This study develops a new template grid – rounding top and saucer – to detect buy signals. Most of the previous studies utilize historical data to derive the template grid, but do not clearly explain how to format weight values of the template grid. This makes the template a “black box” for users since it is difficult to infer the process of the template formation. Therefore, this study proposes a simple and explicit method for deriving the template grid. In addition, to more accurately detect buy signals, the trading rules are developed by capturing reversal of price trend. The empirical results indicate that the template grid and the proposed trading rules developed in this study have considerable forecasting power across tech stocks traded in the US, including MSFT, IBM, INTC, ORCL, DELL, APPLE and HP, since the average returns of the proposed trading rule are greater than the results of buying every day over the sample period. The method proposed here could therefore become an effective component of an expert system to assist investors in investment decisions.
Technical analysis involves the examination of past stock prices to identify patterns that can be exploited to achieve excess profits. Studies of technical analysis mainly look at quantitative indicators, such as relative strength index and moving average (e.g., Brock et al., 1992 and Pruitt and White, 1988). Charting patterns, such as head-and-shoulder, flags, saucers, and rounding tops, have been much less studied, until Lo, Mamaysky, and Wang (2000).
Lo et al. (2000) use kernel regression to identify charting patterns, while Leigh et al., 2002, Leigh et al., 2002, Leigh et al., 2004 and Bo et al., 2005, and Wang and Chan (2007) all implement a variation of the bull flag stock chart using a template-matching technique based on pattern recognition. These studies show that charting patterns can predict stock prices.
The method developed here differs from other studies in three respects. First, no previous study, to our knowledge, utilizes charting patterns (rounding top and saucer) to detect buy signals. In this study, we develop a new template grid to detect buy signals. Second, since previous studies have not clearly defined how to format weight values of the template grid (e.g., Bo et al., 2005, Leigh et al., 2004, Leigh et al., 2002, Wang and Chan, 2007 and Leigh et al., 2002), this makes the template a “black box” for users and might lead to questions regarding data mining. In contrast to the extant literature, this study develops an alternative method for formatting the template grid using a template-matching technique based on pattern recognition. The advantage of the method developed here is that it is simple and explicit, and can generally be applied to other charting patterns’ formation, while avoiding suspicions of data mining. Third, saucer and rounding tops are usually reversal patterns and are typically followed by substantial price movements. To more accurately detect buy signals, the trading rules are developed by capturing reversal of price trend. Moreover, to ensure that the performance is not decided by too few buying signals and that the trading rules have practical application; this study develops trading rules with numerous filter rules to detect buy signals.
This study uses daily stock prices to assess stock market purchasing opportunity. The proposed method is applied to the computer science tech stocks in the US with largest market cap. The selected stocks include Microsoft (MSFT), IBM, Intel (INTC), Oracle (ORCL), DELL, APPLE and Hewlett–Packard (HP). The empirical results demonstrate that trading using conditional trading rules yields significantly better returns than buying every day1 during the sample period. Accordingly, the template grid and the conditional trading rules developed in this study have considerable forecasting power across tech stocks in the US.
The remainder of this paper is organized as follows. Section 2 describes the method used. Section 3 describes the design of the trading rules. Section 4 describes the data and results of the empirical investigation. Finally, Section 5 summarizes the findings and offers conclusions.
This study develops a new template grid – rounding top and saucer – to detect buy signals. Most of the previous studies on this topic utilize historical data to derive the template grid, but do not clearly explain how to format the template grid. This makes the template a “black box” for users since it is difficult for them to infer the process of the template formation. This study proposes a simple and explicit method for deriving the template grid.
In addition, to ensure that the performance is not decided by too few buying signals, and that the trading rules have practical application, we use Trading Rule A to find numerous filter rules with stable profits using NASDAQ index data. These filter rules are further employed to establish Trading Rule B to examine the profit potential for the tech stocks with the largest market cap in US.
The empirical results indicate that Trading Rule B exactly predicts the stock price movements since most of the average returns are greater than buying every day over the sample period. Accordingly, the template grid and the conditional trading rules developed in this study have considerable forecasting power across tech stocks in the US. The method proposed here can therefore be seen as an effective expert system to assist investors in when making investment decisions.
