Journal of Behavioral and Experimental Finance 6 (2015) 42–55

Contents lists available at ScienceDirect

Journal of Behavioral and Experimental Finance

journal homepage: www.elsevier.com/locate/jbef

Full length article

Trading System based on the use of technical analysis: A

computational experiment✩
Thiago Raymon Cruz Cacique da Costa, Rodolfo Toríbio Nazário,
Gabriel Soares Zica Bergo, Vinicius Amorim Sobreiro ∗ , Herbert Kimura
University of Brasília, Department of Management, Campus Darcy Ribeiro, Brasília, Federal District, 70910–900, Brazil

article info abstract

Article history: Previous studies highlight the influence of methods of technical analysis in the search
Received 29 September 2014 for exceptional gains in the context of the financial market. Based on this scenario, the
Received in revised form 3 March 2015 main objective of this paper is to analyze the performance of the Simple Moving Average,
Accepted 4 March 2015
Exponential Moving Average, MACD and Triple Screen techniques in an actual trading
Available online 12 March 2015
system that included 198 stocks traded in the Brazilian. This paper studies the power
of predictability of such methods using various combinations of periods, brokerage fees
Keywords:
Trading system
and a policy of Stop-Loss and compares these with the buy-and-hold strategy. The results
MACD indicate that while the studied techniques lead to a high probability of obtaining a return
Simple Moving Average that exceeds the investment value, they have little power of predictability in the Brazilian
Exponential Moving Average market. In relation to the passive buy strategy, only the smallest part of the obtained returns
Triple Screen outweighs the results of the buy-and-hold strategy.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction and Menkhoff (2010) also found that TA has a large

influence in the financial market. However, there are few
Technical Analysis (TA) involves an attempt to predict studies regarding indicators related to TA in the academic
the movement of future asset prices based on an analysis arena (Teixeira and Oliveira, 2010, p. 6886).
of past prices using qualitative methods, such as graphic Although the importance of TA in the current scenario
analysis, quantitative methods, such as Moving Averages is clear, according to Ming-Ming and Siok-Hwa (2006,
(MA), or a combination of both (Menkhoff and Taylor, p. 145), there is no consensus regarding the effectiveness
2007; Gençay, 1998, p. 937, p. 347). To measure the of TA tools. Furthermore, few studies have attempted to
presence of the TA in the market, Taylor and Allen (1992) investigate the results of these techniques or systems in
conducted a survey in British agencies and concluded that emerging markets such as Brazil, even though, according
at least 90% of the respondents benefit from TA tools to Chang et al. (2004, p. 295), such a market can be
when making decisions related to their portfolios. Studies considered as a good alternative for investors who seek to
such as those of Fabozzi et al. (2007), Menkhoff and diversify their portfolios.
Taylor (2007), Marshall et al. (2008), Hsu et al. (2010), From a practical perspective, autonomous asset trading
programs are transforming the major stock markets
in electronic financial markets (Creamer and Freund,
✩ This document is a collaborative effort.
2010, p. 401). Thus, it is important to note that, according
∗ Corresponding author. to Austin et al. (2004, p. 37), with the development of
E-mail addresses: [email protected] (T.R.C.C. da Costa),
[email protected] (R.T. Nazário), [email protected]
new computational techniques and newly available data,
(G.S.Z. Bergo), [email protected] (V.A. Sobreiro), the construction and use of the Trading System (TS) has
[email protected] (H. Kimura). become increasingly more plausible. A TS can be briefly
http://dx.doi.org/10.1016/j.jbef.2015.03.003
2214-6350/© 2015 Elsevier B.V. All rights reserved.
 T.R.C.C. da Costa et al. / Journal of Behavioral and Experimental Finance 6 (2015) 42–55 43

defined as a set of rules that define necessary conditions On a historical scope, although the TA has gained promi-
to start or exit a negotiation (Chande, 2001, p. 3). nence with the work of Charles Dow in 1884 and William
Within this context, the main objective of this paper is Peter Hamilton in 1900 and in the Wall Street Journal (Van-
to analyze the performance of a TS based on Simple Moving stone and Finnie, 2009, p. 6671), Cowles (1933) is consid-
Average techniques (SMA), Exponential Moving Average ered one of the first empirical studies in the area of TA
(EMA), Moving Average Convergence Divergence (MACD) published in a specialized scientific journal. In this arti-
and Triple Screen in 198 stocks traded in the Brazilian stock cle, the result of an analysis of effort prediction of 45 pro-
market from January 2000 to February 2014, noting various fessional agencies in selecting specific stocks that could
combinations of periods, brokerage fees and a Stop-Loss provide higher than average market return results is pre-
mechanism. More specifically, this paper aims to study the sented. The results show that the average income of agency
following aspects: forecasts was 4% below the overall market average.
Important studies have discussed the power of pre-
1. Which TA tool performs best? dictability of the market. Papers such as Alexander
2. What is the influence of brokerage fees on the (1961), Fama and Blume (1966) and Levy (1967) demon-
performance of TA methods? strate the presence of the random walk in the markets ana-
3. What is the impact of the Stop-Loss mechanism on lyzed in their respective articles. In short, the random walk
operations? is an assumption that involves the Efficient Market Hy-
4. Upon observing a wide range of results, which of the TA pothesis (EMH). In this context, Fama (1991) states that the
methods are capable of overcoming the buy-and-hold EMH assumes that asset prices reflect all available market
strategy? information and follow a random behavior. Adding to the
discussion, Fama and Blume (1966, p. 226) claim that suc-
We posit that the main results of this work contribute cessive price changes are independent, that is, follow the
to or advance the discussions about the effectiveness of TA theory of random walk. This independence implies that the
tools and about the automated systems with respect to the historical series of changes in the price of an asset may not
predictability of the behavior of prices or returns, which be used to predict future movements in the value of the
according to Campbell et al. (1997, p. 27), is one of the main asset in any path in a significant way.
issues in financial econometrics: After the publication of these studies, the EMH was
considered the theory that governs the movement of
‘ONE OF THE EARLIEST and most enduring questions
stock prices (Zhu and Zhou, 2009, p. 521). However, later
of financial econometrics is whether financial asset
influential studies gave greater value to TA in the academic
prices are forecastable. Perhaps because of the obvious
arena. Rejecting the theory of random walk with the
analogy between financial investments and games of
use of weekly stock returns based on a simple test of
chance, mathematical models of asset prices have an
volatility specified, the study of Lo and MacKinlay (1988) is
unusually rich history that predates virtually every
considered one of the major works in the area of TA. Thus,
other aspect of economic analysis. The fact that many
to investigate the predictive capacity of the market, the
prominent mathematicians and scientists have applied
article of Brock et al. (1992), one of the influential studies,
their considerable skills to forecasting financial securities
used the popular strategies of TA, specifically, Moving
prices is a testament to the fascination and the challenges
Averages (MA) and the Trading-Range Breaks techniques,
of this problem.’
on the Dow Jones Industrial Average from 1897 to 1986.
Campbell et al. (1997, p. 27). Excessive returns and the evident power of predictability
This paper is structured as follows. The second section of the method used in relation to the market studied by the
presents a brief description regarding the concepts of TA authors have resulted in increased credibility of the TA.
and a brief historical contextualization of the topic. The Proof of the great influence of the study of Brock et al.
third section is devoted to the presentation of the methods (1992) is the wide replication of his method by several
and the computational model used in the experiment, researchers, such as, Parisi and Vasquez (2000), Kwon
while the fourth section presents the results. The fifth and Kish (2002) and Chang et al. (2004) in different
section highlights the main conclusions of the study. markets. Among these replications is the study of Hudson
et al. (1996), which used as a sample the British stock
market for the period 1935 to 1994. Such research, whose
2. Theoretical reference method used resulted in excessive returns in relation to the
considered market, is widespread in academic studies of
The technical analyst uses past prices and other statis- price behavior, thus contributing to the acceptance of TA
tics when making investment decisions, believing that past in the study of financial markets.
data contain important information about future stock The purpose of MAs, techniques that are based on the
market behavior (Zhu and Zhou, 2009, p. 519). Accord- works of Brock et al. (1992) and Hudson et al. (1996), is
ingly, Murphy (1999, p. 3) defines the three assumptions to identify or signal that a new trend has started or an old
underlying the precepts of TA described below: one has finished (Murphy, 1999, p. 197). Regarding this is-
sue, Zhu and Zhou (2009, p. 521) contend that these are the
1. Prices reflect market events; most popular and simple existing techniques. Therefore,
2. Changes in the prices move in trends; and the MMs were applied in the context of emerging markets
3. Historical prices tend to repeat. in prominent studies, such as Gunasekarage and Power