Is technical analysis profitable on US stocks with certain size, liquidity or industry characteristics?

Abstract

We consider whether popular moving average and trading range breakout technical trading rules are profitable on a subset of the US stocks with certain size, liquidity and industry characteristics. We find these rules are rarely profitable during the period 1990 to 2004, however there is some evidence that they are more profitable for smaller, less liquid stocks. There is no evidence to any industry bias in applying these rules and when a rule does produce statistically significant profits on a stock, these profits tend to be greater for longer decision period rules.

Acknowledgements

The authors wish to thank Rochester Cahan for his help in developing the code for the bootstrap methodology employed in this article. We also thank Tom Smith and the other participants of the Australian National University School of Finance and Applied Statistics seminar series and the anonymous referees for the Financial Management Association Annual Meeting, 2006, for their helpful comments. Finally, we thank the two anonymous referees of this journal for comments that have substantially improved this article.

Notes

¹ See Allen and Taylor (1990), Taylor and Allen (1992) and Menkhoff and Taylor (2007) for excellent surveys of the use of technical analysis by market participants. See Park and Irwin (2004) for a review of the academic literature that tests the profitability of technical trading rules.

² We use this specification of the trading rules for two reasons. Firstly, they are the most prominent in the literature (see Park and Irwin, 2004, for an excellent review). Secondly, researchers who consider rule variations such as imposing a nonzero amount by which the short-term average breaks through the long-term average or maximum/minimum (Bessembinder and Chan, 1988; Brock et al., 1992) find the results generated are qualitatively identical to those generated by the rule we apply.

³ We wish to thank an anonymous referee at the 2006 Annual FMA meeting for this suggestion.

⁴It should be noted that the underlying data fails for the Kolmogorov–Smirnov normality test, so these results should be considered in light of the other results in this article rather than emphasized in isolation.