Large Language Models for Nowcasting Cryptocurrency Market Conditions

Anurag Dutta (), M. Gayathri Lakshmi, A. Ramamoorthy and Pijush Kanti Kumar
Additional contact information
Anurag Dutta: Department of Computer Science and Engineering, Government College of Engineering and Textile Technology, Serampore 712201, West Bengal, India
M. Gayathri Lakshmi: Department of Mathematics, Saveetha Engineering College, Chennai 602105, Tamil Nadu, India
A. Ramamoorthy: Department of Mathematics, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology, Chennai 600062, Tamil Nadu, India
Pijush Kanti Kumar: Department of Information Technology, Government College of Engineering and Textile Technology, Serampore 712201, West Bengal, India

FinTech, 2025, vol. 4, issue 4, 1-15

Abstract: Large language models have expanded their application from traditional tasks in natural language processing to several domains of science, technology, engineering, and mathematics. This research studies the potential of these models for financial “ nowcasting ”–real-time forecasting (of the recent past) for cryptocurrency market conditions. Further, the research benchmarks capabilities of five state-of-the-art decoder-only models, gpt-4.1 (OpenAI), gemini-2.5-pro (Google), claude-3-opus-20240229 (Anthropic), deepseek-reasoner (DeepSeek), and grok-4 (xAI) across 12 major crypto-assets around the world. Using minute-resolution history of a day in USD for the stocks, gemini-2.5-pro emerges as a consistent leader in forecasting (except for a few assets). The stablecoins exhibit minimal deviation across all models, justifying the “ nowcast strength ” in low-volatility environments, although they are not able to perform well for the highly erratic assets. Additionally, since large language models have been known to better their performance when executed for a higher number of passes, the experimentations were conducted for two passes (Pass@1 and Pass@2), and the respective nowcast errors are found to be reduced by 1.2156 % (on average).

Keywords: large language models; nowcasting; cryptocurrencies; market volatility (search for similar items in EconPapers)
JEL-codes: C6 F3 G O3 (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2674-1032/4/4/53/pdf (application/pdf)
https://www.mdpi.com/2674-1032/4/4/53/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jfinte:v:4:y:2025:i:4:p:53-:d:1760695

Access Statistics for this article

FinTech is currently edited by Ms. Lizzy Zhou

More articles in FinTech from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().