umovestСтатистика и ЭкономикаStatistics and Economics2500-3925Plekhanov Russian University of Economics10.21686/2500-3925-2017-3-4-9umovest-1140Research ArticleМЕТОДОЛОГИЯ СТАТИСТИКИMETHODOLOGY OF STATISTICSПрогнозирование рядов динамики рыночных индикаторов на основе нелинейной авторегрессионной нейронной сетиForecasting time series of the market indicators based on a nonlinear autoregressive neural networkДомащенкоД. В.DomashchenkoDenis V.

Кандидат экономических наук, Доцент

Cand. Sci. (Economics)

dendv@rambler.ruНикулинЭ. Е.NikulinEdvard E.edvardnikulin@gmail.comРоссийский экономический университет имени Г.В. ПлехановаРоссияPlekhanov Russian University of EconomicsRussian Federation2017120720170349Copyright © Домащенко Д.В., Никулин Э.Е., 20172017Домащенко Д.В., Никулин Э.Е.Domashchenko D.V., Nikulin E.E.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://statecon.rea.ru/jour/article/view/1140

Современная практика экономических исследований активно полагается на математические модели, позволяющие выявлять в статистических данных скрытые закономерности и строить на их основании прогнозы. Линейные модели прогнозирования рядов динамики, основанные на векторной авторегрессии (VAR) являются наиболее распространенными. Однако связи между рядами динамики в экономике часто имеют сложно идентифицируемый характер, поэтому нелинейные авторегрессионные (NAR) модели показывают более достоверные результаты. Для их реализации обычно используются нейронные сети, которые не предоставляют возможности оценки прогноза в виде математического ожидания и стандартного отклонения. Поэтому предлагаемая в статье модель сочетает в себе два блока: VAR и NAR. NAR используется для построения прогноза на заданное количество точек, а VAR для оценки прогноза в виде математического ожидания и стандартного отклонения. Оценка достоверности модели проводилась на дневных данных валютного курса USD/RUB и цен на нефть марки «Брент» с 1.01.2016 по 1.03.2017. Средняя точность прогнозирования тренда для курса доллара США к рублю составила 54,9%, для цены нефти – 54,0%. При этом относительная ошибка прогнозирования курса доллара составила от 1,09% (для первой точки) до 2,01% (для десятой точки), относительная ошибка прогнозирования цен на нефть составила от 1,28% (для первой точки) до 4,58% (для десятой точки). Таким образом, модель представляет достаточно точные для принятия инвестиционных решений прогнозы, при этом производится оценка прогнозов на основании тестирования NAR блока на исторических данных и на основании прогноза VAR блока в форме математического ожидания и стандартного отклонения.

The modern practice of economic research relies heavily on mathematical models that make it possible to reveal hidden regularities in statistical data and make forecasts on their basis. Linear models based on vector autoregression (VAR) are the most common. However, the relationship between time series in the economy is often difficult to identify, so non-linear autoregressive (NAR) models show more reliable results. Artificial neural networks (ANNs) are usually used for implementation of these models, but ANNs do not provide the possibility of estimating the forecast in the form of mathematical expectation and a standard deviation. Therefore, the model proposed in the article combines two blocks: VAR and NAR. NAR is used to construct a prediction for a given number of points, and VAR for estimating the forecast in the form of a mathematical expectation and a standard deviation. The evaluation of the model was carried out on the daily data: exchange rate USD / RUB and “Brent” oil from 1.01.2016 to 1.03.2017. The average accuracy of forecasting the trend for the dollar was 54.9%, for the oil prices it was 54.0%. In this case, the relative error in predicting the dollar rate was from 1.09% (for the first point) to 2.01% (for the tenth point); the relative error in forecasting oil prices was from 1.28% (for the first point) to 4.58 % (for the tenth point). Thus, the model showed accurate results when predicting dynamic series and can be used to solve other forecasting problems. In particular, it is expedient to use the model as one of the factors when making investment decisions. In addition, the evaluation of forecasts is done on the basis of testing the NAR block of historical data and on the basis of VAR block forecast in the form of mathematical expectation and standard deviation.

искусственная нейронная сетьвекторная авторегрессияпрогнозированиекурс долларацены на нефтьartificial neural networkvector autoregressionforecastingexchange rate USD/RUBoil priceReferencesSims C. A. Macroeconomics and reality // Econometrica: Journal of the Econometric Society. 1980. P. 1–48.Sims C. A. Macroeconomics and reality // Econometrica: Journal of the Econometric Society. 1980. P. 1–48.Johansen S. Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models // Econometrica: Journal of the Econometric Society. 1991. P. 1551–1580.Johansen S. Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models // Econometrica: Journal of the Econometric Society. 1991. P. 1551–1580.Box G. E. P., Jenkins G. M. Time Series Models for Forecasting and Control // San Francisco. 1970.Box G. E. P., Jenkins G. M. Time Series Models for Forecasting and Control // San Francisco. 1970.Engle R. F. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation // Econometrica: Journal of the Econometric Society. 1982. P. 987–1007.Engle R. F. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation // Econometrica: Journal of the Econometric Society. 1982. P. 987–1007.Zhang X. and Frey R. Improving ARMAGARCH forecasts for high frequency data with regime-switching ARMA-GARCH // Journal of Computational Analysis & Applications. 2015. 18(1).Zhang X. and Frey R. Improving ARMAGARCH forecasts for high frequency data with regime-switching ARMA-GARCH // Journal of Computational Analysis & Applications. 2015. 18(1).Kambouroudis D.S., McMillan D.G. and Tsakou K. Forecasting Stock Return Volatility: A Comparison of GARCH, Implied Volatility, and Realized Volatility Models // Journal of Futures Markets. 2016. 36(12). P. 1127–1163.Kambouroudis D.S., McMillan D.G. and Tsakou K. Forecasting Stock Return Volatility: A Comparison of GARCH, Implied Volatility, and Realized Volatility Models // Journal of Futures Markets. 2016. 36(12). P. 1127–1163.Corrêa J.M., Neto A.C., Júnior L.T., Franco E.M.C. and Faria A.E. Time series forecasting with the WARIMAX-GARCH method. Neurocomputing. 2016. P. 805–815.Corrêa J.M., Neto A.C., Júnior L.T., Franco E.M.C. and Faria A.E. Time series forecasting with the WARIMAX-GARCH method. Neurocomputing. 2016. P. 805–815.Leontaritis I. J., Stephen A. Billings. Inputoutput parametric models for non-linear systems part I: deterministic non-linear systems. // International journal of control. 1985. 41.2. P. 303–328.Leontaritis I. J., Stephen A. Billings. Inputoutput parametric models for non-linear systems part I: deterministic non-linear systems. // International journal of control. 1985. 41.2. P. 303–328.Chen S., Billings S. A. Representations of non-linear systems: the NARMAX model // International Journal of Control. 1989. Т. 49. No. 3. P. 1013–1032.Chen S., Billings S. A. Representations of non-linear systems: the NARMAX model // International Journal of Control. 1989. Т. 49. No. 3. P. 1013–1032.Darrat A. F., Zhong M. On testing the randomwalk Hypothesis: A model-comparison approach // Financial Review. 2000. Т. 35. No. 3. P. 105–124.Darrat A. F., Zhong M. On testing the randomwalk Hypothesis: A model-comparison approach // Financial Review. 2000. Т. 35. No. 3. P. 105–124.Jiang C. and Song F. Sunspot Forecasting by Using Chaotic Time-series Analysis and NARX Network // JCP. 2011. 6(7). P. 1424–1429.Jiang C. and Song F. Sunspot Forecasting by Using Chaotic Time-series Analysis and NARX Network // JCP. 2011. 6(7). P. 1424–1429.Diaconescu E. The use of NARX neural networks to predict chaotic time series // Wseas Transactions on computer research. 2008. 3(3). P.182–191.Diaconescu E. The use of NARX neural networks to predict chaotic time series // Wseas Transactions on computer research. 2008. 3(3). P.182–191.Chaudhuri T.D. and Ghosh I. Artificial Neural Network and Time Series Modeling Based Approach to Forecasting the Exchange Rate in a Multivariate Framework // arXiv preprint. 2016.arXiv:1607.02093Chaudhuri T.D. and Ghosh I. Artificial Neural Network and Time Series Modeling Based Approach to Forecasting the Exchange Rate in a Multivariate Framework // arXiv preprint. 2016. arXiv:1607.02093Akaike H. Maximum likelihood identification of Gaussian autoregressive moving average models // Biometrika. 1973. P. 255–265.Akaike H. Maximum likelihood identification of Gaussian autoregressive moving average models // Biometrika. 1973. P. 255–265.Adkinson M. D. et al. Alternative models of climatic effects on sockeye salmon, Oncorhynchus nerka, productivity in Bristol Bay, Alaska, and the Fraser River, British Columbia // Fisheries Oceanography. 1996. Т. 5. No. 3–4. P. 137–152.Adkinson M. D. et al. Alternative models of climatic effects on sockeye salmon, Oncorhynchus nerka, productivity in Bristol Bay, Alaska, and the Fraser River, British Columbia // Fisheries Oceanography. 1996. Т. 5. No. 3–4. P. 137–152.

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