55 - 2001 Summer

SM

A Publication of

Spring-Summer 2001 ● Issue 55

MARKET TECHNICIANS ASSOCIATION, INC.One World Trade Center ● Suite 4447 ● New York, NY 10048 ● 212/912-0995 ● Fax: 212/912-1064 ● e-mail: [email protected] ● www.mta.org

A Not-For-Profit Professional Organization ● Incorporated 1973

MTAJOURNAL

MTA JOURNAL • Spring-Summer 2001 2

MTA JOURNAL EDITORIAL STAFF 3

ABOUT THE MTA JOURNAL 4

MARKET TECHNICIANS ASSOCIATION, INC. 5

2000-2001 MTA BOARD OF DIRECTORS AND MANAGEMENT COMMITTEE 6

EDITOR'S COMMENTARY 7

Henry O. (Hank) Pruden, Ph.D., Editor

NEW APPLICATIONS FOR OPEN INTEREST IN U.S. TREASURY BOND FUTURES:A MONEY FLOW VS. BREADTH APPROACH 9

Sal Greco, CFA, CMT

APPLYING VODOPICH'S INTEGRATION OF ELLIOTT AND GANN TO THELONG-TERM STUDY OF THE DOW JONES INDUSTRIAL AVERAGE 21

Blethyn Hulton

THE METAPHYSICAL IMPLICATIONS OF THE ELLIOTT WAVE PRINCIPLE 29

Jordan E. Kotick, B.A (Hons), M.A., CMT

A CRITICAL STUDY ON THE EFFICACY OF STOP-LOSS 35

William K.N. Chan, CFA

INTERMARKET BREADTH INDICATORS: DOES THE PRICE ACTION OF INTEREST RATESENSITIVE STOCKS PROVIDE CLUES TO TRENDS IN BONDS PRICES? 41

Gary Stone, CMT

WYCKOFF TESTS: NINE CLASSIC TESTS FOR ACCUMULATION; NINE NEW TESTS FOR RE-ACCUMULATION 51

Henry O. (Hank) Pruden, Ph.D.

THE MTA JOURNAL – TABLE OF CONTENTS

SPRING-SUMMER 2001 • ISSUE 55

1

23

4

5

6


EDITOR

Henry O. Pruden, Ph.D.Golden Gate University

San Francisco, California

ASSOCIATE EDITOR

David L. Upshaw, CFA, CMT Jeffrey Morton, M.D. CMTLake Quivira, Kansas PRISM Trading Advisors

Missouri City, Texas

Connie Brown, CMTAerodynamic Investments Inc.

Pawley's Island, South Carolina

John A. Carder, CMTTopline Investment Graphics

Boulder, Colorado

Ann F. Cody, CFAHilliard Lyons

Louisville, Kentucky

Cynthia KaseKase and CompanyAlbuquerque, NM

Charles D. Kirkpatrick, II, CMTKirkpatrick and Company, Inc.

Chatham, Massachusetts

Cornelius LucaBridge Information Systems

New York, New York

Theodore �E. Loud, CMTTel Advisor Inc. of Virginia

Charlottesville, Virginia

John McGinley, CMTTechnical Trends

Wilton, Connecticut

Michael J. Moody, CMTDorsey, Wright & Associates

Pasadena, California

Richard C. Orr, Ph.D.ROME Partners

Marblehead, Massachusetts

Robert B. PeirceCookson, Peirce & Co., Inc.

Pittsburgh, Pennsylvania

Kenneth G. Tower, CMTUST Securities

Princeton, New Jersey

J. Adrian Trezise, M. App. Sc. (II)Consultant to J.P. Morgan

London, England

PRODUCTION COORDINATOR

Barbara I. GompertsFinancial & Investment Graphic Design

Marblehead, Massachusetts

PUBLISHER

Market Technicians Association, Inc.One World Trade Center, Suite 4447

New York, New York 10048

MANUSCRIPT REVIEWERS

THE MTA JOURNAL

SPRING - SUMMER 2001 • ISSUE 55


A NOTE TO AUTHORS ABOUT STYLE

You want your article to be published. The staff of the MTA Journal wants to help you. Our commongoal can be achieved efficiently if you will observe the following conventions. You'll also earn the thanksof our reviewers, editors, and production people.

1. Send your article on a disk. When you send typewritten work, please use 8-1/2" x 11" paper. DOUBLE-SPACE YOUR TEXT. If you use both sides of the paper, take care that it is heavy enough to avoidreverse-side images. Footnotes and references should appear at the end of your article.

2. Submit two copies of your article.3. All charts should be provided in camera-ready form and be properly labeled for text reference. Try

to avoid using the words "above" or "below," but rather, Chart A, Table II, etc. when referring to yourgraphics.

4. Greek characters should be avoided in the text and in all formulae.5. Include a short (one paragraph) biography. We will place this at the end of your article uponpublication. Your name will appear beneath the title of your article.

We will consider any article you send us, regardless of style, but upon acceptance, we will ask you tomake your article conform to the above conventions.

For a more detailed style sheet, please contact the MTA Office, One World Trade Center, Suite 4447,New York, NY 10048.

Mail your manuscripts to:Dr. Henry O. Pruden

Golden Gate University536 Mission Street

San Francisco, CA 94105-2968

The Market Technicians Association Journal is published

by the Market Technicians Association, Inc., (MTA)

One World Trade Center, Suite 4447, New York, NY

10048. Its purpose is to promote the investigation and

analysis of the price and volume activities of the world's

financial markets. The MTA Journal is distributed to

individuals (both academic and practitioner) and li-

braries in the United States, Canada, Europe and sev-

eral other countries. The MTA Journal is copyrighted

by the Market Technicians Association and registered

with the Library of Congress. All rights are reserved.

ABOUT THE MTA JOURNAL

DESCRIPTION OF THE MTA JOURNAL


Members Affiliates■ Invitation to MTA educational meetings ■■ ■■

■ Receive monthly MTA newsletter ■■ ■■

■ Receive MTA Journal ■■ ■■

■ Use of MTA library ■■ ■■

■ Participate on various committees ■■ ■■

■ Colleague of IFTA ■■ ■■

■ Eligible to chair a committee ■■

■ Eligible to vote ■■

Annual subscription to the MTA Journal for nonmembers: $50 (minimum two issues).

Single issue of the MTA Journal (including back issues): $20 each for members and affiliates and

$30 for nonmembers.

✔ ✔

✔ ✔

✔ ✔

✔ ✔

✔ ✔

✔ ✔

✔ ■■

✔ ■■

MARKET TECHNICIANS ASSOCIATION, INC.MEMBER AND AFFILIATE INFORMATION

MEMBER

Member category is available to those "whose professional efforts are spent practicing financial technicalanalysis that is either made available to the investing public or becomes a primary input into an active portfoliomanagement process or for whom technical analysis is a primary basis of their investment decision-makingprocess." Applicants for Member must be engaged in the above capacity for five years and must be sponsored bythree MTA Members familiar with the applicant's work.

AFFILIATE

Affiliate status is available to individuals who are interested in technical analysis, but who do not fully meetthe requirements for Member, as stated above; or who currently do not know three MTA members for sponsor-ship. Privileges are noted below.

DUES

Dues for Members and Affiliates are $200 per year and are payable when joining the MTA and thereafterupon receipt of annual dues notice mailed on July 1. College students may join at a reduced rate of $50 with theendorsement of a professor.

APPLICATION FEES

Applicants for Member will be charged a onetime, nonrefundable application fee of $25; no fee for Affiliates.

BENEFITS OF THE MTA


Director: PresidentPhilip B. Erlanger, CMT

Phil Erlanger Research Co. Inc.978/263-2536

Fax: 978/266-1104E-mail: [email protected]: Vice President

Richard A. DicksonHilliard Lyons502/588-4122

Fax: 502/588-9132E-mail: [email protected]

Director: SecretaryBruno DiGiorgi

Lowry's Reports Inc.561/842-3514


Director: TreasurerAndrew Bekoff

Van Der Moolen LLC212/495-0558


Director: Past PresidentDodge Dorland, CMT

LANDOR Investment Management212/737-1254


DirectorBruce M. Kamich, CMT

Reuters, Inc.212/859-1643

Fax: 212/859-1649e-mail: [email protected]

DirectorCharles Kirkpatrick II, CMT

Kirkpatrick & Co.508/945-3222


DirectorPhilip J. Roth, CMT

Morgan Stanley Dean Witter212/761-6603


DirectorKenneth G. Tower, CMT

UST Securities Corp.609/734-7747


2000-2001 BOARD OF DIRECTORS AND MANAGEMENT COMMITTEEOF THE MARKET TECHNICIANS ASSOCIATION, INC.

Management Committee(4 Officers, Past President and Committee Chairs)

AccreditationDavid L. Upshaw, CFA, CMT

913/268-4708Fax: 913/268-7675

E-mail: [email protected]

Neal Genda, CMTCity National Bank

310/888-6416Fax: 310/888-6388

E-mail: [email protected] of Knowledge

John C. Brooks, CMTYelton Fiscal Inc.

770/645-0095Fax: 770/645-0098

E-mail: [email protected] LearningRichard A. Dickson

Hilliard Lyons502/588-4122


EducationPhilip J. Roth, CMT



Ethics & StandardsLisa M. Kinne, CMT

Salomon Smith Barney212/816-3796


FoundationBruce M. Kamich, CMT

Reuters, Inc.212/859-1643

Fax: 212/859-1649e-mail: [email protected]

IFTA LiaisonMike Epstein

NDB Capital Markets Corp.617/753-9910


Internship CommitteeJohn Kosar, CMT

Bridge Information Services312/930-1511

Fax: 312-454-3465E-mail: [email protected]

Board of Directors(4 Officers, 4 Directors & Past President)

Journal�Henry (Hank) O. Pruden, Ph.D.

Golden Gate University415/442-6583


LibraryDaniel L. Chesler, CTA, CMT

561/793-6867Fax: 561/791-3379


Larry KatzMarket Summary & Forecast

805/370-1919Fax 805/777-0044


Michael N. KahnBridge Information Systems

212/372-7541E-mail: [email protected]

PlacementRick Bensignor



Programs (NY)Bernard Prebor

MCM MoneyWatch212/908-4323


RegionsM. Frederick Meissner, Jr., CMT

404/875-3733E-mail: [email protected]

RulesGeorge A. Schade, Jr., CMT

480/949-8946Fax: 480/429-9491


Nina G. CooperPendragon Research, Inc.

815/244-4451Fax: 815/244-4452



As technical analysts strive to raise their body of knowledge and discipline to higher

standards, the canons of the scientific method and empirical testing will become more

and more commonplace. In earlier Editorials, the case was made for the formulation

and testing of “Theoretical Models.” In this issue there are examples of the two polar

approaches to doing empirical research in the social sciences, hence in behavioral fi-

nance and technical analysis. These two poles can be characterized as “survey research”

findings across a large number of instances to establish statistical validity and reliability

on the one extreme, and the “individual case study” on the other pole. The individual

case study illustrates numerous variables and contingencies operating together in a

real-life situation.

Most of our MTA Journal articles rely upon the “survey” type of technique, which is

appropriate and excellent. But there remains an art form to the execution of trades

and a complexity of interacting indicators and decisions that are better revealed through

the “story” of a case study. In this issue of the MTA Journal the article on “Wyckoff

Rules” is an exposition based upon a case study. Other technical analysts are encour-

aged to consider using the single-case approach to document their studies, perhaps

relating case studies based upon their own experiences.

MULTI-METHOD RESEARCH


EDITOR'S COMMENTARY


ANOTHER PERSPECTIVE

“I have a great respect for orthodoxy; not for those orthodoxies which prevail in particular

schools or nations, and which vary from age to age, but for a certain shrewd orthodoxy which the

sentiment and practice of laymen maintain everywhere. I think that common sense, in a rough

dogged way, is technically sounder than the special schools of philosophy, each of which squints

and overlooks half the facts and half the difficulties in its eagerness to find in some detail the key

to the whole. I am animated by distrust of all high guesses, and by sympathy with the old preju-

dices and workaday opinions of mankind: they are ill expressed, but they are well grounded.

My philosophy is justified, and has been justified in all ages and countries by the facts before

every man's eyes; and no great wit is requisite to discover it, only (what is rarer than wit) candor

and courage. Learning does not liberate men from superstition when their souls are cowed or

perplexed; and, without ‘special’ learning, clear eyes and honest reflection can discern the hang

of the world, and distinguish the edge of truth from the might of imagination.”

George Santayana, Preface to a New Philosophy

APPROACHES TO THE BODY OF KNOWLEDGE INTECHNICAL MARKET ANALYSIS


EDITOR'S COMMENTARY


1INTRODUCTION

The conventional wisdom of technical analysis maintains that Vol-ume (Vol) and Open Interest (OI) studies are breadth indicators.Each measures the continued health, or potential weakening, of agiven trend. We expect increases in Vol and OI as prices move in thedirection of the primary trend, while we tolerate decreases in each asprices move contrary to that trend. This application of Vol and OIdata is clear in the technical analysis literature.

My work suggests that more valuable and timely applications ofOI exist in the exchange-traded futures markets, specifically the USTreasury Bond Futures market. In addition to providing broad con-firmation of trend, or signaling pending divergence to the trend, OIfigures provide vital insight into sentiment levels within particularmarkets. As such, OI provides information which is often contrary innature and more appropriately utilized as a shorter-term sentimentgauge rather than as a longer-term breadth indicator. This paperwill show how OI data analysis can be extended into the area of mar-ket sentiment and contrary opinion, through a greater dissection ofthe Commodity Futures Trading Commission’s Commitment of Trad-ers Reports. It will demonstrate ways to decompose changes in OIthat better reflect the true nature of positions in the UST Bond Fu-tures market, and offer methodologies that can be used to forecastprice action on a shorter-term basis.

TWO FRAMEWORKS FOR OI ANALYSIS

OI analysis can be conducted under two distinct, but complemen-tary, frameworks: a Breadth framework and a Money Flow frame-work. Each framework interprets changes in Total OI as an indica-tion of whether money is flowing into or out of a given market. Ris-ing OI represents money flowing into a market, while falling OI rep-resents money flowing out. From this point of common interpreta-tion, the two frameworks quickly diverge.

A Breadth FrameworkA Breadth framework does not differentiate between the activity

of different participants. All new money (rising OI) supports theprimary trend, regardless of who is establishing new positions and onwhich side of the market they are doing so. So long as new positionsare established, there is fuel available to feed the trend. This under-scores the longer term nature of a Breadth framework, where thegoal is to identify signals of confirmation of or divergence from tothe longer-term trend.

Figure 1 – Formation of OIPurchase Sale OI Effect

New Long New Short Increases

New Long Old Long Unchanged

Old Short New Short Unchanged

Old Short Old Long Decreases

There are several basic principles of OI analysis with which alltechnicians are familiar. These include how OI is affected by buyingand selling in the zero-sum environment of the futures market, aswell as the generalized interpretations of changes to Total OI accom-

panying specific price action. These principles represent the basictenets of a Breadth framework, and are summarized in figures 1through 3.

Figure 2 – Interpretation of Price/Vol/OI InterplayOpen Market

Prices Volume Interest Interpretations

RISING Increasing Increasing STRONG

Decreasing Decreasing WEAK

FALLING Increasing Increasing STRONG

Decreasing Decreasing WEAK

The inherent delay between Breadth divergence and price rever-sals makes it difficult to translate breadth analyses into short-run trad-ing strategies. Some of the more recent work on OI has attempted toimprove upon the Breadth framework in this regard. Hadady (1989)developed the Price Open Interest (POI) Indicator to address thisissue. Hadady characterized percentage changes in Price and OI alonga Bullish-to-Bearish continuum. He reasoned that during those peri-ods in which OI diverged from price action, bullish and bearish sig-nals could be derived from the relative percentage changes of Priceand OI. Though intellectually appealing, application and interpre-tation are difficult for a number of reasons.

Figure 3 – Interpretation of Price /OI PrinciplesCommon Major Interpretation Underlying Reasoning

I. Rising OI in an Uptrend is Bullish New Longs and New Shorts Entering

II. Decling OI in an Uptrend is Bearish Short-Covering and/or Long Selling

III. Rising OI in a Downtrend is Bearsh New Longs and New Shorts Entering

IV. Declining OI in a Downtrend is Bullish Short-Covering and Longs Selling

Painter (1995) addressed the same issue with his On-Balance OpenInterest Indicator (OBOII), a variant of Joseph Granville’s On-Bal-ance Volume (OBV). Painter used the major tenets and principles ofOI to create a time series which provided signals of confirmation anddivergence on a cumulative basis. His methodology incorporatedthe use of dummy variables (not derived from actual market data)which were generated by four different market “states.” He used thesevariables to construct a cumulative index that identified changes tothe intermediate trend. Though it provides an ongoing, cumulativemeasure of breadth, the inherent weaknesses of a breadth frameworkremain.

A Money Flow FrameworkA Money Flow framework distinguishes between the sources of

the money entering the market, and the sources of the money leav-ing the market. It does not assume that all money is created equal –at least over the short-run. In fact, a Money Flow framework reasonsthat bad money will eventually follow good money into a market. Thiscan have dire consequences to a market over the short-term, regard-less of whether the longer term is showing broad participation. Hence,all new money entering a market is not unequivocally good. In itsmost basic form, a Money Flow framework seeks to identify the strongversus weak hands of the market and to determine which faction isdriving price action at any given time.

NEW APPLICATIONS FOR OPEN INTEREST IN U.S. TREASURYBOND FUTURES: A MONEY FLOW VS. BREADTH APPROACH

Sal Greco, CFA, CMT


Rather than characterizing changes to Total OI, a Money Flowframework identifies changes in the composition of Total OI. Thepositioning of similarly categorized participants, over time, in a zero-sum futures market, often yields the presence of strong-handed and/or weak-handed players whose net positions cycle with price action.If such a situation exists, principles of contrary opinion can be usedto generate more effective Buy and Sell signals than could be gener-ated from a Breadth framework. A Money Flow framework providessuperior Buy and Sell signals because of its shorter time frame, andits differentiation of market positions.

For purposes of OI analysis, the separation of strong-handed play-ers from weak-handed players is accomplished through decomposi-tion of the Commodity Futures Trading Commission’s (CFTC) Com-mitment of Traders (CoT) Reports. Since an understanding of theCoT Report and reporting process is critical to understanding theMoney Flow framework, the CoT Report is discussed more fully inAppendix A.

Conventional wisdom suggests that, as a general rule, the LargeTrader category of the CoT Report represents the strong-handed,“smart” money compared with the Small Trader’s weak-handed“dumb” money. Belveal (1985) offered just such a methodology forbroad application of the CoT data. He split gross long and short OIpositions into percentages based on whether the positions were heldby Large (Commercial and Non-Commercial) or Small Traders (Non-Reporting). Belveal’s analyses focused on how changes in OI shiftedthis allocation, and compared the relative positions of Large and SmallTraders over time.

Jiler (1985) analyzed the forecasting ability of the major identifi-able groups of market participants: Large Hedgers (LH), LargeSpeculators (LS), and Small Traders (ST). He used Net open con-tract positions as a percentage of OI to determine the relative posi-tions of participants. Further, in an attempt to remove any seasonal-ity present in the underlying commodity market, Jiler derived nor-malized or average positions to which he would compare actual posi-tions taken from the CoT reports. He viewed material deviationsfrom the norm as a measure of bullish or bearish attitudes on themarket. Jiler found that LH and LS had the best track record forforecasting price action across markets, with the LH consistently su-perior to LS. However, LS performance varied widely from marketto market. ST had the worst record by far, across virtually all mar-kets. Jiler concluded, therefore, that bullish signals on the marketshould be taken when, on a percentage basis, LH are net long morethan normal, LS are net long, and ST are heavily net short by morethan their normalized amount.

A MONEY FLOW FRAMEWORK FOR

TREASURY FUTURES

Decomposition of the Treasury Bond Future CoT ReportThis conventional wisdom, as represented by Belveal and Jiler,

provides an incomplete decomposition of the CoT Report for bonds.Exhibits 1 and 2 separates the bond CoT Report into the traditionalcategories of analysis: Large Traders (LT), Small Traders (ST), Com-mercials (Comml), and Speculators (Specs). Exhibit 1 shows that LTand ST positions are perfect offsets. This is a basic identity propertyof the CoT Report. Since Total Net Positions must sum to zero, andall positions are classified as either LT or ST, these two componentswill always, by definition, offset each other. Comml and Specs, whosepositions are clearly inversely related, are not perfect offsets, sinceeach are included in the LT component of the CoT Report. There-fore, unless the LT/ST division completely reflects behavior, it may

be an incomplete and/or inaccurate decomposition of positioningin the market. This is true for bonds. The CoT Report combines theactivity of participants whose Net Positions (and hence market views)are inversely related, while separating the activity of players who arelike-minded. Therefore, neither a Comml versus Spec nor a LT versusST segregation will accurately reflect the true level of market senti-ment contained in the CoT Report.

Exhibit 1

Exhibit 2


The utility of the CoT Report is derived from two measures: TotalAbsolute Net Positions and the positioning of Comml versus Specs and STs.[I refer to the combined positions of Specs and ST as FADE posi-tions.] Any decomposition must capture and measure both to beeffective. Total Absolute Net Positions (TAbs Net Pos) gauge themagnitude of conflicting opinion in the market by measuring howfar from Neutral the net positions of adversarial participants havemoved, while the relative positioning of Comml vs. Specs/ST depictsthe polarity (long v. short) of the smart money.

Although there is no identity property that states the Spec and STcategories must align, they do so more often than not, especially asTAbs Net Pos increase. For example, the Spec and ST componentsare aligned similarly in 304, or 63%, of the 482 weekly observationsin the CoT Report for bonds between January 1986 and November1998. The observations that do not align occur in periods of rela-tively low TAbs Net Positions. Specifically, of the 178 (37%) observa-tions in which the Net Pos of Specs and NRPT are not aligned, 153,or 86%, occur when TAb Net Positions are less than 15% of SpreadAdjusted OI. As can be seen in Exhibit 3, this is a relatively low read-ing for TAbs Net Pos in the CoT Report for bonds. Simply put, whenthe magnitude of disagreement is small, the relative positioning ofthe players matters less.

With this in mind, a more appropriate CoT decomposition forbonds is provided in Exhibit 4. The first graph shows TAbs Net Posbetween January 1986 and November 1998. The second and thirdgraphs show Net Comml Pos and Net FADE Pos, respectively. Al-though the sum of the absolute values of the second and third graphwill not always equal the first, the latter two graphs are now perfectoffsets. This confirms that the strong- and weak-handed componentsare correctly accounted for, even if they fail to perfectly align on op-posite sides of the market in every observation.

In order to remove the trends caused by dramatic increases in OIover the last few years (Figure 4), each category is restated as a per-centage of Spread Adjusted OI (defined in Appendix A) in Exhibit 5.This adds a degree of relativity to the time series for purposes of his-torical comparison. These OI-adjusted time series are defined in Fig-ure 5. Further, Exhibit 5 clarifies the relationship between TAbs NetPos and CoT%C & CoT%FADE. As TAbs Net Pos increase, conflict-ing opinions/positions in the market are on the rise. As OI is buoyedby this increase in TAbs Net Positions, the market position of thestrong versus weak hand of the market becomes more polar or ex-treme. As the exhibit shows, CoT%C and CoT%FADE can move toeither extreme as TAbs Pos are forced higher.

Exhibit 3

Figure 4 : US Treasury Bond OI (Jan86-Nov98)

Figure 5 : Definition of CoT DerivationsCoT%LT: Net Pos of LT as % of Spread Adj OI

CoT%ST: Net Pos of NRPT as % of Spread Adj OI

CoT%C: Net Pos of Comml as % of Spread Adj OI

CoT%S: Net Pos of Spec as % of Spread Adj OI

CoT%Fade: Net Pos of (Spec + ST) as % of Spread Adj OI

Exhibit 4

Both CoT%C and CoT%FADE are the algebraic equivalent of adiffusion index. A diffusion index measures the difference between3 or more variables that have cyclical properties, making it the mostappropriate measure to determine the relative positioning betweenthe strong and weak hands in the bond market. Although CoT%C isdefined as “Net Commercial Positions as a percentage of Spread Ad-justed OI”, the identity properties of OI assure that CoT%C is pro-


portionally equivalent to the difference between the Net Positions ofCommercials and the Net Positions of the FADE Category (CoT%C –(CoT%S + CoT%ST). Specifically, the CoT%C is equivalent to twicethe magnitude of a pure Diffusion Index which utilizes CoT%C,CoT%S, and CoT%NR.

(CoT%C - CoT%S - CoT%NR)

Spread Adjusted OI

[Appendix B provides an algebraic derivation of the DiffusionIndex (DI)]

Exhibit 5

The definition and interpretation of the DI is relatively simple.Since the sum of the three terms must equal zero, and the sum of twomust be equal and opposite to the third, the DI must vary between 2and -2. These extreme readings represent the largest potential dif-ference in Net Positions between Comml and FADE categories. Thevalue is positive when the Comml’s are on the long side, and negativewhen Comml’s are on the short side. Since CoT%C is equal to one-half of the DI value, it will vary proportionally between +1 and -1 witha similar interpretation. Exhibit 6 shows both CoT%C and CoT DI.Regardless of which measure is used, this formulation represents thebroadest derivation of the relative positioning between the strong andweak hands in the Treasury Futures market provided by the CoT Re-port.

Exhibit 6

Market and Trading ApplicationsThese methodologies can be applied over relatively short time

frames for the purposes of market-timing in the bond market. Appli-cation addresses three issues.

First, the short-term assessment of market opportunity or risk isdriven largely by TAbs Net Pos. TAbs Net Pos represent the primarygauge of market conflict. When TAbs Net Pos reaches a relativelyhigh level, a reversal of price action on a reduction in OI can beexpected. During these periods, execution of positions in the direction of thetrend (purchases in uptrends; sales in downtrends) should be delayed whileexecution of positions or transactions against the trend (purchases indowntrends; sales in uptrends) should be accelerated. Likewise, when TAbsNet Pos reverses from relatively high levels to more moderate or be-low average levels, the degree of conflict that drove prices to recenthighs or lows has likely moderated. Generally, this represents a re-duction in the speculative forces that were driving price action. Asthese forces subside, the trend and the breadth of participation inthe prevailing trend can be reassessed.

Second, CoT%C or the CoT DI is used to determine which waythe market is likely to break, once the speculative forces have ex-hausted themselves. As stated earlier, when the TAbs Net Pos is high,positioning of the players becomes more meaningful. Thus CoT%Cand CoT DI will show whether the strong-handed players are on thelong or short side of the market at these times of maximum positionconflict. If the strong hands are long (CoT%C relatively high), long posi-tions can be taken, or purchases accelerated/sales slowed, in anticipation of arally that ensues as the speculative forces that pushed prices lower are ex-hausted. Conversely, if the strong hands are short (CoT%C relatively low),short positions can be taken, or sales accelerated/purchases slowed, in antici-pation of a sell-off that occurs as the speculative forces that had pushed priceshigher are exhausted. [Box A provides 3 examples of application.]

CoT Diffusion Index (DI) = --------------------------------------------------- = 2* CoT%C


Exhibits 7A

Box A

See Exhibits 7 and 7A. In Jan 98, bonds traded as high as 123.17 on a closing basis. Spread Adj OIincreased by approximately 32% between early Sep and mid-Jan as bonds were making their highs. TAbsNet Pos rose from a neutral 1.0% to 31.6% at the high. Coincident to the increase in TAbs Net Pos,CoT%C moved from .3% to -15.8%, a bearish extreme. This was a clear signal that bonds were spurredhigher on an increase in speculative activity. The increasing conflict between the strong and weak factionsof the market was underscored by a dramatic increase in TAbs Net Pos, and CoT%C maintained theunwinding of conflicting opinion would lead to lower prices. By the time this conflict was unwound inearly Mar, and these indicators moved back to Neutral levels, bonds had fallen 4-3/4 points.

A similar scenario developed in Jan 96 as bonds traded to the year low yield of 5.92%. Bonds rallied 85/8 points between the end of Sep and early Jan 96 on a 28% increase in OI. As OI increased over thatperiod, TAbs Net Positions rose from 6.4% to 21.2%, suggesting increasing conflict and polarity in marketsentiment and positions. CoT%C’s move from 3.1% to -8.2% suggested this would be resolved throughlower Bond prices. Over the ensuing 4 months, Bond prices fell over 14 points as these indicators movedto Neutral levels.

Price action in May 95 illustrates how similar increases in TAbs Net Positions led to dramatic rallies.Bonds were fighting psychological resistance at 5.25% in May 95 following the 1994 Bear Market in bonds.After rallying over 3-1/2 points from the beginning of the year to early Feb, bonds rallied an additional 3-1/8 points by early May 95. OI increased by 7.5% over the first part of the rally, and fell by approximately2.5% over the latter part. As bonds rallied nearly 6-3/4 points between Jan and May95, TAbs Net Positionsincreased from 13.9% to 26%. CoT%C increased from 6.9% in early Jan to 13.0% in early May followingonly a slight moderation in early Feb. In this instance, CoT%C suggested the increased conflict in posi-tioning would be resolved via higher prices since the strong hand of the market had continued to accumu-late positions. The rally that ensued saw bonds rally by more than 9-1/2 points in little over four weeks.

Exhibits 7


Exhibit 8 Exhibit 8A

BOX B

Jul 97 and Oct 98 are two times in which TAbs Net Positions did not mirror thechange in Spread Adjusted OI. [See Exhibit 8 and 8A]. In Jul 97 bond’s were complet-ing a rally begun in mid-Apr97. Spread Adj OI increased by 21% as bond prices climbednearly 8-1/2 points. Coincident to the increase in OI, TAbs Net Pos increased from9.0% to 30.6% at the highs. As bond prices fell by 2-3/4 points in Aug, OI increased by4%. However, as OI rose, TAbs Net Pos moved from 30.6% to 1.0%, in a dramaticmoderation in sentiment. As such, the market was unencumbered by speculative ex-cess as prices began to climb in early September. This set the stage for further gains.By the time TAbs Net Pos revisited the 30.8% area in mid-Dec, bonds were seven pointshigher.

Oct98 represents a recent example of OI and TAbs Net Positions moving in oppo-site directions. In this instance, a dramatic decrease in Spread Adjusted OI was accom-panied by an increase in TAbs Net Pos. During the liquidity crisis of Jul 98-Oct 98bonds rallied smartly as a flight to quality put a bid under the Treasury market. Bondprices rallied 10 points between the end of Jul and early Oct. Spread Adjusted OI,which had already peaked in mid-Jun, fell an additional 15% between the end of Juland the high prices of early Oct. Over the same period, TAbs Net Pos rose to 36.4%from 20.8%. Therefore, as money was leaving the market (OI falling), prices werebeing driven higher by speculative forces (i.e. TAbs Net Pos rising; CoT%C falling).The divergence between prices (up 10 pts) and OI (down 25%) quickly became acutewhen TAbs Net Positions reached the 36.4% level. Prices reversed quickly thereafter.


BIBLIOGRAPHY

■ Barrie, Scott. “The COT Index”. Technical Analysis of Stocks &Commodities. (September 1996).

■ Belveal, L. Dee. Charting Commodity Market Price Behavior (2nd

ed). Dow Jones-Irwin. Homewood, IL. 1985.■ Bianco, James. Various Works. Bianco Research, L.L.C. [An Af-

filiate of Arbor Research & Trading , Inc.] Barrington, IL.■ Hadady, R. Earl. Contrary Opinion. How to Use It For Profits In

Trading Commodity Futures. Hadady Publications, Inc. Pasadena.1983.

■ Hadady, R. Earl. “The POI Index. Market Analysis via Price andOpen Interest Interaction”. MTA Journal (May 1989). MarketTechnicians Association, Inc. New York.

■ Hayes, Timothy W. “Intermarket Sentiment: Using Sentiment inOne Market to Call Prices in Another”. MTA Journal. (Winter1994/Spring 1995). Market Technicians Association, Inc. NewYork.

■ Hyerczyk, James A. “Making a Commitment”. Futures Magazine.(November 1996).

■ Jiler, William L. “Analysis of the CFTC Commitments of TradersReports Can Help You Forecast Futures Prices”. 1985 CommodityYear Book. Commodity Research Bureau. Jersey City. 1985.

■ Murphy, John J. Technical Analysis of the Futures Market: A Com-prehensive Guide to Trading Methods and Applications. New YorkInstitute of Finance. New York. 1986.

■ Painter, William M. “On Balance Open Interest Indicator”. MTAJournal. (Fall-Winter 1995). Market Technicians Association, Inc.New York.

■ Shaleen, Kenneth H. Volume and Open Interest. Cutting EdgeTrading Strategies in the Futures Markets. Probus Publishing Com-pany. Chicago. 1991.

BIOGRAPHY

Sal Greco, CFA, CMT is a Director of Fixed-Income Invest-ments for the Metropolitan Life Insurance Company. He is theSenior Market Strategist for the Department's Cross Sector Rela-tive Group which addresses issues of asset allocation and relativevalue across the company's portfolios. He is responsible for pro-viding both fundamental and technical analyses on the domes-tic fixed-income markets, in addition to providing both short-and longer-term interest rate and yield curve forecasts for thetrading and portfolio management units of MetLife.

Mr. Greco is also the President of SG Research & Strategy,L.L.C., a corporation he formed in 1995 to provide consultingservices, market analyses, and trading strategies to individualsand institutions.

Third, the relationship between TAbs Net Pos and Total OI out-standing can be used to determine whether significant corrections tothe trend are necessary and should be anticipated. TAbs Net Postypically increases as Total OI increases. This dynamic provides theearly signs that bad money is beginning to follow good money intothe market over the short-to-intermediate time frame. However, thisneed not be the case. Increasing OI amidst flat or decreasing TAbsNet Pos would suggest a price move that is sustainable over both theshort-term and the long-term. Increasing OI amidst increasing TAbsNet Pos suggests the market may be suffering the ills of speculativeexcess over the short-term. Whereas the former scenario warrantsrelatively small corrections in price action, the latter warrants moreintense corrections. [Box B provides 2 examples of application.]

CoT%C StudiesIn addition to contemporaneous analysis, technical studies con-

firm the utility of CoT%C as a market timing device. Over the 13year period between January 1986 and November 1998, a CoT%C orCoT DI methodology outperformed a Buy and Hold Strategy forbonds. Using a methodology similar to that used by Hayes (1994) instudying the application of sentiment measures as market-timing de-vices, CoT%C outperformed both a Buy and Hold Strategy for bondsas well as CoT%S and CoT%ST. My study applied volatility bands tosmoothed versions of the CoT%C, CoT%S, and CoT%ST time seriesderived from the CoT Report and used violation of those bands togenerate buy and sell decisions in the US Treasury market over a 13year period starting in January 1986 and ending in November 1998.[see Appendix C for a discussion of the test methodology.] The out-put of this study is summarized in Table 1.

Table 1: Summary of CoT%C Study

HOLDING PERIOD RETURNS AVG ANN COMP RETURNS.50 Std .75 Std 1.0 Std .50 Std .75 Std 1.0 Std

CoT%C 46.1% 53.5% 41.4% 3.2% 3.7% 2.9%

CoT%S 25.9% 26.0% 21.7% 2.0% 2.0% 1.7%

CoT%ST 19.0% 43.6% 39.5% 1.5% 3.1% 2.8%

Buy & Hold 29.6% 2.2%

Despite the usefulness of a CoT%C and TAbs Net Position meth-odology over the last several years, the methodology was misleadingat critical points in the 1990-93 Bull Market in bonds, and the BearMarket which followed in 1994. During the former, the CoT% indi-cator kept one out of the Bull Market for a continuous 23 monthperiod. During the latter, it provided a timely sell signal early in 1994,but reestablished an errant long position for six months in one ofthe worst Bear Markets in a quarter century. This can be seen inExhibits C-2 and C-3 of Appendix C.

CONCLUSIONS

The analysis of OI under a Money Flow framework derives vitalsentiment information directly from the positions of participants ina market. This internal, market-generated expression of sentimentcan be used in conjunction with the more popular survey-based, ex-ternal sentiment measures to gain a more thorough understandingof the underlying dynamics of the market and its price action.

Used in conjunction with traditional charting techniques, thesemethodologies can add significant confirmation to an independentprice action analysis, while providing a quantifiable advantage formarket-timing purposes in the US Treasury market.


Appendix ATHE CFTC COMMITMENT OF TRADERS REPORT

HistoryThe CoT Report is a detailed accounting of total gross open posi-

tions in a given commodity, traded on a specific futures exchange(ie. CBOT T-Bond Futures), categorized by account type. The CoTReport was originated by the US Department of Agriculture (USDA)in an attempt to differentiate between who was selling and who wasbuying futures contracts. The Commodity Exchange Authority (CEA),a division of the USDA, issued a monthly report which segregatedmonth-end OI into Reporting (large) and Non-Reporting (small andforeign) traders. The Reporting category was divided again intoCommercial and Non-Commercial trading account categories. By1976, the Commodity Futures Trading Commission (CFTC) assumedresponsibility for the CoT reporting mechanism in addition to ex-panding coverage into new futures markets. The CFTC publishedthe CoT Reports on a monthly basis until 1982, when they were dis-continued for budgetary and reporting reasons. The reports resumedin early 1983 and the CFTC expanded coverage once again to newcontracts in the rapidly growing futures markets. Between 1983 andDecember 1990, the CFTC released the CoT Reports monthly on ei-ther the 11th day of the month or the first business day which fol-lowed, if the eleventh was a non-business day. Between January 1991and October 1992, the CFTC published the CoT on a twice monthlybasis, reflecting OI holdings on the 15th of the month (or previoustrading day if the 15th was a holiday) and the last trading day of themonth. In October 1992, the CFTC began compiling the OI dataweekly, reflecting holdings as of the close on each Tuesday, releasingthe report every other Friday. This cycle remains in effect today.Therefore, as it now exists, the CFTC CoT Reports are released everyother Friday, reporting OI positions on a weekly basis using the grossopen positions as of the close of each Tuesday.

DecompositionThe CoT Report decomposes OI into Total Gross Positions (long

and short) held by “Reporting” and “Non-Reporting Accounts.” Allclearing members of the exchange, futures commission merchants,and foreign brokers on the exchange are required to make daily re-ports to the CFTC showing each of their customer accounts’ posi-tions on their books, that in any contract month of a commodity,exceeds the reporting level. If the accounts’ position does not ex-ceed the reporting level, the aggregate positions of the customer ac-counts are classified as Non-Reporting for CoT purposes. If they ex-ceed the reporting level, account positions are reported under theReporting category. [The current reporting level for Bonds is 500contracts.]

The Reporting Category is further divided into “Non-Commer-cial” and “Commercial” categories. Non-Commercials, more com-monly referred to as Speculators, are defined as accounts who havepositions in excess of the reporting limit, but do not take positions inthe commodities or securities underlying the futures contract. Com-mercials, more commonly known as Hedgers, are defined as accountswhose positions exceed the reporting limit and take positions in thecash commodities or securities underlying the futures contract. TheNon-Commercial category also contains the positions of “Spreaders”,accounts that have offsetting (long and short) positions in differentcontract months and do not hold positions in the cash commodityunderlying the futures contract.

The CoT Report provides a decompostion of gross open positions(long and shorts) on a total basis as well as into each of these catego-

ries. This results in the reporting of 9 specific cells or segments withinthe CoT Report over time. [Figure A-1 presents the actual CoT Re-port for the period ending 12/29/98 as released through BloombergFinancial Markets.] In addition to the gross number of contractsopen, the CoT Report presents positions in 3 other ways: change inOI since last report, as a percentage of total OI, and as number oftraders or accounts. Aside from the number of traders, the changesand percentages are derived from the gross OI numbers reported atthe top of the report. Therefore, it is that first line of data in the CoTreport as shown in Exhibit A1 that is the most important.

Figure A-1

Table A1: Decomposition of CoT Report

LONG SHORT NET TOTAL

LARGE Comml 424,421 506,241 (81,820) 930,662

Spec 92,484 36,718 55,766 129,202

SMALL (NRPT) 91,287 65,233 26,054 156,520

Spreader 12,650 12,650

620,842 620,842Total Absolute Net Positions 163,640

Open Interest 620,842

Spread Adj Open Interest 608,192

The definitions and terminology of the CoT report can be con-fusing, especially as released by most data vendors. It is more easilyunderstood in the context of Table A1, which re-orders the data pro-vided in the raw report of Exhibit A1. Table A1 removes the Spreadercategory from within the Non-Commercial category and shows it asits own category. Representation of the CoT data in this form allowsgreater clarity and is the basis for reformulation of the CoT data thatunderlies the methodology and indicators I present. Table A2 moreclearly represents the percentage-based measures of positioning inthe CoT Report.

Total OI is defined as one-half of all open positions outstanding.Thus, Total OI can be derived by summing all open long positions orall open short positions in the market. As evident from Table A1,total OI is equally represented by the sum of gross long positions orgross short positions. By definition, the spreader component has anequal number of open contract positions on the long and short sideof the market.


Table A2: CoT Positions as % of Spread Adjusted OI%Long %Short

Large Comml 69.8% 83.2%

Spec 15.2% 6.0%

85.0% 89.3%Small NRPT 15.0% 10.7%

100.0% 100.0%

Several things should become apparent from these two tables.First, despite the fact that total gross positions (long or short) maydiffer greatly between participant categories, Net Position differencesbetween categories are relatively small. For example, Comml grosslong positions are 4.6x larger than Spec gross long positions, whileComml gross short positions are 13.8 times larger than Spec grossshort positions. However, Comml Net Positions are only 1.5x SpecNet Positions (on an absolute basis). A similar, but opposite situa-tion exists between Specs and NRPT.

Second, the total Net Positions of all participants in a given fu-tures market must sum to zero. The zero-sum nature of futures trad-ing assures as much. This is the most basic identity property of OIand the CoT Report. Every purchase of a futures contract must beoffset by a sale of that contract. Further, for every open positionestablished through a short futures sale transaction, there is an openposition established through a a long futures purchase transaction.These two points form the basis of a Money Flow framework and un-derscore its differences versus a traditional Breadth framework.

Third, Spreader activity results in zero Net Positions, thus onlydiluting the effect of Net Positions when they are represented on apercentage basis. Further, the volatile and seasonal nature of spread-ing activity introduces unwanted noise into the OI figures for analy-sis purposes (Breadth or Money Flow). I utilize a measure of OI whichremoves this component. I term this Spread Adjusted OI. It is derivedby summing Comml, Spec, and NRPT positions on either the longor short side of the market. Alternatively, it can be derived by sub-tracting the Spreader component presented on the CoT Report fromTotal OI.

Open Interest =Comml Long + Spec Long + NRPT Long + Spread Long

orComml Short + Spec Short + NRPT Short + Spread Short

Spread Adjusted OI =Comml Long + Spec Long + NRPT Long

orComml Short + Spec Short + NRPT Short

orTotal Open Interest - Spreading OI

Total Absolute Net Positions=

|Net Comml Pos| + |Net Spec Pos| + |Net NRPT Pos|

Fourth, because the sum of total Net Positions must equal zero,the sum of the Net Position of any two of the trader categories mustbe equal in magnitude – but opposite in sign – to the third category.This is a corollary to the tenet that Net Open Positions must sum tozero. However it is often the case – but not the rule – that the sametwo categories will offset the third category especially when TotalAbsolute Net Postions are high. For example, 63% of the 482 weeklyobservations from the CoT Report for Bonds, Specs were aligned onthe same side of the market (short v. Long) with NRPT.

Finally, the importance of Total Absolute Net Positions should beimmediately apparent. Although Net Positions will always sum tozero, Total Absolute Net Positions will vary dramatically. [Total Ab-solute Net Positions in Table A1 are 163,640 cts. On a percentagebasis, that represents 27% of Spread Adj OI according to Table A2.]Total Net Positions represent the degree of conflicting opinion inthe market, as it measures the difference from Neutral of each of thethree participants’ Net Positions. As Total Absolute Net Postions riseon a percentage basis, the degree of conflict between the opinions ofthe players is increasing. It is this degree of conflict that will eventu-ally lead to a significant shift in sentiment (as OI is forced to decline)and defines major turning points in the market. Thus, Total Abso-lute Positions provide a basis for deriving sentiment and applying theprinciples of contrary opinion.

InterpretationThe conventional wisdom offered by Belveal (1985) and Jiler (1985)

suggests the LT component of the CoT will outperform the ST com-ponent; and that Comml can be expected to outperform Specs. Ratherthan address interpretation from the LT v. ST or Comml v. Spec divi-sion, it is more valuable to do so from a more generic “strong-hand”vs. “weak-hand” perspective. This avoids the differences betweenComml and Specs that invariably occurs across markets.

Comml’s represent the strong hands of the market versus the weakhand of the Spec or NRPT Trader, due to their superior market ex-perience and financial strength. Further, the Comml has “qualitiesof judgment, experience, and adequate capitalization” (Belveal, p42) which makes him/her a strong hand.

Specifically, Belveal cites many factors that allow Commercials orHedgers to represent and maintain the strong hand in the commod-ity futures market. Because they are not limited by position limitsand have business requirements and the financial wherewithal to buyor sell in multiples of contract size or market depth, the commercialsare the 100 pound gorilla. The commercial/hedger’s proprietaryinterest in the underlying commodity allows him/her to conduct de-fensive buying or selling when price changes appear to threaten his/hers interests. Finally, the necessity of having to transact in all mar-ket conditions and all price ranges makes the Commercial/Hedgerone of the “best analysts” his/hers market could have. Commercialplayers are always faced with the need to sell in low-priced marketsand buy in high-priced markets. In fact, they are often forced to doone only days or weeks after doing the other. Further, the on-goingnature of hedging operations/activities among commercial accountsattest to their profitability. If it were otherwise, hedging programscould not be justified and would rightly be abandoned by the Com-mercial players inside a market.

Small Traders (NRPT) and Specs are undercapitalized and lackthe information and experience of the larger Commercials. Belvealsums up their weak-handed nature quite simply: “The ‘weak hands’are weak not because they are likely to be wrong in their judgment,but because they perform so badly in the face of vicissitude”. (p. 151)

Such interpretations of positioning among categories of tradersgive the decomposition and derivations of the CoT Report a logicaland applicable anecdotal basis for application.


Appendix BALGEBRAIC DERIVATION OF A COT DIFFUSION INDEX

The identity properties of the CoT Report assure that the sum oftwo of the three components (Comml, Spec, ST) will be equal inmagnitude but opposite in sign to the third component.

CoT%C = (CoT%S + CoT%ST) * (-1)CoT%S = (CoT%C + CoT%ST) * (-1)CoT%ST = (CoT%C + CoT%S) * (-1)

The identity properties can be algebraically manipulated to provethat the CoT Diffusion Index (DI) is another form of the CoT%Cequation. The DI will always be equal to two times CoT%C. Themethodology of the study conducted upon CoT%C, CoT%S andCoT%ST will assure equivalent output between CoT%C, CoT%Fade,and CoT DI. Although their absolute levels may differ, they are equiva-lent on a relative or proportional basis.

Definintion: CoT DI = CoT%C - CoT%S - CoT%STSince, CoT%C = (CoT%S + CoT%ST) * -1, then....CoT%C - CoT%S - CoT%ST = ((CoT%S + CoT%ST) * -1) - CoT%S - CoT%STCoT%C - CoT%S - CoT%ST = - CoT%S - CoT%ST - CoT%S - CoT%STCoT%C - CoT%S - CoT%ST = - 2CoT%S - 2CoT%STCoT%C - CoT%S - CoT%ST = - 2 * (CoT%S + CoT%ST)- 1 * (CoT%C - CoT%S - CoT%ST) / 2 = (CoT%S + CoT%ST)Since, CoT%S + CoT%ST) = CoT%C * -1, then ...-1/2 * CoT Diffusion Index = CoT%C * (-1), and ...CoT Diffusion Index = 2 * CoT%C , or ....CoT%C = 1/2 * CoT Diffusion Index

Appendix CTESTING METHODOLOGY

In order to assess the utility of CoT derived series as market-tim-ing tools, I conducted a study using a methodology similar to thatused by Hayes (1994). [Hayes used weekly Market Vane Sentiment datafrom three markets (Gold, Bonds, and Stocks) to generate Buy and Sell signalsin - and across - those markets to determine if an active contrary opinioninvestment style outperformed passive Buy and Hold Strategies.] My objec-tive was to determine if an active contrary investment style using the3 component series I derived from the CoT Report for Bonds(CoT%C, CoT%S, CoT%ST) outperformed a passive Buy and Holdstrategy for Bonds over the preceding 12 years, and whether one se-ries proved superior to another.

The data requirements for this study were quite simple. TheCoT%C, CoT%S, and CoT%ST series were derived from CFTC datadownloaded from Bloomberg and the CFTC database. The data isclean and reliable. Daily closing prices for the nearby Bond Futurewere taken directly from the Chicago Board of Trade (CBOT) data-base and from Bloomberg. That information was reviewed and foundto be clean and reliable. Weekly close observations were chosen sothat the pricing day corresponded to the last day covered by each ofthe CoT Reports, or to the day closest to the last day covered by theCoT release. There was no constraint that prices be “executable”,since the goal of the study was to determine whether a contrary rela-tionship between Bond prices and CoT Position data existed. Theconstruction of a trading system was not a goal, so the constraints oftrading system design were not applied.

First, a smoothed version of all three CoT indicators was obtainedby deriving 13-week moving averages of the raw CoT%C, CoT%S,and CoT%ST time series from the CoT data. Second, volatility bandswere generated at three standard deviation magnitudes (.50, .75, 1.0)around a 13-week moving average of each of the originally smootheddata. Third, buy and sell signals were generated only after thesmoothed data moved through a band, reached an extreme, and thenreversed back through the same band. Exhibit C-1 shows the smootheddata series for CoT%C and the volatility bands generated at a 1.0 Stdlevel.

Since the CoT%C sentiment series represents “strong” hands or“smart” money, a buy signal is generated when CoT%C moves below theupper band after violating it on a move to an extreme positive reading.A sell signal is generated when CoT%C moves above its lower bandafter violating it on a move to an extreme negative reading.

The CoT%S and CoT%ST series represent “weak” hands or “dumb”money. Therefore, a buy signal is generated when the smoothed CoT%Sand CoT%ST series move above the lower volatility band after violatingit on the way to a an extreme negative reading. A sell signal is generatedwhen the smoothed series moves below the upper band after violating iton a move to an extreme positive reading.The assumption underlying the model is simple. $1.0 mm is avail-

able on 1/2/87. A Buy and Hold strategy is represented by purchaseof a 1 mm Par Bond at the closing price of that day, held over theentire period, and valued at the closing price on the last day of thestudy. The active management style was driven by Buy and Sell sig-nals generated from our three CoT derived sentiment series and theirvolatility bands. The $1.0 mm opening balance remained in cashuntil a Buy signal was generated by the model. The model was inNeutral mode at the start of the study period.

Exhibit C-2 displays the price action of the Nearby Bond Futurealong with the Buy and Sell signals generated by the CoT%C series at


Exhibit C-1

Exhibit C-2

Exhibit C-3

the 1.0 std band level. It outperformed a Buy and Hold Strategy overthe the 12 year period by 1180 bps, or approximately 290 bps on anannually compounded basis. Table C-1 summarizes the output for allthree data series at all three std levels. [This table also appears in thebody of the text as Table 1.]

An active CoT%C strategy would have outperformed a Buy andHold Strategy over the 1986 to 1998 period, suggesting that CoT%Cis a valid sentiment indicator, and that a Money Flow framework forOI Analysis can improve market timing decisions in the Treasury mar-ket. Further, the CoT%C appears superior to both the CoT%S andCoT%ST series. This supports the contention that CoT%C - and itsmany equivalents - is the broadest and most accurate decompositionof the CoT data.

However, two things should be noted in regard to the applicationof this methodology. First, the output exhibits significant varianceStd magnitudes. In fact at higher levels of Std (>1.0), none of theseries returned profitable results versus a Buy and Hold strategy. Thisis a result of timing sensitivity. Clearly, marginal differences in tim-ing of Buys and Sells can change the results dramatically. Since Bondprices can change dramatically week to week, small differences be-tween signals and market highs and lows can inject a considerableamount of variance into this type of study. Exhibit C-3 lists the Buyand Sell signals for each series at each Std level.

Table C-1: Summary of CoT%C Study

HOLDING PERIOD RETURNS AVG ANN COMP RETURNS

.50 Std .75 Std 1.0 Std .50 Std .75 Std 1.0 Std

CoT%C 46.1% 53.5% 41.4% 3.2% 3.7% 2.9%

CoT%S 25.9% 26.0% 21.7% 2.0% 2.0% 1.7%

CoT%ST 19.0% 43.6% 39.5% 1.5% 3.1% 2.8%

Buy & Hold 29.6% 2.2%


INTRODUCTION

In Trading For Profit With Precision Timing, Don Vodopich enhances theanalysis of Elliott Waves by attaching Gann lines to wave starting points.With this approach, Vodopich studies a variety of futures markets in the short-term as is appropriate for trading. The purpose of this research is to applyVodopich’s technique to the long-term study of the stock market, specifically theDow Jones Industrial Average (DJIA), which was the focus of Elliott’s work.The period covered is Supercycle Wave (V) as counted by Frost & Prechter,which began in 1932. The Vodopich analysis breaks down the present cycle ofCycle-degree (i.e., Supercycle Wave (V)) as well as certain Primary, Intermedi-ate, and Minor cycles within it.

BACKGROUND

Vodopich argues that the slope of the line from the start of animpulse wave to its end can be known in advance because the “slopeis particular to the tradable.”1 For example, the slope of an impulsewave of several months’ length in T Bond futures always has a slopeof .10 points (about 3 ticks) per trading day. Such a line is a Gannline since unlike a trendline its slope can be known at its startingpoint, but like a trendline it identifies support and resistance. Chart1 demonstrates how such a Gann line attached to an impulse wavemight work in theory. The Gann line would indicate support upthrough the top of impulse wave 3, would be firmly broken by wave 4,and would indicate resistance thereafter up to the end of the im-pulse wave, the end of wave 5.

Chart 1

Gann Lines Drawn Onto Elliott Impluse Wave

In Vodopich’s work, the slope of the Gann line belonging to anycomponent wave is an integer multiple of the slope of its parent.Further, as Chart 1 shows, the Gann lines of various fifth waves oflesser degrees intersect the parent wave at its end; this intersection islabelled a “Vodopich Convergence” in Chart 1.

Vodopich analyzes both impulse and corrective Elliott waves withGann lines, but this study examines only impulse waves since long-term stock market moves are such waves. Gann lines have tradition-ally been associated with particular angles, such as 45 degrees; butthis is a carryover from the days of manual charting. The angle of atrendline depends on the aspect ratio of the chart and how price andtime are scaled. In the electronic era, when most charts conform tothe aspect ratio of the computer screen or the laser-printed page, the

enduring value of a Gann line is its slope (rise/run, or number ofpoints per trading period), which is determined for each tradablebased on historical observation, or in Vodopich’s words, “The properscale Gann line work is one which has worked in the past.”1 Appen-dix 1 reproduces Gann-line slopes of various tradables, as determinedby Vodopich and John Murphy in their respective works.

AMENDMENT TO VODOPICH’S METHOD

The author has changed the practice of Vodopich’s approach infour ways; generally, the changes relax some of the original rules,improving their application to the study of the stock market. Thefirst change is to start the Gann line at the beginning of wave 3 ratherthan wave 1. In the stock market, the thorough retracement of wave2 would most often slice through any line started at the beginning ofwave 1; and as a result, the Gann line would not as effectively indicatesupport through wave 3. The second change is to accept, using ana-lytic judgment, a Gann-line slope which varies from that dictated byhistorical observation. As will become evident in the examples,Vodopich’s method is more useful during the more volatile 5th wavethan in the 3rd wave. A wave 3 is so strong that it does not need asmuch analytic aid; and so the author works with trial-and-error dur-ing wave 3 to find the line that will work for him thereafter, in waves4 and 5. The third change is to allow the slope of a component waveto be a round multiple (e.g, 1.5x, 2x, 2.5x) of the parent wave, notnecessarily an integer multiple. And finally, the author views theVodopich convergence to be not the end of the trend, but rather apoint in time and/or price where the reward/risk ratio drops dra-matically. As the examples will demonstrate, the momentum builtup through wave 5 may well carr y price beyond the intersection ofthe Gann lines; and if price stalls there, the market may trade side-ways before reaching a new high. In either case, the reward/riskratio falls, suggesting a change in asset allocation. And the Vodopichconvergence alerts one to the greater risk as well as the prospect of areversal.

Chart 2T Bond Futures Contract from April '97 to January '98

Chart 2 demonstrates an amended Vodopich analysis of T Bondfutures from April ’97 through January ’98; the chart plots the 12/97contract through November ’97 and the 3/98 contract thereafter.The impulse wave begins at the 4/11/97 low of 105.788; but the firstGann line, with a slope of .10 points/trading day, is drawn from the

2APPLYING VODOPICH'S INTEGRATION OF ELLIOTT AND GANN TO THELONG-TERM STUDY OF THE DOW JONES INDUSTRIAL AVERAGE

Blethyn Hulton


start of wave 3 at 107.75 on 5/27. Inspecting the chart, one can seethat the “.10 line” did reveal support up through the wave 3 top on8/1, capturing the late June correction. Note that a line drawn fromthe beginning of wave 1 with the same slope would have sliced throughthe ensuing rally, rendering the Gann line useless.

Wave 4 of this rally ended at the 8/26 low of 111.438, firmly break-ing the “.10 line.” Both during wave 4 and after, this same line iden-tified resistance until January ‘98. The second Gann line begins the3rd wave within wave 5, at the 9/11 low of 111.938. It has a slope of.15 points per trading day, 1.5x the slope of the line attached to theparent wave. The “.15 line” indicated support through late Novem-ber ’97. After that line was firmly broken by the 4th wave within wave5, the last Gann line was drawn from the low of 117.78 1 on 12/9/97(using the 3/98 contract). The slope chosen for this line was .20points per day, 2x the slope of the original Gann line and 1.33x theslope of the second line. This slope is chosen in part because it di-rects the line to the intersection of the first and second Gann lines.This intersection, a Vodopich convergence, at 124.5 in mid-January1998, is near the intraday high of 124.28 on 1/12/98 (using the 3/98T Bond futures contract), which remained the high for severalmonths.

This demonstration of Vodopich’s integration of Elliott and Gannsupports its theoretical argument, that an Elliott impulse wave is as-sociated with a Gann line whose slope “has worked for the tradablein the past”1; that Gann lines of component waves have slopes thatare round multiples of the slope of the parent wave; that these Gannlines indicate support through wave 3 and resistance thereafter; andthat the Vodopich convergence, the intersection of the Gann lines,identifies a point in time/and or price (both time and price in thiscase) at which the reward/risk ratio drops substantially. Appendix 2contains the dates and prices of the Elliott waves discussed above withrelevant Fibonacci ratios.

Readers familiar with the T Bond action in 1998 know that afterthe January high, T-Bonds corrected until April ’98 and then ralliedto a historic high in October. This price movement raises the ques-tion whether there was a larger cycle operating which this Vodopichanalysis did not capture. The issue of a hierarchy of cycles is centralto the long-term Elliott Wave analysis of the stock market, the focusof this research and of Elliott’s original study.

APPLYING GANN TO LONG-TERM BULL MARKETS

Referring to a hierarchy of cycles in the Elliott Wave, Frost &Precther stated that “as far as we can determine, then, all waves bothhave and are component waves.”2 Frost & Prechter demonstratedthis idea with the Dow Jones Industrial Average, and Table 1 liststheir hierarchy of waves as well as the labelling of those waves to bediscussed here. If one combines Elliott’s argument for a completehierarchy of waves with Vodopich’s argument that for any “target”wave the Gann-line slopes of component waves are round multiplesof the Gann-line’s slope for the “target” wave, then with any one im-pulse-wave Gann line, one should be able to find the Gann-line slopesfor all parent and component waves throughout the hierarchy.

In search of these Gann lines, this research examines Primary,Intermediate, Minor, and Minute cycles from 1945 to 1997, begin-ning with the present Minute cycle, which began in 1994. This pre-sentation departs from the standard Elliott Wave analysis, which is towork inward from the parent wave to the component waves, in orderto highlight the current cycle, with which readers are most familiar.The largest cycle under study, the cycle of Cycle-degree from the 1932low, will be presented last.

Since this is a long-term study of the Dow, Gann-line slopes are

expressed in relative, percentage change rather than absolute pointchange just as such charts plot price on a logarithmic scale ratherthan a linear scale3; and all percent slopes discussed here are annu-ally compounded. As the reader will see, the slope of the Gann linefor the largest cycle under study (the cycle of Cycle-degree) is 8.5%per year, and the slopes of the Gann lines for all component wavesare round multiples of that. A final note is that for the highs andlows in the DJIA used to mark the endpoints of waves, theoreticalhighs and lows are reported before May 1991 and actual highs andlows thereafter.

Table 1Frost & Prechter's Count of Millennial, Century and

Generational Waves

THE PRESENT MINUTE CYCLE (MINOR CYCLE WAVE

5) - THE BULL MARKET FROM 1994 TO THE PRESENT

Chart 3 plots weekly bars of the bull market from March ’94through March ’98. This bull market, which corresponds to MinorCycle Wave 5, has a Gann-line slope of 34% per year, 4x the slope ofSupercycle Wave (V). Although the Minute Cycle began at the 4/4/94 low of Dow 3552.47, the Gann line begins at Wave iii, 3638.62 on11/23/94; and as Gann lines should, this line indicated support untilthe Wave iii top of 5796.10 on 5/23/96.

After the “34% line” was firmly broken in the mid-’96 correction,Wave v began (Wave iv ended) at Dow 5182.31 on 7/16/96. Thesecond Gann line starts at the beginning of Wave 3 within v, at 5561.46on 9/3/96, drawn with a slope of 51% per year, 1.5x the slope of thefirst Gann line (its parent wave) and 6x the slope of 8.5%. Note thatas Vodopich’s method would indicate, price rose during Wave 3 within5 in between the first, “34% line,” indicating resistance, and the sec-ond, “51% line,” indicating support. Price broke the support of thesecond Gann line after the top of Wave 3 within 5, at Dow 7112.10 on3/11/97. The third and final Gann line of this Minute Cycle (MinorCycle Wave 5) is drawn from the 6356.77 low on 4/14/97, which cor-responds to the beginning of the Wave iii within 5 within v. Theslope of this line is 102% per year, 2x the slope of the second Gannline drawn for this cycle and 3x the first. Appendix 3 contains thedates and prices of the Elliott waves discussed above with relevantFibonacci ratios.


Chart 3The Present Minute Cycle (Minor Cycle Wave 5)Dow Jones Industrial Average from 1994-1997

In summary, Chart 3 demonstrates a Vodopich analysis of a long-term (32 month) Elliott impulse wave of the Dow. The first and sec-ond Gann lines, at 34%/yr. and 51%/yr., indicate support throughthe 3rd wave of their cycle and resistance thereafter. However, thethird Gann line, the “102% line,” performed differently from thefirst two, indicating support through the 3rd wave but never resis-tance. All three lines intersected in late July ’97, near the 8/7/97high of 8299.49. And while this Vodopich convergence did not markthe absolute high, it did indicate the time and price after which pricemovement became more volatile. The Dow did not break the 8299high for six months, in February ’98; and from August ’97 throughAugust ’98, the Dow suffered two significant corrections, -16% inOctober ’97 and -21% in July/August ’98.

The discussion of Minor Cycle Wave 5 is also a practical exampleof how one applies Elliott and Gann to long-term stock market im-pulse waves. Early into what one believes is a 3rd impulse wave, early1995 for the cycle in Chart 3, one draws a Gann line whose slope is around multiple of the slope of the parent-wave Gann line; at the time,this is an academic exercise since the Gann line does nothing to es-tablish that the cycle has begun its 3rd wave. The inability of a Gannline to contribute to the assessment of a 3rd wave is no drawbacksince 3rd waves are the easiest of all waves to spot. In the words ofFrost & Prechter, “Third waves are wonders to behold. They are strongand broad, and the trend at this point is unmistakable. Increasinglyfavorable fundamentals enter the picture as confidence returns . . . .Such points invariably produce breakouts, ‘continuation’ gaps, vol-ume expansions, exceptional breadth, major Dow Theory trend con-firmations and runaway price movement.”4

Relying on other measures, then, to validate the wave 3, one canexperiment with Gann lines of two or three different slopes (e.g.,1.5x, 2x, 2.5x the slope of the parent wave) to find the line that bestindicates support. In the latter half of a 3rd wave, in late 1995 forMinor Cycle Wave 5, one should expect to find the single Gann linethat best indicates support through the 3rd wave. The analyst mustbe careful not to create a Gann line that is too tight, whose slope istoo steep; such a line probably belongs to the 3rd wave within wave 3,not wave 3 overall. To check that the Gann line chosen is properand not too steep, the author draws a price channel, using the Gannline as its lower boundary. The usefulness of this technique is basedon the author’s observation that price channels are common andeasy to see in 3rd waves.

Price firmly breaking the Gann line (as the July ’96 correctionbroke the “34% line”) is reason to suspect a wave 4 correction. Uponresumption of the uptrend, one draws the second Gann line, repeat-ing the evaluative process that led to the first. Now, however, duringthe wave 3rd wave within the 5th wave, one watches for the secondGann line to indicate support while the first line indicates resistance;and this happened in Minor Cycle Wave 5 from September ’96through March ’97. With the third Gann line, drawn after the sec-ond is broken, one is approaching either the end of the impulse wave,or, as is the case in Chart 3, a significant change in trend, reducingthe reward/risk ratio. The third Gann line confirms the intersectionof the first two; and as price approaches the Vodopich convergence,one can adjust one’s equity exposure for the expected higher volatil-ity and lower reward/risk ratio.

The Vodopich convergence is not a reversal signal. As has beendiscussed in the T Bond futures represented in Chart 2 as well as theU.S. equity bull markets shown in Charts 3 and 5, more often thannot, the trend resumes after a Vodopich convergence, albeit withgreater volatility and after some consolidation. This likelihood pointsout a limitation in the author’s use of Vodopich’s integration of El-liott and Gann, since the addition of Gann lines leaves part of theimpulse wave unmeasured. An addendum to this paper examiningthe price action from August ’97 to the present addresses the ques-tion of how to analyze such price action.

THE CURRENT PRIMARY CYCLE (CYCLE WAVE V) -THE BULL MARKET FROM 1974

Chart 4 plots monthly bars of the DJIA from the 1974 low throughMarch 1998 and indicates the particular wave count for this Primarycycle which the Vodopich analysis suggests. For reference, note thatChart 3 (covering the period from 1994 onward) fits in the upperright hand corner of Chart 4. Note also that the “34% line” in Chart4 is the same line of the same slope from Chart 3. The Vodopichanalysis of this period leads to a different wave count than that sug-gested by Frost & Prechter5; and that wave count, in turn, leads totwo important conclusions. The Vodopich analysis suggests first thatthe entire period from 1982 to the present is Primary Cycle Wave((3)) and second that the 1990 high, not the 1987 high, was the topof Intermediate Wave (3) within ((3)). The argument here for therevised wave count is one of “fit” rather than “proof.” The explana-tion fits the market action to date. However, it does not provide aperfect Elliott wave count, nor is it the necessarily the count thatwould have presented itself at various times in the past.

In this count, the Primary Cycle (Cycle-degree Cycle Wave V) be-gins at the 12/9/74 low; and the first Gann line, with a slope of 12.75%,begins at the 8/31/82 low of 769.98, which in this count representsthe start of Primary Wave ((3)). Here again, the integration of El-liott and Gann is iterative; neither the wave count nor the Gann linewere fixed before adding the other. The choice for the Gann-lineslope and the start-date for drawing it (with the wave count impliedby this choice) are made in an effort to find the best wave count andGann line for the price action that followed.

What is fixed in a Vodopich analysis is that the Gann line’s slopemust be a round multiple of the Gann-line slope of the parent wave(12.75%/yr = 1.5 x 8.5%/yr.) and that the Gann line indicates sup-port until it is firmly broken by the 4th wave in the cycle, wave ((4))in this case. As one can see from Chart 4, the “12.75% line” hasremained consistently beneath the market lows after 1982, i.e., thelows in 1984, 1987, 1990, and 1994.


Chart 4The Current Primary Cycle

(Primary Waves ((1)) through ((3)) Cycle-degree Cycle Wave V)Dow Jones Industrial Average from 1974-1997

A wave count of the Intermediate Cycle within Primary Wave ((3))supports the author’s argument for the parent-wave count and is in-cluded in Chart 4. A Gann line with a slope of 17%/yr. begins at the7/25/84 low of Dow 1078.9, which starts Intermediate Wave (3) withinPrimary Wave ((3)). The slope of the line is 2x the slope of SupercycleWave V (8.5%/yr.) and 1.33 x 12.75%, the slope of the parent wave,Primary Wave ((3)). Further, the Gann line indicates support forthe price movement up to a certain point, the 1990 top. However, itdoes not indicate meaningful resistance thereafter; price remains wellunder the “17% line” between 1990 and 1997. (The importance ofthe “17% line” is discussed later).

In the wave count implied by this Gann line, the 1990 top, not the1987 top, represents the end of Intermediate Wave (3), thereby low-ering the significance of the 1987 crash by one degree in the Elliottwave count. This wave count finds validity in the following Fibonacciratios. The ratio of Wave (3) to Wave (1) is 2.63 (+180.3%:+68.7%);and within Wave (3), the ratio of Wave 5 to Wave 3 is .63 (+87.1%:+137.9%). This wave count is unusual in that the high of wave 5(Dow 3024.30) was not much beyond the high of wave 3 (2746.65 inAugust ’87). However, the Vodopich analysis of the T Bond futurescontract yields a wave count that is similar in that the later, marginalhigh with the less dramatic correction emerges as the 3rd wave top.In Chart 2, the marginal high in November immediately precedesthe breaking of the Gann line, thereby suggesting the wave 3 top,even though a more dramatic correction followed the slightly lowerOctober high. The final point to support the argument for the 1990top as the end of wave (3) is that the 1990 correction (IntermediateWave (4) in this count) corresponded to a larger economic phenom-enon, the 1991 recession. Appendix 4 contains the dates and pricesof the Elliott waves discussed above, Primary Wave ((3)) and Inter-mediate Wave (3), with relevant Fibonacci ratios.

Further, as one can see in the upper right hand corner of Chart 4,the “17% line” intersects two other lines, the “34% line” (the same“34% line” from Chart 3) and the line labelled “8.5%/yr.” The “8.5%line” is the Gann line drawn from the 1942 low; and it contributes toa Vodopich convergence, an intersection of two or more Gann lines,around the time and price of the same Vodopich convergence in Chart3, occurring near the Dow’s 1997 high of 8299.49 on 8/7/97. As onecan see in Table 2, which projects the five Gann lines to 8/7/97, allprice projections are within .32% to 3.24% of the actual high. Theaccuracy of these projections, based on five Dow prices from 1942 to1997, demonstrates the insight gained from adding Gann lines tolong-term Elliott impulse waves.

Table 2Projections of Gann Lines to August 7, 1997 and

Comparison with Actual High for Dow Jones Industrial Average

There is no satisfactory Vodopich breakdown of Intermediate Wave(5), the bull market from 1990 to the present; a wave count for Wave(5) is included in Appendix 4 with relevant Fibonacci ratios. Wave(5) is different in that it is concave. In the author’s observation, it ismore common for an impulse wave to look convex, to bow upward,because the 3rd wave advances strongly with less volatility than the5th wave; and it is this convexity that allows a Gann line to indicatesupport through the third impulse wave and resistance thereafter. InIntermediate wave (5), price advanced +70.0% through the top ofMinor wave 3 in January 1994; but from that top of Dow 3985.95,price advanced more, +108%, through the 8/7/97 top of 8299.49.This anomaly is consistent with the unusual economic recovery inthe U.S. after the 1991 recession. The postrecession earnings boomdid not come in 1991 and 1992 as would normally be the case; theboom came in 1994 through 1997.6 And the early, slow advance instock prices reflected the early, modest growth in earnings.

PRIMARY CYCLE (CYCLE WAVE III) - THE BULL

MARKET FROM 1942 TO 1972Chart 5

The Primary Cycle from 1942-1972 (Cycle-degree cycle Wave III)Dow Jones Industrial Average

Chart 5 plots monthly bars of the DJIA from 1942 to 1972, thebull market between World War II and the Vietnam War; and theVodopich analysis confirms Frost & Prechter’s wave count for thisperiod (Cycle Wave III).7 Wave III of Supercycle Wave (V) began atthe 4/28/42 low of Dow 92.92; and from here the Gann line with theslope of 8.5%/yr. is drawn. This line clips the 6/13/49 low of 161.5,which marks the start of Primary Wave ((3)) within Cycle Wave III.And from this point, a Gann line is drawn with a slope of 12.75%/yr.,1.5 x 8.5%/yr., the slope of the parent wave. The “12.75% line” indi-cates support during Wave ((3)), capturing the pullbacks in 1953and 1957, until it is broken in the early 1960s. In this way, the “12.75%


line” validates Frost & Prechter’s count of the Wave ((3)) top in 1958.7The second Gann line for Cycle Wave III is drawn from the start

of Wave (3) within ((5)), at 558.06 on 10/32/62. Its slope of 25.5%/yr. is 2x the slope of the Gann line of the parent wave, and 3x theslope of 8.5%, the slope of the grandparent wave, Cycle Wave III.This second Gann line behaves as a third Gann line often does, notindicating resistance once broken; and no third Gann line emergesfrom the analysis of this period to confirm the intersection of thefirst two. The first two lines do intersect in early 1966 around Dow1250, and this marked Frost & Prechter’s Cycle Wave III top in timebut not in price.

Here again, the Vodopich convergence does not signal a reversalbut rather the beginning of a period in which volatility increases sub-stantially with some advance in price. The 1973 Dow peak of 1051.7exceeds the 1966 peak by 5.7%; but in between the peaks the Dowsuffers two corrections, -25% in 1966 and -36% in 1968-70. Frost &Prechter’s use of complex wave patterns to analyze this period is takenas support for the argument that the portion of a trend after theVodopich convergence is inherently more difficult to analyze withthe Elliott Wave. To analyze the time from 1966 to 1982, Frost &Prechter’s original (1978) count labels the period from 1966 to 1974as an expanding triangle; but in 1982, Prechter favored a double-three wave count, ending Cycle Wave IV in 1982.5

An addendum to this paper analyzes the Dow following theVodopich convergence in August ’97. In this real-time post-conver-gence analysis, the author will suggest how to deal with such a pe-riod.

THE CURRENT CYCLE OF CYCLE-DEGREE

(SUPERCYCLE WAVE (V) ) -THE BULL MARKET FROM 1932 TO THE PRESENT

Chart 6The Bull Market from 1932-Present

Dow Jones Industrial Average

Chart 6 plots quarterly (3 month) bars of the Dow Jones Indus-trial Average from 1910 to the present along with three Gann linesand Cycle Wave turning points. The first two Gann lines, with slopesof 8.5% and 12.75%, are identical to the lines with the same slopes inCharts 5 and 4 respectively. As one can see in Chart 6, the Gann linesapplied to the Dow from World War II onward indicate support andresistance in the same manner as Gann lines in Charts 2 through 4.The first Gann-line, the “8.5% line,” indicates support through CycleWave III; this line is firmly broken by Cycle Wave IV; but the line doesnot indicate any meaningful resistance until the summer of 1997, ashas been discussed previously.

The first two Gann lines in Chart 6 intersect at Dow 29,000 in2012, and this intersection represents a price and/or time target forSupercycle Wave (V). This target is not necessarily the end of thecycle. In Charts 2 through 5, price continued to rise or trend side-ways after a Vodopich convergence. However, in all cases the volatil-ity also increased dramatically, reducing the reward/risk ratio sub-stantially; regardless of the price movement after 2012, one can ex-pect much higher volatility, frustrating long-term market analysis.

This Wave (V) target differs from H. S. Dent’s projection in TheRoaring 2000s of Dow 35,000 in 2008 for a methodological reason.In the chart with the 35,000 target, Dent has drawn a trendline withabout the same slope as the “12.75% line” in Chart 6, although Dentstarts the line from the September ’84 low, not the September ’82low used here. His 35,000 target comes from a parallel, channel linefrom the 1987 high, projected to 2008, the time of the market peaksuggested by his demographic analysis.8 The methodological differ-ence between Dent’s forecast and the one suggested here is that Dentbelieves that the price channel, bound underneath by the trendlinesimilar to the “12.75% line,” continues until the market top. TheVodopich analysis concludes that the trendline will be broken by the4th wave, Primary Wave ((4)), and that once broken it will serve asresistance during the 5th wave, Primary Wave ((5)). As resistancethen, the “12.75%” line marks the upper boundary of price move-ment, not the lower boundary, as it seems to for Dent.

Chart 6 includes a third Gann line, sloping upwards towards Dow29,000 in 2012 at the compound annual rate of 25.5%. Since theDow has yet to complete Primary Wave ((3)), this is a tentative Gannline for Primary Wave ((5)); its purpose is to outline a possible sce-nario for the final wave. The slope of the line is 2x the slope of theGann line of the parent wave, Cycle Wave V, and 3x the slope of the“grandparent” wave, Supercycle Wave (V). As incredible as it mightbe to think of the stock market rising 25% per year for several years,Chart 3, charting the Dow from 1994, shows periods when the mar-ket advanced at a faster pace. Further, in the 1920s, the Dow rose24.9% per year (compounded) from the August ’21 lowest close of63.90 to the September ’29 highest close of 381.17. Thus, it is impor-tant to be open minded to the prospect of the market scoring 20%plus annual returns before dividends during the next decade. Fur-ther, as has been shown in other charts, one can expect the slopes ofcomponent waves to be round multiples of the slope of Primary Wave((5)). For example, using a base slope of 25.5%/yr., the Gann line ofIntermediate Wave (3) within Wave ((5)) might have a slope of38.25%/yr. (1.5x 25.5%/yr).

The long-term application of Vodopich’s integration of Elliott andGann has ignored corrective waves because the author has not founda consistent methodology for them; as a result there is no means inthis analysis to measure how far Primary Wave ((4)) might run. Frost& Prechter’s guideline is that “corrections, especially when they arefourth waves, tend to register their maximum retracement within thespan of travel of the previous fourth wave of one lesser degree, mostcommonly near the level of its terminus.”9 In the Elliott Wave countpresented here, Frost & Prechter’s guideline suggests that PrimaryWave ((4)) will end in the range of Intermediate Wave (4) of ((3))(the 1990 correction), Dow 2344.31 to 3024.26.

This target for Primary wave ((4)) and the “25.5% line” from Chart6 come together in 2002, which is also the bottom of the next 4-yearcycle.10 Note that this intersection represents the start of Intermedi-ate Wave (3) within ((5)). Primary Wave ((4))would end, in thisscenario, in 2000 or 2001, followed by Intermediate Waves (1) and(2) of ((5)).


SUMMARY

The purpose of this research has been to apply Vodopich’s inte-gration of Elliott and Gann to the study of long-term impulse wavesin the Dow Jones Industrial Average. The effectiveness of such ananalysis can be seen by comparing Charts 1 and 6. Chart 1 shows howGann lines should help a wave count; and Chart 6, the DJIA afterWorld War II with Gann lines based on percentage slopes, delineatesa credible wave count for Supercycle Wave (V) and targets its top.Drawing Gann lines from Elliott Wave starting points can help selectthe best among alternative wave counts; but it does not alter the it-erative, trial-and-error process of establishing the count in the ElliottWave. As Frost & Prechter state, “one can use the market action toconfirm the wave count as well use the wave count to confirm themarket action.”11 Plotting a Gann line from the suspected start of a3rd wave is an academic exercise. The Gann line itself does nothingto establish that the cycle has begun a 3rd wave; and only if pricefinds support at a Gann line during a 3rd wave can one conclude thatthe line is appropriate.

The Gann lines contribute to the evaluation of the cycle only dur-ing the latter stage of the 3rd wave and during the 4th and 5th waves.Evidence of the end of the 3rd wave and the start of the 4th wavecorrection comes from price breaking below the first Gann line ofthe cycle. During the 5th wave, one repeats the trial-and-error pro-cess to create the second Gann line, using the slope of the first Gannline as an aid in choosing the slope of the second. The movement ofprice in between the first (now resistance) and second (support) Gannlines during a 5th wave gives the analyst confidence in the 5th wave,which the author has found to be more volatile than the well-chan-nelled 3rd wave, thereby obscuring the wave count. Finally, at theprice and/or time target indicated by the intersection of the two Gannlines (perhaps with the confirmation of a third line), the analyst canreduce market exposure, both confident that the predictable bulk ofthe profit has been achieved and wary of the much greater volatilitythat tends to follow a Vodopich convergence.

A general caution must be added about the reliability of the Gannlines in the way they indicate support and resistance. Price may over-throw (rise above) the Gann lines especially near the top of the cycledue to the momentum built up through the 5th wave; both Frost &Prechter12 and Vodopich13 have observed such “throwover.” Con-versely, price may fall away after breaking below the third Gann linein an impulse wave so that the third line plays no resistance role atall. And especially in smaller impulse waves, one may not be able tosee a third Gann line. However, to the extent that Gann lines con-tribute to finding the correct wave count, they sensitize the analyst toother market measures signalling a reversal.

Finally, the Supercycle Wave (V) target of 29,000 in 2012 must beconsidered preliminary since it lacks the confirmation of the thirdGann line. As we near the top, however, analyzing component waveswith Vodopich's approach will help bring that Supercycle high in theDow into focus.

ADDENDUM - (4/30/99)This addendum covers the DJIA after the Vodopich convergence

in August ’97 (see Chart 3). On a smaller degree, smaller by oneElliott-Wave degree in fact, this period is like the years from 1966 to1973 (see Chart 5). What emerges from analyzing these two periodsis that the price action up to a Vodopich convergence fits well into theElliott Wave hierarchy, but that after the intersection of the Gann lines,price does not fit neatly into a single cycle that itself fits inside a largercycle. The deep retracements of a post-convergence period undercutthe search for a single impulse wave with inter-related Gann lines.

The post-convergence period in the Original Addendum Chartdiffers from the similar period in Chart 5 in that the Dow rises sub-stantially after August ’97, but price does not do so after the 1966high. The current advance in price biases the analysis in favor of animpulse wave count rather than a corrective wave count, such as theone used by Frost & Prechter for 1966 to 1982.5

Original Addendum ChartCycle Extension After July 1997

Dow Jones Industrial Average (Weekly Bars)

As discussed previously, at the 8/7/97 intersection in Chart 3, theDow breaks below the “102% line,” indicating the end of Wave iiiwithin v of the cycle that began in April ’97. What remains is thewave count for Waves iv, v, and beyond. The author prefers to countiv at the 9/11/97 Dow low of 7581.08 and a truncated v at the 10/8/97 high of 8184.70. And this count puts the period in the OriginalAddendum Chart after the October ’97 correction as an extension ofthe present Minute Cycle.

Overall, the entire price action since October ’97 creates an inel-egant Elliott Wave, an expanding diagonal pattern. Frost & Prechterconsider the expanding diagonal invalid, but Vodopich disagrees,describing it as a “100% reliable top formation.”14

The chart shows the first Gann line of the extension starting atthe 1/12/98 low of Dow 7447.39, drawn with a slope of 51%/yr. Theargument for this slope is that since this extension is a componentwave of the Minute Cycle from 1994, its slope must be a multiple of34%/yr., the slope of the Minute Cycle. Validation of that choicecomes from the summer ’98 correction, which finds resistance alongthat line, culminating in the wave 3 top of Dow 9367.84 on 7/20/98.

The author counts the 4th wave to the 10/8/98 low of 7467.75,which is a truncated C-wave in the a-b-c correction from July. Theargument in favor of finding the low here rather than on 8/31/98 isthat the S&P 500 Composite Index made a new low on October 8 asdid various sub-indices (e.g., SOX), and the VIX high in early Octo-ber exceeded the August/September high.

From here, the author’s amended Vodopoch analysis fails in itspurpose, to highlight the best among alternative impulse wave counts.For the 5th wave, the favored Gann line begins at the 10/28/98 lowof 8328.97, beginning the iii within the 5th with a slope of 102%/yr.While this Gann line indicates well resistance and support, especiallyfrom mid-December ’98 to mid-January ’99, the wave count producesa short 3, only +12.6%, and so is likely to violate the rule that “wave 3is never shorter than 1 and 5.15 (Refer to Original Addendum Chart.)

A Revised Addendum Chart presents the “102% line” and “204%line” drawn according to Vodopich’s original rule, from the start ofthe 1st wave rather than the start of the 3rd wave. The success ofthese Gann lines in identifying the best wave count suggests there


may be a legitimate reason for varying the Gann-line start points. Inthe Revised Addendum Chart, the “102% line” begins at the start ofthe extension’s wave i within 5, at the 10/8/98 low. This Gann lineindicates a top to wave iii within 5 at Dow 9647.96 on 1/8/99, fol-lowed by a wave iv triangle. In this count, the 204% line is drawnfrom Dow 9211.5 on 3/3/99. Here the author reverted to theamended approach, starting the “204% line” at the start of 3 within vwithin 5. These Gann-lines indicate well resistance and supportthrough the price action to date.

Revised Addendum ChartCycle Extension After July 1997

Dow Jones Industrial Average (Weekly Bars)

While the purpose of this research is to advocate a single methodin order to enhance the predictive power of the Elliott Wave, theexperience with the “102% line” suggests a flexible approach isneeded. In particular, when the second wave is shallow, it makes senseto draw the Gann line from the start of the 1st wave (Vodopich’s ap-proach), and when the 2nd wave is deep, to draw the Gann line fromthe start of the 3rd wave (the author’s approach).

Both Addendum Charts also include the “8.5% line” from 1942and the “17% line” from 1984, both of which are part of the August’97 Vodopich Convergence (see Table 2). Throughout the exten-sion, the Dow has had difficulty overcoming the resistance indicatedby the “8.5% line” in 1998, and later in the beginning of 1999. Fur-ther, the Dow failed to break above the “17% line” in 1998; and as ofthis writing, the Dow continues to find resistance at this line.

The combination of these Gann lines, especially the “8.5% line”and “17% line,” suggests that Primary Wave ((3)) and its extensionare near their peak; and price crossing back below the “8.5% line”and the “17% line” will contribute to an assessment of the beginningof Primary Wave ((4)).

Appendix 1Gann Line Slopes of Various Tradables

Appendix 2Prices, Dates & Fibonacci Ratios for T Bond Future Contract from

April '97 to January '98


Appendix 3Prices, Dates & Fibonacci Ratios for Minor Cycle Wave 5

ENDNOTES

1. Vodopich, p. 59.2. Frost & Prechter, p. 25.3. Ibid, p. 71.4. Ibid, p. 80.5. Ibid, p. 283.6. Barrons, 6/29/98, p. 46 (Interview with Mark Perkins).7. Ibid, p. 282.8. Dent, p. 294.9. Frost & Prechter, p. 67.10. Pring, p. 255.11. Frost & Prechter, p. 84.12. Ibid, p. 73.13. Vodopich, p. 55.14. Vodopich, p. 47.15. Frost & Prechter, p. 31.

BIBLIOGRAPHY

■ Barron’s, New York: Dow Jones & Co., Inc. 1998.■ Harry S. Dent, Jr., The Roaring 2000s: Building the Wealth and

Life Style You Desire in The Greatest Boom in History, New York:Simon & Schuster, 1998; Touchstone, 1999.

■ The Elliott Wave Theorist, Gainesville, GA: Elliot Wave International,Inc., 1998.

■ John J. Murphy, Technical Analysis Of The Futures Markets: AComprehensive Guide To Trading Methods And Applications,New York: New York Institute Of Finance, 1986.

■ Robert R. Prechter, Jr. & A. J. Frost, Elliott Wave Principle - Key ToMarket Behavior, Gainesville, GA: New Classics Library, 1995.

■ Martin J. Pring, Technical Analysis Explained, New York: McGraw-Hill, Inc., 1991.

■ Don Vodopich, Trading For Profit With Precision Timing,Greenville, SC: Traders Press, Inc., 1984.

BIOGRAPHY

This research, submitted to the MTA in May 1999, fulfilledthe author’s Phase III requirement of the CMT Program. Mr.Hulton is indebted to his CMT Mentor, Gurney Watson, for hisguidance and support.

Mr. Hulton is presently Director of Trader Services at theElectronic Trading Group, L.L.C.


With earth's first clay they did the last man knead,And there of the last harvest sowed the seed.

And the first morning of creation wroteWhat the last dawn of reckoning shall read.

Omar Khayyám

INTRODUCTION

It strikes me that a philosophical discussion of the Elliott Wave Principle isa worthwhile and relevant pursuit, especially since some of its most notedpractitioners, such as A. J. Frost and Robert Prechter, do not hesitate to dis-cuss philosophical issues in their writing nor do they fail to quote variousphilosophers in connection with their own advocations. I believe that Frostwas the first to specifically articulate the connection between philosophy andthe Wave Principle as evidenced by his observation that “[i]t is possible toread into stock market behavior a philosophical significance under the basictenets of the Elliott Wave Principle.”1

There is no shortage of philosophical issues that can be discussed in rela-tion to the Wave Principle, but the limitations inherent in this paper force meto zero in on a select few. As such, I have decided to examine determinism andspecifically whether it should be said from a philosophical point of view thatthe Wave Principle is inherently deterministic. This consideration also seemedto have occurred to Frost: “It is an open question whether or not man is apuppet on a string. In the short term he is not, but over the longer period hemay be”2 and Prechter, “the Wave Principle form shows that a collective systemis...deterministic.”3

This paper is not about the persuasiveness of any particular form of deter-minism nor the validity of the Wave Principle. Instead, my intent is to inves-tigate the compatibility, or lack thereof, between the two. Deciding whether theWave Principle is deterministic is a challenging task since it requires a meta-physical examination of a theory that by its nature, was not directly intendedto address metaphysical issues. There is an important difference between thedeterminism of a particular theory and the more enveloping, less precise no-tion that the world is itself deterministic. This latter view embraces a muchbolder metaphysical view and while it can be supported, it requires more thanjust the consideration of the determinism of a particular theory.

As is the case with any philosophical undertaking, formulating a precisedefinition of the beginning precepts is a daunting, yet not insurmountable,task. However, a discussion of the semantics of the determinism is not theprimary focus of this paper, and so, I will use determinism in a general philo-sophical way while admitting at the outset that there are various versions ofdeterminism that I will omit due to the limitations of this paper. For the pur-poses of this paper, I will examine scientific determinism, and show why Ibelieve that a case can be made that it is compatible with the Wave Principle.

THE ELLIOTT WAVE PRINCIPLE

Ralph Nelson ElliottRalph Nelson Elliott (1871-1948) developed a theory of stock

market behavior that he detailed for the public in a series of 12 ar-ticles written for Financial World magazine in 1939. In 1946, Elliottwrote what he considered to be his definitive work Nature’s Law – TheSecret of the Universe. The grandiose title reflected the confidence El-liott had in his theory which he believed not only encompassed theaction of the stock market averages but also much larger natural lawthat he believed governed all of man’s activities.

As articulated by Frost and Prechter in their work, The Elliott WavePrinciple, the Wave Principle is Elliott’s discovery that crowd behaviortrends and reverses in recognizable patterns. Elliott named and illus-trated thirteen patterns, or "waves" that recur in markets and are re-petitive in form but not necessarily in time or amplitude. He furtherdescribed how these structures link together to form larger versionsof the same patterns and how those in turn become the buildingblocks for patterns of the next larger size and so on. Regardless ofthe size, the form remains constant.

The Wave Principle is the pattern of progress and regress in whichprogress occurs in specific patterns of five waves and reaction occursin specific patterns of three waves or combinations thereof.

Progress ultimately takes the form of five waves of a specific struc-ture. The three waves in the direction of the trend are labeled 1, 3, 5,and are separated by two countertrend interruptions, which are la-beled 2 and 4.

The essential form is five waves generating net movement in thedirection of the one larger trend followed by three waves generatingnet movement against it, producing a three-steps-forward, two-steps-back form of net progress.

Leonardo Fibonacci da PisaLeonardo Fibonacci da Pisa was born between 1170 and 1180.

He was the son of a shipping clerk named Bonaccio (for writing pur-poses, Leonardo nicknamed himself Fibonacci, short for filius Bonacciwhich means son of Bonacci).

Fibonacci wrote three major mathematical treatises; the Liber Abaci(Book of the Abacus) in 1202, Practica Geometriae (The Practice ofGeometry) in 1220 and Liber Quadratorum (Book of Square Numbers)in 1225. Of the three, Liber Abaci is his monumental work. Fibonaccicontinued to expand his mathematical insights and later re-releasedan updated version of Liber Abaci in 1228. While the stated purposeof the book was to introduce Hindu-Arabic numerals to Europe andexplain their usage, it is within the pages of the Liber Abaci that thefamous sequence was introduced by Fibonacci.

3THE METAPHYSICAL IMPLICATIONS OFTHE ELLIOTT WAVE PRINCIPLE

Jordan E. Kotick, B.A (Hons), M.A., CMT

Chart 1


Leonardo Fibonacci discovered (or more accurately rediscovered)what is now commonly referred to as the Fibonacci sequence of num-bers: 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89 .... It begins with the number 1,and each new term from there is the sum of the previous two. Theratio of any two consecutive numbers in the sequence approximates1.618, or its inverse, .618 after the first several numbers. The higherthe numbers, the closer to .618 and 1.618 are the ratios between thenumbers. The number .618034..., is an irrational number that hasbeen referred to historically as the "golden mean," but in this cen-tury as phi (f).

Elliott unified his theory in 1940 when he recognized that theFibonacci sequence was the mathematical basis for the Wave Prin-ciple. The Fibonacci sequence and its corresponding ratios governboth the numbers of waves in a completed Elliott pattern and theproportional relationships between the waves.

DETERMINISM

This metaphysical principle has been understood and assessed invarious ways over the centuries. Discussions concerning determin-ism are often challenging due to the concept involved: what "deter-minism" means. Since the 17th century, it has commonly been ac-cepted as the doctrine/theory that all human thought, action or eventis caused entirely by preceding events. This is to say that all physicalevents and human actions are determined by antecedent factors.Philosophers distinguish between hard determinism (e.g. necessitari-anism,4 fatalism5) and soft determinism (e.g. compatibilism6, liber-tarianism7).

In addressing the validity of determinism, philosophers have of-ten looked to science and scientific theory to represent the best guideto the truth of determinism. Many philosophers have discussed de-terminism in light of opinions about metaphysical topics such as freewill or God. Historically, the principle of determinism applied toboth secular and theistic philosophers. While some thinkers, likeImmanuel Kant, discussed determinism in terms of the science ofthe day, others made it part of their philosophy of nature.

Various writers have often referred to determinism, causality andnatural law as if they were synonymous. They are connected, despitethe fact that they are not fully equivalent and the differences betweenthem can be clarified if one wanted to make his/her analysis meticu-lous enough. For the purposes of this discussion, however, I do notfind it necessary to attempt to sharpen distinctions to the point ofemphasizing all the recognizable differences between the terms. In-stead, I shall use them in an essentially common-sense fashion.

MICRO OR MACRO

In the early 1930s, accomplished historian Edward Cheney stud-ied various historical events in relation to the seemingly influentialactions of certain famous figures. After examining the ostensibly de-cisive effect exercised by individuals at the time and the role theyplayed in helping to bring about these historical events, Cheney con-cluded:

These great changes seem to have come about with a certaininevitableness; there seems to have been an independent trendof events, some inexorable necessity controlling the progressof human affairs. Examined closely, weighed and measuredcarefully, set in true perspective, the personal, the casual, theindividual influences in history sink in significance and greatcyclical forces loom up. Events come of themselves, so to speak;that is, they come so consistently and unavoidably as to ruleout causes not only of physical phenomena but voluntary hu-man action. So arises the conception of law in history. His-

tory, the great course of human affairs, has not been the resultof voluntary efforts on the part of individuals or groups of in-dividuals, much less chance; but has been subject to law.8

Cheney's view strikes me as consistent with the Wave Principle sincethey both advocate that there is a definite pattern of development inthe apparently jumbled story of human history.9 Further, they bothargue that a system's general characteristics can be said to be deter-mined, at least in part, by the structures and characteristics of theconstituents of those systems.

Every human event has a definite place in an unalterable and con-sistent structure of progress and regress as each society passes througha defined series of antecedent changes in order to achieve a subse-quent stage. Though individuals are agents that seem to bring aboutthe specific events of social history, they are simultaneously the in-struments by which certain laws (Elliott waves) and mathematicalprinciples (Fibonacci) relating to the character of social action be-come manifest.

Both Cheney's view of history and the Wave Principle share thecommon premise concerning the impotency of deliberate individualactions to alter the course of social trends. Or as Voltaire poignantlysaid in the passage on "Destiny" in his Philosophical Dictionary, "Every-thing is performed according to immutable laws...in spite of you."10

This line of thought argues that historical changes are the prod-ucts of deep-lying forces which conform to fixed, although not al-ways apparent, patterns of development in mass psychology. DidierSornette and Anders Johansen of the Niels Bohr Institute in Copen-hagen provided evidence for the existence of a macro intelligencewhen they wrote in 1997:

[T]he market as a whole can exhibit an "emergent behavior"not shared by any of its constituent[s]. In other words, wehave in mind the process of the emergence of intelligent be-havior at a macroscopic scale that individuals at the microscopicscale have no idea of.11

Determinism does not necessarily imply that each individual eventis causally determined. In spite of the fact that we apparently maynot insist upon causality for individual events, it seems that there issome sort of regularity, since the apparently individual actions some-how build themselves into a regular pattern. What is behind thisregularity?

The regularity may be formally described by saying that there arelaws of behavior, such as the Wave Principle, although there are notalways laws for individual events. Elliott clearly believed that he dis-covered a law of nature. In his previously mentioned work, Nature'sLaw, his first line reads: "No truth meets more general acceptancethan that the universe is ruled by law." He goes on to claim that "thevery character of law is order...it follows that all that happens willrepeat and can be predicted if we know the law."12

Frost shared Elliott's sentiments about law and order, as evidencedby the following passages: "Law, or Order prevails everywhere and isin and forms part of everything,"13 "Life is ruled by law and not byaccident,"14 and the "Universal law which asserts itself in our every-day affairs."15 Clearly, Elliott and many subsequent Elliotticans likeFrost believed in both laws of nature and the underlying order of theuniverse.

The Wave Principle argues that human history illustrates a single,transculturally invariant law of aggregate human mood and that lawis the Wave Principle. Is this law deterministic? It is the purpose ofthe rest of this paper to find out.


SCIENTIFIC DETERMINISM

A Brief History of Scientific DeterminismAncient cultures often attributed reoccurring phenomena such

as natural disasters, disease and plague to various gods whose behav-ior they could neither predict nor comprehend. As time progressed,people began to observe certain regularities in the behavior of na-ture. One of the first observations was the discovery of the uniformpath of heavenly bodies across the sky thus making astronomy thefirst developed science. The mathematical foundation for this sci-ence was set more than 300 years ago by Newton. His theory of grav-ity is still used today to predict the motion of celestial bodies. Fromthis example of the birth of astronomy, other natural phenomenawere found to obey scientific laws. This led to the advent of scientificdeterminism, which appears to have been first described by PierreSimon Laplace (1749-1827). In his Philosophical Essay on Probabilities(1814), the French mathematician wrote:

We ought then to regard the present state of the universe asthe effect of its anterior state and as the cause for the one whichis to follow. Given for one instance an intelligence which couldcomprehend all the forces by which nature is animated andthe respective situation of the beings who compose it – an in-telligence sufficiently vast to submit these ideas to analysis – itwould embrace the same formula the movements of the great-est bodies of the universe and those of the lightest atom; for it,nothing would be uncertain and the future, as the past, wouldbe present to its eyes.16

By "the forces that animate Nature," I take Laplace to mean the"laws of nature." Laplace's claim is that if one knows the positionsand speeds of all the particles in the universe, one could, in prin-ciple, calculate their position and motion at any specific time in thepast or the future.

This view of determinism holds that the entire future course ofthe universe has already been determined as a consequence of twofactors: natural laws and the state of the universe at any given mo-ment of time. This is not to say that Laplace held that the course ofthe universe was entirely knowable, since he did not believe that manwas capable of discovering or comprehending the totality of all thenatural laws that govern the universe. His claim was intended as anexplanation of the principles that govern our existence, not our abil-ity to embrace such a vast understanding. Our inability to performthe computations notwithstanding, the philosophical implications ofhis view is that our behavior is determined by the laws of nature andthe state of the universe at any given moment.

German physicist Max Planck put forth the idea of quantum me-chanics in 1900. Its implications for determinism were realized in1926 by Werner Heisenberg, also a German physicist, through whatis now known as the Heisenberg Uncertainty Principle. This prin-ciple argues that one cannot measure both the exact position andthe exact speed of a particle. Laplace's vision was based on the neces-sary condition of knowing the exact positions and speeds of particlesin the universe at a given time, so it was seriously undermined byHeisenberg's work.

Einstein was not satisfied with the apparent randomness in natureand this led to his famous phrase "Der Herr Gott würfelt nicht" ("Goddoes not play dice"). He seemed to have felt that uncertainty wasonly temporary and that there was an underlying reality where par-ticles had defined positions and speeds that would evolve as Laplacehad pointed out, according to deterministic laws.

SCIENTIFIC DETERMINISM AND THE WAVE PRINCIPLE

According to scientific determinism, in principle (though not al-ways in practice), whatever happens can be accounted for by citingnatural laws and antecedent conditions. Elliott felt that he had for-mulated a law of nature according to which given events follow fromother things: "[e]ven though we may not understand the cause un-derlying a particular phenomenon, we can, by observation, predictthat phenomenon's recurrence."17

There are traditional criticisms of determinism, specifically deter-minism in human affairs, based on the contention that human his-tory does not exhibit the stability and the regular periodicity of sci-ence; therefore, historical events cannot possibly be elements of adeterministic system. On this view, Newton provided a deterministicsystem since he compiled in his mechanics a schema for mechanicalexplanation of the physical world (for example, his 2nd law that forceon a body is equal to its mass times its acceleration). If one knew bothforce and mass, one could calculate the acceleration. Thus in orderto find out the mechanical explanation of given phenomenon, onehad only to fill in the schema by finding the variables involved. If youknow the laws, and you know the present conditions, you can predictthe future.

This seems to be an unnecessarily narrow application of what adeterministic system must be like. While this criticism may or maynot be persuasive to the idea of historical determinism before El-liott, the Wave Principle seems to provide a system that closely mir-rors the mechanical explanations of many other sciences. If one knowsthe current wave position, one can persuasively extrapolate the forth-coming patterns that are likely to unfold. He/she can deductivelyreason, based on deterministic patterns, what the fractal structurewill look like and this forecasting will be based on the correct identi-fication of the fractal pattern alongside the current position of thepattern. Hence, I believe this should rebut those who might claimthat only "traditionally scientific" systems can be deterministic.

To understand the limited extent to which determinism is im-plied by the Wave Principle, it is important to understand the type offractal pattern it reflects. Traditionally, it has been assumed thatfractals are either self-identical (each component of the pattern isexactly the same as the whole) or indefinite (self-similar to the ex-tent that it is similarly irregular at all levels). Based on Elliott's dis-covery of a third type of self-similarity, Prechter introduced a newtype of fractal, a "robust fractal." This pattern has highly variable com-ponents that fall within a certain defined structure as Prechter notes:"Component patterns do not simply display discontinuity similar tothat of larger patterns, but "[T]hey form, with a certain defined lati-tude, replicas of them." This "latitude" reflects nature's robustnessand variability within overall determined forms. While it may be anopen question whether every nuance of this "latitude" is determined,the Wave Principle unquestionably rests on the premise that certainessential aspects of the design always prevail.

Consider the following example: assume that you believe a wavefour has finished and that wave five is about to begin. Why shouldthis occur? According to Laplace, there are two ingredients in theexplanation:

Ingredient One: Natural Laws1. Stock market prices trend and reverse in recognizable patterns.2. The patterns are repetitive in form.3. Progress takes the form of five waves of a specific structure.4. Three of these waves, which are labeled 1,3,5, actually effect the

directional movement.


5. Waves 1, 3, 5 are separated by two countertrend interruptionswhich are labeled 2 and 4.

6. Wave 2 never moves beyond the start of wave 1.7. Wave 3 is never the shortest wave.8. Wave 4 never enters the price territory of wave 1.

Ingredient Two: Antecedent Conditions1. Wave four has unfolded in three waves.2. Wave a and wave c have traced out five completed waves.3. Waves a and c are equal in length.4. Wave four has not broken the price territory of wave 1.

Of course, this is not an exhaustive list of either the natural lawsinvolved in the Wave Principle nor of the potential antecedent con-ditions, but I believe the point is still made. In terms of the WavePrinciple, the Natural Laws and the antecedent conditions are be-yond our control. The pattern of the markets is entirely determinedby the laws of nature and the antecedent condition. Natural laws,such as the Wave Principle, account for the way the universe appearsto function.18 On this basis, scientific determinism is compatible withthe Wave Principle.

AUTONOMY AND FREE WILL

It would seem then that there is no room for collective free choiceor philosophically speaking, free will, in this macro-depiction of themechanics of the universe (the will is philosophically understood asthe faculty of choice and decision and figured prominently in thewritings of many philosophers such as nineteenth century thinkersSchopenhauer and Nietzsche). Collective behavior (which is whatthe Wave Principle represents) is determined by both the laws of na-ture and antecedent conditions. This does not explain specific actsof individuals within the collective or therefore even the collectiveitself, but instead general sorts of behavioral proclivities, personalitytraits and the like. So while the individuals within the aggregate haveautonomy to choose, to the extent that they participate in the collec-tive dynamic, their behavior is impulsive and therefore determined.

PREDICTION AND RETRODICTION

As argued above, while we cannot predict with any degree of cer-tainty who, for example, the next Prime Minister of Canada will be,we do have strong reasons for believing who it probably will be. Whileour predictions about the future do not exclude all the conceivablealternatives and leave only one possibility, they rule out a huge num-ber of possibilities. This leads to the conclusion that even thoughindividuals who participate in the events of human history (such asthe election of the Canadian Prime Minister) have free choice in theiractions, their collective choice will fall within certain probabilities;the "Wave Principle structurally restricts the number of possible out-comes of social trends."19 The ramification of this realization is thatnot everything that is logically possible is necessarily historically pos-sible at a given time in a given place. There are determining factorsfor both what has happened and will happen throughout human his-tory.

The Wave Principle specifies what the psychological make-up isbehind a bull or bear market, impulsive or corrective waves.20 If his-tory can be understood (on a macro level) through its wave structureand if the future is also predictable due to its unfolding wave struc-ture, then, in principle, one can specify the circumstances under whichcertain events will unfold such as when scientists, mathematiciansand artists are most likely to have creative success.

It is undeniable that humans are continual sources of novelties,

inventions and creation and that the emergence of these are not pre-dictable. No one, for example, could have predicted Count Basie'sApril in Paris or Darwin's Theory of Evolution. What the Wave Principlecan and does do, however, is ascribe the deterministic patterns thatallow for the favourable social conditions that are conducive to boththe undertaking of research and the acceptance of innovative discov-eries and creations.

Whether specific events will take place is a question of probabilitysince the Wave Principle only dictates society's coming characterchanges, but not necessarily specific events. This is to say that theaggregate social mood fashions the character of history, not the spe-cific manifestations.

Individual occurrences and events do not follow deductively fromthe wave pattern. While the environment is certain and determinedby the wave pattern, the individual occurrences within that environ-ment are based on probability.

Notice on Chart 3 that, for example, according to this particularWave count, after the Supercycle Wave 2 low in 1859, it was predict-

Chart 2


able (determined) that Supercycle Wave 3 would begin.21 It was alsopredictable (determined) that the positive environment that charac-terizes a wave three (prosperity and optimism) would also occur.22

Prechter used the social events and conditions of the past to retrodictthe position of the waves. This chart is one potential classification ofthe cultural timeline of Western culture since Roman Times. Whileneither the Industrial Revolution nor the plethora of events thatPrechter cites were determined, according to the Wave Principle, thesocial-mood preconditions that helped give rise to these events were.

As Prechter has often noted, history repeats in mood, but notnecessarily in mode. Or, put another way,

Because fundamental Elliott wave patterns are limited in num-ber... and because the continual expansion of degree impartsuniqueness to every wave, interactive human mentation andbehavior, which produces history, are continually repeated, butnot precisely.23

I have one last point to make. Just as social-mood conditions aredetermined but social events are not, the form of the impulsive andcorrective patterns are determined, but the dimensions of the waveswithin the pattern are not. This leads to an important point regard-ing the nature the predictive power of Elliott waves. The general pat-terns of the Wave Principle are universal in form. But in reality, thepatterns are rarely, if ever, found asserted in precisely the same way(for example, the length of fifth waves are not exactly the same inevery pattern even though they tend to gravitate towards predeter-mined Fibonacci projections), analogous to Einstein's famous expres-sion that "so far as the laws of mathematics refer to reality, they arenot certain. And so far as they are certain, they do not refer to reality."

The consequence is that in applying the generalized patterns to aspecific occurrence, there will be some uncertainty concerningwhether the given situation adheres to the generalizations. The re-sult of this is that aside from being unable to give the initial condi-tions that will result in an exact consequence (the termination ofwave five will end at .618 the net of wave 1-3 if and only if...) one can

only make the probable assertion that the patternwill terminate at a certain point even though onecan be certain that the pattern will terminate.

CONCLUSION

During this paper, I have not attempted to de-fend either the validity of the Wave Principle norscientific determinism. My intention throughouthas been to determine whether the Wave Prin-ciple was compatible with this version of deter-minism.

I have drawn a number of conclusions fromthis work. The Wave Principle seems to be con-sistent with scientific determinism. Aggregatesocial mentation and the fundamental fractal pat-tern of the Wave Principle moves according to adetermined yet robust path that is based on natu-ral laws and antecedent conditions. This is not tosay that one can identify determinism or the WavePrinciple with the potential to predict with un-limited accuracy. At the same time, however, theyare directed towards discovering regularities inthe operations of systems that will empower onewith the ability to formulate various rules that al-low for the prediction or retrodiction of the oc-currence of events.

I believe that based on the determined path of the Wave Prin-ciple, it is reasonable to say that we can predict the future behavior ofa large body of people but not the individuals that compose that body.In other words, explanations of aggregate social mood based on wavepatterns have the structure of a straightforward deductive argumentwhile individual actions do not.

EPILOGUE

Due to the inherent limitations of this paper, I have had to limitthe scope of my examination to specific areas. There remain, how-ever, many other philosophical areas left to investigate.

As we move into the 21st century, technical analysis continues togrow in reputation and stature, so much so that it is slowly beingaccepted by mainstream academia. Hence, a philosophical look atsome of the techniques practiced and the beliefs held by techniciansis an area that I believe holds great potential. It is, in Shakespeare'swords, "undiscovered country." If we do indeed learn atop the shoul-ders of those who came before us, then I believe that some ofphilosophy's greatest minds have valuable insights for us as techni-cians.

When I refer to the dearth in philosophical analysis of the finan-cial markets, I am not referring to Wave Principle specifically (thoughthere is much more to say about philosophy and Elliott Wave) buttechnical analysis in general. For example, the concept of the herd,central to the technicians' view of market movement and market sen-timent, was explored in detail by Nietzsche, Kierkegaard andHeidegger. Aristotle had a lot to say about man as a "political animal"and social creature. Thomas Hobbes' famous belief that a state ofnature was a state of war and that the life of man was "solitary...nasty,brutish and short" certainly seems to ring true with not only the com-petitive, everyone-for-himself nature of trading, but also the typicallongevity of most traders. There is much we can and should learnfrom these thinkers...one step backward, two steps forward.

In short, I believe there is a book waiting to be written. This is abook that will look at the greatest philosophical minds and explore

Chart 3


what they had to say about the state of nature, human thought andour social condition and compare it to what we know and see in thefinancial markets. As the layers of market analysis are peeled backand examined, I have every confidence that one will find the foot-prints of history's greatest philosophers who, if listened to carefully,have something important to tell us.

ENDNOTES

1 Robert R. Prechter, Jr., The Elliott Wave Writings of A.J. Frost andRichard Russell, (Georgia: New Classics Library, 1998), p. 401.

2 Ibid., p. 398.3 Robert R. Prechter, Jr., The Wave Principle of Human Social Behavior

and the New Science of Socionomics, (Georgia: New Classics Library,1999), p. 414. Referred to as Socionomics for the remainder ofthis paper.

4 Necessitarianism holds that humans do not have free will and thatactions are entirely determined by antecedent, external causes.

5 Philosophically, fatalism holds that the suffering and despair arethe inevitable fate of man. It is often used in conjunction withdeterminism since it argues that every event is bound to happenas it does not matter what we do about it. Like necessitarianism,fatalism denies that human actions have any causal efficacy. Adeterminist may believe that a hangover is the effect of a naturalcause, but the fatalist holds that a hangover will occur regardlessof whether one drinks or not.

6 Compatibilism acknowledges that all events, including human ac-tions, have causes. But it allows for free actions when the actionsare caused by one's choices rather than external causes.

7 Metaphysically, the term libertarianism refers to the idea that hu-man beings have free will and thus sees no inherent contradictionbetween determinism and the proposition that human beings aresometimes free agents.

8 Ernest Nagel, "Determinism In History," Philosophy and Phenom-enological Research, Volume XX, September, 1959. p.291.

9 Socionomics is a field of study coined in Prechter's previously men-tioned book, The Wave Principle of Human Social Behavior and theNew Science of Socionomics. It is the examination of social trendsbased on the biologically based patterns of fluctuation in collec-tive mood that are formological in that they have consistent Fi-bonacci-based mathematical properties and produce the WavePrinciple. Prechter argues that the patterning of social moodguides and influences the character of individual and social be-havior resulting in human actions that in turn cause the trendsand events of history.

10 M. De Voltaire, A Philosophical Dictionary, (London: C.H. Reynell,1824), p. 385.

11 Prechter, Socionomics, p. 159.12 Robert R. Prechter, Jr., R. N. Elliott's Masterworks, (Georgia: New

Classics Library, 1994), p. 216.13 Prechter, Frost and Russell, p. 378.14 Ibid., p. 398.15 Ibid., p. 403.16 Pierre Simon Marquis de Laplace. A Philosophical Essay on Prob-

abilities, (New York: Dover Publications, Inc., 1951), p. 3.19 Prechter, Masterworks, p. 216.18 This is an epistemological claim, not an ontological one. I do not

believe there is any compelling reason to believe that if the WavePrinciple is true, it would communicate, as Galileo said, "the lan-guage in which the Book of Nature is written." Perhaps a lan-guage but not the language.

19 Prechter, Socionomics, p. 414.20 The charts on the right hand side of the page illustrate the psy-

chological components and events indicative of impulse and cor-rective waves – taken from A. J. Frost, Robert Prechter, Jr., ElliottWave Principle, (Georgia: New Classics Library, 1998). p. 77-80.

21 Chart on the following page taken from Prechter, Socionomics, p.345.

22 Frost and Prechter, referring to the powerful and positive senti-ments associated with third waves, called them "wonders to be-hold." Frost and Prechter, Elliott Wave Principle, p. 78.

23 Prechter, Socionomics, p.285.

BIBLIOGRAPHY

■ Bernstein, Peter L. Against the Gods, New York: John Wiley & Sons,Inc., 1996.

■ De Voltaire, M. A Philosophical Dictionary, C.H. Reynell: London,1824.

■ Dray, W. H. "Determinism in History." The Encyclopedia of Philoso-phy, Paul Edwards, editor. Volume 2, New York: MacMillan Pub-lishing Co., and Inc. & the Free Press, 1967. Pp. 373-378

■ Frost, A.J. and Prechter Jr., Elliott Wave Principle. Georgia: New Clas-sics Library, 1998.

■ Gies, Joseph and Frances, Leonard of Pisa and The New Mathematicsof the Middle Ages, Georgia: New Classics Library, 1969.

■ Hook, Sidney. Determinism and Freedom in the Age of Modern Science,New York: New York University Press, 1958.

■ Marquis de Laplace, Pierre Simon. A Philosophical Essay on Prob-abilities. New York: Dover Publications, Inc. 1951.

■ McKeon, Richard. The Basic Works of Aristotle, New York: RandomHouse Inc., 1941.

■ Nagel, Ernest. "Determinism in History." Philosophy and Phenom-enological Research, Marvin Farber, editor. Volume XX, New York:University of Buffalo, 1959. Pp. 291-317.

■ Taylor, Richard. "Determinism." The Encyclopedia of Philosophy, PaulEdwards, editor. Volume 2, New York: MacMillan Publishing Co.,and Inc. & the Free Press, 1967. Pp 359-373.

■ Prechter Jr., Robert R. R. N. Elliott's Masterworks. Georgia: NewClassics Library, 1994.

■ Prechter Jr., Robert R. The Wave Principle of Human Social Behaviorand the New Science of Socionomics. Georgia: New Classics Library,1999.

■ Prechter Jr., Robert R. The Elliott Wave Writings of A.J. Frost andRichard Russell. Georgia: New Classics Library, 1996.

BIOGRAPHY

Jordan E. Kotick has an Honours Bachelor of Arts degreewith a double major in Economics and Philosophy, a Master ofArts degree in Philosophy and is a Chartered Market Techni-cian. Jordan was previously employed with CIBC World Marketsas a technical analyst and government bond trader. He is cur-rently Senior Technical Analyst for Elliott Wave Internationaland Vice President of the Canadian Society of Technical Ana-lysts. He can be reached at: [email protected]


INTRODUCTION

The objective of this paper is to study the efficacy of stop-loss mechanisms.We have back-tested two of the most commonly applied stop-loss rules (fixed %and trailing % stop-loss) on both a trend-following and a mean-reversiontrading strategy. The results from our tests indicate a significant negativecorrelation between the value added by applying these stop-loss rules and theprofitability of the trading strategies tested. In other words, there was a ten-dency for these stop-loss mechanisms to undermine the performance of the prof-itable trading strategies and improve that of the unprofitable ones tested.

The results led us to question the expectation that stop-loss will enhancetrading performance over time by reducing the size of losses and argue thatinvestors should carefully examine (e.g. via back-testing) the impact of stop-loss mechanisms on their trading strategies/style prior to adopting them.

In addition, we have discussed several important qualitative attributes ofstop-loss, which we believe should also be considered when assessing the desir-ability of using stop-loss. These include a reduction in the "risk of ruin,"dilemma in re-entr y and the impact of the "fear of regret" syndrome.

We conclude with a warning against a blind acceptance of the utility ofstop-loss, or even worse, an over-reliance on stop-loss. As indicated by our testresults, stop-loss influences primarily the "pattern" of loss rather than the"probability" of loss. To most investors, it is the latter which matters. To effec-tively manage the "probability" of loss, investors should seek to control "risk,"which is an ex-ante parameter, rather than "loss," which is an ex-post param-eter. Risk-control requires a careful management of the size and nature ofexposures. Stop-loss, on its own, is not sufficient to achieve effective risk-con-trol. An over-reliance on stop-loss may even generate a false sense of security,with undesirable consequences.

A CRITICAL STUDY ON THE EFFICACY OF STOP-LOSS

Concept and Objectives of Stop-lossStop-loss refers to the practice of cutting position upon the breach

of a predefined maximum loss limit.In applying stop-loss, investors usually have the following expecta-

tions/objectives in mind:1) Risk control

Stop-loss is expected to minimize the "risk of ruin," which refersto the cases whereby a major loss from one/two position(s) seri-ously undermines the capital base so that trading activities haveto stop.

2) Return enhancementStop-loss discipline forces traders to cut losing positions and ridewinning ones. It is expected to limit losses to small amounts, with-out constraining profit potential, resulting in an enhancement ofreturn over time.The purpose of this study is to analyze the efficacy of stop-loss

mechanisms with respect to these objectives.

Types and Criteria of Stop-lossIn this paper, we focus primarily on two types of stop-loss rules:

1) Fixed % stop-loss■ This is by far the most commonly used stop-loss rule.■ Upon entering a trade, investors specify a maximum loss limit,

the breach of which would trigger a cut of the position.■ The stop-loss trigger applies to the cumulative P&L since incep-

tion. It can be in the form of a % figure or in absolute dollar P&Lterms.

■ Under this rule, the risk of being stopped out is highest when aposition is newly put on. Once it is in the money, the risk of beingstopped out is reduced.

■ The weakness of this stop-loss rule is that it becomes less effectiveif positions are marked to market periodically, such that a big losscould still show up on the monthly (or quarterly) valuations not-withstanding that the cumulative P/L since inception remainspositive.

2) Trailing stop-loss■ Similar to the fixed % stop-loss, the trailing stop-loss rules require

investors to pre-specify a maximum loss limit, at which the posi-tion would be closed out.

■ The key difference from the fixed % stop-loss is that the stop-losstrigger does not apply to the cumulative P&L. Instead, it appliesto the P&L relative to the highest profit achieved. Suppose a 2%threshold is chosen: then the position would be stopped out oncethe mark-to-market P&L drops more than 2% from its highestpoint.

■ Therefore, this stop-loss rule not only seeks to protect capital butalso the profit accrued.

3) Other stop-loss rulesThere are many other different kinds of stop-loss rules. These in-

clude:I. Standard Deviation Stop-loss

Instead of defining the stop-loss in terms of a fixed % or dollaramount, a standard deviation threshold can be used to achieve asimilar objective.

II. Price Level Stop-lossThe stop-loss trigger is set relative to the market price of the in-strument traded rather than the P&L. The trigger is usually set ata price level of technical significance, i.e. a support or resistancelevel. This is very popular among short-term traders whose trad-ing activities are driven primarily by technical analysis.

III.Event Driven Stop-lossFor instance, if a trade position is assumed based on the beliefthat the central bank will intervene to support a certain level. Theabsence of intervention after the breach of the level would trig-ger a stop-loss.Each rule has its theoretical appeals, but for the purpose of this

study, we have only tested the fixed % stop-loss and the trailing stop-loss rules, which are the most popular and widely used ones. As theypossess most of the basic and important characteristics of stop-lossrules, the results should have a meaningful degree of general appli-cability.

Back-Testing the Stop-Loss MechanismsThe total number of tests conducted is 1,680, across different

markets and trading systems. This section provides a description ofthe historical data used and the methodology applied. Though bor-ing, these technical details are provided in order to put the test re-sults into perspective.1) Markets covered:

Equity indices of major stock markets and exchange rates of ma-jor currencies1 against the US dollar are used in the back testingexercise:

A CRITICAL STUDY ON THE EFFICACY OF STOP-LOSS

William K.N. Chan, CFA 4


Equity: S&P 500 FTSE100 NIK225 DAX30 HSI33 (USA) (UK) (Japan) (Germany) (Hong Kong)

FX: $-JPY $-DEM GBP-$ $-CAD AUD-$(Japanese Yen) (Deutschemark) (Sterling) (Canadian $) (Aussie $)

2) Time period tested:Daily market close prices from Jan-1988 to Feb-1999 are used inthe back-testing exercise. The time period was chosen with an in-tention to incorporate the recent 10 complete calendar years ofhistory in the test. It should include enough diversity of marketconditions against which we seek to test the stop-loss mechanisms.

3) Stop-loss rules and parameters tested:Both fixed % stop-loss and trailing stop-loss rules have been tested.To simulate different level of risk tolerance, we have tested a widerange of stop-loss thresholds, ranging from as tight as 1% to aswide as 10%.The full range tested of stop-loss thresholds tested is as follows:1%, 2%, 3%, 4%, 5%, and 10%.

4) Trading styles / strategies tested:Broadly speaking, all trading strategies can be categorized intotwo main styles:

a) Trend-following = Momentum tradesMean-Reversion = Contra-trend tradesStop-loss rules are tested on both styles of trade.

5) Simulated trading rules:The above two trading styles cannot be tested directly. For thesake of back-testing, quantitative trading rules are used to simu-late the two styles.a) Trend-following trades

■ The trend-following trading style is simulated by SimpleMoving Average (MA) Systems, which combine a long dated(slow) moving average with a short dated (fast) one to gen-erate trading signals.■ The formula used is as follows:SMA

t = Σ(P

t, P

t-1, ...., P

t-n+1) / n where t = reference date

n = observation period

■ SMA systems with different parameter inputs are used tosimulate traders/investors of different time-frame orientation(short-dated versus long-dated). The full range of parameterstested is as follows:

Long-dated SMA Short-dated SMA(250 days ; 50 days)(250 days ; 25 days)(100 days ; 50 days)(100 days ; 25 days)(25 days ; 10 days)

■ A buy (sell) signal is generated when the fast moving aver-age crosses over the slow moving average from below (above).■ The trading position will remain unchanged until eitheran alternative trading signal is generated by the MA system ora stop-loss is triggered.■ Once stopped, will remain neutral until a new trading sig-nal is generated.

b) Contra-trend / Mean-reversion trades■ The mean-reversion trading style is simulated by the SlowStochastics indicator, which is an overbought/oversold in-dicator widely used by mean-reversion traders.

■ It is most commonly used to generate return in a range-bound market whereby buy low sell high strategies pay off.

■ The formula of slow stochastics (%D) used is as follows:%D = Σ(%K

t, %K

t-1, ...., %K

t-m+1) / m where m = smoothing factor

%Kt = (C-L) / (H-L) x 100 where C = closing price at time t.

L = lowest price during last n periodH = highest price during last n periods

■ Again, different parameter inputs are used to simulate trad-ers/investors of different time-frame orientation. The fullrange of parameters tested is as follows:

Observation period (n) for % K Smoothing factor (m) for %D

(10 days ; 5 days)(30 days ; 5 days)(30 days ; 10 days)(60 days ; 5 days)(60 days ; 10 days)(60 days ; 30 days)(100 days ; 10 days)(100 days ; 30 days)(100 days ; 60 days)

■ A buy signal is generated when the value of the stochasticsindicator drops below the predefined oversold threshold, andvice versa.■ In the test, we adopt the commonly used overbought andoversold thresholds of 85% & 15% respectively.■ Once stopped out of a position, a neutral position will bemaintained until a new signal is derived.

6) Defining the profitability of a trading system and the value addedby applying stop-loss:a) A trading system/strategy is considered to be profitable if its

cumulative P&L, with NO stop-loss applied, is positive, and viceversa.

b) Application of stop-loss is deemed to add value if it contrib-utes positively to the cumulative P/L of the trading system,and vice versa.

7) Frequency (probability) versus magnitude of the impact of stop-loss:In this study, we focus on the frequency/probability, rather thanthe magnitude, of positive versus negative impacts of stop-loss. Ow-ing to the diversity in the nature and characteristics of the mar-kets tested, aggregating the magnitude of the P&L impacts is likelyto produce bias or unrepresentative results, in the sense that themost volatile markets would dominate the test results. In order togive the results from each market an equal weighting, we haveopted to focus on the frequency (i.e. % probability) rather thanthe magnitude of impact.

SUMMARY OF RESULTS

Results of the test are shown in the appendix. Tables A to H sum-marize the results of individual tests. More important are tables 1a to7b, which contain the key findings. These include:1) The impact of stop-loss on the "pattern" versus the "probability"

of loss■ The use of stop-loss changes the "pattern" of loss by reducingthe size of individual losses, but at the same time it increases thefrequency of losses. The net impact is uncertain and the overallresults are very mixed.■ There is no significant conclusion we can draw from the crude


results of individual tests2 which did not show any consistent im-pact of stop-loss on the final P&L.■ However, a much clearer pattern emerges once we bisect thetest results according to the style and profitability of the tradingstrategies tested (Tables 1a to 7b).

2) Impact of stop-loss on different trading styles (Tables 1a, 1b)■ The application of stop-loss to trend-following trades gener-ated mixed results. There is no significant evidence that it eitherimproves or undermines performance in a consistent manner (seeTable 1a).■ The results from the application of stop-loss on mean-rever-sion trades are much more decisive. Stop-loss consistently improvedthe performance of the mean-reversion trading strategies tested,with a 73% probability. This result is consistent across differentmarkets, parameter values and stop-loss thresholds (see Table 1b).

3) Impact of stop-loss on profitable trading strategies (Tables 4a, 3a,2a)■ Stop-loss undermined the performance of the profitable trad-ing systems with a 64% probability.■ In only 31% of the cases, stop-loss further enhanced the re-sults of the profitable trading strategies tested.■ This observation applies to both mean-reversion and trend-following trading styles.

4) Impact of stop-loss on unprofitable trading strategies (Tables 4b,3b, 2b)■ Stop-loss improved the performance of unprofitable tradingstrategies with a 73% probability.■ In only 22% of the cases, stop-loss further exacerbated the lossof unprofitable trading strategies.■ This observation applies to both mean-reversion and trend-following trading styles. The results are very robust in both cases,with a 84% and 62% probability of return improvement respec-tively.

5) Relative efficacy of trailing stop-loss versus fixed % stop-loss(Tables 7a, 7b, 6a, 6b, 5a, 5b)■ The results are not robust enough for us to draw conclusionsabout the relative efficiency of the two types of stop-loss rules tested(see Tables 7a & 7b).■ However, there is a clear tendency for trailing stop-loss to out-perform in the cases where the underlying trading strategies areunprofitable and underperform in the cases where the underly-ing trading strategies are profitable. This can readily be seen bycomparing table 6a to 5a and table 6b to 5b.

INTERPRETING THE RESULTS

1) The results lend support to the argument that a stop-loss disci-pline is more important to those with a mean-reversion tradingstyle than those who are trend-followers. One key distinction be-tween a mean-reversion and a trend-following trading system isthat the latter has an in-built mechanism to prompt a trader to getout of a loss-making position, the latter does not. The trade signalfrom the overbought/oversold (mean-reversion) trading systemwould only grow stronger as the trend persists against the trader'sposition. By contrast, a trend-following system would automati-cally prompt the trader to close out a loss-making position as thetrend reverses. In addition, stop-loss mechanisms tend to workagainst the discipline of trend-following. Stop-loss can be triggeredby short term volatility (noise) despite an unchanged underlyingtrend. The results of the test on trend-following strategies indi-cate neutral to negative impact from stop-loss. This reflects thatthe impact of missing out significant medium term trend has off-

set or even exceeded the savings from accurate stop-loss.2) The results indicate a significant negative correlation between the

profitability of the underlying trading strategies tested and thevalue added by applying the stop-loss mechanisms tested. Thisserves as a warning against the blind acceptance of the utility ofstop-loss. Indeed, based on the results, an argument can be madethat if a stop-loss is found to work "consistently" with a tradingstrategy, one should query/check the effectiveness of the tradingstrategy itself. An investigation should be made to check if thereare any particular weaknesses or features in the trading strategywhich are responsible for the consistent profitability of the stop-loss. Such weaknesses should be directly overhauled if possiblerather than controlled by stop-loss. In our test, for instance, theapplication of the stop-loss mechanisms tested to the mean-rever-sion trading strategy tested, in equity markets, was found to beconsistently adding value. However, the fact is that the tradingstrategies themselves were counter-productive in many cases (seeTables C & D). All equity markets tested, with the exception ofJapan, have been in a steady uptrend over time. This did not favourthe application of the mean-reversion strategy tested. Applicationof stop-loss, in such cases, improves the performance simply byover-ruling the trading signals from the mean-reversion strategies,which were consistently unprofitable. In cases similar to these, in-stead of seeking help from a stop-loss mechanism, it would makemore sense to seek to improve or alter the trading strategy itself.

3) Relative to fixed % stop-loss, trailing stop-loss tends to have agreater impact on the trading P&L. Owing to its functional speci-fication, trailing stop-loss are triggered more often (or timely) thanthe fixed % rule, given the same % threshold. Thus, by its nature,a trailing stop-loss is a "tighter" stop-loss than an equivalent fixed% one. This explains our test results which showed a tendency forthe trailing stop-loss to outperform the fixed % one in those caseswhereby the underlying trading strategy was unprofitable and viceversa.So far, we have focused on the quantitative attributes of stop-loss

mechanisms. Their qualitative attributes are however no less signifi-cant and should be taken into account when one is consideringwhether to apply a stop-loss. In the next section, we analyze the quali-tative advantages and disadvantages of applying stop-loss.

QUALITATIVE ATTRIBUTES OF

STOP-LOSS MECHANISMS

Advantages of applying stop-losses1) Reduce risk of ruin

The adoption of a stop-loss rule requires one to determine, inadvance, one's maximum "loss-tolerance" in a trade/position. Inthis sense, the discipline of stop-loss will help to reduce the "riskof ruin" regardless of its eventual net impact on the overall profitand loss� 3.

2) Avoid last minute decisionsHaving a stop-loss discipline also helps to avoid the need to makelast minute decisions, which tend to be less well-thought out oreven irrational. When forced to make a last minute decision un-der pressure, investors are more likely to suffer from the "fear ofregret" syndrome. That is to say, having lost money in a position,investors would hesitate to cut the position, fearing that the mar-ket price could reverse course, causing regret. Such "fear of re-gret" often works against one's investment process and disciplineand could significantly exacerbate the loss. A discipline in stop-loss reduces the need to make last minute decisions and thus re-duces the undesirable impact of "fear of regret."


Disadvantages of applying stop-loss1) Dilemma in re-entr y

This is especially problematic for trend-followers. Stop-loss couldbe triggered by short term volatility, despite an unchanged under-lying trend. Once stopped out, one has to face the dilemma ofwhether to re-enter the trade. If one should decide not to re-en-ter the trade despite an unchanged trading signal from the invest-ment process, the integrity of the investment process could beundermined. If, however, one should opt to re-enter the trade,the original objective of applying stop-loss would be defeated.

2) Ineffective risk controlMany investors have "risk-control" in mind when adopting the stop-loss discipline. Unfortunately, stop-loss by itself is not an effectivetool in risk-control. Although stop-loss reduces the risk of a singlelarge loss, it can generate a series of small losses, which in totalcan exceed the single loss. Our test results, for instance, suggestthat the stop-loss mechanisms tested influenced primarily the pat-tern rather than the probability of loss. To most investors, it is thelatter which matters. Without understanding this, an applicationof stop-loss could generate a false sense of security, or even resultin a sub-optimal risk control. In fact, if one's prime objective is tocontrol the "probability" of loss, then, instead of seeking to con-trol "LOSS," which is an ex post parameter, one should seek tocontrol "RISK," i.e. expected tracking error, which is an ex anteparameter.

CONCLUSION

In this paper, we have studied, via both quantitative and qualita-tive analysis, the efficacy of stop-loss mechanisms. We have back-testedtwo of the most commonly applied stop-loss rules (fixed % and trail-ing % stop-loss) on both trend-following and mean-reversion tradingstrategies. The results from our tests indicate a significant negativecorrelation between the value added by applying these stop-loss rulesand the profitability of the trading strategies tested. In other words,there was a tendency for these stop-loss mechanisms to underminethe performance of the profitable trading strategies tested and im-prove that of the unprofitable ones in the test. This observation istrue of both trend-following and mean-reversion trading strategies,although the results are more robust statistically in the latter case.

The results lead us to query the expectation that stop-loss wouldenhance trading performance over time by reducing the size of losses.There are cases, certainly not a negligible minority as far as our testresults are concerned, whereby the tested stop-loss behaves in a fash-ion similar to an insurance policy, i.e. it generates a "loss" when theunderlying trading strategy works well, and vice versa. We thereforeargue against the blind acceptance of the utility of stop-loss mecha-nisms. It is advisable that investors should pre-examine (e.g. via back-testing) the impact of stop-loss mechanisms on their trading strate-gies/style before adopting them.

Besides, based on the test results, we also argue that a consistentvalue added from stop-loss should not lead to complacency. Instead,it should be taken as a warning signal that the trading strategy maycontain a particular weakness which is responsible for the consistentprofitability of the stop-losses. To the extent that is possible, suchunderlying weakness should be overhauled directly rather than con-trolled by a stop-loss.

There are other advantages and disadvantages of applying stop-loss which are not readily quantifiable. Namely, applying stop-lossreduces the risk of ruin and avoids the need to make last minutedecisions. However, in applying stop-loss, investors would also sufferfrom the difficult dilemma of re-entry. As each investor would attachdifferent importance to these qualities, it is difficult to draw any gen-

eralized conclusion on the desirability of stop-loss beyond what wehave commented on based on the results of our quantitative tests.

A common feature of the stop-loss mechanisms tested is that theyinfluenced primarily the pattern rather than the probability of loss.To most investors, it is the latter which matters. Without understand-ing this, an application of stop-loss could generate a false sense ofsecurity, or even result in a sub-optimal risk control.

ENDNOTES

1 Note that the choice of markets to be tested is arbitrary and re-flects the background of the author who specializes in equity andcurrency market investment. The markets included in the test rep-resent over 80% of global equity markets in terms of market capi-talization and over 90% of FX trades in the global currency mar-ket on an average day. The test is readily replicable for investorsspecializing in bond or commodity market investment.

2 Crude results are not shown in the appendix because the numberof data points are very large, and we are not drawing any conclu-sion based on the crude results. All our conclusions are based onthe summarized/bisected results shown on Tables A to H and table1a to 7b.

3 For readers interested in the subject matter, the "Risk of Ruin"article from Kaufman on Market Analysis, April 1994 provides ainsightful quantitative analysis.

BIBLIOGRAPHY

■ Perry J Kaufman, Kaufman on Market Analysis, "Risk of Ruin" April1994 issue and "Stop-Losses," October 1993 issue.

■ Leopold A Hauser,"A Review of Three Risk Control Methods forthe Stock and Futures Markets", MTA Journal Issue 50.

BIOGRAPHY

William K.N. Chan is Head of FX Strategy in HSBC Asset Man-agement Limited. Before taking up his current position in Lon-don, he was a member of the Tactical Investment Unit of thesame company responsible for technical analysis and FX deci-sions. He also produced quantitative analysis on other areas ofinvestment management. One of his research articles on port-folio rebalancing was published in the Pensions & Investmentmagazine.

William holds a M.Sc. degree in Economics from the L.S.E.,University of London and a BSSc. degree in Business Studiesand Economies from the University of Hong Kong. He has beena CFA since 1997.


APPENDIX I (TABLES A-H)


APPENDIX II (TABLES 1A - 7B)


SUMMARY

If markets are efficient, then they should be interrelated. As such, the priceaction of interest rates should be correlated with that of interest-rate-sensitivestocks. Additionally, the "breadth" of the indexes should describe the health ofthe prevailing price trend in interest rates. First this study determines that ofseveral popular interest-rate-sensitive stock indexes, (Dow Jones Utility Aver-age, S&P Electric Company Index, The Philadelphia Bank Stock and UtilityIndexes, The Morgan Stanley Cyclical Index, the S&P Line Insurance Index,and the S&P Bank Stock Index) only the Dow Utility Average, the Philadel-phia Utility Index and S&P Electric Company Index from July 1977 to April1997 are significantly correlated with bond prices (as defined by 100 minusthe bond yield). The study then uses subjective charting techniques and non-subjective computer trading systems with the rules from the Martin Zweig'sZweig Indicator and Gerald Appel's STIX Indicator to illustrate that in factthe breadth of movement in the S&P Electric Company Index contains somevaluable information (forward-looking) for determining the health of bondmarket trends.

I. INTRODUCTION TO THE STUDY

When investors fear a downturn in economic activity, they sell poorquality stocks first (Pring 1991, p.286). Fewer and fewer issues beginto participate in the stock market advance making the upward trendvulnerable. Stock market breadth monitors the internal trend of themarket by tracking the difference between the number of advancingand declining stocks. If markets are efficient in the sense that theydiscount all available information in the price discovery process, thenthey should be interrelated. As such, if we believe the propositionthat the internal trend of the stock market can forewarn stock mar-ket turnarounds, then "bad breadth" of interest-rate-sensitive stocksshould forewarn of a turnaround in interest rates (bond prices).

Generally, analysts look to the bond market for changes in inter-est rate trends that could influence the equity market. A secondaryconsideration of this study is to determine if the price action of inter-est-rate-sensitive stock indexes provide advance signals of changes ininterest rates. This question is raised because of the disproportion-ate influence a few institutions have in the bond market. If moreparticipants mean more price action and more information in theprice discovery process, then interest-rate-sensitive stock indexes mayindeed lead bonds. Credit Suisse First Boston Corporation in a 1999study also questioned bond market efficiency: "With their wide baseof participants and huge liquidity, equity markets resemble piranhas,reacting rapidly to news. A few hundred institutions, moreover, domi-nate global credit markets, of which the top hundred have dispro-portionately large influence. In the U.S. Equity Market, by contrast,individuals accounts for more than two-fifths of transactions, and theproportion is rising due to cheaper trading costs."

This study is broken down into three parts to probe these ques-tions: first is a discussion of breadth indicators, the study of the ad-vance-decline (A-D) line including a summary of other analysts' re-search and work. Second, it is determined through correlation analy-sis if indexes popularly believed to be sensitive to changes in interestrates are in fact correlated with bond prices. Third, the A-D body ofwork is applied through a combination of chart analysis and rules

entered into computer trading system to determine if the breadth ofinterest-rate-sensitive stocks provides any information about the healthof the trend in interest rates (bond prices).

II. BREADTH: AN INTRODUCTION TO THE

ADVANCE-DECLINE LINE

ConstructionThe A-D line was developed in 1926 by Col. Leonard Ayers of the

Cleveland Trust Co. who was attempting to determine the locationof buying and selling climaxes of the Dow Jones Industrial Averages(Merrill, 1988). The A-D line is simply a measure of demand, deter-mining the degree to which the majority of stocks are participatingin the overall market trend. The theory of the A-D line is simple —"as long as the 'army' (breadth of the majority of stocks) stays in stepwith the 'generals' (e.g. the New York Stock Exchange (NYSE) com-posite or any other market index), then the trend, by definition, ishealthy and should sustain. It's when the army shows signs of retreat-ing, or not keeping pace with the generals, that (we become) con-cerned about the viability of the underlying trend." (Shaw, 1998)When indexes make new highs and the majority of stocks are advanc-ing, then internal trend is strong; there is widespread demand forstocks and a greater possibility that more money will flow into themarket.

The two basic A-D indicators are constructed either by some ratioof advances to declines (e.g. A/(A+D)) or a simple difference be-tween the number of advancing and declining stocks (e.g. A-D). Sincethe difference calculation can have extreme volatility, moving aver-ages are usually employed as a smoothing device. However, if thenumber of stocks in the underlying index changes over time, thesimple A-D calculation can be prone to biases making historical com-parisons difficult. To compensate for this bias, some use A/(A+D+U)as the A-D calculation. Table I shows an example of the simple differ-ence calculation in practice.

Table INumber of Number of Difference A-D lineAdvancing Declining (A-D) (Cumulative

Day stocks stocks running total)

1 200 300 -100 -100

2 400 100 300 200

3 300 200 100 300

4 450 50 400 700

5 100 400 -300 400

The ratio or simple difference calculation by itself acts as an oscil-lator with overbought and oversold levels. Like Col. Ayers, analystsattempt to identify the extremes as an indication of buying climaxesor selling capitulation. Contrarians generally use such evidence as asignal that the market has exhausted itself and initiate counter-trendtrades.

All of the indicators and research presented in this section weredeveloped for use in the stock market. These indicators use NYSEcomposite or other composite data in their development and appli-cation. Simple A-D indicators include:

INTERMARKET BREADTH INDICATORS:Does the Price Action of Interest Rate Sensitive Stocks Provide Clues to

Trends in Bonds Prices?

Gary Stone, CMT

5


■ Haurlan Index — an exponential moving average of the simpledifference between advances and declines (A-D). With New YorkStock Exchange (NYSE) data, a short-term (3-day) exponentialmoving average gives a buy signal if it's in excess of 100 and a sellsignal if it falls below –150. Analysts use the trend of a long-term(200-day) moving average to generate signals.

■ Fosback — an activity index because market direction isn't impor-tant. Defined as the absolute value of advances less declines (|A-D|), high readings warn of a change in trend, while low valuesindicate a continuation of trend. Using NYSE data, a cumulative(25-day) running total of the daily readings in excess of 11,000 isbullish for the short-to-intermediate term (one month to one year).

■ The McClellan Oscillator — based upon the simple (A-D) calcu-lation and is one of the most popular indicators in stock marketA-D line analysis. It is a short- to intermediate-term breadth indi-cator based upon the same concept as Gerald Appel's Moving Av-erage Convergence-Divergence Oscillator (MACD). Essentially itis the difference between a 19-period and 39-period exponentialmoving average of advances less declines (A-D). Typically, theMcClellan Oscillator reaches an oversold or overbought extremeprior to a change in the trend of the stock market. An oversoldmarket is identified by oscillator readings between -70 and -100.Oscillator readings of +70 to +100 indicate an overbought mar-ket.Ratio indicators modify the calculation into some sort of ratio of

advances to declines. Ratios make historical comparisons appropri-ate because the ratio calculation transforms the indicator into a mea-sure of extent, force or market conviction of the movement of thestocks advancing (Merrill, 1998). Ratio indicators include:■ Nicoski's A/D — the A-D line is defined as the cumulative 30-day

moving average of the ratio of advances to declines (A/D). Thisratio determines the force of the advance in terms of the strengthof the decline, presenting the "classic struggle of bulls and bears"(Dworkin, 1990). Nicoski's A/D oscillator generates a buy signalif it crosses above 1.0 if it had been below 0.84 and a sell signal ifthe oscillator crosses below 1.0 if it had been above 1.2. The oscil-lator also exhibits recurring cycles and patterns.

■ Martin Zweig's Breadth Indicator — a 10-day moving average ofA/(A+D) and determines the strength of the advance in terms ofthe issues that are being acted on. The indicator oscillates be-tween 0 and 1. Oscillator readings are bullish above 0.66 and bear-ish below 0.367. A derivative of this indicator is the thrust. Zweig'sresearch determined that each of the last four bull markets since1970 began with a thrust — a significant rise in the indicator from(0.40 to 0.615) within the last 10 trading days.

■ Gerald Appel's STIX Indicator — similar to Zweig's (A/(A+D)),except it is smoothed by a 21-day exponential moving average.0.42 to 0.44 is oversold and 0.56 and 0.58 is overbought.

■ The Hughes Breadth Index — a 10-day moving average of (A-D)/(A+D+U) and is bullish above 0.25 and bearish below -0.22.

■ Fosback's Absolute Breadth Index — a hybrid of the Hughes de-fined by |A-D|/(A+D+U) is bullish above 0.40 and bearish below0.15.

■ The Merrill Advance Decline Divergence Oscillator (ADDO) —developed by Arthur Merrill. It employs the ratio (A-D)/U inorder to gauge the internal market conviction of advancing stocks.This indicator attempts to remove the subjectivity of determiningdivergences by calculating the difference between actual valuesand forecast expected values obtained by using least squarestrendlines (statistical regression). Using the DJIA as an example,Merrill would run the regression DJIA

expected = a + b*cumulative

running total of (A-D)/U. The ADDO is calculated from ADDO= (100 * DJIA * (a + b*cumulative A/D volume)) – 100. Theinterpretation is that if ADDO is positive, the DJIA is pulling ahead

of the expected values, thus bearish (the generals are leading thearmy). A negative value is bullish, since the expected values arepulling ahead of the DJIA (the army is leading the generals). Fortrading signals, Merrill uses a 2/3rds Standard Deviation aboveand below the mean as extremes.

Line and Moving Average InterpretationThe A-D line is generally referred to as the cumulative running

total of either the ratio or simple difference of advances and declines.This representation allows the line to have the attribute of having atrend. A-D analysis generally contains three parts: comparison of thecumulative A-D line trend with that of the overall market index; analy-sis of the A-D line's relative value to that of its moving average; mo-mentum and oscillators.

A-D line trend analysis is simply a comparison the trend of the A-D line with that of the overall market index. Both lines should movein the same direction. This analysis is somewhat subjective, as is thecase with most traditional forms of chart pattern analysis. For themost part, analysts attempt to identify divergence between the A-Dline and the index as an indication of a possible deteriorating healthin the trend. A rising A-D line with a declining index is considered abullish or "positive divergence" because the broader market is risingeven though the market index is declining (the army is leading thegenerals). The converse is also true. A declining A-D line with a ris-ing market index is considered bearish or a negative divergence be-cause the broader market is not participating in the index advance(the generals are leading the army). The analysis can be highly sub-jective, because depending upon an analyst's preexisting biases, thesame chart can be open to may different interpretations. (Dworkin,1990).

For the most part, it is the trend of the A-D line that is important.The level of the A-D line is only important to the extent that it con-firms the new highs (or lows) in the index. A new high in an indexshould be confirmed by a new high in the A-D line. Trend weaknessinitially becomes apparent when the index soars to new highs (de-clines to new lows), but the A-D line struggles or fails to make a newhigh (or fails to make a new low). Additional A-D line analysis in-cludes the identification of traditional chart patterns contained inthe movement of the line: triangles, support and resistance zones,can be used to identify breakout points, an indication that the broadermarket is seeing demand (or selling pressure) that may not be appar-ent in the index. Moving averages and identification of reversal pat-terns on the A-D line can also be used to alert technicians of possiblechanges in A-D line trend and ultimately in the underlying index.

Momentum and OscillatorsIn an attempt to reduce the subjective interpretive nature of the

A-D line analysis, analysts have created oscillator indicators. Momen-tum indicators oscillate around 0 and are constructed similar to arate of change. For instance, a 12-day rate of change would be calcu-lated by subtracting the value of the A-D line 12 days ago from thelevel of the A-D line today (A-Dt – A-Dt-12). Momentum can besmoothed by taking a 10-day moving average of the rate of change.Extreme values of the oscillator are considered to be warnings ofreversals. Although these momentum indicators try to remove thesubjective nature of A-D line analysis, research has shown that thedetermination of "extreme" readings is also subjective and, moreover,can change depending upon market conditions. The problem re-searchers point out is that the values of the momentum indicatorsare relative – analysts have to arbitrarily determine the extremes(Downing, 1994). But these indicators are considered more robustbecause the computer, using set rules, decides the turning points.


Bond is the "Discount" price defined as(100 - 30 year U.S. Treasury Bond yield)

III. INTEREST-RATE-SENSITIVE STOCKS AND

INTEREST RATE TRENDS

MethodologyAs discussed, the A-D line is an effective tool because it captures

sector rotation. Investors sell weak stocks because they feel they arevulnerable to an economic downturn. An extension of this theorywould logically be that if stock market participants believe that inter-est rates are going to rise, then they would sell stocks that are vulner-able to changes in interest rates.

Popular interest-rate-sensitive stock indexes are examined in thisstudy. First, a determination is made if these indexes are correlatedwith interest rates. Then breadth indicators are constructed usingthe price action of the stocks underlying the indexes. Using the tra-ditional package of A-D line analysis (analysis of the line itself, mov-ing averages, momentum and oscillators) the breadths of the inter-est rate stock indexes are tested to determine if they have any mean-ingful information concerning trends in interest rates. Daily datafrom July 1979 to April 1997 were used in this study.

Interest RatesAlthough there are several proxies for interest rates, the yield on

the U.S. Treasury 30-year bond is the most preferable for several rea-sons. Other possible proxies include The Dow Jones Bond Averageand the 30-year U.S. Treasury Bond futures contract traded at theChicago Board of Trade.

The Dow Jones Bond Average uses 20 bonds traded on the NewYork Stock Exchange. The problem with this index is it containsconvertible bonds that can distort the actual movement of interestrates. Companies subject to merger and acquisitions rumors thathave bonds with convertible options will respond to factors that havenothing to do with interest rates, but perhaps the convertible optionsbeing reevaluated. Additionally, changes in perceptions of creditquality will affect this index even if expectations of interest rates don'tchange.

The 30-year U.S. Treasury Bond futures contract traded at theChicago Board of Trade has two problems. The first problem is thatalthough it does trade off the underlying behavior of the actual 30-year U.S. Treasury Bond, it is a futures contract, and thus by defini-tion trades off a future expected outcome. The convergence doesn'toccur until the final month of the contract. The second problem isthe issue of the roll. The bond contract is a quarterly contract. Eachquarter, gaps in price occur due to one contract rolling off and an-other starting. These gaps are more pronounced than the "rolls" thatoccur from the Treasury auctioning other bonds.

The proxy we have selected is the yield of the current U.S. Trea-sury 30-Year Bond. This probably is the most accurate proxy for in-terest rates because most analysts tend to look at the bond for trendchanges that could forewarn of changes in the equity market. Sinceclassical chart analysis is based upon price movement, the TreasuryBond yield is converted into a "discount" price, 100-yield (yield andprice move in opposite directions).

Interest-Rate-Sensitive IndexesIt is generally thought that utilities, banks, life insurance compa-

nies and cyclical stocks are greatly influenced by interest rates. Assuch the following eight indexes were analyzed (see Appendix I for alist of the stocks that compose each index):■ The Dow Jones Utility Average, an average of 15 utility companies

listed on the New York Stock Exchange that are involved in theproduction of electrical energy and natural gas;

■ The Philadelphia Utilities Index, a capitalization-weighted index

of 20 utility companies involved in the production of electricalenergy;

■ The S&P Electric Company Index, a capitalization-weighted in-dex of all stocks designed to measure the performance of the elec-trical power utility sector;

■ The Philadelphia Bank Stock Index, a capitalization-weighted in-dex of 24 national, money center and regional lending institu-tions;

■ The S&P Bank Stock index, a capitalization-weighted index of allcompanies involved in the business of banking;

■ The Morgan Stanley Cyclical Index, a capitalization-weighted in-dex of 30 stocks which are cyclical in nature;

■ The S&P Life Insurance Index, a capitalization-weighted index ofall life insurance companies.

Correlation Analysis: Are Interest-Rate-Sensitive StockIndexes Correlated with Bond Prices?

Tables II and III show simple correlation matrixes on daily andweekly time frames for each index with respect to the bond's discount-price from the official start of the index to April 1997. (Note thatdespite the later "official start" date of the index, stocks underlyingthe indexes have price history starting from July 1979 which makesA-D line analysis from 1979-1997 possible.)

This analysis illustrates that the Philadelphia Utility Index and theS&P Electric Company index are highly correlated not only with thebond's discount-price, but also each other across all time frames tested.The Dow Jones Utility Average is also correlated, but there were peri-ods (1996, for example) when the correlation was not significant.(This may be a result of the Dow Jones Utility Average containingNatural Gas Companies, whereas the S&P Electric Company and Phila-delphia Utility Indexes do not).

Chart IPhiladelphia Utility Index and the Bond

Surprisingly, the banks, cyclicals and life insurance indexes werenot correlated with movements in the bond discount-price, but werehighly correlated with each other. The correlation between the S&PBank Index and the Philadelphia Bank Stock Index is not altogethersurprising since they contain many of the same stocks (see appendixI for more detail), but the strong correlation between the banks, lifeinsurance companies and the cyclical stocks is extremely interesting.It means that they are reacting to the same influences, which arewidely believed to be interest rates. Though not tested, one interpre-tation may be that these stocks are more correlated with short-terminterest rates (such as Treasury Bills or 2-Year Notes), which are di-rectly influenced by perceptions of and actual moves in monetarypolicy by the Federal Reserve than bond yields.


Chart I confirms the results of the correlation matrix: the Phila-delphia Utility Index and the bond discount-price tend to move intandem. Upon closer examination, however, it appears that the Phila-delphia Utility Index tends to bottom and peak before the bond dis-count-price. Chart II shows the S&P Electric Company Index andthe bond discount-price. Again, the S&P Electric Company index alsoappears to bottom and peak before interest rates. Chart III shows theDow Jones Utility Average and the bond discount-price. Althoughboth series tend to move in tandem, the Dow Jones Utility Average fitis not as good as either the Philadelphia Utility or S&P Electric Com-pany index as the correlation matrix suggests.

Chart IIS&P Electric Index and the Bond

A-D Analysis: Does Breadth Give Clues to Changes in Trendsof Bond Prices?

Both the correlation matrixes and the accompanying graphs sug-gest that the Philadelphia Utility and the S&P Electric Company in-dexes provide the best fit on a macro level. Therefore, these indexeswere used to see if the breadth of the indexes contains any forwardindications that a change in the trend of interest rates is about tooccur.

As noted earlier, the Philadelphia Utility and S&P Electric Com-pany indexes are extremely correlated. This is not at all surprisingsince the Philadelphia Index is an inclusive subset if the S&P Index.The S&P Index includes nine more electric companies than the Phila-delphia Index (see Appendix I).

Although the indexes are highly correlated across a variety of timeintervals, the interesting question then becomes, does the presenceof the nine extra companies have a material effect on the cumulativeA-D lines generated from the two indexes? Surprisingly, the answeris yes, but only recently.

Correlation of the Philadelphia Utility and S&P Electric Co. Index's A-D lines

7/79 – 4/97 1/79-12/94 1/96-12/96 1/97-4/97

99.2% 99.7% 88.0% 78.0%

The Philadelphia Utility Index A-D line and S&P Electric A-D lineare significantly correlated from July 1979 to December 1994. Therelationship, however, diverges beginning in 1996 suggesting that theprice action in the nine companies included in the S&P Electric Com-pany Index has a material effect on the A-D line and further exami-nation is needed.

Charts IV and V further illustrate this point that although the A-Dlines generated from the two indexes are similar in nature, there aresome striking differences. The period from late 1996 to April 1997 isone such period. The A-D line in the S&P Electric Company Indexappears to continue its steep descent, while the Philadelphia UtilityIndex A-D line appears to be leveling off.

Chart IIIDow Jones Utility Average and the Bond

Chart IVPhiladelphia Utility Index A/D and S&P Electric Company Index A/D

Weekly

Chart VPhiladelphia Utility Index A/D and S&P Electric Company Index A/D

Daily

Bond is the "Discount" price defined as (100 - 30 year U.S. Treasury Bond yield)

Bond is the "Discount" pricedefined as (100 - 30 year U.S.Treasury Bond yield)

(left scale)

(right scale)

(left scale)

(right scale)


A-D Line and Moving Average AnalysisA-D line and moving average analysis can be somewhat subjective

because analysts’ preconceptions can bias the interpretation. Subse-quent analysis is, therefore, made more robust by creating "tradingrules." Rules allow the computer use the A-D ratios and oscillators tomake trading decisions and determine if the rules (the A-D indica-tors) are profitable (profit or positive results are the only true mea-sure of a good rule).

S&P Electric Company Index A-D LineChart VI

Chart VII

Chart VIII

Chart IX

Chart X


Chart XI

Charts VI to XI show the S&P Electric Company Index A-D lineand the bond discount-price. Charts VI and VII show the A-D linefrom late 1981 to 1984. The A-D line appears to reveal little about thetrend in interest rates. In fact, from running a correlation from 1979to 1985 on the Dow Jones Utility Average (the only series which "offi-cially" has data during those dates), bond prices had only a 35.2%correlation. Thus, it is not surprising that A-D line doesn't provideany information. From 1985 to 1990, the A-D line becomes more be-haved, but again not very useful, except for the 1987 top. Chart VIII.In 1987 the A-D line crosses down through its 50 and 200 day movingaverage, officially denoting a downward trend while bond prices con-tinued to consolidate (area A).

The correlations presented in Tables II and III illustrate that from1990 to 1997, bond prices and interest-rate-sensitive stocks are corre-lated. The charts show that the indexes appear to lead changes inbond prices. It also appears that the A-D line leads most trend changesin bond prices. In Chart VIII (1991 to 1993), there are four signifi-cant turning points in bond prices. The A-D line appears to signalthat the trend in bond prices is subject to reversal before prices actu-ally reverse.

In late 1990 (Chart IX, area A), the A-D line begins its trend rever-sal slightly before bond prices begin to recover. The A-D line createsa double bottom and crosses above its 50 and 200 day moving aver-ages, confirming the move in bond prices. In early 1991 (area B),bond prices appear to trace out a head and shoulders reversal forma-tion. When bond prices are peaking at the head, the A-D line doesnot confirm the advance by both failing to make a new and appear-ing to move in the opposite direction as bond prices. The A-D lineappears to bearishly diverge before bond prices fall out of the headand shoulders formation. The A-D line recovers before bond prices,positively diverging. Throughout the rest of 1991, the A-D line con-firms the upward trend in bond prices. The A-D line, however, cre-ates a classic divergence from bond prices in early 1992 (area C):bond prices continue their upward trend while the A-D line abruptlyreverses trend. The A-D line also appears to lead at the bottom ofthe correction in mid-1992 as well. At the bottom of the mid-1992correction, the A-D line is making a new high as bonds are making anew low. At the top in late 1992 (area D), the A-D line clearly is in adowntrend while the bond is completing a head and shoulders topformation.

In Chart X, (late-1993 to early 1996), there are four significantchanges in trend. The first occurs in late 1994, as bond prices con-tinue to make new highs although the A-D line is diverging in a

downtrend, descending through its 50- and 200-day moving averagesbefore bond prices reverse trend. The A-D line spike in early 1994was a false signal. However, had it been followed, bond prices werelittle changed by the time the A-D line declined sharply signalingthat bond prices probably had further to fall (which they ultimatelydid). At the late 1995 (area B) bottom, the A-D line completes a doublebottom, rises above its 50- and 200-day moving averages and is clearlyin an uptrend before bond prices recover. In mid-1996 (area C), bondprices continue to trade higher, even though the A-D line is in a down-ward. Bond prices finally trade lower. It is not completely clear, how-ever, that the A-D line rises above its 50- and 200-day moving averagesbefore bond prices. At a minimum, the A-D line confirms the recov-ery in bond prices, and at a maximum, the A-D line leads the bottomin bond prices by crossing above its moving averages before bondprices start to rise. The A-D line fails to signal lower bond prices inearly 1996 (area D). It is clear that bonds are in a steep decline be-fore the A-D line begins its descent. But once the A-D line catchesup, it is interesting that the A-D line and bond prices appear to betracing out the same "flag" formations as bond prices continue todescend.

Philadelphia Utility Index A-D LineThe Philadelphia Utility Index is analyzed from 1996-97 because

both the Philadelphia Utility and S&P Electric Company A-D lineswere highly correlated with each other until 1996. Chart XI illustrateswhat the correlations show statistically: that the S&P Electric Com-pany Index and Philadelphia Utility Index A-D lines were similar until1996, and then they diverged. Something was occurring in the extranine stocks in the S&P Electric Company Index that caused a shift.Chart XI shows the bond discount-price and the A-D lines generatedfrom the Philadelphia and S&P index. In the beginning of 1996, theS&P Electric Company Index A-D line didn't create any divergencespattern which would have helped identify the intermediate bottomin July 1996 (area A). The Philadelphia Utility Index A-D also failedto diverge at that bottom. Both A-D lines, however, created bearishdivergences, peaking before bond prices in the late 1996, early 1997intermediate top in bond prices (area B). In February 1997, the S&PElectric Company Index A-D line diverged again, portending to thecorrection in bond prices (area C). However, when the A-D line failedto make a new high when bond prices did, then started to decline, itshowed that that the price rise in bonds was merely a correction.Similar patterns occurred with the Philadelphia Utility Index A-Dline, but the action in the S&P Electric Company Index A-D line wasmore dramatic and created clearer signals.

The seven extra stocks in the S&P Electric Company Index ap-pear to have a material effect on the A-D line behavior. The S&PElectric Company Index A-D line appears to be the better indicator,and thus will be the one used in the trading system analysis.

Ratios, Oscillators and Trading SystemsUsing the S&P Electric Company Index A-D data, computer trad-

ing systems were constructed using the Zweig, STIX and McClellanOscillator rules. The computer trading systems generate a more ro-bust analysis because the rules are concrete and the computer em-ploys the rules in a non-subjective manner. Since the majority of NYSEstocks are believed to be interest-rate-sensitive, the rules used by thesystem for the S&P Electric Company Index A-D are the same onesemployed for the NYSE A-D analysis. The buy and sell thresholds ofthe oscillators are not optimized. Table IV summarizes the results.

ZweigThe Zweig rules are based upon the Zweig Breadth Indicator, a

10-day moving average of the ratio A/(A+D). The Zweig Breadth


indicator measures the strength of the advance in terms of the issuesthat actually have supply and demand pressure. Oscillating between0 and 1, Zweig's research determined when using NYSE stock datathat the oscillator is bullish when above 0.66 and bearish below 0.367.

The analysis on the S&P Electric Company Index Zweig A-D ratiois quite interesting. With the exception of 1994-95 period, over 50percent of the trades were winning and had positive net profit andreturn on account. For the most part, however, this system encoun-ters a few problems that would not make it a very good system toblindly follow.

The largest winning trade for the most part is less than the largestlosing trade; the system does not cut losses short. In fact, intradaydrawdowns are quite large, making managing the account somewhatexpensive. By design the system always has a position in the market(long or short). It appears that many of the problems the systemencountered were when it was short. The long trades had the great-est success. The short trades on average were losers, making the sys-tem appear to be unbalanced. Although on average the system wassuccessful, the results from 1994-1995 were not significant. It is veryinteresting to note that the Zweig system was profitable, despite theA-D line's inability to show any clear signals from 1979-1990. Onbalance, the Zweig system is a good confirming indicator, but it doesn'tseem prudent to rely on it signals exclusively.

STIXThe STIX indicator is similar to the Zweig Breadth Indicator, ex-

cept it employs a longer (21-day) exponential moving average. GeraldAppel, the creator of the indicator, identifies 0.42 and 0.44 as over-sold and 0.56 to 0.58 as overbought for NYSE composite stock data.

Across all time frames, the STIX system using the S&P ElectricCompany A-D data for signals to buy and sell bonds is profitable.With the exception of the 1994-95 period, over 50 percent of thetrades in each time period were profitable, generating a significantnet profit and return on equity. Although during the 1994-95 pe-riod, less than 50 percent of the trades were profitable, the systemwas able to cut the losses short, so net profit for the period was posi-tive. Again the computer is always in the market (long or short), butthe difference with the STIX rules is that both long and short tradesare profitable. The system was also profitable during the 1979-1990time period, the same period where the subjective analysis deemedthe A-D line and trends in interest rates to be uncorrelated. The STIXsystem appears to be a more balanced system. Thus the S&P ElectricCompany Index A-D using the STIX rules is a good intermediate-term indicator for bond trends.

McClellan OscillatorThe McClellan Oscillator is a short- to intermediate-term breadth

indicator based upon the same concept as Gerald Appel's MovingAverage Convergence-Divergence Oscillator (MACD). Essentially itis the difference between a 19-period and 39-period exponentialmoving average of advances less declines (A-D). Typically, theMcClellan Oscillator reaches an oversold or overbought extreme priorto a change in the trend of the stock market. An oversold market isidentified by oscillator readings between -70 and -100. Oscillator read-ings of +70 to +100 indicate an overbought market.

It was extremely surprising that the McClellan Oscillator did notreturn any profitable results over the same time periods tested forthe Zweig and STIX indicators. Even when run through an optimizer,the Oscillator was unable to return significant results.

CONCLUSIONS

It appears that the price action of interest-rate-sensitive stocks pro-vide clues to trend changes in interest rates, as defined by the bonddiscount-price (100-bond yield). The S&P Electric Company andPhiladelphia Utility Indexes are significantly correlated with bondprices. Under subjective chart analysis, the indexes appear to leadbond prices at turning points. Further analysis showed that the S&PElectric Company Index held a stronger correlation to bond pricesacross various periods and subsequently was used for the rest of theanalysis.

Using subjective chart analysis, an A-D line generated from theprice action of stocks in the S&P Electric Company index appears tocontain some information concerning the viability of bond pricetrends from 1990-97. Although chart analysis on the A-D line prior to1990 suggests that there is correlation between bond prices and theindexes, analysis using a set of rules applied to a computer tradingsystem disputes that finding. The fact that the computer finds thatthere exists some information that the subjective chart analysis didn'tfind highlights the problem with the A-D line chart analysis: it can bequite subjective – both biased by analysts' preconceptions and inabil-ity to spot information based upon the presentation of the chart.

The computer, employing rules in a non-subjective manner, buysor sells bonds depending upon the S&P Electric Company A-D ratio.The Zweig Breadth and STIX indicators both contain significant in-formation about trends in bond prices. The Zweig Breadth Indica-tor was not successful across all time periods tested and the largeintra-day drawdowns and inability to cut losses short makes the indi-cator/system a poor trading system to use. The Zweig indicator canbe use used as short-term indicator, but confirming indicators shouldalso be employed prior to trading off the signals. The STIX indica-tor, however, offered significant results across all time periods indi-cating it would be an excellent intermediate-term indicator for trendchanges in bond prices.

REFERENCES

■ Burke, Mike (1990). "The Advance/Decline Line" Technical Analysisof Stocks and Commodities, July.

■ Chande, Tushar ( 1984). "Breadth, STIX and Other Tricks" Tech-nical Analysis of Stocks and Commodities, May.

■ Colby, R.W. and T.A. Meyers (1988). The Encyclopedia of Techni-cal Market Indicators, Dow Jones-Irwin.

■ Credit Suisse First Boston Corporation (1999). "Pythons and Pira-nhas" The Global Credit Strategist, July 7.

■ Downing, Daniel (1994). "Advance Decline Line Basics" TechnicalAnalysis of Stocks and Commodities, March.

■ Dworkin, Fay H. (1990). "Defining Advance/Decline Indicators"Technical Analysis of Stocks and Commodities, July.

■ Hartle, Thom (1989). "An Advance/Decline Line for Commodi-ties" Technical Analysis of Stocks and Commodities, May.

■ Merrill, Arthur ( 1988). "Advance Decline Divergence Oscillator"Technical Analysis of Stocks and Commodities, September.

■ Meyers, Dennis (1996). "The Electric Utility Bond Market Indica-tor" Technical Analysis of Stocks and Commodities, January.

■ Mogey Richard (1989). "The McClellan Oscillator" Technical Analy-sis of Stocks and Commodities, September.

■ Murphy, John J. (1982). "The Link Between Bonds and Commodi-ties" Technical Analysis of Stocks and Commodities, May.


■ Murphy, John J. (1982). "The Link Between Bonds and Stocks"Technical Analysis of Stocks and Commodities, June.

■ Murphy, John J. (1982). "Utilities and Bonds" Technical Analysis ofStocks and Commodities, August.

■ Murphy, John J. (1986). Technical Analysis of the Futures Market,New York Institute of Finance.

■ Murphy, John J. (1991). Intermarket Technical Analysis, John Wiley& Sons.

■ Pring, Martin J. (1991). Technical Analysis Explained, McGraw-Hill Book Co.

■ Shaw, Alan R. (1988). "Technical Analysis" Financial Analysts Hand-book, Second Edition, Dow Jones-Irwin.

■ Waxenberg, Howard (1985). "Technical Analysis of NYSI/DJI" Tech-nical Analysis of Stocks and Commodities, December.

■ Williams, Mason (1994). "Advance Decline Line Basics" TechnicalAnalysis of Stocks and Commodities, January.

BIOGRAPHY

Gary Stone is a VP of Business and Content Developmentand Director of Finance and Strategic Business Planning forMulticast Media, a startup company building a broadband broad-cast overlay to the Internet to deliver high-quality streamingmedia to the desktop. Prior to Multicast Media, he spent sevenyears as a trader at Paribas Corporation in the Fixed IncomeDepartment on the financing, short-term arbitrage, and U.S.Agency desks. Mr. Stone was also an assistant economist in theDomestic Research Division and a trader/analyst in the Open-Market Trading Division of the Federal Reserve Bank of NewYork. Mr. Stone has an undergraduate degree in economics andcomputer sciences:mathematics from the University of Roches-ter and received an MBA from the Stern School of Business atNew York University.


PREAMBLE

The Wyckoff Method is a school of thought in technical marketanalysis that necessitates judgment. The analyst- trader acquires judg-ment through experience and through well-guided illustrations ofbasic principles. Although the Wyckoff Method is not a mechanicalsystem per se, nevertheless high reward/low risk entry points can beroutinely and systematically judged with the aid of a checklist of “NineTests.” Each test in the list of “Nine Tests” represents a Wyckoff Prin-ciple.

One purpose of this article is to demonstrate the “Nine ClassicBuying Tests” of the Wyckoff Method at work via a case study of thestock of the San Francisco Company. Although the case name is dis-guised as the San Francisco Company (SF), it does represent an ac-tual company in the energy sector. For the sake of economy, theillustrations in this article feature the bull side of the market, theycan be inverted to illustrate the bear-side of the market.

The classic set of “Nine Classic Buying Tests” (and “Nine SellingTests”) was designed to diagnose significant reversal formations: the“Nine Classic Buying Tests” define the emergence of a new bull trend(See Side Bar #1). A new bull trend emerges out of a base that formsafter a significant price decline. (The “Nine Selling Tests” help de-fine the onset of a bear trend out of top formation following a signifi-cant advance.) These nine classic tests of Wyckoff are logical, time-tested, and reliable. However, the original set of nine tests was notdesigned to include all of those very crucial consolidation periodsthat occur during bull markets and bear markets.

Students of the Wyckoff Method refer to consolidations as re-ac-cumulation or redistribution. There exists a void in the WyckoffMethod with respect to tests to define the trends that emerge out ofconsolidation formations. Thus, a second major purpose of this ar-ticle is an attempt to fill a void in the Wyckoff Method by introducinga new set of “Nine Buying Tests for Re-accumulation.” This new setof “Nine Buying Tests for Re-accumulation” (See Side Bar #2) shallbe illustrated with the same San Francisco Company case study towhich will be applied “Nine Classic Buying Tests” mentioned in thepreceding paragraph.

The San Francisco Company (SF) case study used in this articlereflects an actual trade made by an expert in the Wyckoff Method.This Wyckoff expert used the stock options listed on SF as his tradingvehicle. Vertical line (bar) charts and figure (point and figure) chartsof SF will be used to illustrate both sets of “Nine Classic Buying Tests,”for accumulation and for re-accumulation.

As the reader approaches this case of “Nine Classic Buying Tests,”he/she ought to keep in mind the following admonitions from theReminiscences of a Stock Operator (See Appendix):

“The average ticker hound – or, as they used to call him, tape-worm – goes wrong, I suspect, as much from overspecializationas from anything else. It means a highly expensive inelasticity.After all, the game of speculation isn’t all mathematics or setrules, however rigid the main laws may be. Even in my tapereading something enters that is more than mere arithmetic.There is what I call the behavior of a stock, actions that enableyou to judge whether or not it is going to proceed in accordancewith the precedents that your observation has noted. If a stockdoesn’t act right don’t touch it; because, being unable to tell

precisely what is wrong, you cannot tell which way it is going.No diagnosis, no prognosis. No prognosis, no profit.“This experience has been the experience of so many traders somany times that I can give this rule: In a narrow market, whenprices are not getting anywhere to speak of but move within anarrow range, there is no sense in trying to anticipate what thenext big movement is going to be – up or down. The thing to dois to watch the market, read the tape to determine the limits ofthe get-nowhere prices, and make up your mind that you willnot take an interest until the price breaks through the limit ineither direction. A speculator must concern himself with mak-ing money out of the market and not with insisting that the tapemust agree with him.“Therefore, the thing to determine is the speculative line of leastresistance at the moment of trading; and what he should waitfor is the moment when that line defines itself, because that ishis signal to get busy.”

THE “NINE CLASSIC BUYING TESTS” OF THE

WYCKOFF METHOD

Side Bar #1 lists the “Nine Classic Buying Tests” of the WyckoffMethod. Chart # 1 exhibits a vertical chart of the SF Company; Chart#2 shows a figure chart of SF. A guide to “How to Make Price Projec-tions Using a Figure Chart” is attached as an Appendix to “WyckoffTests.”

This case situation of SF involves a Wyckoff-oriented trader whodiagnosed trading opportunities in SF. While the general marketindex is not shown here, these trading opportunities exhibited goodrelative strength compared to the general market index. The “NineClassic Buying Tests” were passed at the conclusion of the base-build-ing period and the trader elected to buy call options on SF and toenter stop-loss orders (mental) just below prior supports in the trad-ing range. Later, as periods of consolidation come to a halt, the tradercould roll his options forward to a later month and to a higher strikeprice. At the end of the SF case, the option trader is in a position towrap up his campaign, take his profit, and go home.

The First Wyckoff buying test to be passed was Downside (price)objective accomplished. This test was passed at point #4 on the fig-ure chart, which is the $21 level for SF. The preceding top in SFaround point #3 built the cause for the decline, and at $21 the maxi-mum effect of that cause was realized.

The Second Wyckoff buying test was passed at point #8 on the barchart, which was a “secondary test” that occurred on relatively lightvolume and narrowing downside price movement compared to the“selling climax” at point #4. At point #4 the relative increase in vol-ume and the price closing at the high of the day signaled to our Wyck-off-oriented trader that a provisional “selling climax” might be at hand.At point #4, demand was entering the market to absorb the supply ofstock being offered in the vicinity of the downside price objective (buy-ing test one). At this juncture the trader should have covered anyoutstanding short sales on SF at the open of the next day.

The successful secondary test at point #8 revealed that supply wasbeing exhausted for the moment and so the downtrend was stopped,at least temporarily. It was now the job of the trader to sit patientlyon the sidelines until an accumulation base had been formed.

WYCKOFF TESTS: NINE CLASSIC TESTS FOR ACCUMULATION;NINE NEW TESTS FOR RE-ACCUMULATION

Henry O. (Hank) Pruden, Ph.D.6


Chart 1Vertical Chart, Daily

San Francisco Company

Chart 2Point & Figure

San Francisco Company


Side Bar #2New Wyckoff Buying Tests Modified for Re-accumulation

Nine Re-accumulation Tests:

1. Resistance Line Broken (Horizontal Line across the Top of The Trading Range)

2. Activity Bullish (e.g., volume expanding on rallies, shrinking on declines)

3. Higher Lows (price)

4. Higher Highs (price)

5. Favorable Relative Strength (equal to or stronger than the market)

6. Correction Completed in price and/or time (e.g., 1/2 retracement, support line reached)

7. Consolidation pattern formed (e.g., triangular formation)

8. Stepping Stone Count Confirming Count

9. 3-1 Reward to Risk Ratio

Side Bar #1Wyckoff Buying Tests: Nine Classic Tests for Accumulation

Nine Buying Tests (applied to an average or a stock after a decline)*Indication: Determined From:1) Downside price objective accomplished Figure Chart

2) Preliminary support, selling climax, secondary test Vertical and Figure3) Activity bullish (volume increases on rallies and

decreases on reactions) Vertical

4) Downward stride broken (i.e., supply line penetrated) Vertical or Figure

5) Higher supports (daily low) Vertical or Figure

6) Higher tops (daily high prices rising) Vertical or Figure

7) Stock stronger than the market (i.e., stock moreresponsive on rallies and more resistant toreactions than the market index) Vertical Chart

8) Base forming (horizontal price line) Figure Chart

9) Estimated upside profit potential is at least three times Figure Chart forthe loss if protective stop is hit Profit Objective

* Adapted with modifications from Jack K. Hutson, Editor, Charting the Market: TheWyckoff Method (Technical Analysis, Inc., Seattle, Washington, 1986), page 87

Buying Test Three requires judging the volume on the rising andfalling price waves in the trading range. A visual inspection revealsthat by point #16 on the SF chart, volume was expanding on the ral-lies and shrinking on the declines. By the time point #16 was reachedon the vertical chart, SF would have passed the test: “Activity bull-ish.” Turning once again to the figure chart discloses that in thevicinity of point #10 the downward sloping supply line (dashed lineSS) was broken. Thus around point #10, the Fourth Buying Test waspassed. These four foregoing tests, although necessary, were not suf-ficient evidence of accumulation, so the trader had to remain patientuntil all of the “buying tests” clearly revealed that a base had beenformed and that the evidence had accumulated to prove that the lineof least resistance was decidedly upward.

The next two Wyckoff Tests are crucial to the definition of anupward line of least resistance. Buying Test Five is higher lows (highersupports) and Buying Test Six is higher highs (higher tops). Thevertical line or bar chart of SF showed higher price lows along thegradient of points #14, #16, and #18. In a parallel fashion, a series ofrising price peaks appeared at points #12, #13, #15, and #17. At points#17 and #18, the trader-analyst could clearly declare that the higherhighs and higher lows had been reached, and, therefore, WyckoffBuying Tests Five and Six had been passed.

Points #15 and #16, and then again #17 and #18 on the charts,may also be viewed as “Jumps and Backups,” hence legitimate junc-

tures at which to enter a long position. (See January 2001 issue ofthe Active Trader magazine). At point #16 on the charts, and evenmore definitely at point #18, the trader in the SF case concluded thata base had been formed, a cause had been built and a favorable re-ward-to-risk ratio was present. The “count” taken along the $22 lineof the figure chart from point #16 back to beyond point #4 gener-ated a cause of 27 points for upside projections of $47-49, when thatcount was added to the low of the trading range at $20 and to thecount line itself at $22. Moreover, the count along the $25 level atpoint #18 sanctioned price projections as high as $57. As a result ofthese analyses, the trader was justified in concluding that the EighthTest had been passed.

Entering a long position in SF at $25 (point #18) and setting aprotective stop-loss order just below support at $19 would create arisk exposure of $6. The figure chart count along the 25 line equaled31 points of upside potential. Thus, the estimated profit potentialexceeded the indicated risk by over three times, so Buying Test Ninewas also passed. A comparison of the SF chart to its relevant marketindex (not shown) would have revealed that SF was comparativelystronger than the market. Consequently, SF was favored as a candi-date with superior upside prospects. (Buying Test Seven was passed.)

By the time SF had reached point, #18 all of the “Nine ClassicBuying Tests” had been passed. At point #1: the line-of-least-resis-tance had defined itself as upward trending and the trader couldhave entered call option positions with favorable reward to risk pa-rameters. At this stage the trader did purchase SF call options thatwere at the money.

NINE NEW BUYING TESTS FOR RE-ACCUMULATION

In a quest for unity and economy, numerous principles of theWyckoff Method were distilled into “Nine Classic Buying Tests” and“Nine Selling Tests.” As explained above, the nine buying tests wereoriginally designed to define trends coming out of major areas ofaccumulation that followed significant price declines. In addition tothese major reversal formations at bottoms and tops, there also ap-pear many important continuation patterns known by students ofWyckoff as “re-accumulation” and “redistribution.” However, theseimportant consolidation patterns lack an appropriate set of “NineTests” to define the resumption of the upward trend or downwardtrend. Re-accumulation and redistribution areas simply lack a set ofbuying tests /selling tests that are equivalent to the “Classic Nine Tests”for major accumulation or major distribution. Unfortunately, theoriginal set of Wyckoff tests that were used to define departures frombottoms or tops cannot be transferred easily nor applied en toto tozones of re-accumulation or redistribution. Some tests, such as “Pre-liminary Support and Selling Climax and Secondary Test” simply donot apply. The selling climax is good for signaling the onset of a bot-tom after a bear market decline. But re-accumulation zones startafter a price advance, and thus most often commence with a buyingclimax. A straightforward modification of the “climax rule” to fit re-accumulations is made even more ambiguous by the fact that distri-bution after a bull market advance may likewise start with “prelimi-nary supply and a buying climax.”

Similar limitations apply to other tests found in the original list ofnine. For instance, neither “The fulfillment of downside (upside)price objectives” nor the “breaking of downward (upward) slopingprice line” are necessarily relevant for analyzing re-accumulation (re-distribution). In their place it is suggested that we substitute otherWyckoff rules that tell us more clearly that a correction has beencompleted in time and price. These substitute measures are, for ex-ample, the interception by price of the upward sloping demand line


and/or the reaching of the 1/2 re-tracement level.*It is suggested that in place of “ downward stride broken,” the

relevant buying test for re-accumulation become the “breaking of thehorizontal resistance line along the top of the trading range.” Thathorizontal resistance line serves to confine the sideways trend chan-nel in much the same way as the downtrend slanting supply linesconfine a bearish trend channel. Moreover, when a wedge or trian-gular formation appears, the Wyckoff literature advises the studentto enter upon the significant price-and-volume breaking of the resis-tance (support) line.

The “Stepping Stone Confirming Count” measures the amountof potential generated during a re-accumulation trading range. The“stepping stone confirming count” deserves special consideration asa re-accumulation test because it possesses an important Wyckoff tim-ing principle. Thus the trader should be alert to the possible re-sumption of the upward trend when the figure chart has generatedenough re-accumulation potential or “count” to confirm the targetfrom the original base. In the case of SF, this means the trader shouldbe poised for a resumption of the upward trend when the count gen-erated during a consolidation grows large enough to meet the priceobjectives that equal the objectives generated during the original ac-cumulation base. If along the $35 level, for instance, the up anddown price waves during a period of sideways consolidation reach apoint where the figure chart count measures 14 points, thereby pro-jecting to $49, then the trader should become highly alert for thepossible resumption of the upward trend. Remember that the origi-nal base count along the $22 level (point #16) projected to a maxi-mum of $49. If a consolidation projects to the same objectives, thenwe say that it “Confirms” the original count taken along the base.The “stepping stone confirming count” appears as Re-accumulationTest Number Eight on Side Bar #2.

FAILED TESTS

To illustrate the new list of modified Wyckoff Tests for Re-accu-mulation that appear in Side Bar #2, let us return to the case study ofthe SF Company. After the base had been completed, the Wyckoff-oriented trader entered a long option position at point #18 on the SFcharts. The SF stock then moved up sharply from point #18 to point#19, where it encountered enough supply to halt its advance, and soSF entered a period of hesitation and sideways movement starting at#19. This period of hesitation commenced with a “buying climax”around point # 19, which would also have alerted the trader of thepossible onset of re-accumulation before resumption of the upwardtrend or even possibly distribution leading to a reversal of trend. Thetrader, who was actually operating in SF at the time of this case study,recounted his upside figure chart objective to $49 and chose to waitout is interruption in the trend.

At point #20 the trader observed a “Spring” situation and so pre-sumably he could have ventured a long position around the $29 level(see Active Trader magazine, August 2000, for “Springs and Up-thrusts”). At this juncture he could have consulted side bar #2 forthe checklist of Re-accumulation Tests. At point #20, he could haveconcluded that Re-accumulation Buy Tests Number Two and Num-ber Three had been passed. At point #20 the volume had dried upconsiderably and the downside price progress was minimal, which

taken together revealed a lack of supply being pressed upon the mar-ket. Moreover, the interception of the rising support line C-C indi-cated that a sufficient correction in time and price had taken place(Test Six). However, it was not until the subsequent surpassing of theresistance along the $31 level on June 11, on wide upside price move-ment and expanded volume, that SF satisfied several other Re-accu-mulation Tests, such as Test Number One “Resistance Line Broken”and Test Number Four, “Higher High (price).” Then at point #23, apullback to a “Higher Low” was executed (Test Three) and a countof the Figure chart along the 31 level would have projected upwardto $37-39. However, this count was insufficient to confirm the earlierprice target projections of $47-49 taken along the $22 level. Hence,Re-accumulation Test Number Eight was not passed. Moreover, a tradetaken at 31 also would have fallen short of the 3-1 reward-to-risk mini-mum because a stop would need to have been placed 3 points away at28, and the re-accumulation count was only 8 points. Thus, Test Num-ber Nine also failed. Presumably a pattern analyst could have labeledthe consolidation from #19 to #20 a “pennant” (Test Seven).

RE-ACCUMULATION TESTS PASSED

With two tests already failed our trader chose to pass up adding tohis position at the point #23 juncture on the charts. Shortly thereaf-ter the SF stock shot up from point #23 to point #27 and underwenta more prolonged correction. This complex correction would havebeen a challenge to the pattern recognition skills of most WyckoffAnalysts. Nonetheless, the Wyckoff expert who was operating in thestock identified it as a large wedge or apex (often called a “one-eyed-Joe” by Wyckoffians), which thus fulfilled Re-accumulation Test Num-ber Seven. He took a count taken across the $35 level back to thezone around point #22. That count indicated a re-accumulation thatwas sizable enough to reach the $47-49 target that was first estab-lished at point #16, and in the process it flashed a “Stepping StoneConfirming Count” (Re-accumulation buy signal Number Seven).

As price broke out of this wedge formation, it burst through the(downward sloping) “Resistance Line” connecting points 27 and 30,thereby triggering a passage of Re-accumulation Test Number One.On balance, the volume tended to expand during the rallies andshrink during the declines, while the SF stock was in the triangulartrading range (passage of Re-accumulation Test Number Two). Priceregistered a series of higher lows from point #23 to point #28 to point#31 (passed Re-accumulation Test Number Three). These series ofhigher lows by SF contrasted sharply with series of lower lows thatwere occurring in the general market index at that time (passage ofRe-accumulation Test Number Five). Moreover, at point #28 and #31price met support near the 1/2 retracement level of the move frompoint #20 to point #27 (“1/2 ” mark on Chart 1), thus fulfilling Re-accumulation Test Number Six. At either point 28 or point 31, thetrader would have had a better than 3-1 reward to risk ratio (14 countvs. 3-4 points of risk) for the passage of Re-accumulation Test Num-ber Nine.

The trader under the foregoing re-accumulation circumstancesshould have (and did) roll his options contract forward to a laterexpiration and higher strike price. He simultaneously increased thesize of his line. The passage of all nine re-accumulation tests hadcreated a compelling enough case for him to roll his option con-tracts forward at the $35 strike and to add to his position.

CONCLUSION

When SF reached the $49 level, the trader exited his SF optionsposition. He judged that the relatively high volume occurring in theprice-objective zone around $49 was sufficient reason to exit. To make

* Examples of these and other tests for re-accumulation are available in the Wyckoffliterature. In Basic Lecture Number 12 of the SMI/Wyckoff course, for instance, thenarrator counsels the student to place resting buy orders at the 1/2 re-tracement levelin order to add positions during corrections in a bull market. Elsewhere in the Wyck-off literature the student is admonished to purchase when the price intercepts andencounters support along an important upward slanting demand line.


the case for exiting even more enticing, the general market indexhad started to weaken and diverge from the higher price set by SFaround $49.

There were targets outstanding at $51-$57, but this Wyckoff-ori-ented trader elected to take his profits at $49 because that was themaximum effect of the cause built during the re-accumulation step-ping-stone-count along the $35 line (point #22 to point #31). Hereckoned that he would have to weather another sideways to downcorrection/consolidation as further preparation for the final advance.He further reckoned that the risk did not justify waiting to capturethe final 8 points available beyond $49. Of course, as we can seeretrospectively, he exited prematurely because SF promptly advancedto $54. (Upon further reflection, this Wyckoff trader said that hewould do the same thing again because “bulls make money, bearsmake money, and pigs get slaughtered.”)

The case study of the San Francisco Company (SF) demonstratedhow, with the help the Wyckoff “Nine Classic Buying Tests,” an op-tion trader could have entered favorable reward-to-risk long positionsjust as the line-of-least resistance became defined with the passage ofthe “Nine Classic Tests” for accumulation and as the stock was leav-ing the base formation. This case study also demonstrated how anoption trader could have later employed a new set of the “Nine Re-accumulation Tests” to both roll his contracts forward and to add tohis position. The fulfillment of the “stepping stone confirming count”nature of this re-accumulation consolidation gave the trader addedreason to hold on to his positions until his longer-term base targetswere being reached at $49. Furthermore, the “stepping stone con-firming count” provided an additional compelling reason for him toexit his long options on the burst of strength as SF reached the $49level.

In general, the Wyckoff “Nine Classic Buying Tests” and the set of“Nine New Tests for Re-accumulation” can help investors and tradersto advance forward in their quest to control risk, ride winners andtake home maximum profits.

APPENDIX TO “WYCKOFF TESTS”

“How to Make Price Projections Using a Figure Chart”by Prof. Hank Pruden, Ph.D., Golden Gate University

“The average ticker hound – or, as they used to call him, tape-worm – goes wrong, I suspect, as much from overspecialization asfrom anything else. It means a highly expensive inelasticity. Afterall, the game of speculation isn’t all mathematics or set rules, how-ever rigid the main laws may be. Even in my tape reading, somethingenters that is more than mere arithmetic.”

- Reminiscences of a Stock Operator

The Wyckoff Method rests upon three main laws: (1) the law ofsupply and demand, (2) the law of effort vs. result, and (3) the law ofcause and effect.

According to the Wyckoff Law of Cause and Effect, the trader-investor-analyst measures the extent of the cause built up during atrading range and then projects a price objective the potential effectof that cause. The relationship between the cause and the subse-quent effect is one-to-one, which means that every unit of cause thatis measured horizontally in a trading range translates into an expectedone unit of vertical effect.

The cause is created during the up and down buying and sellingwaves that occur during a trading range.* The cause is measuredand projected on the figure chart according to the Wyckoff “Countguide.” The Wyckoff Count Guide is stated as follows (Source: Wyck-off/Stock Market Institute):

■ After having identified a Sign of Strength (SOS) on the verticalline chart, locate the last point at which support was met on areaction – the Last Point of Support – (LPS). Locate this point onyour figure chart also and count from right-to-left, taking yourmost conservative count first and moving further to the left as themove progresses.

■ In moving to the left, turn to your vertical line chart and dividethe area of accumulation into phases, adding one complete phaseat a time. Never add only part of a phase to your count. Volumeaction will usually show where the phase began and ended.

■ As the moves progress you will often see a lateral move forming ata higher level. Very often such a move will become a “SteppingStone Confirming Count” of the original count. Thus, as such anlevel forms, you can often get a timing indication by watching theaction of the stock as the potential count begins to confirm theoriginal count. A resumption could begin at such a point.

■ For longer term counts one should add his/her count to the ex-act low, or a point about one-half way between the low and thecount line. You will thus be certain that the most conservativecount is being used.

■ Counts are only points of “Stop, Look and Listen,” and shouldnever be looked upon as exact points of stopping and turning.Use them as projected points where a turn could occur, and usethe vertical line chart to show the action as these points are ap-proached.

■ In the case of a longer-term count, often the Last Point of Sup-port (LPS) comes at the original level of climax, and this levelshould be looked at first in studying the longer term count. Theclimax itself indicated a reversal, with the subsequent action be-ing the forming of the cause for the next effect. For the LastPoint of Support (LPS) to come at such a level of climax usuallymakes it a more valid count. Very often the climax is preceded bypreliminary support and the Last Point of Support often occursat the same level as the preliminary support.

■ A #3 Spring or the Secondary test of a #2 Spring, quite often con-stitutes the Sign of Strength and the Last Point of Support in thesame action which is reached at the same point and at the sametime. Usually a Spring will be followed by a more important Signof Strength and the reaction following that Sign of Strength isalso a valid Last Point of Support.

■ Frequently, long term counts on three- and five-point charts areconfirmed by subsequent minor counts on the one-point chart asthe move progresses. Watch for this confirmation very carefullyas it often indicates when a move will be resumed.

■ In case of three-point or five-point charts, the same count lineshould be used as for the one-point chart.Analysts who wish to use the Wyckoff Count Guide must appreci-

ate and comprehend certain philosophies and procedures unique tothe Wyckoff figure chart. Four key elements of Wyckoff figure chartanalyses are as follows:1. Figure charts play a special supplementary and complementary

role in the Wyckoff Method. The key law of Supply and Demandrelies upon the vertical chart to diagnose the present position andfuture trend of the market. The figure chart is not used per se fordetermining the trend of the market, because the volume infor-mation of the vertical chart makes it a superior tool for determin-ing the trend. Philosophically, Wyckoff analysts believe the verti-

* For readers who recall their high school physics lessons, the law of the cause andeffect can be likened to Hooke's Law of Elasticity. Hooke’s Law declares the agita-tions up and down build up energy, the cause (e.g. agitating a metal coat hanger backand forth) and the resultant effect (bend the hanger out of shape) expends energy inan exactly one-to-one proportion to the preceding energy built up.


cal chart ought to be used for trend analysis; however determin-ing the potential extent of the move is the special provence of thefigure chart, sometimes referred to as the “cause and effect chart.”

2. Procedure. The building blocks of the figure chart are box size,intraday data, number of reversal points and full-unit crossing.Most commonly the box size is one point. Hence, intraday priceaction must meet or exceed the full price levels to trigger a figurechart entry.Reversal points are normally one point or three-point. For the“one point figure” chart, a very special consideration to keep inmind under the Wyckoff figure chart procedure is the necessityof having at least two entries in any column. Many software pro-grams change columns when price changes direction, even if onlya single entry exists in a column. To compensate for this, theanalyst must shift prices to create a column with at least two en-tries in a column before price can move to the next column. Hencea quick down, up, down of one point each would remain in a singlecolumn.For larger moves, the analyst has the option to either relying uponthe three-point reversal or an increase in the box size.

3. Perspective. The analyst can visualize horizontal counts as fittingwithin a saucer appearing bottom and a dome looking top. Thefirst count line should be conservative, nearest the lows, and beconsidered as the minimum possible. The next count line willusually be within the trading range, broader, and considered thelikely objective. Finally, the pullback following the upside “jump”or valid/breakout creates the widest count and the highest up-side count, and is thus the least conservative measurement (this isthe last-point-of-support that follows after a more important sign-of-strength).

REFERENCES

■ Forte, Jim, CMT, “Anatomy of a Trading Range,” MTA Journal,Summer-Fall 1994

■ Hutson, Jack K., Editor, Charting The Market: The WyckoffMethod, Technical Analysis, Inc., 1986

■ Mathis, David, “Santa Fe: A Classic,” audio tape and charts, StockMarket Institute, 1978

■ Pruden, Henry O. (Hank), Ph.D., “Trading the Wyckoff Way:Buying Springs and Selling Upthrusts,” Active Trader magazine,August 2000

■ Pruden, Henry O. (Hank), Ph.D., “Wyckoff Axioms: Jumps andBackups,” Active Trader magazine, January-February 2001

■ __________, Introduction to the Wyckoff Method of Stock Mar-ket Analysis –Text Exhibits and Illustrations, Stock Market Insti-tute, 1983

■ __________, “Basic Lecture No. 12,” audio tape and charts, StockMarket Institute, 1968

BIOGRAPHY

Henry O. (Hank) Pruden, Ph.D., is Professor of Business andis Executive Director of The Institute for Technical Market Analy-sis at Golden Gate University, San Francisco, CA, and he is alsoEditor of the Market Technicians Association Journal. Hank canbe reached at [email protected], phone 415/442-6583 andwww.hankpruden.com.

This article was reviewed, edited and approved by Mr. DavidUpshaw, CFA, CMT, Associate Editor, MTA Journal.

55 - 2001 Summer

Documents

Transcript of 55 - 2001 Summer