建構混合式匯率預測模型

Abstract

所謂匯率是指兩國之間兌換貨幣的比率。匯率可以用來判斷一個國家的經濟狀況，匯率變動也會影響企業的營收，此外，有部分投資者利用匯率的變動，採取買低賣高的兌換貨幣策略來賺取其中的差額；而一般國人出國旅遊觀光時，也必須兌換該國的貨幣…等等，可以說不論是個人或是團體，匯率與我們生活周遭的經濟活動息息相關，因此準確地預測匯率模型對個人或是財務管理者而言是很重要的。目前國內外已有許多預測匯率的研究，文獻上所提出之常見的預測方法有時間序列分析、類神經網路或是灰預測方法。本研究為提高現有在匯率預測之準確性，應用遺傳規劃法與滾動灰色預測模型，發展了一個混合式(hybrid)匯率預測模型，以有效提升匯率之預測準確度。

Exchange rate refers to the ratio between the currencies of two countries. Exchange rates can also be used to determine a country’s economic state and the change in exchange rates can affect the revenues of a company. Investors often use the change of exchange rates to gain profit by exploiting the price difference between purchases and sells of currency. Tourists also need to exchange their currencies to local currencies based on the exchange rates. Therefore, the exchange rate is an essential part of the financial activities, and an accurate prediction model for exchange rate is important for individuals and financial managers. There are numerous studies on predicting the exchange rate utilizing time-series analysis, artificial neural models or grey system theory to construct prediction model. This study develops a hybrid prediction model for exchange rate by combining genetic programming and rolling grey prediction model to enhance the prediction accuracy.