基于加权损失函数下广义指数预报因子模型的汇率预测  

Foreign Exchange Rates Prediction Based on Generalized Exponential Predictor Models with Weighted Loss Function

作　　者：尹伟[1] 严威[1] 缪柏其[1] 

机构地区：[1]中国科学技术大学统计与金融系,安徽合肥230026

出　　处：《数理统计与管理》2012年第5期799-804,共6页Journal of Applied Statistics and Management

摘　　要：本文提出在加权损失函数下构建汇率预测的广义指数预报因子模型。该方法首先选取有限个不同滑动参数构造指数预报因子,同时基于绝对值损失和平方损失的提出加权损失函数作为变量筛选的准则,然后在该准则下将指数预报因子进行线性组合,建立汇率预报的广义指数预报因子模型。本文最后用英镑/美元单周汇率数据与文献中的一些已有方法做比较,实证分析表明本文提出的方法在汇率预测效果上有较大改进。The genenralized exponential predictor models for exchange rate forecasting based on weighted loss function is proposed.This method construct some exponential predictors through different smoothing parameters firstly, and then the weighted loss function based on absolute loss and square loss was proposed to select vaxiable,under which we combine exponential predictors to construct genneralized predictor model.At last compare with some existing methods,the models we proposed improves forecast precision.

关 键 词：指数加权滑动平均 加权损失 变量筛选 汇率预测 

分 类 号：F830.92[经济管理—金融学] O212[理学—概率论与数理统计]