Journal of the Korea Institute of Information and Communication Engineering (한국정보통신학회논문지)

The Korea Institute of Information and Commucation Engineering (한국정보통신학회)
권호 표지
DOI QR코드

DOI QR Code

Utilizing On-Chain Data to Predict Bitcoin Prices based on LSTM

On-Chain Data를 활용한 LSTM 기반 비트코인 가격 예측

  • An, Yu-Jin (Department of Fintech, Sungkyunkwan University) ;
  • Oh, Ha-Young (College of Computing & Informatics, Sungkyunkwan University)
  • Received : 2021.05.26
  • Accepted : 2021.07.09
  • Published : 2021.10.31
Download PDF
⟨ Previous Next ⟩

Abstract

During the past decade, it seems apparent that Bitcoin has been the best performing asset class. Even without a centralized authority that takes control over, Bitcoin, which started off with basically no value at all, reached around 65000 dollars in 2021, showing a movement that will definitely go down in history. Thus, even those who were skeptical of Bitcoin's intangible nature are stacking bitcoin as a huge part of their portfolios. Bitcoin's exponential growth in value also caught the attention of traditional banking and investment firms. Along with the spotlight Bitcoin is getting from the investment world, research using macro-economic variables and investor sentiment to explain Bitcoin's price movement has shown progress. However, previous studies do not make use of On-Chain Data, which are data processed using transaction data in Bitcoin's blockchain network. Therefore, in this paper, we will be utilizing LSTM, a method widely used for time-series data prediction, with On-Chain Data to predict the price of Bitcoin.

최근 10여 년 동안 가장 가파르게 가치가 상승한 자산군을 꼽자면 단연 비트코인이라고 할 수 있을 것이다. 특히 비트코인은 중앙통제 기관이 없음에도 불구하고 첫 등장을 한 2009년의 사실상 0달러에서 2021년 최고점인 65,000 달러 수준까지 치솟아 역사에 길이 남을 가치 상승을 보여주었다. 이에 따라 비트코인의 가능성에 대해서 반신반의 했던 상당수 투자자들의 포트폴리오에도 비트코인이 상당한 비중을 차지하는 경우가 많아졌으며, 제도권 내의 금융권에서도 이런 비트코인의 움직임에 주목하고 있다. 비트코인에 대한 관심과 더불어 비트코인의 가격에 거시경제 변수나 센티멘트가 비트코인의 가격이 어떻게 움직이는가에 대한 연구 또한 상당히 진전되었다. 하지만, 이들 연구에서 활용한 변수들은 비트코인만의 특징적인 데이터라고 할 수 있는 블록체인 내의 데이터를 취합하여 가공한 온체인 데이터를 적극적으로 활용하지는 않았다. 따라서, 본 논문에서는 시계열 데이터 예측에 적극적으로 활용되고 있는 LSTM을 기반으로 온체인 데이터를 활용하여 비트코인의 가격을 예측해보고자 한다.

Keywords

Acknowledgement

I would like to thank the Department of Fintech, Sungkyunkwwan University (SKKU) for providing me with the environment necessary to accomplish this research. Special thanks should be given to CryptoQuant for giving me the opportunity to use its state-of-the-art data. This work was supported by the BK21 FOUR Project. Following are results of a study on the "Convergence and Open Sharing System" Project, supported by the Ministry of Education and National Research Foundation of Korea

References

  1. T. Bradshaw. (2021, May). Apple ad for alternative pay ments job signals cryptocurrency interest. Financial Times. [Internet]. Available: https://www.ft.com/content/2b30e69d-6662-40e8-9735-eaea8662906e.
  2. A. Aggarwal, I. Gupta, N. Garg, and A. Goel, "Deep learning approach to determine the impact of socio economic factors on bitcoin price prediction," 2019 Twelfth International Conference on Contemporary Computing (IC3), pp. 1-5, 2019. doi: 10.1109/IC3.2019.8844928.
  3. S. Pyo and J. Lee, "Do FOMC and macroeconomic announcements affect bitcoin prices?," Finance Research Letters, vol. 37, 2020.
  4. J. Lee, K. Kim, and D. Park, "Empirical analysis on bitcoin price change by consumer, industry and macro-economy variables," Journal of Intelligence and Information Systems, vol. 24, no. 2, pp. 195-220, 2018. https://doi.org/10.13088/JIIS.2018.24.2.195
  5. K. Lee, S. Cho, G. Min, and C. Yang, "The determinant of bitcoin prices in Korea," Korean Journal of Financial Studies, vol. 47, no. 4, pp. 393-415, 2019.
  6. E. Kim and J. Hong, "The prediction model of cryptocurrency price using news sentiment analysis and deep learning," Korea Society of IT Services, 2020 Fall Conference, pp. 122-126, 2020.
  7. B. Jafar. (2021, April). Bitcoin is not a bubble, BTC is like digital gold, says Bill Miller. Finance Magnates | Financial and business news [Internet]. Available: https://www.financemagnates.com/cryptocurrency/news/bitcoin-is-not-a-bubble-btc-is-like-digital-gold-says-bill-miller/.
  8. S. McNally, J. Roche, and S. Caton, "Predicting the price of bitcoin using machine learning," 2018 26th Euromicro International Conference on Parallel, Distributed and Network-based Processing (PDP), pp. 339-343, 2018.
  9. Z. Chen, C. Li, and W. Sun, "Bitcoin price prediction using machine learning: an approach to sample dimension engineering," Journal of Computational and Applied Mathematics, vol. 365, 2020.
  10. CryptoQuant. CryptoQuant Data API (1.3.0) [Internet]. Available: https://cryptoquant.com/docs.
  11. Y. Ng. (2019, January). Machine learning techniques applied to stock price prediction [Internet]. Available: https://towardsdatascience.com/machine-learning-techniques-applied-to-stock-price-prediction-6c1994da8001.