financeФинансы: теория и практика/Finance: Theory and PracticeFinance: Theory and Practice2587-56712587-7089Financial University under The Government of Russian Federation10.26794/2587-5671-2020-24-1-14-23finance-950Research ArticleСОВРЕМЕННЫЕ МЕТОДЫ ИССЛЕДОВАНИЯMODERN RESEARCH METHODSСравнительный анализ прогнозных моделей ARIMA и lSTM на примере акций российских компанийComparative Analysis of ARIMA and lSTM Predictive Models: Evidence from Russian Stockshttps://orcid.org/0000-0001-8944-5679АлжеевА. В.AlzheevA. V.

Андрей Вадимович Алжеев —  студент магистратуры Департамента анализа данных, принятия решений и финансовых технологий

Andrei V. Alzheev —  Master’s student, Department of Data Analysis, Decision Making and Financial Technology

alzheev@gmail.cohttps://orcid.org/0000-0003-3186-3901КочкаровР. А.KochkarovR. A.

Расул Ахматович Кочкаров —  кандидат экономических наук, доцент Департамента анализа данных, принятия решений и финансовых технологий

Rasul A. Kochkarov —  Cand. Sci. (Econ.), Assoc. Prof., Department of Data Analysis, Decision Making and Financial Technology

Финансовый университетРоссияFinancial UniversityRussian Federation2020070320202411423Copyright © Алжеев А.В., Кочкаров Р.А., 20202020Алжеев А.В., Кочкаров Р.А.Alzheev A.V., Kochkarov R.A.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://financetp.fa.ru/jour/article/view/950

The article aims to find the best time series predictive model, considering the minimization of errors and high accuracy of the prediction. The authors performed the comparative analysis of the most popular “traditional” econometric model ARIMA and the deep learning model LSTM (Long short-term memory) based on a recurrent neural network. The study provides a mathematical description of these predictive models. The authors developed algorithms for predicting time series based on the “Rolling forecasting origin” approach. These are Python-based algorithms using the Keras, Theano and Statsmodels libraries. Stock quotes of Russian companies Alrosa, Gazprom, KamAZ, NLMK, Kiwi, Rosneft, VTB and Yandex for the period from June 2, 2014 to November 11, 2019, broken down by week, served as input data. The research results confirm the superiority of the LSTM model, where the RMSE error is 65% less than with the ARIMA model. Therefore, an LSTM model-based algorithm is more preferable for the better quality of time series prediction.

ARIMALSTMпрогнозные моделиакциианализпрогнозирование котировокалгоритмыARIMALSTMpredictive modelsstocksanalysisstock quote predictionalgorithmsРабота выполнена при поддержке гранта РФФИ 18–00–01103. Финансовый университет, МоскваThe article was supported by the RFBR grant 18–00–01103. Financial University, MoscowReferencesМагнус Я. Р., Катышев П. К., Пересецкий А. А. Эконометрика. Начальный курс. М.: Дело; 2007. 504 с.Magnus Ya.R., Katyshev P. K., Peresetsky A. A. Econometrics. Beginner course. Moscow: Delo; 2007. 504 p. (In Russ.).Schmidhuber J. Habilitation. System modeling and optimization. Postdoctoral thesis. Munich: Technical University of Munich; 1993. 209 p. (In German).Schmidhuber J. Habilitation. System modeling and optimization. Postdoctoral thesis. Munich: Technical University of Munich; 1993. 209 p. (In German).Hochreiter S., Schmidhuber J. Long short-term memory. Neural Computation. 1997;9(8):1735–1780. DOI: 10.1162/ neco.1997.9.8.1735Hochreiter S., Schmidhuber J. Long short-term memory. Neural Computation. 1997;9(8):1735–1780. DOI: 10.1162/ neco.1997.9.8.1735Brownlee J. Time series prediction with LSTM recurrent neural networks in Python with Keras. Machine Learning Mastery. 2016. URL: https://machinelearning mastery.com/time-series-prediction-lstm-recurrent-neuralnetworks-python-kerasBrownlee J. Time series prediction with LSTM recurrent neural networks in Python with Keras. Machine Learning Mastery. 2016. URL: https://machinelearning mastery.com/time-series-prediction-lstm-recurrent-neuralnetworks-python-kerasGers F. A., Schmidhuber J., Cummins F. Learning to forget: Continual prediction with LSTM. Neural Computation. 2000;12(10):2451–2471. DOI: 10.1162/089976600300015015Gers F. A., Schmidhuber J., Cummins F. Learning to forget: Continual prediction with LSTM. Neural Computation. 2000;12(10):2451–2471. DOI: 10.1162/089976600300015015Кратович П. В. Нейронные сети и модели ARIMA для прогнозирования котировок. Программные продукты и системы. 2011;(1):95–98.Kratovich P. V. Neural networks and ARIMA models for predicting quotes. Programmnye produkty i sistemy = Software & Systems. 2011;(1):95–98. (In Russ.).Garcia F., Guijarro F., Moya I., Oliver J. Estimating returns and conditional volatility: A comparison between the ARMA-GARCH-M models and the backpropagation neural network. International Journal of Complex Systems in Science. 2012;1(2):21–26.Garcia F., Guijarro F., Moya I., Oliver J. Estimating returns and conditional volatility: A comparison between the ARMA-GARCH-M models and the backpropagation neural network. International Journal of Complex Systems in Science. 2012;1(2):21–26.Krauss C., Do X. A., Huck N. Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on the S&P 500. European Journal of Operational Research. 2017;259(2):689–702. DOI: 10.1016/j.ejor. 2016.10.031Krauss C., Do X. A., Huck N. Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on the S&P 500. European Journal of Operational Research. 2017;259(2):689–702. DOI: 10.1016/j.ejor. 2016.10.031Chung J., Lee D., Seo Y., Yoo C. D. Deep attribute networks. NIPS Workshop on Deep Learning and Unsupervised Feature Learning. 2012;3. URL: http://www.eng.uwaterloo.ca/~jbergstr/files/nips_dl_2012/Paper%2011.pdfChung J., Lee D., Seo Y., Yoo C. D. Deep attribute networks. NIPS Workshop on Deep Learning and Unsupervised Feature Learning. 2012;3. URL: http://www.eng.uwaterloo.ca/~jbergstr/files/nips_dl_2012/Paper%2011.pdfFischer T., Krauss C. Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research. 2018;270(2):654–669. DOI: 10.1016/j.ejor.2017.11.054Fischer T., Krauss C. Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research. 2018;270(2):654–669. DOI: 10.1016/j.ejor.2017.11.054Эконометрика. Елисеева И. И., ред. М.: Финансы и статистика; 2006. 576 с.Eliseeva I. I., ed. Econometrics. Moscow: Finansy i statistika; 2006. 576 p. (In Russ.).Armstrong J. S., Collopy F. Error measures for generalizing about forecasting methods: Empirical comparisons. International Journal of Forecasting. 1992;8(1):69–80. DOI: 10.1016/0169–2070(92)90008-WArmstrong J. S., Collopy F. Error measures for generalizing about forecasting methods: Empirical comparisons. International Journal of Forecasting. 1992;8(1):69–80. DOI: 10.1016/0169–2070(92)90008-WHyndman R. J., Athanasopoulos G. Forecasting: Principles and practice. 2nd ed. Melbourne: OTexts; 2018. 382 p.Hyndman R. J., Athanasopoulos G. Forecasting: Principles and practice. 2nd ed. Melbourne: OTexts; 2018. 382 p.

The authors declare that there are no conflicts of interest present.