ScholarWorks@성균관대학교

초록

This study investigates the efficiency of the won/dollar foreign exchange market by analyzing daily won/dollar exchange excess returns and employing five commonly used technical trading rules representative of algorithmic trading strategies. The results, based on averaging the returns of 436 sub-trading rules to mitigate data mining concerns, reveal that the statistical significance of excess returns diminishes steadily throughout the 1990s, 2000s, and 2010s. Combined bootstrap methods and tests on technical trading rules demonstrate that excess returns from these technical strategies are not statistically distinguishable from those generated by null models such as RWD, AR(1), and CAPM in recent years, implying increased support for the adaptive market hypothesis or the efficient market hypothesis. The in-sample buy-sell returns from the five technical trading rules, which are generated by rolling 500-sample windows 6781 times, exhibit similar patterns. Notably, the one-period-ahead out-of-sample forecast results based on identical rolling regression windows indicate that CAPM, which attributes significant portions of excess returns to systematic risk, achieves superior performance in comparison to both the five technical trading rules and other null models evaluated.

키워드