多因子動態計量貨幣交易系統-獲利與預測能力實證研究

Abstract

在過去的文獻發展歷程中，已經有許多學者先進試圖在複雜的外匯市場中找出影響匯價的因素。從最早的市場結清理論告訴我們外匯市場是開放總體經濟各個市場中最後結清的市場，到現在五花八門的匯價走勢預測模型如總體經濟為基礎的利差模型、購買力評價模型或是以線圖為基礎的動能交易模型、基因演算法等，目的都是為了能夠有效預測並解釋匯價的走勢。歷來的研究也告訴我們不論基本面或是技術面分析，或多或少都可以解釋一部分的匯率波動並預測未來匯價的走勢，但是效果普遍不是那麼明顯或是一致。 有鑑於此，引入新的外匯分析方法似乎有其必要，Ilmanen, A. (1997)將動態調整影響因子的概念應用在債券市場的交易上面。同樣的，本篇論文引入這個概念到外匯市場上，外匯市場的影響因子來自股市、債市、及總體經濟面等多方面力道影響，且隨時間經過會被不同的因素主導，這特性使動態調整因素權重的預測模型同樣能夠在外匯市場有不錯的表現。 本篇論文試圖從學理及實務兩方面著手，不僅僅只是從經濟邏輯上面驗證代理變數的正當性，更希望能夠提供台灣金融機構在進行外匯操作時的參考依據，進而達到理論與實務結合的美好目標。

In the past literatures, there are many predecessors research in the foreign exchange area and try to find out the factors driving the complex market. From the earliest market clearing theory to the modern macro-based and technical-based theories, all these models are trying to forecast the market movement correctly. No matter the fundamental and technical theories, both of these can explain part of the market movement and forecast the future direction. But the performances are not always significant or consistent. Therefore, introducing new analysis tools seems inevasible. Ilmanen, A. (1997) introduced dynamic weighted concept into bond market. In this research, we try to inject this concept into foreign exchange markets. The factors in foreign exchange markets are coming from many sources such as stock market, bond market and macro economy performance, the movements in each currency pair is driven by different factors in specific period. This characteristic is consistent with the concept of dynamic weighted forecasting model and also has good performance in FX market. This research tries to analyze from both academic and practice. Not only to verify the proxies from academic theories, but also can provide a trading reference for the financial institutions in Taiwan.