Volatility forecasting using deep recurrent neural networks as GARCH models

Di-Giorgi, Gustavo; Salas, Rodrigo; Avaria, Rodrigo; Ubal, Cristian; Rosas, Harvey; Torres, Romina

doi:10.1007/s00180-023-01349-1

Cited by 3 publications

(4 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…where f (λ) is defined as in ( 2) and is associated with the parameter set Θ of the ARFIMA model defined in (1). The log-likelihood function of the process Y is provided by…”

Section: Methodsmentioning

confidence: 99%

“…Time series analysis and forecasting are essential in many areas of application, such as finance and marketing [1], air pollution [2], electricity consumption [3], and weather forecasting [4,5], among others. However, selecting the appropriate model strongly depends on the degree of predictability of the time series [6].…”

Section: Introductionmentioning

confidence: 99%

“…The Hurst exponent was used to measure the long-term dependency of time series, and LCNN was found to improve classification performance and to be suitable for small datasets. Recently, Di-Giorgi et al [1] proposed the application of deep recurrent neural networks for volatility forecasting as GARCH models. However, the works mentioned above have not addressed the need to create a method that can accurately and efficiently estimate and predict the time-varying long memory index of the time series.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Predicting the Long-Term Dependencies in Time Series Using Recurrent Artificial Neural Networks

Ubal,

Di-Giorgi,

Contreras-Reyes

et al. 2023

MAKE

Self Cite

View full text Add to dashboard Cite

Long-term dependence is an essential feature for the predictability of time series. Estimating the parameter that describes long memory is essential to describing the behavior of time series models. However, most long memory estimation methods assume that this parameter has a constant value throughout the time series, and do not consider that the parameter may change over time. In this work, we propose an automated methodology that combines the estimation methodologies of the fractional differentiation parameter (and/or Hurst parameter) with its application to Recurrent Neural Networks (RNNs) in order for said networks to learn and predict long memory dependencies from information obtained in nonlinear time series. The proposal combines three methods that allow for better approximation in the prediction of the values of the parameters for each one of the windows obtained, using Recurrent Neural Networks as an adaptive method to learn and predict the dependencies of long memory in Time Series. For the RNNs, we have evaluated four different architectures: the Simple RNN, LSTM, the BiLSTM, and the GRU. These models are built from blocks with gates controlling the cell state and memory. We have evaluated the proposed approach using both synthetic and real-world data sets. We have simulated ARFIMA models for the synthetic data to generate several time series by varying the fractional differentiation parameter. We have evaluated the proposed approach using synthetic and real datasets using Whittle’s estimates of the Hurst parameter classically obtained in each window. We have simulated ARFIMA models in such a way that the synthetic data generate several time series by varying the fractional differentiation parameter. The real-world IPSA stock option index and Tree Ringtime series datasets were evaluated. All of the results show that the proposed approach can predict the Hurst exponent with good performance by selecting the optimal window size and overlap change.

show abstract

“…where f (λ) is defined as in ( 2) and is associated with the parameter set Θ of the ARFIMA model defined in (1). The log-likelihood function of the process Y is provided by…”

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Predicting the Long-Term Dependencies in Time Series Using Recurrent Artificial Neural Networks

Ubal,

Di-Giorgi,

Contreras-Reyes

et al. 2023

MAKE

Self Cite

View full text Add to dashboard Cite

show abstract

“…Как доказано в [16], результаты сравнения полученных прогнозов показывают, что гибридные модели с GDT явно превосходят прогнозируемые результаты с моделями семейства GARCH. Аналогичные выводы представлены в работах [17][18][19][20]. Считаем перспективным направлением разработку моделей прогнозирования волатильности российского биржевого рынка акций с использованием искусственных нейронных сетей, созданных российскими разработчиками.…”

Section: выводыunclassified

Forecasting the Volatility of the Russian Stock Market in the Context of International Economic Sanctions

Glebova,

Kovaleva

2024

jour

View full text Add to dashboard Cite

The article is devoted to the study of trends in the development of the Russian stock market in the context of international economic sanctions. The purpose of the study is to make forecasts of the volatility of the Russian stock market using a scenario approach. Statistical data of the Moscow Stock Exchange were used for calculations. The authors have made a forecast of the volatility of the stock exchange market of the Russian Federation. The basis of the forecast calculations is the dynamics of the Moscow Exchange Index (IMOEX), taken as a key indicator of the Russian organized securities market, for the period from June 2013 to July 2022. Based on the basic historical dynamics of the Moscow Stock Exchange Index, negative (international economic sanctions are being tightened) and positive (implies the easing and/or lifting of some sanctions) scenarios for the development of the stock market of the Russian Federation are compiled. The scientific novelty is the authors’ assessment of the convergence of the volatility forecast under negative and positive scenarios to a certain level of volatility in 2023. The results of the calculations showed that under different scenarios of the situation, volatility tends to the same value at different assumed values of the Moscow Exchange Index, which allowed us to draw a new and practically significant conclusion that over time the economy of the Russian Federation stabilizes regardless of the tightening or easing of international economic sanctions — this may be due to the implementation of the country has a policy of import substitution, the formation of national production in most areas of the economy and the development of the domestic market. The work carried out by the authors contributes to the development of theoretical and applied economics in terms of making forecasts for the development of the stock market and using the results of forecasting to make economically sound decisions.

show abstract

Dynamics in Realized Volatility Forecasting: Evaluating GARCH Models and Deep Learning Algorithms Across Parameter Variations

Akgun,

Gulay

2024

Comput Econ

View full text Add to dashboard Cite

The modeling and forecasting of return volatility for the top three cryptocurrencies, which are identified by the highest trading volumes, is the main focus of the study. Eleven different GARCH-type models were analyzed using a comprehensive methodology in six different distributions, and deep learning algorithms were used to rigorously assess each model’s forecasting performance. Additionally, the study investigates the impact of selecting dynamic parameters for the forecasting performance of these models. This study investigates if there are any appreciable differences in forecast outcomes between the two different realized variance calculations and variations in training size. Further investigation focuses on how the use of expanding and rolling windows affects the optimal window type for forecasting. Finally, the importance of choosing different error measurements is emphasized in the framework of comparing forecasting performances. Our results indicate that in GARCH-type models, 5-minute realized variance shows the best forecasting performance, while in deep learning models, median realized variance (MedRV) has the best performance. Moreover, it has been determined that an increase in the training/test ratio and the selection of the rolling window approach both play important roles in achieving better forecast accuracy. Finally, our results show that deep learning models outperform GARCH-type models in volatility forecasts.

show abstract

Volatility forecasting using deep recurrent neural networks as GARCH models

Cited by 3 publications

References 31 publications

Predicting the Long-Term Dependencies in Time Series Using Recurrent Artificial Neural Networks

Predicting the Long-Term Dependencies in Time Series Using Recurrent Artificial Neural Networks

Forecasting the Volatility of the Russian Stock Market in the Context of International Economic Sanctions

Dynamics in Realized Volatility Forecasting: Evaluating GARCH Models and Deep Learning Algorithms Across Parameter Variations

Contact Info

Product

Resources

About