基于分形理论的国际金价波动长记忆性识别及预测研究  

Identification and Prediction Research on the Long-memory Price Fluctuation of International Gold Based on Fractal Market Hypothesis

作　　者：楼晓东[1] 张良[1] 

机构地区：[1]浙江工商职业技术学院,浙江宁波315012

出　　处：《上海金融》2013年第6期80-84,118,共5页Shanghai Finance

摘　　要：分形理论为国际金价定量描述提供了新的研究思路。研究结果表明,识别长记忆性时,V/S方法最贴近实际,计算出的Hurst指数证实了国际黄金现货(周线/月线)存在着显著的长记忆性,表明黄金市场不是弱势有效的,因此在国际金价预测中运用统计分析是有效的;随后通过分数差分将长记忆识别与分形预测模型有机联结了起来,构建的ARFIMA、FIGARCH与ARFIMA-GARCH等模型能够很好地刻画国际金价的内在波动规律,具有良好的定量预测功能。The fractal market hypothesis provides a new research method for quantitative description of the international gold price.The result shows that the V/S method is more practical while identifying the long-memory volatility.The Hurst index demonstrates that there exists prominent lone-memory volatility in the international gold stock(weekly chart/ monthly chart),meaning the gold market is not weak and efficient.Therefore,it is effective to apply statistic analysis in the prediction of international gold price.Additionally,by combining the long-memory identification and the fractal prediction model with fractional difference,the article constructs models such as ARFIMA,FIGARCH,and ARFIMA-GARCH,which show the fluctuation pattern of the international gold price well and function well in quantitative price prediction.

关 键 词：长记忆性 分形理论 R S分析 ARFIMA模型 

分 类 号：F830.9[经济管理—金融学]