This paper investigates the profitability of candlestick charting patterns. The holding periods are one, three, five, and ten days. Two exit strategies are studied. One is the Marshall-Young-Rose (MYR) exit strategy (Marshall, Young, & Rose, 2006) and the other is the Caginalp-Laurent (CL) exit strategy (Caginalp & Laurent, 1998). The MYR sets a specific day to exit the market. In contrast, the CL sets an exit price equal to an average of close prices over the holding period, assuming that investors make an equal-weight exit within this period. The data include daily open, high, low and close prices of component stocks of the SET50 index for a ten-year period from July 3, 2006 to June 30, 2016. This study tests the predictive power of bullish and bearish candlestick reversal patterns both without technical filtering and with technical filtering [Stochastics (%D), Relative Strength Index (RSI), Money Flow Index (MFI)] using the skewness adjusted t-test (Johnson, 1978) and the binomial test. The statistical analysis finds little use of both bullish and bearish candlestick reversal patterns since the mean returns of most patterns are not statistically different from zero. Even the ones with statistically significant returns do have high risks in terms of standard deviations. The binomial test results also indicate that candlestick patterns cannot reliably predict market directions. In addition, this paper finds that filtering either by %D, RSI or MFI generally do not increase profitability nor prediction accuracy of candlestick patterns.
Keywords: technical analysis, trading rule, candlestick, charting pattern
- Hits: 920