://www.scopus.com/inward/record.url?scp=85217737086&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85217737086&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85217737086
T3 - Proceedings - International Conference on Computational Linguistics, COLING
SP - 401
EP - 406
BT - Joint Workshop of the 9th Financial Technology and Natural Language Processing, FinNLP 2025, the 6th Financial Narrative Processing, FNP 2025, and the 1st Workshop on Large Language Models for Finance and Legal, LLMFinLegal 2025
A2 - Chen, Chung-Chi
A2 - Moreno-Sandoval, Antonio
A2 - Huang, Jimin
A2 - Xie, Qianqian
A2 - Ananiadou, Sophia
A2 - Chen, Hsin-Hsi
Y2 - 19 January 2025 through 20 January 2025
ER -