Yasemin Ulu
Department of Economics, Eastern Michigan University, Ypsilanti, MI 48197, USA.
*Corresponding author: Yasemin Ulu
Abstract
In this study, we construct hybrid models that are based on the combination of different Deep Learning models like the Long-Short-Term Memory Model (LSTM), Bi-Directional Long-Short Term Memory Model (BiLSTM), and the conventional GARCH model. The aim is to forecast the volatility of stocks traded in the financial, transportation, communication, and petroleum sectors, in the BIST30, Turkish Stock Market Index. Specifically, we construct (BiLSTM) and Long-(LSTM) models that utilize forecasts from conventional GARCH models to forecast one day a headstock volatility of the stocks for the sectors considered. We use Mean Absolute Percentage Error (MAPE) and Symmetric Mean Absolute Percentage Error (SMAPE) as forecast evaluation criteria. The period we consider covers the COVID-19 crisis, allowing for further comparison. Although the forecasting performance of the models uniformly seems to be lower for the COVID-19 period judged by forecast evaluation criteria, we find that hybrid models that utilize deep learning and GARCH forecasts perform better in forecasting the volatility of the stocks considered. We highly recommend utilizing the hybrid Deep Learning BiLSTM-GARCH model to forecast the volatility of the stocks in the corresponding sectors considered.
References
[1] Sezer OB, Gudelek MU, Ozbayoglu AM. Financial time series forecasting with deep learning: A systematic literature review: 2005-2019. Applied Soft Computing. 2020;90:106181.
[2] Wang W, Li W, Zhang N, Liu K. Portfolio formation with preselection using deep learning from long-term financial data. Expert Systems with Applications. 2019.
[3] Gao T, Chai Y. Improving Stock Closing Price Prediction Using Recurrent Neural Network and Technical Indicators. Neural Computation. 2018;30(10):2833-2854.
[4] Vo NN, He X, Liu S, Xu G. Deep learning for decision making and the optimization of socially responsible investments and portfolio. Decision Support Systems. 2019;124:113097.
[5] Sambas A, He S, Liu H, Vaidyanathan S, Hidayat Y, Saputra J. Dynamical analysis and adaptive fuzzy control for the fractional-order financial risk chaotic system. Advances in Difference Equations. 2020;674(1):1-12.
[6] Min HP, Dongyan N, Yerin K, Jang HK. CBOE Volatility Index Forecasting under COVID-19: An Integrated BiLSTM-ARIMA-GARCH Model. Computer Systems Science and Engineering. 2023;47(1).
[7] Duan Y, Liu L., Wang Z. COVID-19 sentiment and the Chinese stock market: Evidence from the official news media and Sina Weibo. Research in International Business and Finance. 2021;58(4):101432.
[8] Kim HY, Won C. H. Forecasting the volatility of stock price index: A hybrid model integrating LSTM with multiple GARCH-type models. Expert Systems with Applications. 2018;103(1):25-37.
How to cite this paper
Forecasting Stock Volatility via Hybrid Deep Learning and GARCH Family Models: A Case Study from BIST30
How to cite this paper: Yasemin Ulu. (2024) Forecasting Stock Volatility via Hybrid Deep Learning and GARCH Family Models: A Case Study from BIST30. Journal of Applied Mathematics and Computation, 8(4), 280-285.
DOI: http://dx.doi.org/10.26855/jamc.2024.12.001