Among lots of literatures about prediction of exchange rate, the models being concerned are mainly divided into two kinds: the usual models and the unusual ones. The usual models predict the future exchange rate only based on the information provided by the history values such as time series modeling. The unusual models forecast the future values according to correlations between exchange rates and other economic element, variables (e.g. interest rate, inflation rate etc.) for example the financial models and so on. In this paper we discuss two criterions belong to the usual models for prediction of growth rate. One is symmetric entropy loss function, the other is Linex loss function. The limiting development of the trend of the valuation compared with the real development is studied for i.i.d growth rate. In this paper we aim at the trend of the limitation ratio of prediction value to true value under the criterion of non-square loss functionsThe main contents of this thesis can be summarized as follows:Chapter 1 introduces the fundamental conceptions of statistical decision theory and entropy.Chapter 2 is about the model fo increase of exchange rate and describes some propositions in preparations.Chapter 3 respectively gives the prediction increase satisfied the symmetric entropy loss function and Linex loss function, and then study the limitation trend of ratio of prediction value to true value with corresponding proofs. |