本文主要針對新臺幣兌換美元的外匯匯率, 採用有效排除資料窺視偏誤的Superior Predictive Ability檢定來探究技術交易策略運用於臺灣匯市時 , 是否具有日內交易表現較好, 且隨著決策時間愈短表現愈佳之特性 ;並藉此驗證為何在現實生活中, 投資人會隨著決策時間的縮短而使用更高比例的技術交易法則。我們的實證結果顯示:在現有資料長度限制下 , 技術交易策略於新臺幣/美元外匯市場上的日間交易, 具有十分顯著的獲利能力;反之, 在日內交易方面, 不管資料頻率為何 (30分鐘、15分鐘、10分鐘、或5分鐘), 皆不存在任何顯著的獲利性。整體而言, 在臺灣的外匯市場中, 我們沒有找到隨決策時間愈短, 技術交易策略表現愈佳的證據。透過進一步的分析, 我們發現兩個有趣的結果:第一、技術交易策略在日間交易與日內交易的獲利能力差異與匯價走勢的變化程度有關。第二、移動平均法則適用在馬可夫轉換模型的環境中。因此, 在投資標的物價格可以由馬可夫轉換模型配適良好的前提下, 投資人採用移動平均法則來從事投資決策應能有不錯的獲利表現。

This paper investigates whether technical trading rules, when applied to Taiwan’s foreign exchange market, perform better in intraday trading and gain better performance as the intraday decision horizon is shortened. To avoid data snooping bias, we adopt the Superior Predictive Ability (SPA) test to obtain robust results. Our empirical results show that technical trading rules enjoy significant profitability for the daily trading in the NTD/USD exchange rate market. In contrast, when applied to intraday data at various frequencies, none of the technical trading rules wetest are significantly profitable. Overall, on Taiwan’s FX market, we did not find evidence that technical trading rules are more profitable with a shorter time horizon. Further analyses lead to two interesting findings. First of all, differential performance of technical trading rules in daily trading versus intraday trading is related to the magnitudes of price-changes of foreign exchange rates. Second, the essence of the moving average trading rule can be captured by a Markov switching model. The implication is that, if the targeted as set prices could be better fitted by a Markov switching model, traders may obtain decent profitability by adopting the moving average trading rule in their investment strategy.