Optimal forecast combination based on ensemble empirical mode decomposition for agricultural commodity futures prices

• Author: Saeed Heravi,Shangjuan Wu,Bo Guan,Yongmei Fang
• DOI: http://doi.org/10.1002/for.2665
• Date: 01 September 2020
• Published date: 01 September 2020

RESEARCH ARTICLE

Optimal forecast combination based on ensemble empirical

mode decomposition for agricultural commodity futures

prices

Yongmei Fang

1,2,3

| Bo Guan

3

| Shangjuan Wu

1

| Saeed Heravi

3

1

College of Mathematics and Informatics,

South China Agricultural University,

Guangzhou, China

2

College of Economics and Management,

South China Normal University,

Guangzhou, China

3

Cardiff Business School, University of

Cardiff, Cardiff, UK

Correspondence

Saeed Heravi, Cardiff Business School,

University of Cardiff, Cardiff CF10 3EU,

UK.

Email: heravis@cardiff.ac.uk

Abstract

Improving the prediction accuracy of agricultural product futures prices is

important for investors, agricultural producers, and policymakers. This is to

evade risks and enable government departments to formulate appropriate agri-

cultural regulations and policies. This study employs the ensemble empirical

mode decomposition (EEMD) technique to decompose six different categories

of agricultural futures prices. Subsequently, three models—support vector

machine (SVM), neural network (NN), and autoregressive integrated moving

average (ARIMA)—are used to predict the decomposition components. The

final hybrid model is then constructed by comparing the prediction perfor-

mance of the decomposition components. The predicting performance of the

combination model is then compared with the benchmark individual models:

SVM, NN, and ARIMA. Our main interest in this study is on short-term fore-

casting, and thus we only consider 1-day and 3-day forecast horizons. The

results indicate that the prediction performance of the EEMD combined model

is better than that of individual models, especially for the 3-day forecasting

horizon. The study also concluded that the machine learning methods out-

perform the statistical methods in forecasting high-frequency volatile compo-

nents. However, there is no obvious difference between individual models in

predicting low-frequency components.

KEYWORDS

forecast combination, future prices, hybrid model, support vector machine

1|INTRODUCTION

The first standardized futures contract in China was

made in|May 1993 for wheat. There are currently around

20 categories of agricultural futures listed in China, with

a trading volume of 978 million and turnover of 34.89

trillion yuan.

The Chinese agricultural futures market has a signifi-

cant impact on the world futures market, with a market

share of 58% in trading volume in the global agricultural

market in 2011. As reported in the United States Futures

Association in 2014, half of the top 20 trading volume of

agricultural futures and options products are from China.

Among them, vegetables, soybeans, sugar, natural rubber,

and palm oil ranked in the top five products and soybean

oil, eggs, cotton, yellow soybeans, and rapeseed oil were

listed 7th, 9th, 10th, 13th, and 18th, respectively. The

trading volume of the above 10 categories amounted to

more than 940 million, which is approximately 70% of

the total trading volume of global agricultural futures and

Received: 1 October 2019 Accepted: 12 January 2020

DOI: 10.1002/for.2665

Journal of Forecasting. 2020;39:877–886. wileyonlinelibrary.com/journal/for © 2020 John Wiley & Sons, Ltd. 877