Donchian 4 W PDF
Donchian 4 W PDF
Donchian 4 W PDF
of
Technical
Analysis
Issue 59
Winter-Spring 2003
SM
Associate Editor
Michael Carr
Cheyenne, Wyoming
Manuscript Reviewers
Connie Brown, CMT J. Ronald Davis, CMT Jeffrey Morton, MD, CMT
Aerodynamic Investments Inc. Golum Investors, Inc. PRISM Trading Advisors
Pawley's Island, South Carolina Portland, Oregon Missouri City, Texas
Matthew Claassen, CMT Cynthia Kase, CMT Kenneth G. Tower, CMT
The Technical View Kase and Company CyberTrader, Inc.
Vienna, Virginia Albuquerque, New Mexico Princeton, New Jersey
Julie Dahlquist, Ph.D. Michael J. Moody, CMT Avner Wolf, Ph.D.
University of Texas Dorsey, Wright & Associates Bernard M. Baruch College of the
San Antonio, Texas Pasadena, California City University of New York
New York, New York
The JOURNAL of Technical Analysis is published by the Market Technicians Association, Inc., (MTA) 74 Main Street, 3rd Floor, Woodbridge, NJ 07095. Its
purpose is to promote the investigation and analysis of the price and volume activities of the world's financial markets. The JOURNAL of Technical Analysis is
distributed to individuals (both academic and practitioner) and libraries in the United States, Canada and several other countries in Europe and Asia. The JOURNAL of
Technical Analysis is copyrighted by the Market Technicians Association and registered with the Library of Congress. All rights are reserved.
1
A Theoretical Foundation for Technical Analysis
Gunduz Caginalp, Ph.D. and Donald Balenovich, Ph.D.
Abstract third by using a model which generalizes the classical theory of adjustment to
include supply and demand changes which depend on price derivative in addi-
Using a dynamical microeconomic model which generalizes the classical tion to price [Caginalp and Balenovich (1991), (1994b)]. Within this approach,
theory of adjustment to include finite asset base and trend-based investment the classical (value-based) motivation for purchasing the equity is augmented
preference, we develop a foundation for the technical analyis (or charting) of with a trend-based strategy of buying as a result of rising prices. The analysis
securities. The mathematically complete system of (deterministic) ordinary of supply, demand and price, based on money flow, as a function of time leads
differential equations that has provided a quantitative explanation of the labo- to a system of differential equations, which may be deterministic or stochastic.
ratory bubbles experiments generates a broad spectrum of patterns that are The model incorporates, in a natural way, the conservation of asset principles
used by practitioners of technical analysis. The origins of many of these pat- which prevent a bubble from continuing indefinitely, and includes the effects
terns are classified as (i) those that can be generated by the activities of a of finite markets that are absent in the classical theory. These equations have
single group, and (ii) those that can be generated by the presence of two or been useful in understanding the non-classical behavior of real and experimen-
more groups with asymmetric information. Examples of (i) include the head tal markets such as the formation of bubbles and subsequent fall in prices. In
and shoulders, double tops, rising wedge while (ii) includes pennants and sym- particular, the parameters in the equations can be calibrated with a single ex-
metric triangles. The system of differential equations is easily generalized to periment such as those of Porter and Smith (1989), (1994); the remaining ex-
stochastic ODE’S. Application is also made to Japanese candlestick analysis. periments are then predicted with no adjustable parameters. Statistical tests
confirm that the predictions are overwhelmingly more likely than the alterna-
1. Introduction tive hypothesis, either in the form of the efficient market hypothesis or in the
trivialization of the trend based component.
TECHNICAL ANALYSIS METHODOLOGY The use of technical analysis has always posed an interesting question for
Technical analysis or charting is a technique which uses the patterns of the the efficient market hypothesis as the latter implies that such methods could
price history of a financial instrument (commodity, currency, stock or compos- not be successful. In particular, the weak form of the efficient market hypoth-
ite average) in order to provide indications on the future behavior of prices esis maintains that prices incorporate all public information so that an analysis
[Edwards and Magee (1992), Myers (1989), Pring (1993)]. Technical analysis of price pattern cannot produce any profits. Moreover, significant return from
actually consists of numerous methods with a common set of basic principles, technical analysis, even in conjunction with valuation methods, tends to argue
and is the chief alternative to fundamental analysis which strives to assess the against the efficient market hypothesis. Consequently, there is a close link
true value of financial instruments. The philosophical basis of fundamental between the validity of technical analysis and the inefficiency of the market.
analysis is quite close to classical economic theory which stipulates that prices For example, the bubbles that have been observed in laboratory experiments
will move in a direction to alleviate the discrepancy between the current price are an example of inefficiency. At the same time, it would be easy to use mo-
and the true value. Thus, the investor tries to discover a stock, commodity or mentum indicators, for example, to profit from these bubbles if one were par-
financial instrument whose current price does not yet reflect its improved for- ticipating in these experiments. Also, the necessity for technical analysis in
tunes. The investor who trades on such a fundamental basis thereby becomes securities marketing, e.g., secondary offerings, becomes evident if the relevant
part of the process by which prices are returned to equilibrium while profits are market is not very efficient. We discuss some recent examples of market inef-
made in the process. Traders who prefer technical analysis usually agree that ficiency below.
the fundamental value will eventually be attained. However, they contend that The second question of statistical testing based upon patterns is not ad-
funds can remain unproductive for the intermediate term (from days to months) dressed in this paper, though it has been studied in others. Brock et al. (1992)
while awaiting the long term process of reaching the equilibrium or fundamen- review the literature and conclude that many simple rules have found no statis-
tal value which may take years. Furthermore, they contend that even the direc- tical validity. In particular, they test the hypothesis of the moving average trad-
tion of prices need not be uniformly in the direction of fundamental value. The ing rule that consists of a buy signal when the price moves above a particular
implicit assumption is that the dissemination of information and reassessment moving average and a sell signal when the price crosses below the average.
of value is a slow process which may be overshadowed by sellers of underval- Using the data set of the Dow Jones average for several decades, they found
ued securities who are either unaware of true value or hesitate to rely on the almost no net gain for using either the buy or the sell signal. White (1993)
optimizing behavior of others (see Section 3) and are attempting to limit losses. analyzed neural networks on 1000 closing prices of IBM stock that was used to
The immediate direction of prices, they argue, is determined solely by the sup- make predictions on the next 500. Roughly speaking, White’s procedures in-
ply and demand of the financial instrument, and that the changes in supply and volved simply finding the optimal fit for the past three days of closing prices
demand are generally evident upon examining recent price trends. and utilizing it to predict the next day. The procedure failed to produce a prof-
Unlike fundamental analysis, technical analysis has no basis in classical itable trading strategy. To technical analysts these failures are not surprising.
economic theory. For example, a decline in prices for an undervalued stock An examination of simply closing prices during three days is unlikely to cap-
can only be attributed to random fluctuations in the context of basic theory. ture the emotion of the previous days. Also a relatively small sample such as
Thus, the widespread practical use of methods of technical analysis which are 500 data points for one stock will not have the capability to uncover particular
contradictory to classical economic theories provides three important puzzles patterns.
for mathematical economics. First, can one explain the key patterns of techni- However, a recent study by Blume et al. (1994) explores technical analysis
cal analysis based on a generalization of the classical theory of adjustment? as a component of agents’ learning process. Focusing mainly on the informa-
Second, can one verify using statistical hypothesis testing that such patterns tional role of volume, they conclude that sequences of volume and price can be
actually appear in real markets? Third, do such patterns have a predictive value informative and argue that traders who use information contained in the market
within the context of a model or in real markets? statistics attain a competitive advantage.
In this paper we focus attention mainly on the first issue and partly on the A different approach, adopted by Caginalp and Laurent (1998) involves
Rising (or Falling) Wedge: This is a rather common reversal pattern that
differs in a subtle way from the pennant and symmetric triangle formations
which we discuss as part of the two group patterns below. The price rises into
the wedge which has positive slope (i.e., the rising wedge in Figure 4 and
analogously for a falling wedge in Figure 5) as opposed to the symmetrical
triangle or pennant which have zero or negative slope respectively. The nega-
tive prediction rendered by the rising wedge is evident upon a comparison with
a rising channel. The rising wedge is essentially a deteriorating channel as the
progressive new highs are unable to maintain the same ratios as the new lows
(on the bottom trendline), thereby indicating that selling occurs earlier than
one would expect, and sending a signal that a topping out process may be un-
derway.
TWO-GROUP PATTERNS
Symmetrical Triangles and Pennants: A common pattern in charts ap-
pears as a series of oscillations with diminishing amplitude. A pennant is formed
by the two lines which join the successive tops and bottoms respectively. If the
two slopes have equal magnitude and different sign, then the pattern is called a
symmetric triangle (Figure 6). In general, pennants are regarded as consolida-
tion formations which appear at regular intervals as a trend takes a pause. How-
ever, they can also be a reversal formation (Figures 7, 8) as shown in the next
section. A breakout usually occurs before the vertex is reached as the trend is
resumed. The pennant sometimes points down (i.e., the bisector of the two Flags: The flag is a consolidation pattern which is similar to the pennant in
enveloping lines has negative slope) during an uptrend and points up in a which the oscillation occurs between two parallel lines until a breakout occurs
downtrend. Volume is often diminishing until shortly before the breakout point, (Figure 6).
at which time it increases dramatically.
(3.3)
Basic
Candlestick
Symbols
Piercing Line
Advanced Block
From the perspective of the theory, each of these patterns represents a small
oscillation about a linear Pa(t), for example, as shown in Figure 24. Hence, in
the head and shoulders pattern [Figure 14] any of the five maxima and minima Three White Soldiers and Advance Block: The three white soldiers pat-
would be similar to the continuous analog of the morning or evening star pat- tern (Figure 26) occurs in a downtrend and is defined by three days of trading
terns. that are all white bodies in which the open and close enclose each day’s trad-
Piercing Line and On Neckline: A pattern which appears to be quite dif- ing. The second and third days both open within the prior day’s body. This
ferent but is again similar to an oscillation about a constant Pa(t) is the piercing defines “three white soldiers.” A pattern similar to this is generated by V-shaped
line (Figure 25). The definition is specified by a downtrend followed by two Pa(t). The overlapping bodies are essentially oscillations about an increasing
days consisting of Pa(t). The temporal length of the transition provides very strong confirmation
i) A long black body for the first day. for this reversal pattern.
ii) A white body for the second day which opens below the low of the previous A pattern that appears to be superficially similar, but is in fact a strong
day and closes above the midpoint of the first day’s body. reversal sign, is the “advance block” (see Figure 26), which occurs during an
uptrend and is defined by the following
The opposite pattern for a bearish reversal is called a dark cloud cover.
Various modifications of these occur depending on the extent of the second i) All three days have white bodies with each of the last two opens within the
day’s rebound. If all features are the same as above except that the close of the prior day’s body.
second day is at the low of the first day then one has an “on neckline” which is ii) The last two bodies are smaller than the first.
a continuation pattern rather than a reversal. iii) The second day has an upper shadow which is much longer than the first
Distinguishing these two patterns, as well as other related patterns that are day.
in between in terms of the rebound (e.g. above the close but below the midpoint iv) The closing prices of all three days are all near one another.
of the body), is accomplished most simply by assuming a (locally) linear Pa(t) While the stock price moves up during each day the rally is actually losing
and examining the oscillation in P(t) as a function of the slope of Pa(t). If the steam as the new closing highs are not significantly higher than the previous,
slope is zero near the relevant time, for example, then an initial overvaluation and consequently it is a reversal signal. From the perspective of the theory, the
will result in overshooting Pa(t) followed by a rebound. On the other hand, a
rebound that does not go far into the previous day’s is easily produced with Pattern Breakdown
either a negative slope or with an increased supply of stock submitted for sale.
Figure 25 shows the differences in the extent of the rebound for a declining
Pa(t) versus one which has leveled off. Thus the extent of the rebound provides ➔
an indication of the perceptions of underlying value.
Other patterns which are related include “kicking” in which a black body in
a downtrend is followed by a white body gapping upward, while both have a
range defined by the body — i.e. no shadows (called “marubazu”). This is
similar to the piercing line except in the abruptness of the transition. In terms
of the model, the same values of q1 and q2 that produced a solution path with
constant Pa(t) will produce one that is consistent with kicking only if Pa(t)
itself exhibits a strong reversal. This is compatible with the designation of
kicking as a strong reversal sign which requires no confirmation.
Preface breakouts, N-Day rules, etc. Despite the minor time frame modification to the
original Four-Week Rule, the system tested herein will still be respectfully ad-
Every commodity trader has to decide on a market analysis technique to dressed as The Four-Week Rule.
generate buy and sell signals. Given the plethora of techniques to choose from,
it is not surprising that the eventual choice does not always stem from a thor- Applying Four-Week Rule Trading System to Bridge/
ough analysis of all the available options. Simplicity, convenience, and con- CRB Index Commodities
vention each have intrinsic appeal, but we are ultimately interested in a workable
tool that is successful under a variety of market conditions. The Bridge/CRB (Commodity Research Bureau) Index was first calculated
by the Commodity Research Bureau, Inc. in 1957 and made its inaugural ap-
Introduction pearance in the 1958 CRB Commodity Year Book3. The Index, originally
twenty-eight commodities, was designed to provide a dynamic representation
Most of today’s market analysis techniques have been derived from funda- of broad trends in overall commodity prices.
mental analysis, technical analysis, or a combination of the two. Simply put, The Bridge/CRB Index comprises seventeen commodities which are sepa-
fundamental analysis (FA) concerns itself with the causes of market move- rated into six groups: energies, grains, industrials, livestock, precious metals
ment: commodity supply and demand, macroeconomic data, association pro- and softs. All these commodities have respective futures contracts which are
duction surveys, foreign production and consumption, consumer trends, climate listed on various U.S. futures exchanges.
change, and countless other factors. The weakness of FA is its very sensitivity,
which renders it susceptible to intraday price noise, rumors, boredom, greed, Group Commodity Exchange Contract Months
and fear. Technical analysis, on the other hand, focuses on the effects of market Energy Natural Gas (NG) NYMEX All 12 Calendar Months
movement. Those who favor this approach believe that all the data one needs
Crude Oil (CL) NYMEX All 12 Calendar Months
are derived from the study of the market itself1: historical prices, volume, open
interest, volatility, trend lines, price patterns, moving averages, and momentum Heating Oil (HO) NYMEX All 12 Calendar Months
measures. Grains Corn (C) CBOT Mar, May, Jul, Sep, Dec
Technical analysts have developed various mechanical trading systems to Soybeans (S) CBOT Jan, Mar, May, Jul, Aug, Nov
handle these data and eliminate a trader’s natural tendency to overreact to mar- Wheat (W) CBOT Mar, May, Jul, Sep, Dec
ket fluctuations. In their simplest form, such trading systems initiate buy and Livestock Cattle (LC) CME Feb, Apr, June, Aug, Oct, Dec
sell signals on the basis of predetermined criteria. The Four-Week Rule, just
Hogs (LH) CME Feb, Apr, June, Aug, Oct, Dec
one of hundreds such mechanical systems, is both popular and simple to use.
Precious Metals Gold (GC) COMEX Feb, Apr, Jun, Aug, Dec
The Four-Week Rule Silver (SI) COMEX Mar, May, Jul, Sep, Dec
Platinum (PL) COMEX Jan, Apr, Jul, Oct
The Four-Week Rule was developed by Richard Donchian2 several decades Industrials Copper (CP) COMEX Mar, May, Jul, Sep, Dec
ago and has maintained its status as one of the best mechanical trading systems
Cotton (CT) NYBOT Mar, May, Jul, Dec
available because of its robustness, simplicity, and versatility. It forms the
basis of the channel breakouts, moving average crossover systems and other Softs Sugar #14 (SE) NYBOT Mar, May, Jul, Oct
trend-following methods. The Four-Week Rule can be stated as follows: Orange Juice (OJ) NYBOT Jan, Mar, May, Jul, Sep, Nov
Cover shorts and go long when the closing price exceeds the highs of the Coffee (CF) NYBOT Mar, May, Jul, Sep, Dec
preceding four full calendar weeks, and liquidate longs and go short when Cocoa (CC) NYBOT Mar, May, Jul, Sep, Dec
the closing price falls below the lows of the four preceding calendar weeks.
In other words, the Rule compares Friday’s closing price against the previ- These commodity futures provide an excellent pool of historical data for
ous four full calendar weeks’ highs and lows. If the four-week high or low is back-testing a mechanical trading system. The contract histories of the Bridge/
penetrated, the system dictates buying or selling, respectively. Entry and exit CRB Index commodities have experienced the gamut of fundamental price
signals become market orders for execution on the following opening session. shocks. For example, international wars, trade embargoes, economic booms,
Although Donchian’s Four-Week Rule is still applicable today, a more dy- recessions, droughts, floods, food scares, and unpredictable domestic and for-
namic system will be tested for this analysis. Since a four week period is syn- eign demand/supply scenarios are just a few catalysts of the price swings en-
onymous with twenty days, the past 20 days will be back-tested for signals. countered by these commodities over the years.
The advantages of this minor modification are that breakouts are recognized For this analysis, The Four-Week Rule has been back-tested for each con-
earlier and signals may be generated any day of the week. The modified Four- tinuous commodity contract for the last 20 years, thus approximately 5,000
Week Rule will remain a stop-and-reverse (SAR) system and the signals are as trading sessions. The 20-year time period was selected to provide long-term,
follows: equitable evaluation of this trading system over multiple commodities. For the
Cover shorts and go long when the closing price penetrates the previous contracts with a shorter trading history, the back-test will commence with the
twenty day highs and liquidate longs and go short when the closing price contract’s initial trading date. Each of the continuous contracts were constructed
falls below the previous twenty day lows. to rollover three days prior to expiration to allow for the normal volume, li-
quidity shift.
This system also reflects other technical analysis tools and time frames:
Since trading system evaluation encompasses numerous statistics, this analy-
20-day moving average, monthly cycles, weekly price channels, channel sis will present only the “industry accepted” ones. The Four-Week Rule trad-
GC SI PL CP CT OJ CF CC
Closed Net Profit (22,000) (53,050) (16,110) (19,675) (81,330) (24,232) (65,887) (62,790)
Total Trade Count 95 178 166 83 129 61 181 101
Percent Long 51 50 50 51 50 49 50 50
Average. Duration 38 29 31 61 32 61 29 50
Average. Profit (201) (294) (72) (231) (601) (397) (364) (622)
Average. Win 1,101 1,892 1,228 1,823 2,460 1,470 2,243 1,837
Maximum Win 3,650 13,150 6,735 10,637 10,195 13,410 16,387 12,030
Average. Loss (1,232) (1,221) (789) (1,222) (2,241) (1,527) (1,502) (2,168)
Maximum Loss (31,850) (5,750) (3,025) (4,825) (49,710) (36,255) (6,412) (71,310)
Max Closed Drawdown (31,850) (14,400) (8,050) (23,787) (55,010) (37,222) (21,712) (76,880)
Max Drawdown Amount (41,690) (73,200) (28,785) (34,600) (95,200) (39,789) (107,456) (81,020)
Max. Drawdown Duration 1,064 4,931 3,665 3,753 1,246 3,020 4,432 2,132
Profit To Max Draw (0.46) (0.72) (0.42) (0.55) (0.81) (0.61) (0.61) (0.77)
Profit Loss Ratio 0.71 0.66 0.86 0.72 0.59 0.58 0.65 0.53
Percent Winners 44.21 29.78 35.54 32.53 34.88 37.70 30.39 38.61
Remove To Neutral 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
ing system evaluation herein does not employ “pyramiding,” thus the position following systems may augment returns to various degrees. Optimally, the
size will remain constant. most profitable systems will achieve significant profits in the shortest time
The aforementioned results of the Four-Week Rule will vary per commod- frames thereby reducing the risk exposure. Furthermore, most profitable sys-
ity for multiple reasons. First, the 20-year time period back-test may consist of tems will identify losing trades faster thereby cutting losses sooner. For ex-
varying amount of trading sessions per contract. Second, the important con- ample, NG, CL, C and LH each had average wins almost twice their respective
tract indicators of open interest, daily volume, historical volatility, commercial average losses.
hedging interest and speculative fund participation will impact the successes of
trading systems. For example, the Four-Week Rule system appears to have Conclusion
more successes with contracts (i.e., NG, CL, C and LH) that have large open The success of Donchian’s Four Week Rule with the Bridge/CRB Index
interest, volume, and high speculative fund activity. commodities is a positive testament of the application of mechanical trading
The Remove-To-Neutral (RTN) measures what percentage of all trades may systems among multiple and unrelated markets. The Four-Week Rule trading
be removed prior to becoming a break-even system. High RTN indicates more system, in its most basic form, has proven that it is a profitable, effective sys-
robustness, reliability and system effectiveness. Percent Winners is another tem, especially in four of the sixteen contracts.
important system statistic that appears to also positively correlate with higher
RTN. Thus, the contracts with a higher RTN and Percent Winners were also Suggestions
the most profitable.
The fact that all the winning trade percentages were greater than 30% so- The Four-Week Rule and other mechanical systems may also be applied to
lidifies the technical analysis foundation that markets move in trends and trend over-the-counter (OTC) contracts, fundamental data and virtually all types of
Footnotes
1. Edwards, Robert M.; John Magee [1992]. Technical Analysis of Stock
Trends, pg. 4
2. Murphy, John J. [1986]. Technical Analysis of the Futures Markets, pg.
268.
3. Bridge/CRB Futures Indices, http://www.crbindex.com Background and
History.
References
■ Burke, Gibbons [February 1998]. Futures, “Turtle redux”
■ Bridge/CRB Futures Indices, http://www.crbindex.com
■ Edwards, R. and Magee, J. [1992]. Technical Analysis of Stock Trends,
Sixth Ed., New York Institute of Finance.
■ Kaufman, Perry J. [1998]. Trading Systems and Methods, John Wiley &
Sons, Inc.
■ Lowenstein, Roger [2000]. When Genius Failed, Random House.
■ Lukac, Louis B., Brorsen, Wade and Irwin, Scott [1980]. A Comparison of
Twelve Technical Trading Systems, Traders Press.
■ Murphy, John J. [1999]. Technical Analysis of the Financial Markets, New
York Institute of Finance.
■ Ruszkowski, Art [2000]. Mechanical Trading System Vs. The S&P 100
Index, Market Technicians Association Library.
■ Turtle Trader, http://www.turtletrader.com/trader-donchian.html “Richard
Donchian”
■ Logical Information Machines, Inc., Profit/Loss User Manual.
■ Sharpe, William F. [1999]. Investments.
■ Natenberg, Sheldon [1994]. Option Volatility & Pricing.
Notes
1. The Commitments of Traders (COT) reports provide a breakdown of each
Tuesday’s open interest for markets in which 20 or more traders hold posi-
tions equal to or above the reporting levels established by the CFTC. The
weekly report for Commitments of Traders in Commodity Futures is re-
leased every Friday at 3:30 p.m. Eastern time.
2. The Four-Week Rule and other mechanical trading systems may be back-
tested through the application of spreadsheet software or complex trading
system software. The author’s results were computed and validated by uti-
lizing CQG and the Logical Information Machine (LIM) XMIM software.
Biography
Geoff C. Hicks, CMT, is a trader with El Paso Merchant Energy of Hous-
ton, Texas and a former floor trader at the Chicago Mercantile Exchange.
He holds the following degrees: M.B.A with distinction, DePaul Univer-
sity; M.S. thesis, Texas A&M University; B.B.A., Texas Tech University.
This research paper was authored to complete the Chartered Market Tech-
nician (CMT) designation from the Market Technicians Association (MTA).
Over a large part of the past 30 years, the discipline of finance has been
under the aegis of the efficient market hypothesis. But in recent years, enough
The adoption-diffusion life cycle model, which is widely used in social
science and in marketing research, can also be modified to fit the stock market.
Using the model, Figure 3 shows how the four major parameters of technical
analysis – price, volume, time, and sentiment – are interrelated. Further, the
model can be used to specify and interrelate indicators to measure those param-
eters. It also reveals how these parameters can combine to form continuation or
reversal patterns.
anomalies have piled up, cracking its dominance of the field. As a consequence,
the arrival of new thinking to explain market behavior has warranted attention, Four Elements
and its name is behavioral finance.
Behavioral finance proponents believe that markets reflect the thoughts, The four major parameters are a distinct aspect of the technical condition of
emotions, and actions of normal people as opposed to the idealized economic the US stock market. Since the data for each parameter is independent from
investor underlying the efficient market school as well as fundamental analy- that of the others, the indicators representing them can be combined. This fea-
sis. Behavioral man may intend to be rational, but that rationality tends to be ture of price, time, volume, and sentiment is very important; it gives a more
hampered by cognitive biases, emotional quirks, and social influences. complete conclusion regarding the market’s present position and probable fu-
Behavioral finance uses psychology, sociology, and other behavioral theo- ture trend.
ries to explain and predict financial markets. It also describes the behavior of Further, indicators that represent each parameter can be arranged to provide
investors and money managers. In addition, it recognizes the roles that varying a more indepth and reliable understanding of each parameter, as suggested in
attitudes play toward risk, framing of information, cognitive errors, self-con- Figure 4. This can be used as a worksheet to aid in model development and
trol and lack thereof, regret in financial decision-making, and the influence of testing. As can be seen, each element is broken down into three levels of analy-
mass psychology. sis. The columns titled “Indicators” and “Weighting” are filled in by the ana-
Assumptions about the frailty of human rationality and the acceptance of lyst, with the entry depending on the market being analyzed and time frame
such drives as fear and greed have long been accepted by students of technical selected.
analysis. Indeed, in his Stock Market Behavior: The Technical Approach To Joe Granville’s “tree of indicators” concept comes into play during the build-
Understanding Wall Street, Harvey Krow classified technical analysis in the ing and testing of complex models. In this notion, rather than simply relying
behaviorist school of thought. upon price trend and sentiment, the analyst can add together indicators such as
price pattern, Elliott wave count, point-and-figure proportion, on-balance and
Integrating Technical Indicators total volume, and put-call ratio to fully exploit the technical information avail-
able. Using the adoption-diffusion model, analysts can make sense of how these
Conceptual models stemming from behavioral finance can help the techni- various parameters are tied together.
cal analyst construct and test systems of technical analysis. Further, the techni- Depending upon the trader-analyst’s time horizon and confidence in certain
cian can harness his or her intuitive grasp of crowd psychology for practical indicators, an arsenal of specific technical indicators can be judiciously se-
application through the use of a modified version of a life cycle framework. lected. If, for example, an analyst is an intermediate-term options or futures
The framework I have in mind is the adoption-diffusion model of crowd trader, then he or she might wish to examine the price parameter using stochastics
behavior, illustrated in Figures 1 and 2. These figures show how a society adopts or relative strength index (RSI) to study momentum, an hourly Dow Jones chart
an innovation over time. The graph takes on the shape of a bell curve to repre- to count Elliott waves, and a point-and-figure chart of the DJIA to measure the
sent the number of people adopting the change each period and looks like an S- potential extent of moves. These price indicators are seen positioned along the
curve when representing the number of people on a cumulative basis. S-shaped curve.
For examples of crowd behavior, we can look at such classic cases as With respect to volume, the analyst might include total daily New York
Tulipmania, the South Sea Bubble, and the Mississippi Scheme. The herd in- Stock Exchange (NYSE) volume, a measure of overall upside vs. downside
stinct of financial markets in these cases can be analyzed using the adoption- volume, and perhaps also a further refinement of an on-balance volume study
diffusion model. The crowd phenomenon is also observable in market cycles of of the 30 stocks in the DJIA. Volume is appropriately viewed under price on the
much shorter duration and smaller magnitude. Finally, recent observations in bell-shaped curve.
the news media reinforce the logic behind the use of a life cycle model of crowd Sentiment can be seen as measuring both the opinion and the behavior of
behavior, drawing the link between technical analysis and behavioral finance various market participants. Sentiment indicators of opinion are captured by
models. the feedback loop and indicators of sentiment behavior fit into the bell-shaped
adoption curve. Here, the analyst might choose to evaluate market opinion by
Curves, S-shaped and Bell-shaped using the Investor’s Intelligence ratio of bulls to bears. In addition, he might
evaluate prevailing sentiment using the headlines and leading stories from news-
The herd instinct is reflected by the S-shaped curve of the life cycle model
papers and magazines. He can then appraise speculative behavior by calculat-
(Figure 1), while the bell-shaped model shows how groups of market partici-
ing OEX put/call open interest and volume ratios. Finally, the intermediate-term
pants may be positioned and interrelated, ranging from the smart money to
investor might utilize the fourth major parameter, time, by analyzing a 10- to
those who enter the market last (Figure 2). Together, the two form a cycle model
13-week trough-to-trough cycle, the duration spent in a given trend, and sig-
that can be used to organize indicators to gauge technical market conditions
nificance of seasonal influences or special days in the month.
and to predict crowd behavior. In fact, economic theorist Theodore Modis ar-
By framing the indicators into the model shown in Figure 3, the trader can
Figure 2: Adopter
Categorization. The
innovation dimension,
as measured by time at
which an individual
adopts an innovation, is
continuous. However,
this variable may be
partitioned into five
adopter categories by
laying off standard
deviations from the
average time of
adoption.
INDICATORS
Three deep at every position
Four elements: Price, volume, time, and sentiment Figure 4: Three Deep
Three levels of analysis, or three units of analysis for every element At Every Position.
This form shows each
Element Unit Indicators* Weighting element broken down
into three levels or
Bullish-Bearish units of analysis.
+4 +2 +1 -1 -2 -4 Entries under
“Indicators” and
Price Momentum “Weighting” depend
Extent on the market and time
Form frame being analyzed.
Volume Total
Upside/Downside
On-balance
Time Cycle
Duration
Season
Sentiment News
Opinion
Speculation
* Indicators chosen by the analyst. Depending on the time frame used and the market studied, each technician
can systematically select an array of specific technical indicators to represent each element of the model.
better judge when the odds are optimal to buy an upside breakout. The com- Further Reading
bined picture of price-volume-sentiment-time appears different in the lower-
left quadrant (accumulation) of the model than in the upper right (distribution). ■ Bazerman, Max [1998]. Judgment In Managerial Decision Making, fourth
One would want to buy every high-volume upside breakout in the former case, edition, John Wiley & Sons.
but not when the latter circumstances appear to prevail. Further timing clues ■ Granville, Joseph [1976]. New Strategy Of Daily Stock Market Timing,
are given by the classic bottom-reversal patterns in the lower left and the clas- Prentice-Hall.
sic top-reversal patterns in the upper right.
■ Hartle, Thom [1998]. “On a new market paradigm: Henry Pruden of Golden
The model also gives the trader grounds for establishing numerical bench-
Gate University,” interview, Technical Analysis of STOCKS & COMMODI-
marks for entry and exit signals. These benchmarks can come from backtesting
TIES, Volume 16: September.
and real-time experience. Certain indicators can be given more weight, and the
threshold levels between bullish, very bullish, bearish, and very bearish de- ■ Investor’s Intelligence/Chartcraft, 30 Church St., New Rochelle, NY 10801.
pend on the analyst’s choice of indicators, beliefs about the market, and expe- ■ Krow, Harvey [1969]. Stock Market Behavior: The Technical Approach To
rience. Understanding Wall Street, Random House.
The adoption-diffusion life cycle model allows the technical trader to use ■ LeBon, Gustave [1995]. The Crowd, Isis Large Print.
the rich array of indicators available in software packages, at the same time
avoiding being overwhelmed by data. As can be seen in Figure 5, small S- ■ Leonard, Brent L. [1996]. “Answering the bell of sentiment indicators,”
shaped life cycle curves build into larger ones. Thus, the model provides a Market Technicians Association Journal, spring-summer .
systematic way of viewing and interrelating the daily, short-term, intermediate ■ MacKay, Charles [1995]. Extraordinary Popular Delusions And The Mad-
and long-term trends of the stock market. ness Of Crowds, Crown Publishing.
■ Modis, Theodore [1994]. “Life cycles: Forecasting the rise and fall of al-
Gaining an Edge most anything,” The Futurist, September-October.
The field of technical trading has become too competitive for a trader to ■ Peters, Edgar E. [1991]. Chaos And Order In The Capital Markets, John
blindly rely on a system that may no longer be feasible. Trading in the markets Wiley & Sons.
is fast approaching the levels of competition found in professional sports; we ■ Pruden, Henry O. [1995]. “Behavioral finance: What is it?” Market Techni-
need something to help us gain that extra edge. Behavioral finance models can cians Association newsletter and MTA Journal, September.
help us frame our technical information to gain that advantage. ■ Rogers, Everett M., and F. Floyd Shoemaker [1971]. Communications of
Innovations, Free Press.
Biography
Henry Pruden, Ph.D. is professor of business and executive director of
the Institute for Technical Market Analysis, Golden Gate University, in San
Francisco. He can be reached at Golden Gate University, 536 Mission Street,
San Francisco, CA 94105, phone 415/442-6583, fax 415/442-6579, e-mail:
hpruden@ggu.edu
noates
noates
noates
noatesnoates
noates
noates