摘要
碳排放交易市场中不同期货价格波动及其相互影响较为复杂,价格趋势预测也在金融投资领域占有重要地位。针对碳交易市场中非线性预测问题,选取欧盟配额期货与碳排放核证减排量期货相关参数作为研究对象,运用协整关系检验确定其是否具有长期协整关系,采用Granger因果检验确定其领先滞后关系,将具有领先关系的期货参数作为部分输入变量,建立遗传算法改进的运用不同小波函数的神经网络模型,对具有滞后关系的期货价格趋势进行预测,并与改进前的BP小波神经网络模型预测结果进行对比。实验结果表明,碳排放交易市场中期货价格之间存在长期均衡协整关系,改进的模型可以有效刻画期货价格序列变化趋势,为碳排放交易提供良好的投资建议。
At carbon trading market,different futures prices volatility and their interaction is complicated. Price trend forecasting plays a key part in financial investment. This paper,aiming at the non-linear forecasting issues in carbon trading market,selects EUA futures and CER futures to study their long term cointegration relation,uses Granger causality test to determine its leading or lagging. The leading futures as input variable was used to set up neural network model of difference wavelet function that is improved by genetic algorithms. The lagging futures is forecasted and compared with neural network＇s forecasted results from pre-improved BP wavelet neural network model. The results show a long term balanced cointegration of futures in carbon trading market. The improved model can effectively forecast the trend of futures,proving investment suggestions for carbon emission trading.
出处
《资源与产业》
2016年第2期111-120,共10页
Resources & Industries
基金
国家自然科学基金(71173199)
关键词
时间序列预测
遗传算法
小波神经网络
GRANGER因果检验
碳市场
temporal series forecast
genetic algorithms
wavelet neural network
Granger causality test
carbon trading market
