Commodity futures price forecast based on multi-scale combination model

Yijia Liu, Yukun Gao, Yufeng Shi, Yuxue Zhang, Li Li and Qimeng Han
Additional contact information
Yijia Liu: Institute for Financial Studies, Shandong University, Jinan 250100, Shandong, P. R. China
Yukun Gao: University of Toronto, Toronto, Ontario, Canada M5S 1A1, Canada
Yufeng Shi: Institute for Financial Studies and Shandong Province Key Laboratory of Financial Risk, Shandong University, Jinan 250100, Shandong, P. R. China
Yuxue Zhang: Institute for Financial Studies, Shandong University, Jinan 250100, Shandong, P. R. China
Li Li: Zhongtai Futures Co., Ltd, Jinan 250001, Shandong, P. R. China
Qimeng Han: Zhongtai Futures Co., Ltd, Jinan 250001, Shandong, P. R. China

International Journal of Financial Engineering (IJFE), 2022, vol. 09, issue 04, 1-32

Abstract: Along with developing the commodity futures market, its promoting effect on China’s economic development has gradually increased. Research on the price prediction of commodity futures has important practical significance to society and enterprises. However, commodity futures price series often show nonstationary and nonlinear characteristics In this paper, a new multi-scale combined prediction model is proposed, which combines variational mode decomposition (VMD), long short-term memory neural network (LSTM), and improved self-attention mechanism (XNSA). First, VMD decomposes futures prices into several components to reduce their nonstationarity. Then, the LSTM model with an improved self-attention mechanism (XNSA) is used to model and optimize the decomposed sub-sequences so that the model can concentrate on learning more important data features and further improve the prediction performance. In order to verify the effectiveness of this method, this paper takes No. 1 Soybeans Futures, Corn Futures, and Soybean Meal Futures daily closing price series from Dalian Commodity Exchange as representatives to predict their future return trend. The results show that compared with the existing combination forecasting models, the proposed multi-scale combination model (VMD-LSTM-XNSA) has better forecasting performance. It lays the foundation for developing a corresponding quantitative investment strategy.

Keywords: Commodity futures price forecast; variational mode decomposition; long short-term memory neural network; improved self-attention mechanism (search for similar items in EconPapers)
Date: 2022
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.worldscientific.com/doi/abs/10.1142/S2424786322500311
Access to full text is restricted to subscribers

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:wsi:ijfexx:v:09:y:2022:i:04:n:s2424786322500311

Ordering information: This journal article can be ordered from

DOI: 10.1142/S2424786322500311

Access Statistics for this article

International Journal of Financial Engineering (IJFE) is currently edited by George Yuan

More articles in International Journal of Financial Engineering (IJFE) from World Scientific Publishing Co. Pte. Ltd.
Bibliographic data for series maintained by Tai Tone Lim ().