Ethereum Price Prediction Employing Large Language Models for Short-term and Few-shot Forecasting

2025-03-29

AI Summary

This work addresses the significant challenge of short-term Ethereum price forecasting under few-shot learning conditions. We propose a temporal transfer learning framework leveraging pretrained large language models (LLMs), marking the first cross-modal adaptation of LLMs—originally pretrained on natural language and image data—to cryptocurrency price prediction. Our method introduces temporal data adaptation and a hierarchical selective freezing strategy to align LLMs with financial time series, thereby enhancing generalization and few-shot adaptability. Extensive experiments demonstrate that our approach consistently outperforms classical statistical models (e.g., ARIMA), deep temporal models (e.g., LSTM, N-BEATS), and state-of-the-art methods across multiple metrics—including MSE, MAE, and RMSE—validating the efficacy and robustness of LLMs in financial time-series forecasting. This work establishes a novel paradigm for few-shot financial prediction.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Cryptocurrencies have transformed financial markets with their innovative blockchain technology and volatile price movements, presenting both challenges and opportunities for predictive analytics. Ethereum, being one of the leading cryptocurrencies, has experienced significant market fluctuations, making its price prediction an attractive yet complex problem. This paper presents a comprehensive study on the effectiveness of Large Language Models (LLMs) in predicting Ethereum prices for short-term and few-shot forecasting scenarios. The main challenge in training models for time series analysis is the lack of data. We address this by leveraging a novel approach that adapts existing pre-trained LLMs on natural language or images from billions of tokens to the unique characteristics of Ethereum price time series data. Through thorough experimentation and comparison with traditional and contemporary models, our results demonstrate that selectively freezing certain layers of pre-trained LLMs achieves state-of-the-art performance in this domain. This approach consistently surpasses benchmarks across multiple metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE), demonstrating its effectiveness and robustness. Our research not only contributes to the existing body of knowledge on LLMs but also provides practical insights in the cryptocurrency prediction domain. The adaptability of pre-trained LLMs to handle the nature of Ethereum prices suggests a promising direction for future research, potentially including the integration of sentiment analysis to further refine forecasting accuracy.

Problem

• Predict Ethereum prices using LLMs for short-term forecasting
• Address data scarcity in time series with pre-trained LLMs
• Improve accuracy over traditional models with selective layer freezing

Innovation

• Adapts pre-trained LLMs for Ethereum price prediction
• Freezes selective LLM layers for optimal performance
• Outperforms benchmarks in multiple error metrics