Efficient market hypothesis

The efficient-market hypothesis (EMH) contradicts the basic tenets of technical analysis by stating that past prices cannot be used to profitably predict future prices. Thus it holds that technical analysis cannot be effective. Economist Eugene Fama published the seminal paper on the EMH in the Journal of Finance in 1970, and said "In short, the evidence in support of the efficient markets model is extensive, and (somewhat uniquely in economics) contradictory evidence is sparse."^[59]

Technicians say^[who?] that EMH ignores the way markets work, in that many investors base their expectations on past earnings or track record, for example. Because future stock prices can be strongly influenced by investor expectations, technicians claim it only follows that past prices influence future prices.^[60] They also point to research in the field ofbehavioral finance, specifically that people are not the rational participants EMH makes them out to be. Technicians have long said that irrational human behavior influences stock prices, and that this behavior leads to predictable outcomes.^[61] Author David Aronson says that the theory of behavioral finance blends with the practice of technical analysis:

By considering the impact of emotions, cognitive errors, irrational preferences, and the dynamics of group behavior, behavioral finance offers succinct explanations of excess market volatility as well as the excess returns earned by stale information strategies.... cognitive errors may also explain the existence of market inefficiencies that spawn the systematic price movements that allow objective TA [technical analysis] methods to work.^[60]

EMH advocates reply that while individual market participants do not always act rationally (or have complete information), their aggregate decisions balance each other, resulting in a rational outcome (optimists who buy stock and bid the price higher are countered by pessimists who sell their stock, which keeps the price in equilibrium).^[62] Likewise, complete information is reflected in the price because all market participants bring their own individual, but incomplete, knowledge together in the market.^[62]

Random walk hypothesis[edit]

The random walk hypothesis may be derived from the weak-form efficient markets hypothesis, which is based on the assumption that market participants take full account of any information contained in past price movements (but not necessarily other public information). In his book A Random Walk Down Wall Street, Princeton economist Burton Malkielsaid that technical forecasting tools such as pattern analysis must ultimately be self-defeating: "The problem is that once such a regularity is known to market participants, people will act in such a way that prevents it from happening in the future."^[63] Malkiel has stated that while momentum may explain some stock price movements, there is not enough momentum to make excess profits. Malkiel has compared technical analysis to "astrology".^[64]

In the late 1980s, professors Andrew Lo and Craig McKinlay published a paper which cast doubt on the random walk hypothesis. In a 1999 response to Malkiel, Lo and McKinlay collected empirical papers that questioned the hypothesis' applicability^[65] that suggested a non-random and possibly predictive component to stock price movement, though they were careful to point out that rejecting random walk does not necessarily invalidate EMH, which is an entirely separate concept from RWH. In a 2000 paper, Andrew Lo back-analyzed data from U.S. from 1962 to 1996 and found that "several technical indicators do provide incremental information and may have some practical value".^[56] Burton Malkiel dismissed the irregularities mentioned by Lo and McKinlay as being too small to profit from.^[64]

Technicians say^[who?] that the EMH and random walk theories both ignore the realities of markets, in that participants are not completely rational and that current price moves are not independent of previous moves.^[30][66] Some signal processing researchers negate the random walk hypothesis that stock market prices resemble Wiener processes, because the statistical moments of such processes and real stock data vary significantly with respect to window size and similarity measure.^[67] They argue that feature transformations used for the description of audio and biosignals can also be used to predict stock market prices successfully which would contradict the random walk hypothesis.

The random walk index (RWI) is a technical indicator that attempts to determine if a stock’s price movement is random in nature or a result of a statistically significant trend. The random walk index attempts to determine when the market is in a strong uptrend or downtrend by measuring price ranges over N and how it differs from what would be expected by a random walk (randomly going up or down). The greater the range suggests a stronger trend.

Empirical evidence

Whether technical analysis actually works is a matter of controversy. Methods vary greatly, and different technical analysts can sometimes make contradictory predictions from the same data. Many investors claim that they experience positive returns, but academic appraisals often find that it has little predictive power.^[50] Of 95 modern studies, 56 concluded that technical analysis had positive results, although data-snooping bias and other problems make the analysis difficult.^[14] Nonlinear prediction using neural networksoccasionally produces statistically significant prediction results.^[51] A Federal Reserve working paper^[22] regarding support and resistance levels in short-term foreign exchange rates "offers strong evidence that the levels help to predict intraday trend interruptions," although the "predictive power" of those levels was "found to vary across the exchange rates and firms examined".

Technical trading strategies were found to be effective in the Chinese marketplace by a recent study that states, "Finally, we find significant positive returns on buy trades generated by the contrarian version of the moving-average crossover rule, the channel breakout rule, and the Bollinger band trading rule, after accounting for transaction costs of 0.50 percent."^[52]

An influential 1992 study by Brock et al. which appeared to find support for technical trading rules was tested for data snooping and other problems in 1999;^[53] the sample covered by Brock et al. was robust to data snooping.

Subsequently, a comprehensive study of the question by Amsterdam economist Gerwin Griffioen concludes that: "for the U.S., Japanese and most Western European stock market indices the recursive out-of-sample forecasting procedure does not show to be profitable, after implementing little transaction costs. Moreover, for sufficiently high transaction costs it is found, by estimating CAPMs, that technical trading shows no statistically significant risk-corrected out-of-sample forecasting power for almost all of the stock market indices."^[17] Transaction costs are particularly applicable to "momentum strategies"; a comprehensive 1996 review of the data and studies concluded that even small transaction costs would lead to an inability to capture any excess from such strategies.^[54]

In a paper published in the Journal of Finance, Dr. Andrew W. Lo, director MIT Laboratory for Financial Engineering, working with Harry Mamaysky and Jiang Wang found that:

Technical analysis, also known as "charting", has been a part of financial practice for many decades, but this discipline has not received the same level of academic scrutiny and acceptance as more traditional approaches such as fundamental analysis. One of the main obstacles is the highly subjective nature of technical analysis – the presence of geometric shapes in historical price charts is often in the eyes of the beholder. In this paper, we propose a systematic and automatic approach to technical pattern recognition using nonparametric kernel regression, and apply this method to a large number of U.S. stocks from 1962 to 1996 to evaluate the effectiveness of technical analysis. By comparing the unconditional empirical distribution of daily stock returns to the conditional distribution – conditioned on specific technical indicators such as head-and-shoulders or double-bottoms – we find that over the 31-year sample period, several technical indicators do provide incremental information and may have some practical value.^[55]

In that same paper Dr. Lo wrote that "several academic studies suggest that ... technical analysis may well be an effective means for extracting useful information from market prices."^[56] Some techniques such as Drummond Geometry attempt to overcome the past data bias by projecting support and resistance levels from differing time frames into the near-term future and combining that with reversion to the mean techniques.^[57]

Thomas DeMark's indicators enjoy a remarkable endorsement in the financial industry. A recent work ^[58] has investigated the predictive power of three DeMark indicators (Sequential, Combo and Setup Trend), over 21 commodity futures markets and 10 years of data. Market entry signals have been tested by comparing conditional returns (i.e. conditioned on the entry signals) to unconditional returns. For the period from Jan. 2004 to Jan. 2014, the tests suggest statistically significant predictive power on a wide range 

Empirical evidence

Whether technical analysis actually works is a matter of controversy. Methods vary greatly, and different technical analysts can sometimes make contradictory predictions from the same data. Many investors claim that they experience positive returns, but academic appraisals often find that it has little predictive power.^[50] Of 95 modern studies, 56 concluded that technical analysis had positive results, although data-snooping bias and other problems make the analysis difficult.^[14] Nonlinear prediction using neural networksoccasionally produces statistically significant prediction results.^[51] A Federal Reserve working paper^[22] regarding support and resistance levels in short-term foreign exchange rates "offers strong evidence that the levels help to predict intraday trend interruptions," although the "predictive power" of those levels was "found to vary across the exchange rates and firms examined".

Technical trading strategies were found to be effective in the Chinese marketplace by a recent study that states, "Finally, we find significant positive returns on buy trades generated by the contrarian version of the moving-average crossover rule, the channel breakout rule, and the Bollinger band trading rule, after accounting for transaction costs of 0.50 percent."^[52]

An influential 1992 study by Brock et al. which appeared to find support for technical trading rules was tested for data snooping and other problems in 1999;^[53] the sample covered by Brock et al. was robust to data snooping.

Subsequently, a comprehensive study of the question by Amsterdam economist Gerwin Griffioen concludes that: "for the U.S., Japanese and most Western European stock market indices the recursive out-of-sample forecasting procedure does not show to be profitable, after implementing little transaction costs. Moreover, for sufficiently high transaction costs it is found, by estimating CAPMs, that technical trading shows no statistically significant risk-corrected out-of-sample forecasting power for almost all of the stock market indices."^[17] Transaction costs are particularly applicable to "momentum strategies"; a comprehensive 1996 review of the data and studies concluded that even small transaction costs would lead to an inability to capture any excess from such strategies.^[54]

In a paper published in the Journal of Finance, Dr. Andrew W. Lo, director MIT Laboratory for Financial Engineering, working with Harry Mamaysky and Jiang Wang found that:

Technical analysis, also known as "charting", has been a part of financial practice for many decades, but this discipline has not received the same level of academic scrutiny and acceptance as more traditional approaches such as fundamental analysis. One of the main obstacles is the highly subjective nature of technical analysis – the presence of geometric shapes in historical price charts is often in the eyes of the beholder. In this paper, we propose a systematic and automatic approach to technical pattern recognition using nonparametric kernel regression, and apply this method to a large number of U.S. stocks from 1962 to 1996 to evaluate the effectiveness of technical analysis. By comparing the unconditional empirical distribution of daily stock returns to the conditional distribution – conditioned on specific technical indicators such as head-and-shoulders or double-bottoms – we find that over the 31-year sample period, several technical indicators do provide incremental information and may have some practical value.^[55]

In that same paper Dr. Lo wrote that "several academic studies suggest that ... technical analysis may well be an effective means for extracting useful information from market prices."^[56] Some techniques such as Drummond Geometry attempt to overcome the past data bias by projecting support and resistance levels from differing time frames into the near-term future and combining that with reversion to the mean techniques.^[57]

Thomas DeMark's indicators enjoy a remarkable endorsement in the financial industry. A recent work ^[58] has investigated the predictive power of three DeMark indicators (Sequential, Combo and Setup Trend), over 21 commodity futures markets and 10 years of data. Market entry signals have been tested by comparing conditional returns (i.e. conditioned on the entry signals) to unconditional returns. For the period from Jan. 2004 to Jan. 2014, the tests suggest statistically significant predictive power on a wide range 

Combination with other market forecast methods

John Murphy states that the principal sources of information available to technicians are price, volume and open interest.^[10] Other data, such as indicators and sentiment analysis, are considered secondary.

However, many technical analysts reach outside pure technical analysis, combining other market forecast methods with their technical work. One advocate for this approach isJohn Bollinger, who coined the term rational analysis in the middle 1980s for the intersection of technical analysis and fundamental analysis.^[47] Another such approach, fusion analysis, overlays fundamental analysis with technical, in an attempt to improve portfolio manager performance.

Technical analysis is also often combined with quantitative analysis and economics. For example, neural networks may be used to help identify intermarket relationships.^[48]

Investor and newsletter polls, and magazine cover sentiment indicators, are also used by technical analysts.

Systematic trading

Neural networks[edit]

Since the early 1990s when the first practically usable types emerged, artificial neural networks (ANNs) have rapidly grown in popularity. They are artificial intelligence adaptive software systems that have been inspired by how biological neural networks work. They are used because they can learn to detect complex patterns in data. In mathematical terms, they are universal function approximators,^[40][41] meaning that given the right data and configured correctly, they can capture and model any input-output relationships. This not only removes the need for human interpretation of charts or the series of rules for generating entry/exit signals, but also provides a bridge to fundamental analysis, as the variables used in fundamental analysis can be used as input.

As ANNs are essentially non-linear statistical models, their accuracy and prediction capabilities can be both mathematically and empirically tested. In various studies, authors have claimed that neural networks used for generating trading signals given various technical and fundamental inputs have significantly outperformed buy-hold strategies as well as traditional linear technical analysis methods when combined with rule-based expert systems.^[42][43][44]

While the advanced mathematical nature of such adaptive systems has kept neural networks for financial analysis mostly within academic research circles, in recent years more user friendly neural network software has made the technology more accessible to traders. However, large-scale application is problematic because of the problem of matching the correct neural topology to the market being studied.

Backtesting[edit]

Systematic trading is most often employed after testing an investment strategy on historic data. This is known as backtesting. Backtesting is most often performed for technical indicators, but can be applied to most investment strategies (e.g. fundamental analysis). While traditional backtesting was done by hand, this was usually only performed on human-selected stocks, and was thus prone to prior knowledge in stock selection. With the advent of computers, backtesting can be performed on entire exchanges over decades of historic data in very short amounts of time.

The use of computers does have its drawbacks, being limited to algorithms that a computer can perform. Several trading strategies rely on human interpretation,^[45] and are unsuitable for computer processing.^[46] Only technical indicators which are entirely algorithmic can be programmed for computerised automated backtesting.

Industry

The industry is globally represented by the International Federation of Technical Analysts(IFTA), which is a federation of regional and national organizations . In the United States, the industry is represented by both the Market Technicians Association (MTA) and the American Association of Professional Technical Analysts (AAPTA). The United States is also represented by the Technical Security Analysts Association of San Francisco (TSAASF). In the United Kingdom, the industry is represented by the Society of Technical Analysts (STA). In Canada the industry is represented by the Canadian Society of Technical Analysts.^[35] In Australia, the industry is represented by the Australian Technical Analysts Association (ATAA),^[36] (which is affiliated to IFTA) and the Australian Professional Technical Analysts (APTA) Inc.^[37]

Professional technical analysis societies have worked on creating a body of knowledge that describes the field of Technical Analysis. A body of knowledge is central to the field as a way of defining how and why technical analysis may work. It can then be used by academia, as well as regulatory bodies, in developing proper research and standards for the field.^[38] The Market Technicians Association (MTA) has published a body of knowledge, which is the structure for the MTA's Chartered Market Technician (CMT) exam.

History tends to repeat itself

Technical analysts believe that investors collectively repeat the behavior of the investors that preceded them. To a technician, the emotions in the market may be irrational, but they exist. Because investor behavior repeats itself so often, technicians believe that recognizable (and predictable) price patterns will develop on a chart.^[10] Recognition of these patterns can allow the technician to select trades that have a higher probability of success.^[31]

Technical analysis is not limited to charting, but it always considers price trends.^[1] For example, many technicians monitor surveys of investor sentiment. These surveys gauge the attitude of market participants, specifically whether they are bearish or bullish. Technicians use these surveys to help determine whether a trend will continue or if a reversal could develop; they are most likely to anticipate a change when the surveys report extreme investor sentiment.^[32] Surveys that show overwhelming bullishness, for example, are evidence that an uptrend may reverse; the premise being that if most investors are bullish they have already bought the market (anticipating higher prices). And because most investors are bullish and invested, one assumes that few buyers remain. This leaves more potential sellers than buyers, despite the bullish sentiment. This suggests that prices will trend down, and is an example of contrarian trading.^[33]

Recently, Kim Man Lui, Lun Hu, and Keith C.C. Chan have suggested that there is statistical evidence of association relationships between some of the index composite stocks whereas there is no evidence for such a relationship between some index composite others. They show that the price behavior of these Hang Seng index composite stocks is easier to understand than that of the index.^[