Review Asking If Technical Trading Is Profitable Finance Essay
The purpose of this paper is to review the evidence on the profitability of technical analysis in the stock markets. The empirical literature is divided into two groups, including developed stock markets and developing markets, according to their target markets. Empirical evidence suggests the technical analysis can generate excess return in US stock market at least until 1980s (for other developed countries at least until 1990s). Moreover, the empirical evidence suggests technical trading rules can generate significant profit irrespective to time periods in the emerging stock markets, but the profit is declining as the market become more efficient.
The report aims to review existing literatures regarding to the profitability of technical analysis in both emerging markets. Main methodology (bootstrap procedure) is developed in USA stock market, and then extends to other developed and developing markets to test the profitability of technical analysis.
Methodology developments in USA stock market mainly focus on Brock et al., (1992)’s model based bootstrap which successfully solve or mitigate some of the early studies’ limitation (data snooping, test procedures etc). Later modern studies in stock market always developed based on this procedure and give more precise estimation of the profitability of technical analysis. In addition, the modern studies in USA stock show a clear trend.
The bootstrap methodology extends to other developed markets by Mills (1997). Through applying more advanced model, the general evidence is similar as USA which the profitability of technical analysis in developed stock markets disappear in later 1990s.
Finally, Bessembinder and Chan (1995), Ito (1997), Ito (1999) and Ratner and Leal (1999) advance the Brock et al., (1992)’s methodology to estimate the profitability of technical analysis in emerging markets. The general evidence suggests technical analysis can generate excess return in emerging markets during 1990s-2000s (except India). Moreover, there is a clear trend the profitability of technical analysis will decrease as the market become more efficient.
For further study, the control of country specific factors (policy factors etc), the measurement of transaction cost and more advanced model should be considered.
In the simplest term, technical analysis uses information about historical price movements, summarized in the form of price charts to forecast future price trends. Pring (2002) detailed define the term as ‘The technical approach to investment is essentially a reflection of the idea that prices move in trends that are determined by the changing attitudes of investors toward a variety of economic, monetary, political, and psychological forces. The art of technical analysis, for it is an art, is to identify a trend reversal at a relatively early stage and ride on that trend until the weight of the evidence shows or proves that the trend has reversed’ (cited by Parker and Irwin, 2007). It includes several forecasting methods, including chart analysis (heads and shoulders), cycle analysis, and computerized technical trading systems.
Technical analysis has a long history of widespread use by investors in speculative markets. It is widely accepted by trading advisors as the single dominant strategy to generate abnormal returns. Financial survey suggests more than 60% of the commodity trading advisors (CTA) rely heavily on computer guided technical trading system to identify the price trends (Billingsley and Chance, 1999). In contrast to the practitioners in speculative markets, academics still holds sceptical opinion about the technical analysis’ ability to generate excess return. Kendall (1953) suggest the technical analysis as ‘ the series look like a wandering one, almost as if once a week the demon of chance drew a random number…and added it to the current price to determine the next week’s price’.
The essay will discuss the technical analysis’ ability to generate excess return through reviewing the existing literature in developed and developing countries’ stock markets. The main mythology is developed in the USA and then extends to other developed and developing countries. Developing countries will be paid more attention because the emerging markets are supposed to exhibit higher volatility than developed markets and higher persistence in excess returns even after adjusting for risk factors. The remaining part of essay will be divided into 4 sections
In section 1 (theoretical source of excess return), the theoretical source of excess return of technical analysis will be explained. In section 2 (developed market), the development of studies, and the studies extend in the USA will be discussed. Then, the clear trend of profitability change in USA market will be discussed and the reason of it will be explained with recent studies. In section 3, the profitability of technical analysis in other developed countries will be discussed. In section 4, some improvement in methodology to test profitability in emerging markets will be described and the general evidence will be discussed.
Theoretical source of the excess return
Literatures always connect the excess return of technical analysis with the market efficiency hypothesis. Fama (1970) provide the definition of the market efficiency which is ‘in an efficient market, the price of the security should already reflect all available information of the security’. As a result, the technical analysis which focus on examining past price pattern will not generate any excess return. Jensen (1978) extends Fama’s definition, and he subdivided the market efficiency into 3 groups (weak-form, semi-strong, and strong). Base on the weak form market efficiency, there should be no excess return for technical analysis after taken risk and transaction cost into account.
However, there are some market frictions exists in the market which may leads excess return of technical analysis. In theoretical models, technical trading profits may arise because of market ‘frictions’, including noise in current equilibrium prices, traders’ sentimes, herding behavior, market power or chaos (Park and Irwin, 2007). Follow the rationale of market efficiency hypothesis and the existence market frictions, the profit of technical trading rules may likely to be generated in an immature market (lots of market frictions), and decrease as the market become more efficient (less market frictions).
Methodology development in US stock market
2.1 Early empirical evidence 1950s~1980s
Generally, the early studies, which focus on technical trading rules between 1950s-1980s in developed world, always suggest that there is no excess return for technical trading. Fama and Blume (1966) is the key study which exam the daily closing prices of 30 individual stocks in the Dow Jones Industrial Average (DJIA) over 1956-1962, and they find that only 3 of all technical trading rules generate higher than mean profit. After taking account risk and transaction cost (like bid-ask spread, brokers’ commission etc), these technical trading rules may even perform worse than buy-and-hold strategy. Similarly, James (1968) finds that moving average technical trading rules do not yield excess returns. Moreover, Jensen and Benington (1970) refute the past studies (Levy, 1967 is the key one) which suggest positive net profit of technical analysis, and they argue the excess return these studies find is due to some model and data problems. In their study, two trading rules (relative strength and portfolio upgrading) generate 0.31% and 2.36% less return than the buy and hold strategy. The results also support by Van Horne and Parker, (1967).
The early studies which suggest no excess return for technical trading rules seem to support the stock markets are weak-form efficient. However, these ‘no excess return’ results may not be caused by no market frictions in the stock market over 1950s-1980s (when the markets in developed countries were immature and inefficient), but caused by the problems exist in data, model and test procedures. Parker and Irwin (2007) conclude from early studies that these studies suffer the problems, including limited trading systems are tested, parameter optimization and out of sample verification are not employed, poor statically test, and data snooping. They further argue that these limitations cause falsely estimating the net profit of technical trading rules.
2.2 Modern empirical evidence 1980s~present
Modern studies show different results after solving some of the early test procedures, data and model problems. Majority of the key studies and methodologies are developed based on US market. Most of them support these is positive excess return in pre-1980s periods, but no excess return in post-1980s period in US stock market.
Brock et al., (1992) is the most influential modern study which first uses model-based bootstrap procedures and a large part of studies developed base on this one. This method typically analyzes part or all of the trading rules and overcome the weaknesses of traditional t-test. In addition, they mitigate the data snooping problems through using a long time period data (1897-1986 Dow Jones Industries Average) across non-overlapping sub-periods. Through using the bootstrap procedure and reduce the impact of data snooping problem, they find buy (sell) signals from moving average and trading range break-out technical trading rules significantly outperform the buy-and-hold strategies (ibid). However, the returns reported in the study are gross returns which are not adjusted for transaction cost, and the data snooping problems and the return measurement problem still exist. Thus, it still cannot conclude that the technical trading rules can generate abnormal returns.
After controlling some of the Brock et al (1992)’s limitation, recent studies develop the null models, and then provide a clear trend of profits of technical trading in the US stock market.
Bessembinder and Chan (1999) follow Brock et al (1992)’s method and take transaction cost into account. Instead of using DJIA data which may underestimate the actual return, they use DJIA after dividend data over 1926-1991 and find the average profit across all trading rules is 4.4% per year including 0.39% transaction cost in 1926-1976 sub-sample period. On the other hand, gross return of technical trading rules over 1976-1991 is 0.22% which is less than the transaction cost (0.29%).
Data snooping problems
Ready (2002) finds the profit of technical analysis is overestimated in the Brock et al., (1992)’s study due to data snooping problems through comparing the Brock et al., (1992)’s best trading rules with the technical trading rules formed by generic programming model. He suggest the excess return of technical trading rules disappear since 1975 in the US market
Sullivan et al. (1999) mitigate the data snooping problem in the Brock et al., (1992)’s study by applying the bootstrap reality check method which can directly quantify the effect of data snooping through evaluating the performance of the best trading rule in the context of all rules. They find the 5-day moving average is the best technical trading rule which generate 17.2% annual return (risk adjust return) over 1897 to 1987 sub-sample period. The general evidences suggest technical trading can generate excess returns until 1980s in the USA. However, the technical trading rules fail to generate excess return in the 1987-1996 sub-sample periods.
Day and Wang (2002) argues the inclusion nonsynchronous prices in the closing index level to at which trades are assumed to be executed imparts a positive bias to the estimated trading profits that depends on the precise sequencing of the observed return series. These positive bias causes significantly overestimated profit of technical analysis. After taking the transaction costs and the impact of nonsynchronous price on the reported closing level of the DJIA, they find the excess return of technical trading rules (pre-1986 period) reported in the early modern studies (Brock et al., (1992) and Sullivan et al., (1999), and Sullivan et al., (2001) etc) should be reduced by a large extent. These problems exist in modern studies lead overestimate of the excess return of technical analysis in pre-1986 sample period, but the profit still exists.
As the trade volume dramatic increase in the late 1980s, the nonsynchronous prices in the closing index level start to reflect the true price the level. Thus, Day and Wang (2002) find similar result (no excess return) as Sullivan et al., (1999), and Sullivan et al., (2001) as the positive bias on estimated trading profits disappear in the post 1980s sub-sample periods.
2.3 Reason behind the trend
There are two main reasons behind the trend
Baber (2000) argues dramatic increase in transaction cost significantly cast the profit of technical trading rules in the post 1980s periods. For frequently trading strategies, the cost might take account for 50%~70% of the turnover.
Sullivan et al. (2003) argues that the phenomena ‘excess return until 1980s and no excess return after 1980s’ follow the rationale of market efficiency. In detail way, there is a dramatic increase in trade volume in USA stock market in late 1980s and the market become much more efficient than the before 1980s one
Other developed markets
The methodology developed in the USA market could extend to the other developed countries. General evidence suggests in developed countries, the technical trading rules cannot generate excess return after taking transaction cost into account after 1990s.
Mills (1997) is the first study extends the Brock et al., (1992)’s method to other market (UK stock market), and they find the daily return of moving average trading rules do not outperform the buy and hold strategy over the 1975-1994.
Parker and Irwin (2007) conclude from existing literature that most studies focus on developed countries (including Japan, UK, and South Korea etc) show no excess return after late 1990s.
Developing stock countries
The use of the bootstrap methodologies appears to have become a common approach to the examinations of the technical trading rules. The studies in the emerging stock markets always follow the more advanced techniques (already taking transaction cost and return measurement problem into account) developed base on Brock et al., (1992)’s null model. The general evidences suggest that the technical trading rules can generate significant excess return in emerging markets regardless of time periods, exclude India stock market.
Bessembinder and Chan (1995) is the first study extends the technique to emerging markets and the study reports that moving average trading rules (the same as Brock) have predictive power to the future price change in 6 markets in Asia (Hong Kong, Japan, Korea, Malaysia, Thailand and Taiwan) over the period 1975-1991. Moreover, the trading rules are more successful in predicting stock price movement in emerging markets. In Malaysia, Thailand and Taiwan, the rules generate 0.167% per day return. On the other hand, in the other more developed markets, the rules generate excess return 0.028% per day return. However, they explain 3 possible measurement error may lead overestimate the predictive power of technical trading rules, including nonsynchronous reporting of prices induce spurious positive autocorrelation in portfolio (as Day and Wang argue above), underestimate of transaction cost of technical trading rules, and the impacts of not using time varying returns (the first two are tested, but the third one is not). The inclusion of the first two problems will significantly reduced the excess return of the technical trading rules by a large extent, but the excess return still positive and significant (ibid).
The exclusion of time varying return is inappropriate in assessing the profit of technical analysis. Ito (1997) argues the nonsynchronous price and transaction cost alone cannot completely explain away the excess return of technical trading rules. The two problems, what are (1) simulated p-values Brock and Bessembinder and Chan used are based on the random walk with a drift, which do not allow for any time variation of expected returns, (2) the trading rule profits are not tested against any specific international asset pricing models nor the impacts of capital market segmentation on the time-varying expected returns, should be developed.
Moreover, Ito (1999) evaluates the profitability of technical trading rules in the Pacific-Basin stock market by using equilibrium asset pricing models with time-varying expected returns. After controlling the time varying returns (reduce some return), they still find the evidence that technical trading rules have significant predictive power in 5 markets (Japan, Canada, Indonesia, Mexican and Taiwanese), but not in USA market over 1980 to 1996 and the rules generate higher return in emerging markets (ibid). This is to say, after controlling the 3 possible causes of predictive power of technical analysis, the market inefficiency (market segmentation factors and country specific factors) may be the possible cause of the excess return.
Ratner and Leal (1999) report strong evidence of forecasting ability for moving average rules in 10 emerging equity markets in Latin American and Asia using daily, inflation-adjusted, index level returns from 1982~1995. However, the technical analysis does not generate excess return in India stock market. Gunasekarage and Power (2001) and Chang et al., (2004) find similar results and they explain the reason as the India market is a relative large market among emerging markets, with relative large foreign capital inflows which leads more efficient market.
However, recent studies find that the excess return in emerging also show a decline trend as the market become more efficient (use China as a example).
Tian et al., (2002) argues the methods of testing for successful technical trading rules considered to suffer from the potential problem of data mining because the rules are imposed ex post by the testers. Given the technical analysts use different technical trading rules according to different market conditions in real world, Brock’s arbitrarily selection cannot reflect the actual condition. Thus, they expand the universe of 26 rules to 412 rules to avoid the data mining problem. They find that during the 1990s the technical strategies can generate excess return in Chinese stock market after taking transaction cost into account.
Cai et al., (2006) extends the Tian et al., (2002)’s study through analyzing the change of technical trading rules’ profit in the Chinese stock market during the 1990s. They find that while technical trading rules have short term predictive ability and profitability, this lessened as the decade progressed.
Moreover, Fifield and Jetty (2008) apply parametric and non-parametric variance ratio test to the daily data of 370 shares over 1996-2005. They find the profit of the technical analysis still exist but continue declining as both Chinese A share and B share become more efficient. The reason which used to explain the excess return of technical trading rules in US stock market disappear in late 1980s could also apply the Chinese case. As the Chinese market more integrate to world market and more efficient than before (less market friction), the profit significantly decreases.
The trend of profitability of technical analysis also applies to other developing countries stock markets which has increasing inflow of foreign capital. Coutts and Cheung, (2000) and Krausz et al., (2009) find similar results in Indonesia, Malaysia, Hong Kong (not developing market), and Thailand over 1980s~2007.
Summary and conclusion
The studies which examine the profitability of technical analysis are mainly developed in the USA. The early studies always suggest there is technical analysis cannot generate excess returns. However, the negative results are caused by the limitation (data snooping problems, no statics test etc) exist in the early models. The modern studies always follow and develop the Brock et al., (1992)’s bootstrap methodology which controls some of the early studies’ limitation. After applying more advanced model and methodology, general evidence technical analysis do generate excess return in the pre-1980s sample periods in the USA stock market, but the profit disappear in the post-1980s period. More efficient market is the main reason behind the trend. In addition, the most of the technical analysis in other developed market also generate insignificant profit after taking transaction cost into account. With respect to developing countries, the general studies suggest the technical analysis can generate in the emerging stock markets, however, the profit decrease as the market become more efficient.
Need help with your literature review?
Our qualified researchers are here to help. Click on the button below to find out more:
In addition to the example literature review above we also have a range of free study materials to help you with your own dissertation: