國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2 Does Market Volatility Improve Profitability of Technical Analysis?
2.1 Introduction
Technical analysis, based on the premise that price movement will repeat itself and then display regular or recurring pattern, is a method of forecasting future price movement relying only on the information of past prices and/or volumes. This method has a long history among invest-ment professionals (Smidt, 1965; Allen and Taylor, 1992; Billingsley and Chance, 1996; Lui and Mole, 1998; Oberlechner, 2001; Gehrig and Menkhoff, 2004; Covel, 2005; Lo and Hasan-hodzic, 2009). The essence of technical analysis and its widespread adoption by practitioners conflict with the central idea of the efficient market hypothesis. Thus, whether investors can get statistically significant economic profits by technical analysis has drawn a lot of attention and discussion since Alexander (1961). Recent studies find more and more evidences of profitabil-ity of technical analysis even by examining more trading systems or adopting stricter statistical tests solving data snooping bias problems (Brock et al., 1992; Chan et al., 1996; Neely et al., 1997; Sullivan et al., 1999; LeBaron, 1999; Okunev and White, 2003; Hsu and Kuan, 2005;
Schulmeister, 2008). The consensus of these studies appears that using technical analysis helps investors to forecast the market.
In addition to technical analysis, market volatility investigation is also an important issue in the financial literature and there exists a lot of papers on modeling and estimating it. Fur-thermore, the market volatility is viewed as the valuable information and acts as a key input in numerous financial researches such as the portfolio diversifications, the hedging strategy inves-tigations, and the asset pricing models. Many value-at-risk models also require the estimation of volatility parameter to measure the market risk. Despite the accumulating evidence on the value of technical analysis by examining different trading systems and the increasing importance of the market volatility as a key input in financial studies, little is known about the value of market volatility in technical analysis. What is the role of market volatility playing in technical analy-sis? Is it valuable for enhancing the profitability of technical analysis if it is used as an input in
‧
In this paper we are interested in examining a number of related questions. First, we ask whether market volatility is useful in enhancing the profitability of the technical analysis in which market volatility acts as an important input in the technical trading systems. We adopt the variable moving average (VMA), a variant of the popular moving average system proposed first by Chande (1992). The VMA rule automatically adjusts its effective length of the moving average according to the changing market conditions measured by the level of market volatility.
It is viewed as the representative of the technical analysis comprising the information of market volatility in this study. Market volatility in the VMA rule is built to detect whether the market price makes big moves in up or down direction or whether it moves in a narrow range. In other words, market volatility here helps the moving average trading strategy to detect the market trend more timely. We apply five most tested technical trading systems (filter, moving average, support and resistance, channel breakout, and momentum strategies in volume), and compare them with the VMA rule on the Dow Jones Industrial Average (DJIA) index over 1928/10/1-2010/6/28. In order to minimize data snooping bias, the Superior Predictive Ability Test (SPA) proposed by Hansen (2005) is performed. Our SPA results both in full sample and several subsample periods clearly demonstrate that the VMA rule outperforms others with statistically significant greater profitability in the DJIA market.
Since we find strong evidence for the existence of statistically significant larger excess return to the VMA rule, we then consider what the source of the larger excess returns of the VMA rule might be. That is we want to discuss the possible reason why market volatility betters the performance of the technical analysis. For any technical trading rule to be profitable, the stock return must be predictable. The higher predictability of stock returns permits the possibility of larger excess return of technical analysis. Therefore, we carry out the market timing tests described in Cumby and Modest (1987) to examine the comparative predictability of the best VMA rule. This test can be used for investigating the ability of a trading rule to predict the sign of the one-period-ahead excess return. In this study, we modify this test by regressing the excess return to three dummy variables, which represent three forecast positions of the rule, long, short
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
and no positions. By the modification, we can study the market timing ability of a trading rule in forecasting future price to rise or descend without losing the essence of the market timing tests. Since the VMA rule is the variant of the moving average system, we compare the market timing ability of the best VMA rule with that of the best MA rule. We also make comparisons between the market timing ability of the best VMA rule and that of the second best rule in our SPA results, which is the momentum strategies in volume (MSV) rule. There is strong evidence indicating that all of these three best rules can forecast the upward direction of future price movement well, but only the best VMA rule possesses predictive ability for future downward trends.
Moreover, we are also interested in examining whether the market timing ability of the trading rules is different due to different market conditions. In this study, different market conditions are measured by bull and bear markets, which are identified with the dating algorithm of Pagan and Sossounov (2003). Our results show that the market timing ability of the best VMA, best MA and best MSV rule are all asymmetric in different market conditions. They are good at detecting price rising trends in bull markets rather than in bear markets, while they do possess predictive ability of price decreases only for bear markets not for bull markets. Besides, the best VMA rule gains higher daily profit and suffer less daily loss as it signals long (short) and short (long) positions in bull (bear) markets, respectively. As a whole, the best VMA rule outperforms the best MA and best MSV rule both in bull and bear markets.
The remainder of the paper is organized as follows: Section 2 outlines the technical trading rules used in this study. The SPA test of Hansen (2005) is presented in Section 3. Section 4 contains the data description and the SPA results of full and several subsample periods. Section 5 provides evidence for the market timing ability of the best VMA, best MA and best MSV rule.
Section 6 discusses whether these three best rules have differential performance in different market conditions, and Section 7 concludes the paper.