The Information Content of Charts: Algorithm and Evidence from Hong Kong Stock Market

This study introduces an improved method for nonparametric regression and pattern identification in the research of technical analysis based on the concept of stopping time. We are interested in the efficacy, predictability, and profitability of 10 patterns from chart analysis. Using Hong Kong stock data, we attempt to identify these patterns and test their predictability and profitability. We find that most of them demonstrate predictability with regard to future price trend. Some of them even show profitability against the null hypothesis of random walk. Furthermore, we find some connections between price and trading volume behavior in one or two patterns.