Abstract:

In financial industry, one of the major analytical methods for security investment is Technical Analysis (TA). TA relies on trading signals (i.e., buy and sell signals) generated by technical trading rules. Although TA is widely adopted by security market participants, there is still a debate on the efficacy of TA amongst academics. The subjective nature of TA makes its definitions inconsistent in different studies and there is no uniform framework available for assessing reliability of technical trading rules. To fill the research gap, we design a scientific reliability analysis framework for technical trading rules, which is based on statistical methods (e.g., two tailed t-test) and conditional random field model (a powerful model for dealing with a time series with hidden states). Compared with the trading strategy without TA (i.e., buy-and-hold strategy), the effectiveness of TA can be tested by employing this framework. This framework also improves the reliability of technical trading rules by combining data mining techniques. Real market data of Hong Kong stock market can be used to evaluate the framework.