A Study on Cryptocurrency Price Movement Prediction Using Prompt Engineering 


Vol. 14,  No. 10, pp. 813-824, Oct.  2025
https://doi.org/10.3745/TKIPS.2025.14.10.813


PDF
  Abstract

Recently, the application of Large Language Models (LLMs) in financial market forecasting has garnered significant attention. This study employs LLMs to predict the directional movement of cryptocurrency prices and quantitatively compares the performance of LLM-based approaches with numerical and deep learning–based trading strategies. The dataset used in the experiments consists of daily open price time series for four cryptocurrencies—BTC, BNB, SOL, and XRP—collected throughout 2023, and the task is formulated as a binary classification problem distinguishing upward (Positive) and downward (Negative) movements. For comparison, we included numerical trading strategies such as the Simple Moving Average (SMA), Short–Long Moving Average crossovers (SLMA), Moving Average Convergence–Divergence (MACD), and Bollinger Bands (BB), as well as deep learning–based methods including Long Short-Term Memory (LSTM), Informer, AutoFormer, and TimesNet. On the LLM side, we evaluated models from the phi-4, Llama3, and Qwen3 families, applying various prompting strategies such as Chain-of-Thought (CoT), Counterfactual Chain-of-Thought (CCoT), and Role Prompting (RP). The experimental results show that the best performance among numerical strategies was achieved by MACD with an F1 score of 51.22, while the best-performing deep learning method, AutoFormer, reached 56.27. In contrast, the proposed LLM-based approach, employing Counterfactual Chain-of-Thought (CCoT) and Role Prompting (RP), achieved a substantially higher F1 score of 69.03. This confirms that structured reasoning prompts are effective in enhancing predictive performance. Overall, this study empirically demonstrates that LLM-based reasoning strategies can outperform both numerical and deep learning–based approaches in highly volatile cryptocurrency markets and highlights the potential for future research on advanced LLM forecasting systems through prompt engineering.

  Statistics
  Cite this article

[IEEE Style]

H. Jeong and Y. Shin, "A Study on Cryptocurrency Price Movement Prediction Using Prompt Engineering," The Transactions of the Korea Information Processing Society, vol. 14, no. 10, pp. 813-824, 2025. DOI: https://doi.org/10.3745/TKIPS.2025.14.10.813.

[ACM Style]

Hwangil Jeong and Youhyun Shin. 2025. A Study on Cryptocurrency Price Movement Prediction Using Prompt Engineering. The Transactions of the Korea Information Processing Society, 14, 10, (2025), 813-824. DOI: https://doi.org/10.3745/TKIPS.2025.14.10.813.

Forms

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)

Old Journals

Indexing

All publications of TKIPS is indexed in DOI, EBSCO, Google Scholar, Crossref, and CrossCheck.

TKIPS is also selected as the Journal for Accreditation by NRF (National Research Foundation of Korea).

Related Journals