Foreign Exchange Rates Prediction Based On Generalized Exponential Predictor Models With Weighted Loss Function

Since 1970s, when the bretton woods system callapsed, the volatility of international foreign exchange market became more capricious. The factors which involves exchange rates fluctuation is becoming more and more complex, these factors also makes exchange rates forecast more difficult than ever before. Our country accessed to the WTO in 2001, the financial openness continues to improving, interest rate marketization steadily. Therefore, exchange rates forecast becomes an important and challenging research subject for academic and practice area.The generalized 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 variable,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.