摘要
自回归移动平均模型(ARMA)及其扩展模型被广泛用于资产价格预测.然而相关已有研究要么忽略了多个资产之间的交叉影响,要么未能合理地分配目标资产自回归项与其他资产自回归项的权重.提出了一种新的多元交叉平衡ARMA模型(CB-ARMA),模型合理地平衡了目标资产与其他资产的自回归项的权重,并利用极大似然法估计CB-ARMA的参数矩阵.的实证研究基于中国股指期货30分钟价格序列数据,并使用滑动窗口在线预测方法来模拟投资.实证结果表明,提出的CB-ARMA模型优于所有对比模型.
Autoregressive Moving Average model(ARMA) and its extended models are widely used for asset prices prediction.However,previous researches either ignored the cross terms which measure the cross impacts between the target asset and other assets,or failed to deal with the weights of the cross terms reasonably.This paper proposes a new cross balanced ARMAmodel(CB-ARMA) that gives proper attention to both the weights of the autoregressive terms and the cross terms.A maximum likelihood method is applied to estimate the parameter matrices ofCB-ARMA.The empirical study is based on the 30-minute price sequences of Chinese stock index futures.We simulate the real-world investments using a sliding-window online prediction approach.The results show that our proposed model outperforms all the baseline models.
作者
付志能
徐维军
罗艺旸
刘勇平
FU Zhi-neng;XU Wei-jun;LUO Yi-yang;LIU Yong-ping(School of Business Administration,South China University of Technology,Guangzhou 510641,China;School of Economics and Management,South China Normal University,Guangzhou,510631,China;Risk Management Department,Guangzhou Rural Commercial Bank,Guangzhou 510663,China)
出处
《数学的实践与认识》
2021年第11期91-102,共12页
Mathematics in Practice and Theory
基金
国家自然科学基金资助项目(71771091)
国家自然科学基金国际(地区)合作与交流重点项目(71720107002)
科技部科技创新2030-“新一代人工智能”重大项目(2020AAA0108404)
广东省基础与应用基础研究基金(2019A1515011752)。
关键词
股指期货
ARMA
交叉项
多元
极大似然估计
stock index futures
ARMA
cross terms
multivariate
maximum likelihood estimation
