Artificial Neural Networks for Realized Volatility Prediction in Cryptocurrency Time Series

  • Ryotaro Miura
  • Lukáš PichlEmail author
  • Taisei Kaizoji
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 11554)


Realized volatility (RV) is defined as the sum of the squares of logarithmic returns on high-frequency sampling grid and aggregated over a certain time interval, typically a trading day in finance. It is not a priori clear what the aggregation period should be in case of continuously traded cryptocurrencies at online exchanges. In this work, we aggregate RV values using minute-sampled Bitcoin returns over 3-h intervals. Next, using the RV time series, we predict the future values based on the past samples using a plethora of machine learning methods, ANN (MLP, GRU, LSTM), SVM, and Ridge Regression, which are compared to the Heterogeneous Auto-Regressive Realized Volatility (HARRV) model with optimized lag parameters. It is shown that Ridge Regression performs the best, which supports the auto-regressive dynamics postulated by HARRV model. Mean Squared Error values by the neural-network based methods closely follow, whereas the SVM shows the worst performance. The present benchmarks can be used for dynamic risk hedging in algorithmic trading at cryptocurrency markets.


ANN MLP LSTM GRU CNN SVM HARRV Ridge regression Realized volatility 


  1. 1.
    Andersen, T.G., Bollerslev, T., Diebold, F., Labys, P.: Modeling and forecasting realized volatility. Econometrica 71, 579–625 (2003)Google Scholar
  2. 2.
    Kaggle, Bitcoin historical data. Accessed 1 Dec 2018. Released under CC BY-SA 4.0 license
  3. 3.
    Moews, B., Herrmann, J.M., Ibikunle, G.: Lagged correlation-based deep learning for directional trend change prediction in financial time series. Expert Syst. Appl. 120, 197–206 (2019)Google Scholar
  4. 4.
    Cao, J., Li, Z., Li, J.: Financial time series forecasting model based on CEEMDAN and LSTM. Phy. A: Stat. Mech. Appl. 519, 127–139 (2019)Google Scholar
  5. 5.
    Mallqui, D.C.A., Fernandes, R.A.S.: Predicting the direction, maximum, minimum and closing prices of daily Bitcoin exchange rate using machine learning techniques. Appl. Soft Comput. 75, 596–606 (2019)Google Scholar
  6. 6.
    Lahmiri, S., Bekiros, S.: Cryptocurrency forecasting with deep learning chaotic neural networks. Chaos, Solitons Fractals 118, 35–40 (2019)Google Scholar
  7. 7.
    Nakano, M., Takahashi, A., Takahashi, S.: Bitcoin technical trading with artificial neural network. Phys. A: Stat. Mech. Appl. 510, 587–609 (2018)Google Scholar
  8. 8.
    Rosenblatt, F.: Principles of Neurodynamics Perceptrons and the Theory of Brain Mechanisms. Spartan Books, Washington (1961)Google Scholar
  9. 9.
    LeCun, Y., et al.: Back-propagation applied to handwritten zip code recognition. Neural Compu. 1(4), 541–551 (1989)Google Scholar
  10. 10.
    Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9(8), 1735–1780 (1997)Google Scholar
  11. 11.
    Cho, K., van Merrienboer, B., Bahdanau, D., Bengio, Y.: On the properties of neural machine translation: encoder-decoder approaches. In: 8th Workshop on Syntax. Semantics and Structure in Statistical Translation, pp. 102–111. Association for Computational Linguistics, Doha (2014)Google Scholar
  12. 12.
    Vapnik, V.N.: The Nature of Statistical Learning Theory. Springer, Heidelberg (1995). Scholar
  13. 13.
    Hoerl, A.E., Kennard, R.W.: Ridge regression: biased estimation for nonorthogonal problems. Technometrics 12, 55–67 (1970)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.International Christian UniversityMitakaJapan

Personalised recommendations