journal of technical analysis (jota). issue 42 (1993, winter)

WINTER 1993~SPRING 1994 ISSUE 42

A PUBLICATION OF THE MARKET TECHNlClANS ASSOCIATION

ONE WORLD TRADE CENTER, SUITE 4447 . NEW YORK, NEW YORK 10048 . (212) 912-1064

MARKET TECHNICIANS ASSOCIATION JOURNAL

Issue 42

Editor

Henry 0. Pruden, Ph.D. Golden Gate University

San Francisco, California

Associate Editor

George A. Schade, Jr., CMT

Scottsdale, Arizona

Manuscript Reviewers

Winter 1993 - Spring 1994

John A. Carder, CMT Richard C. Orr, Ph.D. Topline Graphics Chronos Corporation Boulder, Colorado Lexington, Massachusetts

Ann F. Cody Invest Financial Corporation

Tampa, Florida

Eugene E. Peroni, Jr.

Janney Montgomery Scott Inc.

Philadelphia, Pennsylvania

Don Dillistone, CFA, CMT Richardson Greenshields of Canada

Winnepeg, Manitoba

David L. Upshaw, CFA, CMT

Waddell and Reed, Inc.

Shawnee Mission, Kansas

Robert I. Webb, Ph.D. Associate Professor and

Paul Tudor Jones II Research Fellow

McIntire School of Commerce University of Virginia

Charlottesville, Virginia

Printer

Tritech Services

New York, New York

Publisher

Market Technicians Association

One World Trade Center, Suite 4447

New York, New York 10048

MTA JOURNAL i WINTER 1993 - SPRING 1994 1

2 MTA JOURNAL / WINTER 1993 - SPRING 1994

MARKET TECHNICIANS ASSOCIATION, INC.

Member and Affiliate Information

ELIGIBILITY: MEMBERSHIP is available to those “whose professional efforts are spent practicing

financial technical analysis that is either made available to the investing public or becomes a primary

input into an active portfolio management process or for whom technical analysis is the basis of their decision-making process.” Applicants for membership must be engaged in the above capacity

for five years and must be sponsored by three MTA members.

AFFILIATE category is available to individuals who are interested in keeping abreast of the field

of technical analysis, but who don’t fully meet the requirements for membership. Privileges are

noted below.

DUES: Dues for Members and Affiliates are $150.00 per year and are payable when joining the

MTA and thereafter upon receipt of annual dues notice mailed on July 1. College students may join at a reduced rate of $50.00 with the endorsement of a professor.

APPLICATION FEES: Applicants for membership will be charged a one-time application fee of $25.00; no fee for affiliates.

Benefits of MTA

Regular Members Affiliates

Invitation to Monthly MTA Educational Meetings - No charge Yes Yes

Receive Monthly MTA Newsletter Yes Yes

Receive MTA Journal Yes Yes

Use of MTA Library Yes Yes

Participate on Various Committees Yes Yes

Colleague of IFTA Yes Yes

Eligible to Chair a Committee Yes No

Eligible to Vote Yes No

Annual Subscription to the MTA Journal for non-members - $50.00 (minimum two issues).

Single Issue of MTA Journal (including back issues) - $20.00 each for members and affiliates,

and $30.00 for non-members.

STYLE SHEET FOR THE SUBMISSION OF ARTICLES

MTA Editorial Policy

The MARKET TECHNICIANS ASSOCIATION JOURNAL is published by the Market Technicians Associ-

ation, One World Trade Center, Suite 4447, New York, NY 10048 to promote the investigation and analysis of price and volume activities of the world’s financial markets. The MTA Journd is

distributed to individuals (both academic and practitioner) and libraries in the United States, Canada, Europe and several other countries. The Journal is copyrighted by the Market Technicians Associa-

tion and registered with the Library of Congress. All rights are reserved.

Style For The MTA Journal

All papers submitted to the MTA Journal are

requested to have the following items as pre-

requisites to consideration for publication:

1. Short (one paragraph) biographical presenta-

tion for inclusion at the end of the accepted article upon publication. Name and affiliation

will be shown under the title.

2. All charts should be provided in camera-ready

form and be properly labeled for text reference.

3. Paper should be submitted double-spaced if

typewritten, in completed form on 8% by 11

inch paper. If both sides are used, care should be taken to use sufficiently heavy paper to

avoid reverse side images. Footnotes and refer-

ences should be put at the end of the article. Sub- mission on disk is encouraged by arrangement.

4. Greek characters should be avoided in the

text and in all formulae.

5. Two submission copies are necessary.

Manuscript of any style will be received and

examined, but upon acceptance, they should be

prepared in accordance with the above policies.

Mail your manuscripts to:

Dr. Henry Pruden

PO. Box 1348

Ross, CA 94957

MTA JOURNAL /WINTER 1993 - SPRING 1994 3

MARKET TECHNICIANS ASSOCIATION Board of Directors, 1993-94

Offkers/Offke Manager

President Vice-President/Long Range Vice-President/Seminar Kenneth Tower, CMT Mike Epstein Ned Davis UST Securities Corp. Sherwood Securities Ned Davis Research Inc. 5 Vaughn Drive 1 Exchange Plaza, 21st Fl. PO. Box 1287 Princeton, NJ 08543-5209 New York, NY 10006 Nokomis, FL 34274-1287 6091734-7747 2121482-2454 8131484-6107

Treasurer Secretary James Bohan Andrea Neumann

MTA Offke Manager

Merrill Lynch, No. Tower CM&M Futures Shelley Lebeck Market Technicians Association

World Finl. Center, 19th Fl. New York, NY 10281-1319

140 Broadway, 17th Fl. 1 World Trade Center, Suite 4447 New York, NY 10005 New York, NY 10048

2121449-0552 2121825-9302 212/912-0995 FAX: 2121912-1064

Committee Chairpersons

Accreditation William Raftery Smith Barney Shearson, 27th Fl. 1345 Avenue of Americas New York, NY 10105 2121698-6003

Foundation James Stewart, Jr., CMT NatWest Fin’l. Markets Group 10 Exchange Place, 22nd Fl. Jersey City, NJ 07302 201/547-2910

Membership Philip Roth, CMT Dean Witter Reynolds 2 World Trade Center, 63rd Fl. New York, NY 10048 2121392-3516

Computer John Bollinger, CFA, CMT Bollinger Capital Mgmt. PO. Box 3358 Manhattan Beach, CA 90266 310/798-8855

IFTA Conference 1993 David Krell, CMT New York Stock Exchange 11 Wall Street, 23rd Fl. New York, NY 10005 2121656-2865

Newsletter Jack Cahn, CMT CahnlVince & Company 100 Fountain View Terrace, # 13 Lake St. Louis, MO 63367 3141639-5315

Derivatives/Forex Ralph Vince 279 North Street Chagrin Falls, OH 44022 2161247-0073

IFTA Liaison Bruce Kamich, CMT MCM, Inc. 71 Broadway, 11th Fl. New York, NY 10006 2121908-4326

Placements Fred Schutzman, CMT BFF Trading Group Inc. 103 Water Street, #3L New York, NY 10005 2121425-1440

Education Dodge Dorland LANDOR Investment Mgmt. 103 East 75th Street, X4FiE New York, NY 10021 2121737-1254

Journal Dr. Henry Pruden P.O. Box 1348 Ross, CA 94957 4151459-1319

Programs Pamela King Brean Murray, Foster Set 633 Third Avenue, 11th Fl. New York, NY 10017-6795 2121476-0756

4 MTA JOURNAL /WINTER 1993 - SPRING 1994

Ethics & Standards Paul Desmond Lowry’s Reports, Inc. 631 U.S. Highway 1, #305 No. Palm Beach, FL 33408 4071842-3514

Library Cay Lee, CMT B.T.C. PO. Box 1230 Valrico, FL 33594 8131681-2996

Regions Philip Erlanger, CMT Phil Erlanger Research Co. PO. Box 2680 Acton, MA 01720 5081263-2536

TABLE OF CONTENTS

West Coast Boys.. . . . . . . . . . . . . . . .ll

Wayne H. Wagner “We are all ‘West Coast Boys’ now”, asserts Wagner in this

entertaining trip down memory lane. He contrasts the intuitive, experiential and

technical type he calls the “East Coast Boys” with the analytical, academic and quantitative type he dubs the “West

Coast Boys”.

A Comparison of

Three Approaches

to Price Channels . . . . . . . . . . . . . .15

JR. Davis The Gann Swing method,

the Channel (Dennis) method and the 5-Count Method were subjected to

empirical and statistical tests. J.R. Davis offers conclusions and implica- tions based on these findings.

Enhanced Candlepower . . . . .19

Theodore E. Loud, C.l?P The predic-

tive power of the Japanese Candlestick

Method can be enhanced by combining it with other techniques. To enhance the power of candlesticks during uptrending

markets, Ted Loud explores the ability of an uptrend to swiftly correct speculative excesses and its ability to sustain

critical levels of momentum.

The Impact of Bond Yield

Gaps on Currencies-

An Intermarket View . . . . . . . . .37

R. John Slatter, C.FIA., M.S.TA. By examining the relationship between

government bond yields and currency, John Slatter extends John J. Murphy’s pioneering work in intermarket analysis.

It was concluded that bond yield spreads provide an important strategic overview

of the currency markets.


The Derivative Membership

Oscillator: and Affiliate

A New Approach Information . . . . . . . . . . . . . . . . . . . . . . . .2

for an Old

Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Connie Brown Traditional momen- Style Sheet for

turn indicators fail to solve the problem the Submission

of incorrect or premature buy/sell of Articles . . . . . . . . . ..s.......*...... 3

signals caused by the Elliott Wave Prin- ciples complex corrective patterns. In MTA Offkers this article, Connie Brown introduces a new approach to handling this problem:

and Committee

an oscillator which is a triple smoothed Chairpersons . . . . . . . . . . . . . . . . . . . . . .4

derivative of RSI plotted as a histogram. Real time performance evaluations of this oscillator are provided.

Editor’s Commentary.. . . . . . . . . .7

Letters to the Editor.. . . . . . . .8 Don Dill&one, C.l?A., C.M.?: John Kozey, CMT

6 MTA JOURNAL I WINTER 1993 - SPRING 1994

Editor’s Commentary

The Manuscript Review Process by Henry 0. Pruden, Ph.D., Editor

An important cast of characters work busily behind he scenes of technical analysis. On the W&Z Street Veek television program these characters were cast LS “The Elves”. The noun “Elf’ conjures up the im- lge of a small, mischievous fairy with magical bowers. Presumably the elves work furtively on ethnical indicators and then mysteriously generate narket signals.

At the MTA Journal these hard working, behind he scenes, characters are the cast of manuscript ,eviewers. In this commentary I want to take you on 1 tour of the manuscript review process. Some of the iighlights we’ll see are “refereed journals”, “the blind *eview “, “two heads are better than one”, “revise and (hen revise again”, and “the final acceptance”.

rejecting it. As a result, several cycles of revision and review may be necessary in order to help the author meet the MTA Journal’s standards for publication.

ACCEPTANCES-Ultimate authority and respon- sibility for accepting or rejecting an article rests with the Editor. To guide his judgment, the Editor relies heavily upon the written and even verbal inputs of the manuscript reviewers. Through the inclusion of manuscript reviewers, the Editor can tap expertise in many areas of technical analysis. Through these collaborative efforts the Editorial Staff can work to continually improve the quality of the MTA Journal.

REFEREED JOURNAL-A manuscript review pro- :ess raises the statures of the MTA Journal. Having n-titles submitted to a panel for critial review and jossible rejection is the sine quo non of a scholarly ournal. Hence, the referees or manuscript reviewers are engaged in upholding and even the raising the standards of technical analysis as a discipline. Pos- sessing a refereed journal is an indispensable step .n advancing the status of technical analysis.

BLIND REVIEW-Most articles submitted for publication are subjected to a “blind review” process. The names of the authors or other tell-tale references are removed from the manuscript to assure anonym- ity. These blind copies are then sent out to one or more (usually two) manuscript reviewers for their written comments and recommendations. The referees furnish specific suggestions for revision and Dffer their judgments concerning the suitability of the article for publication in the MTA Journal.

TWO HEADS-seeking the judgments of two reviewers concerning quality and suitability reflects the belief that two heads are better than one. Usu- ally an article is sent simultaneously to two of our regular reviewers. In an effort to secure an expert opinion, articles are occasionally sent out for review to persons who are not regular referees.

REVISION-It is the philosophy of the Editorial Staff of the MTA Journal to assist authors as much as possible. We seek to salvage an article rather than


Letters To The Editor

To the Editor:

There was compelling evidence of a rather serious breakdown in communication recently in the sum- mer 1993 edition of the MTA Journal. This was evident in the letter to the editor from Elliot Lapan criticizing equation 2 in John Kozey’s article in the winter 1992-93 Journal (page 201 and Kozey’s reply. Equation 2 originally read:

TSI (y,z) = lOO* EMAz*EMA,*(LS-LS.1)

EMAz*EMAy*(ILS-LS.11)

where the EMA’s are moving averages and LS and LS.l are sequential stock prices. Lapan correctly pointed out that as written, EMAy and EMAz vanish from the equation which thus simplifies:

TSI (y,z) = 100” (LS-LS.1)

OLS-LS.11)

As Lapan states the expression

(LS-LS.1)

(ILS-LS.11)

has only three solutions: - 1,1 or meaningless. So the original equation must always provide one of only three answers; 100, - 100 or meaningless depending on the values of LS and LS.1.

In his rebuttal, Kozey changed the notation to read

TSI (y,z) = 100” [

EMAz (EMAv (LS-LS.1)) 1 EMAz (EMAy (1 LS-LS.11) 1

But this is the same equation. Once the operations (LS-LS. 11 and ( I LS- LS. 11) have been performed, both equations become completely commutative (two times three times four is the same as three times four times two) so the additional brackets are algebraically irrelevant. So Lapan’s comments have not been refuted and his hopes that the problem was due to a typographical error have been dashed. In both of Kozey’s equations, except for sign and the

constant 100, the numerator and denominator are identical and the only information that could possibly be derived from them is whether the stock went up, down or sideways.

Curiously, in his rebuttal, Kozey makes the statement: “The key point in this (new) notation is to em- phasize that there is a double smoothing of the change in price in Formula 2 and not a multiplica- tion of moving averages with the change (in price), which is nowhere stated in the article.” But that is precisely what the expression in the numerator EMAz (EMAy (LS-LS.l)) means: Multiply one moving average times the change in price times a second moving average.

Equation 3 is also fatally flawed.

(3) TSI (y,z) = lOO* EMAz* [

(EMAy (Ls-Ls’l)) (EMAy (ILS-LS.11)) 1

The only difference in this equation is that the only possible solutions are lOOEMAz, - 1OOEMAz or meaningless.

If Kozey actually utilized the equations as he described them in his computer he would only have been provided with answers of 100, - 100, lOOEMAz, - 1OOEMAz or an error message. Yet according to his article, he was able to generate 310 trades. Clearly he got different answers. What went wrong?

The only solution I can think of is that Kozey did not properly translate the operations carried out by his computer into standard algebraic notation. The result was a breakdown in communications between Kozey and his readers.

I have one further, trifling point. Lapan states that when LS = LS.1, TSI would equal infinity. I believe that to be only one of an infinite number of solutions. The absurd equation T = zero divided by zero can be re-written in an equally meaningless truism where zero times T = zero. By inspection, T can thus have any value between minus infinity through to plus infinity.

Don Dillistone, CFA, CMT


To the Editor:

This letter is to respond to the comments of a letter sent to you by Don Dillistone, and forwarded to me.

The letter states that the adjusted equation of the True Strength Index (TSI) of my April 27,1993 letter to you has only three solutions: -1, 1 or meaningless.

The notation that I used is consistent with the notation that the developer of the TSI, William Blau, had used in two articles published in Tech- nical Analysis of Stocks and Commodities magazine. Dillistone states that, “Once the operations (LS-LS.1) and /(LS-LS.l)l have been performed, both equations become completely commutative.” This is correct if this ratio was calculated first in the formula. Here is the calculation process that was used in the study.

The formula is:

TSI (y,z) = lOO* EMAz (EMA, (LS-LS.l))

EMAz (EMAy I(LS-LS.l)l) 1 The calculations are performed in the following order:

I. Numerator calculation: (a) Exponentially smooth, over y periods,

(LS-LS.1). (b) Exponentially smooth, over z periods, the

results of step I(a).

II. Denominator calculation: (a) Exponentially smooth, over y periods, the

absolute value of (LS-LS.l). (b) Exponentially smooth, over z periods, the

results of step II(a).

III. Final TSI Value: (a) Divide the value of I(b) by the value of II(b). (b) Multiply the value of III(a) by 100.

As you can see, the entire numerator and denominator are calculated first and their quotient is calculated next. I agree that it would make no sense to double smooth the ratio of (LS-LS.1) to its absolute value first. This notation, again, is taken from the original articles on TSI published in Technical Analysis of Stocks and Commodities, and, as far as I can tell, is commonly used in that publication.

John Kozey, CMT


West Coast Boys by Wayne H. Wagner

In 1970 I gave a speech to the Q group on Modern Portfolio Theory in Practice. I described how we at Wells Fargo had constructed the world’s first index fund. When I sat down, the man next to me said “This is going to put you on the dinner speech cir- cuit.” He was right: only it took 23 years until I made it to the dinner podium.

You know what a heretic is? My associate Mark Edwards says a heretic is a prophet with a bad sense of timing. Let’s just say I was ahead of my time. (I guess a bad sense of timing is not much of a creden- tial for a Market Technician!)

Surely, I have the credentials as one of the original efficient marketeers, a random walker, a quant. Let me not deceive you: for most of the 30 years in this business I have accepted the view that debunked technical analysis. I believed in what went by the detestable name of Modern Portfolio Theory. Oscar Wilde said “Nothing is so dangerous as being too modern. One is apt (spelled A-P-T) to grow old fashioned quite suddenly.”

I like the thought that says we learn more from the amusing than the wise. So I am here to enter- tain and not to evangelize at 9:00 Saturday night.

Still, we share an intense interest: how do markets work, and how do we work markets? We see the market from different perspectives, and I suspect we both have something to learn from each other.

I am reading a book called, How We Know What Isn’t So by Thomas Gilovich, a psychologist at Cornell. His first example is about basketball play- ers unshakable faith in streaks, which can easily be shown to be illusory. The longer the streak, the more likely it is to be broken, not continue. Gilovich’s thesis is that confirming evidence is far more mem- orable than non-confirming evidence. The difference is between “I can believe this” and “MUST I believe this?” The personal relevance here is that I’ve seen some hard to swallow evidence in our own work that contradicts my long held belief in random walk. But more about that later.

To begin, I want to tell you a story. Once upon a time there was a great market crash. During the crash a well regarded quant was working for

Salomon Bros. During the crash, the quant stares at the Quotron in fascination, trying to grasp the cosmic significance of so massive a market move. At the same time, or so the story goes, Stanley Shopkorn and John Gutfreund are down on the trade floor writing tickets.

Within a short period of time the quant no longer works for Salomon. He just didn’t fit the culture. He was, in the very words of Salomon, a “West Coast Boy.” This, of course, is a standard put-down - like egghead or wonk. Interestingly, our West Coast Boys goes on to form a very successful money management firm - on the west coast, of course!

Here we contrast two different attitudes, two different thought processes, two different paradigms. Who are the West Coast Boys and the East Coast Boys? Why do they seem so different Where did they arise? What do they mean for the future of the market? How long ‘til this speech is over?

In few words, the East Coast Boys are the tradi- tionalists, and the West Coast Boys are the some of us now grey haired young Turks. So how do these cultures differ? East Coast Boys come from a Guild system. Truth is handed down to apprentice from journeyman and master. The NYSE specialist handed his seat to his son - and then on to HIS son. The system changes only slowly. There’s no outside impetus - there’s no place from which to come to a new perspective, no different paradigm on which to stand. There simply was no place significantly different from the center. It worked that way for decades.

But the system was highly vulnerable to change brought in from outside. As my friend Chris Keith, who worked many years on NYSE data systems, is fond of saying, Napoleon wasn’t French, Stalin wasn’t Russian, and Hitler wasn’t German. Funda- mental change always comes from outside.

As in so many industries today, the impetus comes from advancing technology. The East Coast Boys were as smug as the US auto industry, although perhaps a bit less adaptable. West Coast Boys had to wait a generation to rise to influence. This is not a takeover, it’s a war of attrition.


r How do you find the West Coast? East Coast Boys

think it starts at the west bank of the Hudson! The West Coast Boys phenomenon did not start

on the west bank or the Pacific coast, it started on the west coast of Lake Michigan. The West Coast Boys started in the early ‘60’s at the U of Chicago when Larry Fisher and Jim Lorie founded the Center for Research in Securities Prices.

They gathered together prices of every exchange stock from 1926 to 1960. No one had ever organized the data before. Indeed, East Coast Boys never thought it was of any value at all! But now it was possible to study whether some quantitative idea might work.

Of course, it was not the data itself that created the change. This is important. What facilitated new insights was the viewpoint that the data provided. When you stand in a different place, you see things from a different perspective. Which raises at least the possibility of having an original thought.

What the West Coast Boys studied was an abstraction, a model, freed from the pesky nuances of real life data. To the Establishment, the East Coast Boys, most of these ideas came through as nonsense from heretics.

East Coast Boys believed they knew the truth, West Coast Boys saw nothing but superstition.

East Coast Boys believed in loyalty, West Coast Boys believed in proofs.

East Coast Boys derived knowledge from emersion, West Coast Boys learned from abstraction.

East Coast Boys believed in tradition, West Coast Boys believed in change.

East Coast Boys believed the old ways were best, West Coast Boys looked for ways to improve.

East Coast Boys believed the market was there to make money in. West Coast Boys viewed it as a means of solving a problem.

East Coast Boys intuited, West Coast Boys computed. East Coast Boys believed risk was something you felt

in your gut, West Coast Boys measured and manipulated risk.

East Coast Boys trusted personal relationships, West Coast Boys trusted mathematical relationships.

East Coast Boys made good dinner companions, West Coast Boys. . .well “Do go on about modern financial theory, but with somebody else.”

East Coast Boys came from the heart, West Coast Boys from the head.

East Coast Boys smoked cigarettes, West Coast Boys smoked pipes.

East Coast Boys believed in technical analysis, West Coast Boys believed in random walk.

that 1970 Q-Group Speech on “Efficient Market Portfolio?” I described the construction and trading of the first index fund, which changed the face of money management big time.

Sitting in the audience was the CIO of Bank of America’s Trust Department - a very powerful man of the time. After the speech he called my boss’s boss’s boss’s boss and told him that he had a heretic - not a prophet - on the loose, spouting nonsense, and that they had better muzzle me. Of course, Wells Fargo was delighted, for we were becoming the first West Coast Boys.

And so the heresy spread. The man I worked for was Mac McQuown, who, in a more religious age, would have been voted most likely to be burned at the stake. Mac’s idea was to decompose the investment process and build tools to better process the relevant knowledge, ending up with a completely disciplined approach.

Oh, we were cocky all right. We had our gods, Fama, Markowitz and Sharpe. Down with the old gods of John B. Williams and Ben Graham, to say nothing of Dow and Elliot.

We toiled away for years, exploring every blind alley we could find, spending glorious amounts of bank money. Almost everything we did was a failure, yet at the end we created the index fund, the one of the most successful new investment idea of all time in terms of money managed. We lost every battle, but we won the war. Unfortunately, there were no survivors.

But the genie was out of the bottle. Soon we weren’t the only West Coast Boys. Many fine academics sprouted at Chicago and Stanford and Berkeley. Many consulting firms sprung up on the West Coast: Russell, Callan, Wilshire, BARRA, Fong, Plexus Group. . .the list goes on and on.

Out of this new passion came monumental changes to financial markets: Index funds, options, risk control/hedging, basket trading, passive trading, DOT, spiders, crossing, and implementation analysis,

1 my current favorite.

1. Stock picking came first: maximize returns. Dis- passionate modeling would discover important relationships which would lead to superior performance. Still a major tool.

2. Then came risk and diversification. This was the big academic breakthrough. Once the portfolio took the ascendancy, individual stocks became a vehicle for capturing risk-adjusted returns. Still a major tool.

3. Third came an understanding of stock relationships that would improve on quickly outdated

Neither had much use for the other. Remember ) portfolio strategies. This gave rise to option based


strategies, including the ill-fated portfolio insurance. Nonetheless, still a major tool.

4. Still, few are able to beat the market. Why? There is a fourth key area: an understanding of implementation costs and market mechanics. Here’s where the rubber meets the road. This is the area we at Plexus specialize in. We believe this is the key insight of the 90’s.

I want to share some startling numbers with you from our work.

A year ago we aggregated 64,000 trade desk orders from 22 institutions. This data was an eye- opener. Most of transaction costs were NOT in impact and commissions, but in timing and opportunity - trades delayed or abandoned. This was not the standard wisdom. When we looked at the trades the managers were trying to implement, we saw the problem was one of liquidity, not impact.

Let me give you some key numbers about institutional trading:

050% of assets are institutional (slide) 040% of institutional orders exceed one day’s

volume--2/3 exceed 5’~ day’s volume

The cost of this process typically leaks 2-3% roundtrip.

l Institutions tend to act together, and they swamp market liquidity

*These orders are not simply presented to the market. A whole orchestration takes place to get these monsters completed.

*Orders carry over from one day to the next. In the process, they generate autoserial correlation, i.e. trends. Not random walk.

*They remain active until the profitability is eked out of them. Even then, there’s more potential volume behind them. One out of four shares ordered is never executed.

Bottom line: with the increase in institutional management, we can expect to see some amplified liquidity shortages, with the attendant serial correlation and volatility.

Prof Larry Harris from USC identified the following types of traders:

Value traders Informed traders Market makers Upstairs traders Parasitic traders Electronic proprietary traders Pure arbitragers Statistical arbitragers Technical traders Bluffers Uninformed investors Hedgers Gamblers Fledglings Cross-subsidizers Inefficient traders

And everybody’s favorite trading partners, the Pseudo-informed traders and Victimized traders.

Larry’s interest was who wins from whom? In- stitutionalization means we probably have more sav- vy traders and fewer fledglings, victims, and other natural prey.

Are your technical models adequate to predict the constantly mutating reactions of all these traders? Technical trading is difficult, indeed!

The market is like a flu virus - as soon as you think you have it pegged, it mutates into something else. Yet this continuing pressure from one buyer is not the basic grist of technical analysis. The core of technical analysis seems to be that there are undy- ing truths to which the data speak.

Now, let me be right up front: I have a healthy skepticism toward technical trading. I read Nobel laureate Paul Samuelson’s paper entitled “Proof That Properly Anticipated Prices Fluctuate Randomly.” So when I hear statements about “. . . two Fibonacci retracement points overlapping” frankly, it sounds to me more like astrology than science.

When my friend David Lienweber first came in- to this industry he bought Edwards and Magee’s thick book of technical analysis signals. David accepted them, but he didn’t have a religious fervor about them. But David’s computer couldn’t generate predictive value for him. Some other skill was required to interpret the signal. So technical analysis failed a basic criteria of science. It was not replicable. An econobot cannot do technical trading.

Yet there is evidence in our own work of factors that will allow some kinds of technical trading may work, at least over short-term horizons. Certainly, it has been my life experience that there are always new things to be learned.

I suspect the market is no longer one of sheep, but one of increasingly frequent surprises!

One of the West Coast Boys’s problems is that their’s is a world of theory. When they move into practice they enter the East Coast Boys’s harsh reality. There can be econobot investing, but there is no such thing as econobot trading!

Our core business at Plexus focusses on “implementation shortfall:” the inability of either East Coast Boys or West Coast Boys theorists to achieve returns even close to their expectations. The problem: theory occurs in a vacuum; and when it comes to performance, vacuums suck, pun intended! So Plexus has come up with a way to bring together the West Coast Boys theories and the East Coast Boys realities into something that fits the market.

Because I myself come from a West Coast Boys world, I’ve had to open my mind to a lot of new ideas. Turns out some of these new-to-me ideas are as old

MTA JOURNAL / WINTER 1993 - SPRING 1994 13

the hills! Better put, the only new things in the world are the history you never bothered to learn.

One of our clients sent me a copy of the NEW MARKET WIZARDS (he’s featured in the book.) The interviews in this book come closer to the core of technical trading. Nearby lie some ‘truths’ of technical trading that are valuable to any manager or trader, West Coast Boy or East Coast Boy.

We see the primary challenge in institutional portfolio management as the ability to get trades done. In an effort to cut costs, most buy side traders have become West Coast Boy believers, and are often frozen by the workings of the East Coast Boys’ markets. They would do well to learn a few things from the New Market Wizards:

can be immensely helpful here. Record your decision signals and track what happen when you act or fail to act on that decision.

Quality, that is consistent success, is not a random occurrence. It is always the result of deep understanding and skillful execution. A process has quality when all the links of the decision chain carry the requisite weight.

1. Discipline! Don’t worry about the market acting in some predetermined way. Instead focus on knowing exactly what to do when the market acts in a certain way. It doesn’t care about any rules or individual; it’s going to do whatever it does. Don’t fight it, learn how to ride it.

2. Adaptivity! Don’t ever stop looking/learning/lis- tening for something better. Technical trading goes back to the 1800’s, and it is still evolving. As Lewis Carroll’s Red Queen said: “In this country, one must run as fast as they can simply to stay in one place.” The same holds for all of investment science; today’s rocket science is tomorrow’s fossilized rocks.

5. Beware the curse of success. East Coast Boys and the West Coast Boys have enormous egos; the West Coast Boys is based on academic excellence, the East Coast Boys is based on a grand old tradition. There’s wisdom in both approaches, but nothing works forever.

The New Market Wizards are an eclectic group that mix both cultures; statisticians and quants with the souls of gunslingers.

A successful gunslinger is cold and dispassionate, constantly looking for the edge that will ensure his being around for another day. The gunslinger needs to resist the subtle notion that his ‘skills’ will always win out. Once that happens, he’s dead.

6. Finally, beware the consequence of success: size. A man’s got to know his limits, As Eastwood says. What worked well with $100 million may not work with a billion or ten billion. Transaction cost, more specifically liquidity failure, is what killed Portfolio Insurance. This is our world at Plexus, and believe me, its a lot uglier and more dangerous than taught in school. Except, of course, the school of hard knocks.

3. Breadth of understanding. Be careful there’s not something more in your strategy than what immediately meets the eye. Beware, for example of the unintended short straddle which pays off a little if the price moves a little, but loses a lot in big price moves. It’s surprisingly easy to build a short straddle without knowing it. It may work for a long time, only to hit an ultimate disaster.

The ill-fated Portfolio Insurance idea was the classic example. Ride the gains, but back off when your hand gets cold. This was the PI’s problem: When prices plunged, there was no liquidity at any price. There was no one for the Econobot to sell to. Great in theory, lousy in practice. West Coast Boys are not infallible.

4. Depth of understanding. Make sure that you know when your strategy works and when it does not. Even more fundamental, make sure you know that it works at all, and that you have not deluded your- self into the basketball player’s belief in streaks.

I heard today statements such as “It seems to have worked well for us.” Consider how much more power is in a statement like “Of 999 signals, 666 added value.”

Implementation analysis, the Plexus approach,

To close, we come from different backgrounds, you and I. Yet we head toward common ground.

We all believe in Markowitz, at least as much as we ever believed in Ben Graham. We all use com- puters and databases and communications as soon as we can get our hands on them.

The East Coast Boys realizes the West Coast Boys aren’t heretics, and the West Coast Boys see the wisdom in the old intuition.

We are all heading toward the same center, but the center isn’t where it was in 1960. It’s moved off the East Coast, across the Hudson. Maybe Chicago. Maybe somewhere near San Antonio.

Who knows, maybe the technical traders are the unacknowledged original heretics - sort of the pre- historic West Coast Boys. Indeed, We are all West Coast Boys now.

Wayne Wagner is a founding Partner of Plexus Group, a Santa Monica based Znuestment Aduisor. He is author and editor ofThe Complete Guide to Securities Trans- actions: Improving Performance and Reducing Costs; co- author, with Jack Treynor, ofthe Chapter on Execution in the CFA Guide, Managing Investment Portfolios. He writes and speaks frequently on trading and inuestment subjects.


A Comparison of Three Approaches to Price Channels by J.R. Davis

Abstract The author tested three channel break-out trading systems across at least one commodity from each of the major commodity groups. The results were tested using the Student’s T-test to determine whether the differences in profitability was systematic or random. The T-test showed two systems to be systematically distinguishable from each other, while neither was systematically distinguishable from the third system.

Trend following techniques include a variety of “break-out” approaches. A “break-out” occurs when prices or volatility or some other measure of market activity exceeds certain boundary conditions. Price channels establish the boundary conditions by measuring price over a period of time. One method is to simply track the n day high and the n day low. Donchian presented a subset of this approach with his 4-week rule.’ Donchian’s system is always in the market. When prices exceed the highest high of the prior four calendar weeks, go long. Stay long until the low is lower than the lowest low of the prior four calendar weeks, then liquidate longs and go short. The initial position can be long or short.

Richard Dennis used a variation of this approach with his “Turtle” traders.2 The basic approach is to go long if prices exceed the 40-day high, or short if prices decline below the 40-day low. Once in a trade, exit a long position if prices decline below the ten- day low, and close a short position if prices rise above the ten-day high. Dennis also is said to have used a variety of money management techniques.

In this study, I used the basic 40-day entry, ten- day exit system. Only one contract was traded, and no other money management techniques were used.

W.D. Gann Swing Charts promulgate another approach to setting boundary conditions for break- out trades. Three or more consecutive higher highs, followed by three or more consecutive lower lows establishes a “swing high.” Three or more consecutive lower lows followed by three or more consecutive higher highs establishes a “swing low.” When prices exceed a swing high, a long position is instituted, and any shorts are liquidated. When

prices decline below a swing low, any longs are liquidated and a short position is taken. I used this technique as the second of the channel break-out systems I tested.

The third variation I tested is related to Gann’s Swing approach! Set initial high and low count to zero. A new high increments the high count, and a new low increments the low count. When either counter reaches five, both counters are reset to zero. A high count of five followed by a low count of five establishes a swing high at the highest high of the last high count of five. Reversing the procedure establishes a swing low. When prices exceed a swing high, a long position is instituted, and any shorts are liquidated. When prices decline below a swing low, any longs are liquidated and a short position is taken.

Other approaches to price channel break-out systems have been investigated. One method uses closing prices rather than intraday highs and lows to define the channels! Another uses two sets of channels, like the Dennis system, but defines the exit channel differently than does Dennis? A third uses a combination of a price channel and yester- day’s high and low price.6

The systems were tested across my entire data range, ending first quarter of 1990. The time frame tested represents as little as six years for crude oil or eight years for S&P 500 contracts, to 22 years each for wheat and soybeans. Commodities which trade only four contracts yearly had each contract tested. Contracts which trade more than four contracts had every other contract tested. 135 days were used in each case: from the 150th day prior to expiration to the 15th day prior to expiration. This approach yielded considerable overlap, and both profits and losses will be overstated, roughly by a factor of two- and-a half. However, the relative profitability of the three methods is not affected, inasmuch as all traded the same data.

No allowance was made for slippage or commissions. However, the data set shows the number of trades and the average profits or losses per trade, so appropriate adjustments can be made. Trades were

MTA JOURNAL /WINTER 1993 SPRING 1994 15

r

TABLE A: COMPARISON OF SYSTEM TEST RESULTS

CHANNEL (DENNIS) PROFITS WORST SINGLE RESULT FCT NUMBER TRADW PROFIT/

COMh4ODITY LONG SHORT PER TRADE LONG SHORT YEARS WINS TFUDES YFAR YEAR

COFFEE 5267,084 9119,012 51,893 ($4,248) ($7,380) 16 50% 204 12.75 $24,131

DMARKS $83,975 $118,188 5818 (92,513) ($2,475) 14 59% 247 17.64 $14,440

BONDS 576.050 $63,180 $503 ($3,220) ($4,130) 12 42% 277 23.08 $11,603

SUGAR $121,632 $67.782 $746 ($5,208) ($7,302) 18 48% 254 14.11 3 10,523

GOLD $50.850 $91,340 5624 (55,310) ($3.9001 15 49% 228 15.20 $9,479

SOYBEANS s 153,940 $35,385 $516 (57,000) ($lO,DOO) 22 43% 367 16.68 $8,606

TBILLS $66,500 $48,000 5413 ($4,950) ($4,825) 14 46% 277 19.79 $8,179

CRUDE $16,950 $31,580 5426 (51,550) ($1,510) 6 54% 114 19.00 98,088

CO-I-TON $77,370 $23,000 5429 ($4,300) (54,035) 13 47% 234 18.00 $7,721

CORN $31,100 $24,740 5235 (51,270) ($1.260) 20 50% 238 11.90 $2,792

LIVE COWS $36,200 $2,500 5120 (52,140) (51,540) 20 41% 323 16.15 51.935

WHEAT $18,185 $23,940 9152 (52,350) (92,370) 22 39% 277 12.59 $1,915

S&P (57,600) 57,075 ($3) ($9.500) ($7,750) 8 31% 187 23.38 ($66)

GANN SWING PROFITS WORST SINGLE RESULT F’CT NUMEGR TRADW PROFlT/

COMMODITY LONG SHORT PER TRADE LONG SHORT YEARS WINS TRADES YEAR COFFEE $96,480 $17,438 $1,280 (57,380) ($5,490) 16 44% a9 5.56 sgY20

SOYBEANS $69,800 $56,485 $784 ($11,250) ($12,500) 22 50% 161 7.32 55.740

CRUDE $19,700 S 14,060 $767 ($1,510) (52,050) 6 52% 44 7.33 $5,627

TBILLS $50,425 $25,525 5623 (56,725) ($4,525) 14 45% 122 8.71 $5,425

S&P $72.190 (530,225) 5545 ($17,200) ($14.325) 8 40% 77 9.63 $5.246

GOLD $29,510 $36,900 $772 ($4,350) ($3,250) 15 42% 86 5.73 54,427

SUGAR $31,450 $41,518 $676 ($6,720) ($2,565) 18 42% 108 6.00 54,054

DMARKS $26,075 $23,388 $556 ($3,413) ($3,663) 14 56% 89 6.36 $3,533

COTTON $32,570 810,555 $396 (84,610) ($4.900) 13 46% 109 8.38 53,317

CORN $21,080 516,950 $377 ($2,120) ($990) 20 52% 101 5.05 $1,902

LIVE COWS $36,116 $484 $263 (S2.400) ($1,952) 20 42% 139 6.95 $1,830

BONDS $11,660 $2,780 $129 N8‘840) (87,400) 12 38% 112 9.33 $1,203

WHEAT $9,780 620,895) ($82) ($2,680) ($3,280) 22 33% 136 6.18 ($505)

5-COUNT

PROFITS WORST SINGLE RESULT rc~ NUMBER nbmw Prom/

COMMODITY LONG Si iORT PER TRADE LONG SHORT YEAR3 WINS TRADEis YEAR YEAR

BONDS 9235,610 341,110 51,821 ($5,380) ($5.9401 12 45% 152 12.67 $23,060

DMARKS $69,188 $82.475 5766 ($3,063) ($2,788) 14 56% 198 14.14 $10,833

COFFEE $114,192 $40,493 5905 ($8,406) ($14,058) 16 40% 171 10.69 99,668

TBILLS $66.800 534.325 $508 ($4,750) (94,225) 14 47% 199 14.21 57.223

SOYBEANS $90,765 549,980 $510 ($11,250) ($12,500) 22 45% 276 12.55 $6,398

GOLD $33,690 $50,960 $498 ($6,200) ($5,300) 15 42% 170 11.33 $5,643

CRUDE 99,740 $18,150 $306 (52,510) ($2,200) 6 44% 91 15.17 $4,648

COlTON $39,630 $11,900 $310 ($4,965) ($4,175) 13 42% 166 12.77 $3,964

SUGAR $19,141 $34.238 $281 ($12,432) ($6,754) 18 39% 190 10.56 $2,966

CORN $19,675 $20,490 $214 ($2,370) ($1,350) 20 45% 188 9.40 52,008

LIVE COWS 926,032 51,456 $123 (52,380) (52,040) 20 40% 224 11.20 $1,374

WHEAT $13,555 $10,010 $106 (53,650) ($2,730) 22 38% 223 10.14 $1,071

S&P 930,240 (994,100) ($440) ($18,575) ($9,575) 8 33% 145 18.13 (57,983)


TABLE B Commodity by commodity comparison of profitability. Differences were or were not significant, as marked.

PROFIT PROFIT VS GANN PROFIT vs sx!ouNT COMMoDrry CHANNEL CANN AT95%? 5COUNT AT 95% 7

COPFEB s24.131 7.120 Y $9.668 Y

IMARKS s14.440 3.533 Y $10.833 N

BONDS 511.603 I.203 Y fu.060 Y

SUGAR $10.523 4.054 Y t2= Y

GoLn 59.479 4.427 Y 55.643 Y

soYBsANs S8.606 5.740 N S6.398 N

TBILLS S8.179 5.425 N f7.223 N

CRUDE $8.088 3.627 N t4.w N

CO’ITON 57,721 3.317 Y $3.964 N

CORN S.-J= I.902 N SZW8 N

LIVE cows $1,935 I.830 N $1.374 N

WHEAT $1.915 WV N s1.071 N

s&P (566) 3.246 Y (S7.983) Y

COMMODITY

BONDS

DMARKS

coFFBB

TBLLLS

SOYBBANS

GOLD

CRUDE

COlTON

SUGAR

CORN

LlvBcOws

WHEAT

s&P

PROPIT

5 COUNT

82%~

$10.833

s9.668

57.223

56.398

55,643

w.698

$3,964

82565

SW38 51.374

SI.MI

67.983)

ScouNr PROFIT VS GANN

CANN AT 95% 7

Sl.203 Y

$3.533 Y

s7.120 N

55,423 N

55,740 N

S4.427 N

35.627 N

$3.317 N

$4.054 N

s1.902 N

$1.830 N

(SW N

SS.246 Y

considered to have occurred at the break-out boun- rank-order testing. Lukac, supra, found the data to dary unless the opening price was beyond the break- be normally distributed; therefore, a two tailed T- out price, in which case the opening price was used. test was applied to the total profit sequence of each Results may have been distorted because the pro- method. The difference in average profits between gramming allowed trading on lock-limit days, which the channel with channel stop (Dennis) model and is improbable. The effect between systems and the the Gann Swing model was statistically significant overall effect on profitability is likely to be negligi- at the 95% confidence level. Differences between the ble. The results are summarized in the Table A. Gann Swing and the 5count models were not Table A is organized by descending profitability both statistically significant, nor were the differences be- to allow the reader to see which combination of tween the Channel (Dennis) and 5count models. In system and commodity worked best and to allow for other words, the difference in profits between the


Dennis method and Gann method is unlikely to be random. The difference in profits between the 5count method and the other two could well be a chance occurrence.

I tested the results on a commodity-by-commodity basis. Those results, Table B, are in accord with the test of the overall results. Here again, differences at the 95% confidence level (labeled “Y”) are almost certainly systematic. Differences which have lesser confidence levels (labeled “N”) could easily be random or chance differences.

Finally, I tested the number of trades per year. The Gann Swing method was distinguishable from the 5-Count method at the 95% level of certainty, and from the Channel (Dennis) method at the 99% level of certainty. The Channel (Dennis) method was distinguishable from the 5-Count method at the 90% level of certainty. This means that the difference in the number of trades produced by the Gann Swing method and the other two was systematic rather than a chance occurrence. The difference between the 5-Count method and Channel (Dennis) method is likely to be systematic, but does not meet the usual statistical standard (95%) to be declared “systematic”

The Gann Swing method results are distinguishable from the Channel (Dennis) method in all respects tested. This suggests that either Swing trading is significantly different from Channel trading or that being in the market continuously produces significantly different results from being in the market only part of the time, or both. The Channel (Dennis) method and the 5Count method are not distinguishable statistically. In Table A, the results are sorted by profits per year, from highest to lowest. With the exception of Sugar, grouping commodities by sets of three or four produces identical sets in the Channel (Dennis) and 5-Count methods, a further indication of their similarity. There is no such similarity between the Gann Swing groupings and the other two methods.

Differences between the approaches may result in part from the way each approach sets its exit point. In both Gann and 5-Count, a short term reversal is needed to establish an exit (stop and reverse> price. Strong trends are often accompanied by price action which is nearly an exponential curve. Prices may move 30% or more of the total move without a correction. Such price action makes the Gann and 5-Count approaches vulnerable to giving back an excessively large fraction of their gains. The Dennis methodology avoids this problem to a fair degree by allowing the stop loss price to move up as quickly

either the Gann or the 5Count. Visual examination of the “Winning %” sug-

gests that success ratings of under 40% are likely to yield unprofitable results. Except for the 5count sugar results, the under 40% group produced profits per trade of less than $200. Slippage and commissions would reduce the profit yet further, making both safety and profitability questionable. One inter- esting question is whether other approaches would yield average profits of more than $200 per trade but have fewer than 40% winning trades.

Those commodities which consistently showed better results long than short, or vice-versa, should be examined for persistence of long term trends. Live cattle would be a good candidate to examine using a “buy and hold” strategy. Commodities such as sugar which show mixed results might be traded significantly more profitably by using some sort of mixed system algorithms. Finally, items such as wheat probably should be traded only by some other approach.

REFERENCES

1. Commodity Trading Systems and Methods, PJ. Kaufman, 1978, Wiley.

2. “Turtle Madness”, Gibbons Burke, Futures, January, 1993.

3. J.R. Davis, “Swing Charts,” Technical Analysis of Stocks and Commodities, 1991.

4. Technical Analysis in Commodities, PJ. Kaufman, Wiley, 1980.

5. “Similarity of Computer Guided Technical Trading Systems,” Lukac et al, Journal of Futures Markets, February, 1988, repub- lished by Traders Press as A Comparison of Twelve Technical Trading Systems.

6. Lukac, ibid.

BIBLIOGRAPHY

Burke, Gibbons, “Turtle Madness,” Futures, January, 1993.

Davis, J.R., “Swing Charts:’ Technical Analysis of Stocks and Commodities, August, 1991.

Gann, W.D., Hou! to Make Profits in Commodities, Lambert-Gann Publishing, 1978.

Gann, W.D., The U?D. Gann Commodity Trading Course, Lambert- Gann Publishing, 1978.

Kaufman, F?J., Technical Analysis in Commodities, John Wiley & Sons, New York, 1980.

Kaufman, P.J., Commodity Trading Systems and Methods, John Wiley & Sons, New York, 1978.

Lukac, Louis P, Brorsen, B. Wade, and Irwin, Scot H., “Similarity of Computer Guided Technical Trading Systems,” Journal of Futures Markets, Vol. 8, No. 1, February, 1988, reprinted as A Comparison of Twelve Technical Trading Systems, Traders Press, Inc., 1992.

JR. Davis is Senior Vice President, Senior Technical Analyst at Golum Investors, Inc. He is currently editor of Stocks & Investment and former editor of MarketLife Recommendations. He has also published articles in several publications, including; Technical Analysis of

as the ten day lagged lows. Whipsaw losses are prob- Stocks and Commodities, MarketLife Recommenda-

ably not the cause of weaker performance inasmuch tions, and MTA Newsletter. He also wrote text and soft-

as the Dennis approach trades more often than ware for the trading primer, First Profits.


Enhanced Candlepower by Theodore E. Loud, C.F.F?

Introduction Much has been written in the past several years about the predictive value of the Japanese candlestick method, a price charting technique first used in the 1600s for trading rice futures and one of the earliest attempts to identify and interpret patterns in the operation of a financial market. The candlestick method provides a graphic depiction of the relationship between a market high and low for a particular time period with its open and close, producing an image that resembles a candlestick. Price movements produce different kinds of candlestick shapes, and devotees look to patterns of these shapes to confirm market trends and make reversal predictions.

The validity of the candlestick method remains,

appropriately to produce a whole new series of charts. My goal is develop a more comprehensive method that might be more useful in predicting the direction and termination of a sustained stock market trend. By being able to manipulate numbers, it is possible to depart from the limitations imposed by the visual interpretation of candlestick patterns. I have chosen an uptrend for my analysis, which is a time series for the Dow Jones Industrial Average for 113 weeks beginning l/1/91 and ending 2126193. In order for the reader to understand the refinements I propose, I shall begin with a brief description of the Japanese candlestick method and analysis of its value as set forth by other market technicians.

however, open to question. The general consensus of The Japanese Candlestick Method those who have looked into the subject is cautious and Its Value support. It seems safe to conclude that a pattern of During medieval times in Japan, rice was not candlesticks should not in itself be used to predict merely a staple, but a medium of exchange. Samurai stock market reversals, but may be valuable in con- warriors were paid in rice and peasants were taxed firming underlying trends discerned using other in rice, and so determining-and anticipating-the methods. value of rice was crucial to merchants as well as

In an effort to provide a stronger, more com- government leaders. In the 1600s the Japanese prehensive impression of market activity, the tradi- economy had become complicated enough that for- tional candlestick method was combined with ward trading in rice had become commonplace. another technique. In a December 1990 article in Candlestick charts, and the patterns they produced, Stocks & Commodities magazine, Greg Morris’ sug- were developed to help traders predict price move- gested that the candlestick method could be comments. Crucial to making market forecasts was their bined with Richard Arms’ Equivolume techniqueZ belief that the relationship of the open and close had to provide a more complete picture of trading activity. overwhelming significance. (Arms added volume to price range in his charts to The candlestick is composed of several parts. The create rectangles of various shapes and sizes, but real body is a rectangle bounded on the top and bot- with no regard for the open or close price.) Morris tom by the market open and close. The real body is added the dimension of volume to the body of black when the close is lower than the open, white the candlestick and called his new technique “Can- when the close is higher than the open. Thin lines, dlePower charting.” called shadows, extend from the top and bottom of

In this article, I attempt to take Morris’ idea the real body, indicating the market high and low. a step further by supplementing the pattern-iden- The relationship between the real body (the open titication that remains the essence of the Candle- and close) and the shadows (high and low) give the Power method with well-known statistical tech- candlesticks their characteristic shapes. A long black niques. In order to accomplish this, the numerical body indicates that prices moved within a large values for price change and volume that underlie range and the market opened near its high and Candlepower charting are segregated and combined closed near its low. A long white body is the reverse


Entire Time Series Plotted Using Traditional Candlestick Method

4ocm t First Ouartllr I sacond aualtilr I Third Ouariilr I Fourth Ouatiila

3900 3900

3800 3800

3700 3700

3600 3600

3500 3500

3400 3400 I I c

Entire Time Series Plotted Using Arms Equivolume Method

. . . . . - . . - - . - . - - . . . . . - . . . - - - 1 . . - - - - - - - . - . - _.._..._..____.___ .___.. . - ._._.__..___-.-- _ - -_ . ._ . . . __.

1991 1992

FIGURE 16 SWRCE TEL.ADVtSDRS

20 MTA JOURNAL /WINTER 1993 . SPRING 1994

Entire Time Series Plotted Using Morris Volume Addition Method

1991

FlGURElC

WUnin( Period

Period-

Dark Cloud Cover

I , Doji

8 Engulfing Pattern

b Piercing Pattern

$ Spinning Tops

L+tltttl++t+l++tl+t+tlt+t~ttlt.t I,*.+ ltt+ltttlttt 1, l +tt . ,.--..--..--...-- .___.--__.--_.-__..--__------. . . _ _ . ___ 1992

SOURCETEL.ADVISORS

of a long black body. The market opens near its low and closes near its high.

A second characteristic shape is the spinning top. In a spinning top the real body is relatively small because the market open and close are relatively close together. Spinning tops are important in conjunction with other shapes. The third important shape is called a doji line. In doji lines, the open and close are identical, and the real body is flattened into a simple horizontal line. Quite often both shapes are found at market tops and bottoms. They are very important, for purposes of this report, because they always represent caution.

Devotees of candlestick charting maintain that the patterns these shapes make over a period of time provide material for analysis. They may signify that a trend will reverse itself, continue, or is un- stable. For instance, an engulfing pattern is formed when the real body in a given time period brackets the real body of the preceding time period. It can signal a major reversal. Piercing lines and dark cloud c0ueF.s also signal reversals. In a piercing line,

. I

a bullish pattern, a white body follows a long black body. The white body opens lower than the close on the previous day, but closes above its midpoint, With a dark cloud cover, the black body follows a long white body, and the trend, of course, is bearish (see Figure IA for an example of traditional candlestick charting).

The Japanese candlestick method has continued to be refined over the years, and has recently been examined by several analysts in the United States. Larry Williams3 has tested several well-known patterns including the dark cloud cover and found that some worked better than others depending on the security or commodity being traded. Gary Wagner and Bradley Matheny4 swear by an approach they call the three C’s of candlesticks (confirmation, con- tinuation, and confluence), in which patterns are used to confirm an outcome that has been predicted independently. But the consensus seems to be that candlesticks on their own should not be relied on for anything other than short-term trading. This is the conclusion that Steve Nison: who produced a


series of influential articles and publications about candlesticks, also endorses.

In an attempt to improve upon the candlestick technique, Greg Morris combines it with Equivol- ume charting. (See Figure 1B for an example of Equivolume charting applied to the time series in Figure 1A). The resulting hybrid technique allows Morris to create powerful charts that combine price range and trading volume over the course of a trading period (see Figure 10. Morris uses his Candlepower technique to examine the September 1990 Treasury bond market and the 1990 New York Composite Index and finds that several candlestick patterns, such as the engulfing pattern, still seem predictive, but admits that his conclusions must be considered tentative. Morris does not attempt to exploit the unique possibilities opened by the Candlepower method to establish predictive patterns of its own. Instead, he relies on established candlestick interpretations. Nor does he move beyond the constraints of pattern recognition as will be under- taken in this report.

Finding the Proper Place for Candlepower Charting

Before we can begin to appreciate the significance of patterns formed with Candlepower Chart- ing, we must attempt to reach some conclusions about the conditions under which Candlepower charting and the patterns it produces have the most validity for the market technician. For the most part, the data which is used to produce Candlepower patterns, price change and volume, can help us approach an answer to this question.

As already mentioned, the time series chosen for this study is the Dow Jones Industrial Average for a period of 113 weeks beginning l/1/91 and ending 2126193. The basic time period I used was a week and, for simplicity’s sake, each volume unit equaled 25 million shares (unless there was holiday, when

the volume unit becomes 20 million). Changes in price during the week were considered positive or negative depending on whether the close was above or below the opening price.

The application of statistical tests to the Candle- Power patterns created during this period, simply considering color (which indicates direction of price movement), reveals that the statistical significance of being able to predict the color of a candle on the basis of candles that preceded it is limited. A chi- squared test was performed to see if it was possible to predict the color of a subsequent candle knowing the color of the previous one, two, and three rectangles (see Table 1 and Figure4 for the calculations that support this table). Probability theory states that the deviation from a normal or symmetrical distribution that can be measured determines the confidence level of any prediction. (For our purposes, a confidence level of. P=.OO5 was necessary in order to make valid predictions.) Otherwise, such predictions will turn out in the long run to be no better than flipping a coin! The test registers statistical significance only in the case of a trigram, where the colors of two previous candles were known (we shall take a closer look at these results in the next section), and that there seems to be no significance to dig-rams of two candles and tetragrams of four candles.

Therefore, taking our analysis of the dependency and stationarity questions much farther does not seem to be a productive exercise, although the histograms discussed in the following section on momen- turn would suggest that this time series is not sta- tionary (see Figures 4A-4D). Also, a comprehensive autocorrelation test of this time series, supervised by Don Richards, a professor of mathematics and statistics at the University of Virginia, concluded that there was no level of dependency whatsoever between a predicted rectangle color and any number of previously known colors.’

Clearly, then, in face of a firmly established

Candies Detrended For Entire Time Series lMi91 - 2l26/93

I.. . , . ,, . .,,“Y”“.,I.*.“.l...~...~l...~“...”..”.~.l~.. “.. -..‘.--...-“...1..““““-“.““.----.~- .I.. -.-1.1-m--*SW- Volume/Time

FIGURE 1D SWRCETELKWISORS

22 MTA JOURNAL/WINTER 1993 - SPRING 1994

:): >>;: > :.. 2. ./.... :;., ;,:.: :.,.: :...:j .::..+ . . .:. :.j.::.j:.:;,:..::. ,.:, j ,: ., : .,, ::. .:.:. I::.: . . :.:: . . .:.. ,.. :. . . .:-::...:.: ._..,. . .., . . .,:.:.). ,.,,,.

~~~~~~f:;~“~~~~~~y:;ot-f~~t:.~~sults:Jow:VEirlousl,,~hservationi,C.ombi~~]ons-~~~

Critical Values

Type Observation

Digram (1,>2) Trig ram (1,2,>3)

Tetragram (1,2,3,>4)

Actual Result P=.O5 P=.OO5

1.15 7.81 12.84 36.70 14.07 20.28

10.66 25.00 32.80

SOUrm: MaVlemaGcs of Tech&U&s&: Clifford J. Sherq

uptrend, such as we find in the time series under study, the use of Candlepower charting to predict outcomes is clearly limited. In such trends, most patterns are not reliable in predicting a trend reversal. However, as excesses build up or there is a critical loss of momentum, they may become more significant. As a corollary, Candlepower patterns may also become more important as age becomes a factor.

Self Correction It has long been said that “bull markets climb

a wall of worry.” This is another way of saying that their longevity is tied to the degree of caution that persists throughout their life cycle. Markets must self-correct in order to rid themselves of speculative pressure, but these corrections must not undermine the momentum that sustains the uptrend. The first question that Candlepower charting can help us determine is the ability of the market during this period to self-correct.

Certainly, a cursory look at a detrended candlestick chart for the period (Figure 1D) reveals a market with the capacity for self-correction. More than 40 percent of the time, the market corrected itself. Moreover, those weeks in which the market closed down are spread evenly over the entire period under study. This seems to indicate that the market was generating its own liquidity internally to a meaningful degree.

Another way of corroborating this fact is to look at the doji and spinning top patterns. Both of these symbols indicate caution, because there is a balance of trading activity between bulls and bears. Trans- lating dojis and spinning tops into numerical terms, we define them as shapes in which the product of the price and volume figures that determine the size of their real bodies is k41.34. We have chosen this number because there is a gap on the plus and on the minus side between it and the next largest integer, providing us a crisp border for identification. (See Exhibit IA). Our analysis shows that these symbols are evenly distributed throughout the last three

quartiles of this time series, revealing the caution that characterizes a self-correcting market.

A third method of assessing the self-correcting tendency of the time series under analysis is to refer back to our analysis of the trigram pattern, which we introduced in the preceding section. For purposes of illustration, black real bodies were given a value of 1 and white real bodies were given a value of 2. When the expected outcomes of a trigram pattern are compared with those actually observed, we find that the black bodies have predictive significance (see Table 2). The predictive value of black bodies, which indicate negative price movements, seems to support the hypothesis that this trend is self-correcting internally to a significant degree. It is apparently not developing excesses, which might be the case if the tetragram chi-squared test had statistical significance (see Table 1). If the 2,2,2 trig-ram supporting the uptrend was the cause of statistical significance, rather than the l,l,l, trigram, one could hypothesize that excesses were developing.

The ability of the market to self-correct during this time series again suggests that Candlepower patterns do not have great significance during the period under study.

Momentum Candlepower charts, by combining price change

and volume, are excellent reflections of the momentum built up in a market. Statistically manipulating the data that underlies them, however, provides an even better understanding of market dynamics. Con- sider the following hypothetical example: bears carry the day in three of the five time periods considering price alone, but when volume is added, an entirely different picture emerges (see Table 3). When the products of price change and volume are totaled, the bulls have a 21 point advantage. Market momentum is on the upside. In the short series depicted in this example, our results might seem distorted, but I believe that over time the law of large numbers eliminates any bias that is bound to appear.


(Observed -

-- (observed- Expected)Y

Trigram Observed Expected Expected Expected)2 Expected I;

l,l,l 17 4.71 12.29 151.08 32.08 32.08

1,1,2 17 14.13 2.88 8.27 0.59 32.67

121 18 23.54 -5.54 30.71 1.30 33.97

122 15 14.13 0.88 0.77 0.05 34.03

2,1,1 17 14.13 2.88 8.27 0.59 34.61

z12 17 23.54 -6.54 42.79 1.82 36.43

2,2,1 16 14.13 1.88 3.52 0.25 36.68

222 5 4.71 0.29 0.08 0.02 36.70

sag: Malhemalw Clifford J. Sher$

Time Period Price Change

1 4

2 -2

3 -7

4 5

5 -1

CumulativeTotal -1

Volume Product

5 20

3 -6

3 -21

5 25

3 3

NA +21

Source: TEL Advisors, Inc. of Wrginia

%” .k ‘-:zx::$:;::.ii; )xz::z::,:> ,, . . .+:::.~.:.:i . . . . I,.> .: L :. ,. .:...: :;::::“;i . ,: .,. . . ..: .:,:, ~ ,.:,:, :,. .,. .,. .,., ~, ,,.

~~~~~~~~.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ia~~~~~~~niiitj;irttt

1 . ..‘-?.*~&...~ . . . . . . . . ;*.;::*.$ . . . :,,.

Product Count I II III IV

More than 400 5 5 2 1

Between 200 & 399 5 0 1 4

Between 100 & 199 4 6 6 5

Between 0 & 99 2 4 7 7

Between 0 & (-99) 4 4 6 4

Between (-)100 & (-)199 3 5 2 2

Between (-)200 & (-)399 4 3 4 6

More than (-) 400 1 1 0 0

Total 28 28 28 29 Source: TEL Advisors. Inc. of Virginia (EXHIBIT 1A)


Bull Markets: Ending Years -Since 1900

-

. . . I . . .

Previous Bull Markets Current Market

SW End

sewrter19m .h?l931 omum (ywluw) hillat Ftnd xl%t-lga 10 54 78 U.UX NmsmbellOm FebruaryWC6 27 u 103 127.277. Nmomber 1907 L%mlter1R19 26 Y 100 %.19x

Sep(amber1911 &kkl912 14 73 94 26.77% Decee%r1914 NObwk.?rl916 24 54 110 la370%

Dpember1917 t4c+wlhr1919 24 68 115 69.12%

Aqsll921 March1923 20 65 105 61 54% June1924 FhN?l~1926 21 90 170 %.89X

thy1926 seplenaarl92¶ 41 150 390 lW.W%

uyl932 WI=4 a 40 110 175.00x

FIGURE semter1934 Mad1937 31 85 190 123.5s Msrch1938 2B An61946 srplember1939 19 im @Id1942 50 95 210 160 121.05% m.m%

An8 1949 JanuarvlW 43 1% 2% 6369%

se@mter1953 Apill 32 270 510 88.89x ccktarl957 h.my1960 27 410 6% %.29% oCkhl9B) NGwlterl%l 14 %a 720 24.14% h81962 F&wyl%6 45 540 IWO 85.19%

OcbbM1966 -Cl%8 27 7% 975 a.m%

May 1970 Janaq 1973 32 560 1050 90.91%

December1974 seplehfl976 22 5m 1025 79.82%

kidll9Bo A&d 1981 13 7% 1020 26.00%

krgurcl962 J-WI= 18 7% 1300 m.56* MY1984 Aup 37 cim 27% 15000?4 CCktWlW krluslW 33 1620 %25 86.73%

Anrag8: 26.8 6468x 0: 10.3 3973%

'om*trdJmalAvenp~shd

Statistical Significance Data

FIGURE 2A

settual

VurPkR 0.501753

ZRssun

02617% 0229621

slngdbm 0.3%649 obwvams 26

1 1.038910 1.036910 8.076801 O.soOSm 24 3.537~ 0.128629 FIGURE 25 4.12son 2c

- StmdudEnu t slaudtc P-nlU Law %% 4P-m

0.3OOU% 0.199238 1507768 O.lU148 a11mw2 0.711611 0 197900 0.66e4rn 2.841971 0.879700 05118co 0.341620

SWRETU-NNGORS

nmth cab VdW xChv*lr - -r193l 2350.00 - 1 -r19Sl 2 JaMr)r1991 3 Febnnyl%l 4 til991 5 &tlll%l

263366 12.07% 273639 18447. 2662.18 22.65% 2913.86 23.99% 2687.87 2289x xez7.50 2883% 29X.75 2389% 2972.50 .x49* m43.w 29.51% X116.77 28.37% 3cc4.82 27.87X

6 May 1991 7 .hm I%1 8 My 1991

13 c!ecmbfl%l 14 Jt4WPfl992 15 Feh!arfW% 16 hWill%Z 17 Ml%2 18 tiny1992 19 Am1992 20 .hiy1%2 21 AyFa(l992 22 6epaeell992 n oclobcrl992 24 Nmmber 1992 25 Dscarbar1%2 26 Jalwy1%3

2a94.68 2318% 3168.83 3484% 3223.29 3716%

3267.67 39.05% 3235.24 3767X 3359.12 42.94% 3396.88 4455X 331851 41 21% 339378 U42X 3267.61 39 05% 3271.66 3922% 3226.28 37.2% 3X6.16 4065% m1.11 4047X 33lo.m 4085%


Applying this method to our time series, the picture gained is that of a market whose underlying uptrend remains intact, although the momentum that sustains this trend appears to be waning rapidly. To some degree, this an be seen simply by observing the slight penetration of the trend line since October 1992 (see Figure 10. However, a clearer picture of this waning momentum is seen in Figure 3, which depicts a moving nine-week unweighted sum of products. This unweighted moving sum clearly shows that after the initial surge took place in early 1991, each additional surge has failed to improve upon the thrust of its major predecessor, while sinking to a new low prior to making its subsequent attempt. Even the latest upward move in January-February 1993 appears to have stalled before reaching the previous high set in December-January 1992.

Another approach to the question of momentum is to consider the product of price change and volume for the entire time series by quartiles, comparing the behavior of the product in each quartile with the behavior in the other quartiles and then to the whole (see Exhibits 3A-3D and the Addendum for another way of analyzing these same quartiles). What becomes immediately apparent is the loss of volatility from quartile to quartile (see Table 4). There is a progressive decline in the number of positive product extremes as well as negative product extremes. (This type of behavior has been analyzed by John BollingeP in his Bollinger Bands. He has concluded

that such a diminution of volatility eventually results in an explosive breakout in one direction or another.)

This diminution of volatility can also be seen in the histograms of products for each quartile (see Figures 4A-4D). Once again, they clearly show the market’s loss of momentum, although there has been some modest improvement during the last part of the fourth quartile. However, this improvement has not been sufficient to restore the momentum generated on balance during the first three quartiles (see Figure 5). In this figure, the product of price change and volume for the entire time series has been plotted cumulatively on a semi-log scale. As can be clearly seen by the penetration of the long-term uptrend line, momentum broke down at the beginning of the fourth quartile but at the end of the period was in the process of gradual restoration.

The Age of the Market The waning momentum clearly evident from our

analysis of price change and volume takes on added significance when considered in light of the age of the market. Although there should not be too high a level of confidence placed on the regression line shown in Figure 2A, which traces the course of bull markets since 1900, there is a meaningful correlation between percent price change and duration, as will be seen from the material presented below.

In Figure 2A, not only has the regression line

Nine Week Summation Index


First Quartile Product Frequency Distribution

-----------------------

U

FIC -

Second Quartile Product Frequency Distribution


Fourth Quartile Product Frequency Distribution

Mdl.. .‘I 3.l

0 05 -----------...-----I-------~------,-,~~

P”,d”Cl Value SajRiEIEt.MvM45 ---_ FIGURE 4D

been plotted and adjusted statistically, but the appears unlikely. All actual observed data points means and standard deviations for price change and beyond 37 months (one elapsed time standard devia- elapsed time are plotted as well. Superimposed on tionl are significantly above the Z/2/26/92 level of this diagram is the course of the current market up- 3370. The closest point (the bull market ending in trend since its inception, along with its own regres- 19531 would suggest a level on the Dow of 3836 in sion line appropriately extended. What is not shown, the 43rd month (July 19941. Indeed, after the mean but what must be kept in mind, is that fact that elapsed time period of 26.8 months has been there is a significant drop in confidence in the overall reached, only two of 13 uptrends fall more than one regression line after the mean elapsed time period standard deviation below the regression line (1953 for bull markets after 1900 of 26.8 months has been and 1968). According to some abbreviated work by reached. This means that the standard deviation Arthur Merrill’ in 1989, the probability that this lines are considerably wider than those shown for has occurred by chance alone is less than one in the time period beyond the mean time period. one hundred. This high level of probability suggests Although this observation suggests that the current that unless the current uptrend’s momentum im- uptrend could easily last 40 months or two standard proves, it will terminate before it reaches the age deviations below the overall regression line without of 43 months, and perhaps much sooner. a 10 percent or greater correction, such an outcome In other words, the age of the market adds

Cumulative Product For Entire Period -Actual vs. Ranked


significance to our analysis of price-change-volume product that showed market momentum waning and suggests that followers of our time series should begin to look for significance in Candlepower patterns.

Conclusion The analysis of this time series shows suggests

that the predictive value of Candlepower patterns, while limited while the upturn remains strong, should take on added meaning as the trend ages. As a result, we can tentatively assign some meaning to the patterns that emerge in Figure 1C or 1D. Here each major upward thrust (as labeled) can be easily seen in terms of follow through. Prior to the most recent one, each major upthrust of white real bodies had some follow through in that it was followed by succeeding white real bodies. The last upthrust,

value so far for volume is six for the entire series (see Exhibit lA), suggesting that a minimum weekly negative price change of $100 will be necessary in order to signal the end of this bull market.

Nonetheless, it should be evident that an important value of the Candlepower method as a forecast- ing tool in a sustained uptrend may be gained by converting the Candlepower figures into numerical values for the purpose of measuring the two key dynamic elements that promote uptrend longevity. These key elements are the ability to correct any buildup in speculative excesses swiftly and the ability to sustain a critical level of momentum. With this information in mind, combined with data on the relative age of the uptrend, an analyst may be able to make a better judgment about the application of Candlepower charting and the patterns that it creates.

which was preceded by a period of extreme caution, was immediately covered by two black real bodies that almost entirely negated its upward movement, suggesting that the trend line is weakening and possibly signaling the beginning of a reversal.

These conclusions are necessarily tentative, because the time series under examination continues to maintain an upward course. A definitive answer to the question of the proper time to apply the Candlepower method will require a comparison with other bull markets or an extension of this time series to the actual point of reversal. However, no confirmation of the patterns we might observe in this sequence should occur until we have a negative product value in excess of 600. Such a value is well below two standard deviations from the long-term mean (see Figure 6). In addition, this confirmation should occur on sharply increased volume. The greatest unit

REFERENCES

1. Greg Morris, “East Meets West: Candlepower Charting,” Technical Analysis of Stocks and Commodities, December 1990, pp. 16-20.

2. Richard Arms, Jr., “Market Timing Through Equivalence Charting,” in Volume Cycles in the Stock Market (Dow-Jones- Irwin), chap. 15.

3. Larry Williams, “Candlestick Patterns: How Reliable Are They?” Futures, July 1991, pp. 14-18.

4. Gary S. Wagner and Bradley L. Matheny, “Candlesticks as a Leading Indicator,” Technical Analysis ofStocks and Commodities, November 1992, pp. 100-6.

5. Steve Nison, Japanese Candlestick Charting Techniques (New York Institute of Finance, 19911, chaps. 10-15.

6. Clifford J. Sherry, The Mathematics of Technical Analysis Pro- bus Publishing, 1992), chaps. 2,3.

7. Tests performed during the week ended February 19, 1993.

8. John Bollinger, “Using Bollinger Bands,” Technical Analysis of Stocks and Commodities, Bonus issue, 1993, pp. 84-89.

9. Arthur Merrill, “Significance: What Is It?’ MTA Journal, Winter 1988.89, pp. 41-43.

-.


wosmav-131 3:xmos

I io ZOL 1, I ;I 9; 00 9c 00s 1008

lSB69 1169 lPPll

co L6.81 iOOOCP- 105 /0098- 116/80/11

co LZCE

C06ZfB

COLOl6 LL lLP6 LL 9L96 LZISIOI LZ SLPOI

?ZSLLOl LZ E9011

LZRZEII

?ZLESiI

rz01911

LZBCOZI

LL CSZZI

LZ 19PZI

1Z 69921

LZ PLBZI

55 8LOCI

zz BLZEI

zz CLPC,

zz B99Cl

zz 8PECI

zz CZOPI

ZZ SGIPI

zz SCCPl

iz OSPPL

zi 095Pi

~OOZOP- 09

'009LC 09 PL WC 0 3 oossc OS OS PLC 5s 00 *zc- s P ooooc 0 P ooaaz 0 P 005sz OS OOPSZ- 05 OOEZZ- OP ooaiz 09 oosiz- OS

jOOEOZ~ OP ooaoz 0 P 00 50z 0 s

/SGCOZ 05

lOOOi3~ OP

:OOSLi OS

'OOZLl OP

,OOOPi OP

'oosll- ,05

,oooii 15s

,OOROl IOP

6L 09. 1 z: F6/80/lC

;oo IL- ~ ZWZl/9C

1;; ;;: 16/1c/sa

16/81101

io0: E:- Z6/81/60 ZGIPIIEO

IOOW ~16/1Z,90

\OE os- C6/6l/ZO

100 LS- z6/S0/90

loo ac- 16/6i/to

,OOCP lW61,LO

00 zs- 'ZWIEILO

100 2s. ZWaWzO

IO0 Iv- 16/PO/IO

16L Ot- EwziIzO 00 OP- zwzo/o1

100 6'2 116/LZ/60 1006C- '16!011so

,oo sr- z6/su6o

,oo sc- 16/Pl/90

00 LL lZ6/OZlCO 100 IZ- 1i6/90/60

slu",o~ a[iueq~ e,=a

v1 lIElIHX3

II1 ZZCLBPI

01) IL CLGPI

6OL 01 ZLOFl

801 01 L91SI LOI 01 ZPZSI 901 01 OICSI SOL 09LLFSI

I WI 09LEPSI CO& 09L6PSI

tot 09 LSSSI

101 09 il9Si

00, 09IS9SI

66 0916951

86 099ZL51

16 01 PPLSI 16 01 GSLSL

I 56 OI69LSi

P6 01 6LLSi

E6

I

01 SBLSi

Z6 01 06LSl

b6 01 PLLSI

I 06 01 ESLSI

68 OI EELS1

88 LILSI

L8 01 L89SI

98

I 01

01 LS95I

00 001 05

8RH6 09 00% OS OOSL- oc

0089~ OP

OSL9. 5P

0009. OS

0009 09

0009. 09

00 r5 1st

/OOOZ- ~ZS/LiIiO~ EB 1016PSS1

BP91-

0061.

00 SZ-

OOLl

0051

OOZI~

00 01

00 01

OOZI-

00 Ol-

00 0,

00 1

00 s-

'00 c-

1:: ",I

00 2~

00 I-

00 P

00 B

00 s

00 L

00 5

00 9

100 8

EWUIO zn 9L LOSSI

26/PUlO LB Iso9psI

lwzuIi 99 iS9Offil

Z6/iUZO IL ISBPCSI

ZWL i/PO 81 9P PEZSI

16/c0/s0 LL SPSOZSI

iw8o/co Of 96LFiSL

IG/z7JzO SL 96ZSOSI

16/z0/90 PL 96ZL6Pi

Z6/IUIO CL 96268PI

16IOU60 ZL 9P96LPl

zwcoIPo LL 9P ZOLPI

26!01/L0 Of 9PZ09PI

ZWPO/ZI 69 9P ZOSPI

26/90/11 89 9P86FPI

Z6lLIILO LO 9PC6ZPI 1&!92190 99 9P88iPI

zwcl/ii $9 YP OBOPI

Z6/il/zi PO 9PZf6EI

1 WCZ/EO co 9PZ99Cl

ZW61/90 z9 SPZPLEI

Z6/10/80 L9 966i9Cl

Z6I6ZISO 09 966BtEl

'16/1O/ii 6E

I-

CO OSCCL

‘6lLZJCO BP CO56iFi

!iLlP OS OOZS OP 0009 0 P sow oc OOSL OS

OS 9L SB

0008 OS

oooe 0 P

0008 0 P

asps sr

80096 09

:00001 OP

'OOOOl OS

OOWI OP

oosoi oc

00*01 OS

OOBOI OP

OOBOl SP

00011 OS

OOOZI OC

OSLZI SC

OOOFI OS

C66Ci Of

81 Cl SP6

OOCI

OOSI

SC IL

OOSI

006 0091 00 oz

lOOOZ 100 IL ,oo 91

00 IL

00 LZ

00 PZ

00 zz

00 OP

00 SE

00 9z

6661

'C6/10110 5s C09L9ZI

ZSILZIII ps CO6EFZI I WSo/fO cs E000czl

: zw1cJz0 ZS coollzl

j ZWSUZ I LS COZl6ll

ZEJOrIOI 05 COZOLli

16K1/21 6P COL9PiI

ZG/CUOI 8v COILZII

ZWSI/SO fV C"CPO,l zw1 o/so iw co SO801

lt,";::SP FP co 19501 Pp 9ZROEOi

'i6/SO/CO FD 9ePiool

iW6Ull zv 98 6996

Z6/9ZIBO L, 986lC6 LWZUSO Ot 99Z6E9 16/9O/Zl 6t 98 EWE

ZGIIUEO SE 98 oooa 16/LW90 fC SHOCSL iw9uPo DE 9EOZOL

zw9U90 SE 986059

~ZWPUPO vc 386P6S

ZEJCWLO EE 986PIS

C6&1/10 ZE 00 19ZP

loo L81 ‘55 00691 Of

00061 OS

00961 OS OOOlZ OS OOSIZ OS 00912 OP ooazz 09 OOECZ Of

00 PPZ 0 P

LLzSZ OL

00 tc lwsuoI fZ 00 LZ 16/ZI/PO az

00 EC 16Kl/LO SZ

00 cc L6/9Z/LO ?Z

00 ZP 1643uc0 CL

00 CP 126/91/01 zz

00 PS ~LWPIIZO 1z

00 BE ~ls/Ei/Io OZ

00 bC 16,lO/CO 61

ZWOUll 81

W6UIO Lb cw9z/z0 et I6/ZZ/CO FL

lwSOILO Vb

16/10/20 EL

Z6IPZILO ZI

iG/LUZI 6

26/01/10 8

16/S2/10 L

00 001 iws0,Zn 0

86 921 cwso/zo E

00 toz IwtzISO t

00891 16/11110 E

OOZ6l 16/91/80 z

Actual Entire Series Data:

PI.&* Cumulative Price Cumulative Price Week

Cumulative Date

PIiCe Cumulative Chmga Volume Product Product Week Date Change Volume Product Product

1 Week me Change Volume Product Product Week Date

01/04Al -41001 50 -205OOi -20500 29 Change Volume Product Product

07/19Al 4300 2

50 -21500 346000 57 OlI3lA2 -10 00 40 -4000 Ol/llAl

627250 85 08114A2 168 00'

-7200 60 100600~ 803 00 30

40 -28800 676800 07R6Al 33 00 60 198 00 365600 58 02107/92 1500

J 40 60 00

O111tlAl 633250 86

3800 OSr2lA2 27 00

60 40 10900

22800, 687600

103100 31 08m219i -1200 45 -54 00 4

3604 00 59 02/14AZ 5400! 40 21600 654650 87 OlR5Al 0000, 70

OWBA2 56000/

2600 40 104 00 159llJO 32 08109Al

6980 00 -2000 55 -Ii000 3494 00 60 02/21A2 -17 oo! 40 6900

5 02miAl 648050 88

6100)

09m4A2 12 00 30 70 427001

36 00 701600

201800 31 08/16Al 19200 60 115200 464600 61 0208192 -5200 40 6

20800 02/08Al 100 00'

627250 89

I

09/llA2 23 00 80 8OOOOl

50 .11500 690 I 00 281800 34 06R3Al 4 00 40 1600 466200 62 03rn6A2 2500 40 l0000

7 637250 SO

02/15Al OY/lBA2

-4300 -7500

40 .I7200 40 .30000 66OIOO

264600 35 08r30/91 -3500 40 -14000 452200 63 03/13A2 4500 40 180 00 655250 91 8 02122Al -1000

09t25A2 -4500, 60

40 -18000 -60 00 258600 36 09rn6Al -2100

642100 50 -10500 441700 64 03/2OA2 -27 00 40

8 10900

03/OlAl 34 00, 6444 50 82 lom2A2 4000 50 -20000

70 23800 282400 37 09/13Al 622100

33 00 50 165 00 458200 65 03127A2 8001 40 3200 10 03rn8Al

647650 93 iom9A2 -1000

29 00 60

60 -6000,

174 00 276400 38 09ROAl -1000

6395 00 40 -4000 454200 66 04/03?92

11 -700' 50 -3500

03/16A, 644150 04

-6700 60 ~40200' 10116192 4300 50 21500

236200 39 09/77Al -39 00 661000

50 -19500 434700 67 M/l0192 112001 40 44800 12 03RZAI

6889 50 95 10,23A2 69 00

1600 50

50 34500

80 00 270700 44 lOm4Al 3100

6690 00 50 155 00 450200 68 M/17/92 -15001 45 -6750

1J 03i79Al 682200 96 lOi3OA2

42 00 1500

50 210 00 50 75 00

291700 41 iO/llAl 676500

94 00 50 47000 497200 69 04t24A2 3500' 35 122 50 694450 97 14 04m5Ai

llrn6A2 1300

-200 40

50 5200 I I

-1000 675500 296900 42 10118191 -7200 4 5 -32400 464800 70 05IOlA2

15 04/12Al 27 00 2000' 40 7024 50 OE

70 11113A2 -100 50

189 00 -5 00

315800 ,J lOr25Al 34 00 55 675000

18700 483500 71 05/OSA2 -2500 16 04/19Al -38 00,

6924 50 99 llROA2 61 00 1 ;i : ! ! I ;

24400' 60 -22000

40 293000 44 1 lmiA1 6 00 50 3000

6994 00 486500 72 05/15A2 7000

17 04i26Al 2200 700450 100 flR7A7

50 945 50

11000 47 25

304000 45 i immi 86 00 7041 25

50 -43000 443500 73 05122A? 3500 710950 101 18 05/03Al 1200

12m4A2 300, 50 .6000

50 1500 298000 46 11/15Al -6300

702625 60 -37800 405700 74 05/29A2 500 60 713950 102 12/llA2

19 05/lOAl -39 00 50 -19500 4 00 40i ,600~

779500 47 llR2Al -2500 30 704225

-7500 396200 75 06105A2 -57 00 20 05/17Al 2100

691150 103 12/lSA2 45 94 50 ( 6% i;;;

1600 601 96 00' 287950 48 iin9Ai 2000

713925 50 10000 408200 76 0602192

21 -7100 655650 104 12/25A2

05r24Al 204 00 45 21 35

91800 30 64 05

319750 49 I2rn6A 1 2100 720230

50 10500 418700 77 06/19/92 500, 22 05i31Al

657650 105 OlmlA3 -59 00

13 78 55 -324 50 341300 50

7243 64 12/13Al 9 00 85 76 50 426350 IS 06126A2 4000 30 I20 00

23 06/07Al 669650 106 Oll08A3

24 00 45 -60 79

10000 358100 51 12ROAl 182 00 :: A",:::1 6878 90

65 llE300 544650 70 07lO3A2 50 17500:

2600 50 13000 662650 107 01115A3 139 93' 24 06/14Al -3500

1999 70 340600 52 l2ff7Al

701883 102 00 50 51000 595650 80 07/10/92 -5 00 35 -1750

25 06/2lAl 680900 108 01/22A3

-5300 1646

50 ~76500 60 -9888

314100 53 0 im3A7 5 00 80 4000 691995

599G 50 81 26

07/17AZ .4 00 25 -10 00 06r29A I -200 30

679900 109 01/79A3 -6 00,

36 II 70 252 77 313500 54

7172 7: 01/1OA2 7300, 70 51100 650750 82

27 07l24A2 11100~ 40 444 00

07,05AI 10 00 724300 110 02m5A3 I?696

50 70

350 00 89886

34R5 00 55 8061 56

Ol/l7A2 2000 50 100 00 640750 a3 07DlA2 52 00 40 206 00 28 07/12Al 38 00

703500 111 02112A3 50 19000

-40 79 50 20395 ,675"" 56 0 lT24A? ,900 50 -95 00

765763 631?50 84 08,07,97 700 30 7100 705600 112 07/19A3 50 DO SO 754 00 iG03 G3

113 OLr..GA3 4890 60 293 40 7897 03

Mwn 1144 491 6989 Mwhan 8 00 5 00 3600

0 55 I? I Ii .,07 O?

EXHIBIT IR SOlJRCE I tL~ADVISORS

-

Ranked Quartile Data For Entire Series:

First Quartile Second Quartile Third Quartile Fourth Quartile Price Cumulative Price Cumulative P&C

Weak Date Cumulative Price ’

Change Volume Product Cumulative

Product Date Change Volume Product Product Date Change Volume Product Product 1

Dne 01/11/91 16800 60 100800

Change Volume Product Product 100800 12/20/91 182 00 65 118300

2

1183 00 04/10/92 11200

05/24/91

40 44800

204 00 91800 448 00 02/05/93 12698 70' La3886 6m86

1926 00 08/16/91 19200 60 115200 2335 00 07/24/92 11100 3 ::I 80000/

40 444 00 02/08/91

892 00 02/26/93 4890 100 00

60 29340 118226 2726 00 01/10/92 7300 70 511 00

4

2846 00 02114192 40 21600 110800 01/29/93 01/25/91 8000/

36 11 70 56OOOi

70 25277 14350:

328600 12/27/W 10200 50 51000 3356 00 03/13/92 5

401 18000 02/01/91 6100' 70 427001

1288 00 11/20/92 40; 24400 50! 47000

167903 371300 10/11/91 94 00 3826 00 07/03/92 50

6 13000

07/05/91 70001 35000' 501 21500' 189403

4063 00 07:26/91 3300 60 19800 4024 00 M/24/92 7 21 345001

35 12250 03122191 69 001

174 001 440800 10/25/911

60 34 00 55

206803

a 18700 4211 00 06/26/92 4000

701 23800: 30 12000

03101191 1660 50 01/15/93

34 001 139 931

464600 09/13/W 70

3300 50 220796

16500 437600 05/22/92 3500 30 10500 9

1765 50 08121192 60) 22800

40' 01118/91 3800

10800i 231596 4874 00 10/04/91 3100 50 155.00 4531 00 03/06/92 25 00

IO 40 100 00 186550 08/28/92

03/29191 4200 50 21000 40 10400~ 241996

5084 00 12/06/91 21 00 50 10500 4636 00 05/01/92 2000 40 11

80 00 07112191

1945 50 12/18/92 3800j

1600 50

60 19000

96 00 251596 5274 00 11/29/91 2000 50 100 00 4736 00 05/15/92 2000

12

40 8000 04/12191 27 001

2025 50 10/23/92 70 189 00 5463 00 12/13/91 9 00 85 76 50 4812 50 02/07/92 1500 4.0

13 22001 50 11000 6000 2085 50 10/30/92

Oh:26191 557300 01/03/92 500 80 4000 485250 03/27/92 800 40 14

3200 06/07/91

2117 50 12125192 24 00 45 10800 5681 00 ll/Ol/Sl

15 2147 50 11/27/92

05/17/91 21 00 45 94 50

18 04/05/91 1300 40 5200 j7 06/28/91 -2 00 30 -600

18 02122191 -1000 60 -6000 5761 50 11/22/91 -25 00 30 -7500 4729 50 07/10/92

19 03/08:91 -1oso 60 -6000 5701 50 01/24/92 -1900 50 -95 00 4634 50 04/0X92 20 05/03/91 -1200 50 -6000 5641 50 01/17/92 -20 00 50 -10000

21

4534 50 01/31/92 02/15/91 4300 40 ~17200 5469 50 09/06/91 -21 00 50 -10500 442950 04/17/92

22 06/14/91 -3500 50 -17500 5294 50 OW/OS/Sl -20 00 55 -11000 4319 50 02121192

23 05/10/91 -3900 50 -19500 5099 50 08/30/91 -35 00

i-i:;"9:9":1 -:;:i 60 -22800 466650 07/19/91 -4300 50 -21500 376950 02/28/92 % :o" ::,":i 153450 02112193 i,";: i$ -ilHI!(

40 -14000 4179 50 05/08/92 25 00 40 -10000 24

1850 50 09125192 -4500 50 -20500

2451 72 4894 50 09127191 -39 00 50 -19500 3984 50 03/20/92

25

1742 50 10102~92 2251 72

26 06:21/91 -53 00 50 -26500 204777

4401 50 10/18/91 -7200 45 ~324 00 3445 50 07/31/92 52 00

27 40 -20800 1326 50 02119i93

05131191 -59 00 5080

5 5 -32450 5 0 ~254 00 1793 77

4077 00 11115/91 -6300 60 -37800

28

3067 50 06/05/92 -57 00 40 -22800 03/15/91

109850 08114192 -67 00

-7200 60 -40200

40, -26800 150577 367500 11/08/91 -86 00 50 -43000 2637 50 06112192 -71 00 50 -35500 743 50 09/10/92, -7500 4 OS -30000 1205 77

Mean

OliOWi93

21 50'

-60 79 60 -364 74

5431 131 25

841 03

1336 5 29 94 20 789 4 00 2655

Median 21 50 5 00 3 30 4 93 29 00'

101 25 4 50 500 2300 6 00 4 00 25 50 _ .~___. ..-

64 so 110 1200 5 00 41341

cl 36281 -, _--

66 24 116 _._I___. - ~_.-

383 18 43 19 069 ___ --.--

17662 4332 1 13 241 46;

EXHIBIT 2A SOUR(;t ltLADVIS0R~

Actual Quartile Data For Entire Series:

First Quartile Second Quartile Third Quartile Fourth Quartile Pric# CUlllllltiiV~ Price Cumulative PIiCe Cumulative Price Cumulative

Wook Dt4tO Change Volume Product Product DdO Change Volume Product Product Date Change Volumo Product Product Date Change Votumo Product Product 1 01/04/91 -4100 50 -20500 -205 00 07/19/91 -43 00 50 -21500 -215 00 01/31/92 -1000 40 -4000 -40.00 08/14/92 -7200 40 -28800 -266 00

2 01/11/91 16800 60 100800 803 00 07/26/91 3300 60 19800 -17 00 02/07/92 15.00 40 6000 20 00 08/21/92 2700 40 10800 -180 00

3 01/18/91 3800 60 22800 1031 00 08/02/91 -1200 45 -54 00 -71 00 02/14/92 54 00 40 21600 23600 08/28/92 2600, 40 104 00 -7600

4 01/25/91 80 00 7.0 560.00 1591.00 08/09/91 -2000 55 -11000 -181 00 02/21/92 -17.00 40 -6800) 16800 09/04/92 1200 30 3600 -40.00

5 02/01/91 61.00 70 42700 2018.00 08/16/91 192 00 6.0 115200 97100 02/28/92 -52.00 40 -20800 -4000 09/11/92 -2300 50 -11500 -15500

6 / ')2/08/91 10000 80 80000 281800 08123191 400 40 1600 987 00 03106192 25 00 40 10000, 60 00 09/18/92 -7500 40 -30000 -45500

7 02/15/91 -4300 4 0 -17200 2646 00 08/30/91 -3500 40 -14000 847 00 03/13/92 4500 40 18000; 24000 09/25/92 -4500 40 -18000 -63500

8 02/22/91 -1000 60 -6000 2586 00 09/06/91 -21 00 5.0 -105.00 742 00 03120192 -2700 4.0 -10800 132 00 10/02/92 -4000 50 -20000 -83500

9 03/01/91 34 00 70 23800 2824 00 09/13/91 3300 5.0 165 00 907 00 03127192 800 40 3200 164 00 10/09/92 60 17400 -661 00

10 03/08/91 -10.00 60 -6000 276400 09120191 -1000 4.0 -4000 867 00 04!03/92 -7 00 50 -3500 12900 10/16/92 ;; ti 50 21500 -44600

11 03/15/91 -67.00 60 -40200 2362 00 09/27/91 -39 00 50 -19500 672 00 04/10/92 11200 40 44800 57700 10/23/92 -36600

12 03/22/91 69 00 50 34500 270700 10/04/91 31 00 50 15500 827 00 04/17/92 -1500 45 -67 50 6";;,; ;~om;~~~ !;; ; I i; E;i -291 00

13 03/29/91 4200 50 21000 2917 00 10/11/91 94 00 50 47000 129700 04/24/92 3500 35 12250 -301 00

14 w/05/91 13 00 40 5200 296900 10/W/91 -7200 45 -32400 973 00 05/01/92 2000 40 80 00 712.00 11/13/92 -1 00 50 -500: -306 00

15 04/12191 27 00 70 18900 3158.00 10/25/91 34 00 55 187.00 116000 05lO8192 -2500 40 -10000 612.00 11/20/92 61 00 40 244 00 -62OG

16 04119191 -38.00 60 -22800 293000 11/01/91 600 5.0 3000 1190 00 05/15/92 2000 40 80 00 692 00 11/27/92 945 50 4725 -14 75

17 04/26/91 2200 50 110.00 3040 00 11/08/91 -86 00 50 -43000 76000 05122192 3500 30 10500 79700 12/04/92 -300 50 -1500 -2975

18 05/03/91 -1200 50 -6000 2989 00 11/15/91 -6300 60 -37800 382 00 05129192 500 60 3000 827 00 12/11/92 400 40 1600 -13 75

19 05/10/91 -39 00 50 -19500 2785 00 11/22/91 -2500 30 -7500 307 00 06/05/92 -57 00 40 -22800 599.00 12/18/92 1600 60 96 00 82 25

20 05/17/91 21 00 45 94 50 2879 50 11/29/91 2000 50 10000 407 00 06/12/92 -71 00 50 -35500 244.00 12125192 21 35 3.0 64 05 14630

21 05/24/91 204 00 45 91800 3797.50 12/06/91 21 00 50 10500 512 00 06/19/92 500 40 2000 264 00 01/01/93 1378 30 41 34 18764

22 05/31/91 -59 00 55 -324 50 3473 00 12/13/91 9 00 85 7650 56-J 50 06/26/92 4000 30 12000 38400 01/08/93 -6079 60 -364 74 -17710

23 06/07/91 24 00 45 10800 3581 00 12/20/91 18200 65 118300 1771 50 07/03/92 2600 50 13000 514 00 01/15/93 1999 70 13993 -3717

24 06/14/91 -3500 50 -17500 3406 00 12/27/91 10200 50 51000 2281 50 07/10/92 -5 00 35 -1750 49650 01/22/93 -1648 60 -98 88 -13605

25 06/21/91 -53 00 50 -265 00 3141 00 01/03/92 500 80 4000 2321 50 07/17/92 -4 00 25 -1000 486 50 01/29/93 3611 70 25277, 116 72

26 06/28/91 -2 00 30 -6 00 313500 01/10/92 73 00 70 51100 2832 50 07/24/92 111 00 40 44400 930 50 02/05/93 12698 70 88886i 100558

27 07/05/91 7000 50 35000 3485 00 01/17/92 -2000 50 -10000 2732 50 07/31/92 -5200 40 -20800 722 50 02/12/93 -40 79 50 -203951 801 63

28 07/12/91 3800 50 19000 367500 01/24/92 -1900 50 -95.00 2637 50 08/07/92 700 30 21 00 74350 02/19/93 -5080 50 -254 001 54763

I 02/26/93 4890 60 29340 84103

Moan 21 50 5431 131 25 1336 5 29 94 20 789 4 00 26 55 330 4 93 29 00

Median 21 50 500' 101 25 4 50 500 2300 6 00 400 25 50 1200 5 00 41 34

0 64 90 110 36281 66 24 116 383 18 4319 0 69 17662 4332 1 13 24146

EXHIBIT 28 SOURCE 1 EL-ADVISORS

Fourth Quartile Cumulative Product-Actual vs. Ranked

&&iii 1.1.1

1.13

12.1

133

2.1.1

2.13

2.2.1

2.23

Exhibit 4

(0bsri-wd-

Obwl~rd- (Observrd- Erpcdrd)‘/

12 II 3767 -3 67 ,344 036 0 59

:I 3s 3767 -1 67 III OIY 1 OS

:1 211 lSS3 1.17 I 36 007 1.15

(Obsmrck

Oherved- (Obuwwl- Esprclecl~ I lrigrm Obwmrd Exlw’rd EX~WCIEd Eqwclrd)’ Exlwr~cd x

I.11 17 4 II 1219 IS1 IIS 32 OS 32us

I I.2 17 II I3 1 ss s27 0 '9 32 67

:.2.1 1s 2351 .s 51 307, I 30 33 97

1.x I 14.13 " ss 0 77 ous 3103

II I 17 II 13 2 ss sz7 OSY 3161

:I2 17 13 '4 -6!1 42 7Y I S? 36 43

., __I 16 I4 I3 I ss 3 52 0 2s 36 6S

.T, -.__ s 171 02Y 0 OS 0 02 36.5U

(ObwwJ

OLwncB (OlnwwL Ex~rclrd)‘/

Y

Y

,I


/ ADDENDUM TO EXHIBITS 3A-D ( (

These four exhibits provide another way to look at market momentum. Each chart has the same vertical scale, so that the quartiles can be compared each to the other. I have discussed in the main body of this report the fact that market momentum currently is waning rapidly. The slope of the line between the starting and ending point for each quartile easily sup- ports this conclusion visually. However, if there is some reader skepticism, perhaps the following table will give additional support to my conclusion.

A Comparison Between Ihe Maximum & Endlng Product Values

Maxnlum 96 Ending Value

Quadile ProductValue Ending Value lo Maximum Value

1 5627.50 3675.00 63.06

2 4969.50 2637.50 52.66

3 2166.50 743.50 33.97

4 2675.60 641.03 29.25

swuu TEL kMlul. Inc o( MW ,wmR 11,

On the other hand these charts are also designed to illustrate a buildup in speculation. If the vast majority of actual product activity takes place above the starting to ending value line, speculative excesses are building up within the market. This is an appropriate conclusion, if the actual course of the market approaches the high to low ranked course over the time period, as is the case during much of the first time quartile (see Exhibit 3A). The closer the actual course parallels the ranked course of the market, the greater the potential for the downward momentum to develop towards the end of the time period. Just the opposite is true if the actual course is below the starting to ending value line for most of the time period (see Exhibits 3B & D). Looking at these exhibits together with the above table, one is again left with the impression that the bulls are gradually losing the war, despite some modest improvement in the fourth time quartile.

Theodore E. Loud is a graduate of Yale University CBA) and the Wharton School, University of Pennsyluania (MBA). He has nearly thirty years of investment experience, and has been president of TEL-Advisors, Inc. of Virginia since 1980. This small sophisticated investment management firm on several occasions has been ranked as one of the top equity managers in the U.S. by Nelson’s Directory and the Money Manager Review, most recently in their 1994 publications.


The Impact of Government Bond Yield Spreads on Currencies-An Intermarket View by R.J. Slatter

Summary This article extends John J. Murphy’s pioneering and exploratory work in intermarket technical analysis. More specifically it examines the relationship between government bond yield spreads and currency movements. Government bond yield spread is here defined as the yield on a lo-year U.S. Treasury bond minus the yield on a lo-year foreign government bond. Four currency markets were examined in relation to the American dollar (viz. Deutschmark, Yen, Sterling and Canadian dollar). It was concluded that bond yield spreads frequently trend in the same direction as, and frequently lead turns in, the relevant currency markets. Conse- quently the intermarket analysis of bond yield spreads provides an important strategic overview of the currency markets.

l Bonds lead turns in the stock market. l The Dow Utilities follow the bond market and lead

stocks. . The U.S. bond and stock markets are linked to

global markets. l Some stock groups (such as oil, gold mining, cop-

per and interest-sensitive stocks) are influenced by related futures markets.

This article will introduce and examine a new hypothesis concerning intermarket analysis. More specifically, it is proposed that the analysis of government bond yield spreads provides an important insight into currency movements.

Currency movements are precipitated by a myriad of factors which may be summarized as follows:4

Introduction Inter-market analysis has been defined by Murphy’

as,

“An additional aspect of technical analysis that takes into consideration the price action of related market sectors. The four sectors are currencies, commodities, bonds and stocks. International markets are also included. This approach is based on the premise that all markets are interrelated and impact on one another.”

Although Murphy2 has described intermarket analysis as “both pioneering and exploratory” he has identified various principles and relationships, the most important of which are as follows:3

l The U.S. dollar usually trends in the opposite direction of the gold market.

l The U.S. dollar usually trends in the opposite direction of the Commodity Research Bureau Index.

l Gold leads turns in the CRB Index in the same direction.

* The CRB Index normally trends in the opposite direction of the bond market.

l Bonds normally trend in the same direction as the stock market.

“There are essentially two types of interacting factors affecting exchange rates: statistical and emotive These are the key elements which influence the international flow of funds and, consequently, determine the type and timing of movements in par- titular currencies’ exchange rates. The monitoring of these factors over a period of time can give both an understanding of the contemporary exchange-rate environment and a reasonable indication of the major future trends in exchange-rate movements.

“Statistical factors include: relative inflation rates, money supply figures, balance-of-payments, current account and capital account flows of funds (including, in particular, the growth or contraction of official reserves as a measure of capital flows), interest rates, corporate overseas investment, and international financial investment, productivity and overall economic growth, national budget deficits and surpluses, savings and consumption rates and central bank intervention.

“Emotive factors include: politics, market momentum and sentiment, and psychological factors that affect national habits in, for example, savings and consumption. (Savings and consumption are also affected by national economic policy.) Dramatic one- off economic or political events, such as wars or sharp increases in the price of key commodities like oil, have a major, albeit often transient, influence on exchange rates.”


r 3.60 _ FIGURE 1

1 1 _ 5.00

3.40 1 U.S. . Genmn yield spread : 4.50

3.20 I 14.00

3.00 I I3.50

2.80 I : 3.00

2.60 I : 2.50

2.40 : marks per U.S. dollar : 2, o.

2.20 : 11.50

2.00 I Il.00

I.80 I : 0.50

1.60 ; 1977 ’ 1978 ’ 19’3 ’ 1980 ’

-0 1 g,q I ’ 138% ’ 1983 ’ 1984 ’ 1985

- DEUTSCHMARKS PER U.S. DOLLAR - U.S. - GERMAN WEID SPREAD (RIGHT HAND SCALJZ) Source: DATASTREAM

Although it is acknowledged that currency movements are precipitated by a multitude of interacting and variable factors, it is proposed that one of the more important factors is relative interest rates. Indeed, it is hypothesized that the yield spreads of government bonds have an important impact on currency movements.

Monthly observations were made of the relevant yield spreads and currency markets. For clarity of exposition, in each case the period since 1977 was divided into two charts of nine years. The vertical scales of the charts were then adjusted for maximum resolution.

1. U.S.-German government bond yield spread

Hypotheses

l Government bond yield spreads frequently trend in the same direction as the relevant currencies.

l Government bond yield spreads frequently lead turns in the relevant currency markets.

Analysis of Government Bond Yield Spreads and Currencies

Government bond yield spread is here defined as the yield on a lo-year U.S. Treasury bond minus the yield on a lo-year foreign government bond.

Because of the importance of the American financial markets, four currency markets were examined in relation to the American dollar (viz. Deutsch- mark, Yen, Sterling and Canadian dollar).

The analysis examined the period since 1977, by which time the currency markets had been liberalized and were characterized by the “managed float- ing” system rather than the former “adjustable peg” system.5

The U.S.-German government bond yield spread is depicted in Figures 1 and 2.

Before the currency markets were liberalized in the early 1970’s there were approximately 4 Deutsch- marks to the dollar. Subsequently, however, the dollar weakened significantly in value.

Starting in mid-1977 the U.S.-German yield spread increased from 40 basis points to 200 basis points and then fell back to 40 basis points by mid-1979. This rally and fall in the yield spread coincided with an end to the dollar’s downtrend.

As shown in Figure 1, from mid-1979 to mid-1984 the yield spread increased from 40 basis points to 480 basis points in a major uptrend which displayed the classic pattern of ascending peaks and troughs. The start of the uptrend in the yield spread anticipated by one year the start of the uptrend in the dollar. Moreover, this five-year period was characterized by the following sequence of events: the yield spread increased, the dollar strengthened, the yield spread closed somewhat,


the dollar weakened, and the yield spread then increased to a new high.

In mid-1984 the yield spread peaked at 480 basis points and then over the next year fell to the upward trendline which started in mid-1979. The dollar peaked against the Deutschmark in early 1985. However, in mid-1985 the yield spread broke its trendline, thereby indicating the end of its six-year upward trend. Moreover, the new and clear downward trend in the yield spread provided an early warning that the peak in the dollar in early 1985 was not merely an intermediate correction but the end of its five-year uptrend. Subsequently the yield spread and the dollar trended sharply lower in unison.

As shown in Figure 2, in early 1987 the yield spread halted its downtrend by rallying from 160 basis points to 300 basis points. Over the next two years the yield spread produced a consolidation which might have anticipated the subsequent consolidation in the dollar.

In early 1989 the yield spread broke the lower line of its triangle. This corroborated the bearish view presented by the dollar’s rising wedge formation. Indeed, six months after the yield spread broke the lower line of its triangle the dollar broke the lower line of its wedge formation. Subsequently the yield spread and the dollar trended lower in unison.

As Figure 2 shows, since 1991 the yield spread has potentially given highly reliable warnings of impending changes in the dollar-mark.

-

3.20 _

3.00 I

2.80 I

2.60 I

2.40 I

2.20 I

2.00 1

1.80 I

1.60:

1.40 1

FIGURE 2

2. U.S.-Japanese government bond yield spread The U.S.-Japanese government bond yield spread

is depicted in Figures 3 and 4. Before the currency markets were liberalized in

the early 1970’s there were approximately 360 yen to the dollar. Subsequently, however, the dollar weakened significantly in value.

As shown in Figure 3, starting in mid-1977 the U.S.-Japanese yield spread went from -0.5% (i.e. Japanese yields were higher than American yields) to 1.8% and then back to -0.5%. This rally and fall in the yield spread coincided with an end to the dollar’s downtrend.

From mid-1979 to early 1982 the yield spread increased sharply from -0.5% to 5.6% before col- lapsing to 1.8% by the end of 1982. In this cycle the yield spread, it could be argued, anticipated the strong rally in the dollar by eighteen months. On the downside, the yield spread broke out of a broadening top formation in mid-1982, thereby preceding by 3 months a collapse in the dollar.

From late 1982 to mid-1984 the yield spread rallied from 1.8% to 5.5%, coinciding with a halt to the fall in the dollar. However, the yield spread peaked in mid-1984 just below the previous peak in early 1982 (i.e. it “failed”). This provided an early warning that the rally in the dollar would probably not persist. Indeed, the dollar peaked in early 1985 below the previous peak in late 1982 (i.e. it too “failed”). Subsequently the yield spread

:4

13

12

I

:0

:-1

1.20 ’ 1985 ’ 1986 ’ 1987 ’ 1988 ’ 1989 ’ 1990 ’ 1991 ’ 1992 ’ 1993 ’ --

r 2

- DEIJTSCHMARKS PER US. DGLIAR - U.S. - GERMAN YIELD SPREAD (RIGHT HAND SCALE) Source: DATASTREAM

--

MTA JOURNAL/WINTER 1993 - SPRING 1994 39

1

28Ow FIGURE 3 1 -6

270:

:5 260:

250: :4

240:

230: :3

220s :2

210:

200: qamse yen per U.S. dollar :1

19c: 10

180: - Japanese yield spread

I'@- t

1977 ’ 1978 ’ 1979 ’ -1

198rl ’ 1981 ’ 1982 ’ 1983 ’ lq84 I 1985 - JAPANESE YEN PER U.S. DGLhR - U.S. - JAPANESE YIELD SPREAD (RIGHT HAND SCALE)

and the dollar trended sharply lower in unison. As shown in Figure 4, in early 1987 the yield

spread halted its downtrend by rallying sharply from 2.3% to 4.9%. Over the next two and a half years the yield spread produced what turned out to be a horizontal triangle which might have anticipated the subsequent sideways movement in the dollar.

26Ow.

2401

2201

200:

180:

1601

1401

120:

100;

Source: DATASTREAM

In mid-1989 the yield spread broke the lower line of its symmetrical triangle. This provided an early warning that the weak rally in the dollar would probably not persist. Indeed, one year later this weak rally in the dollar peaked and turned downwards.

From late 1990 to early 1993 there has been

_ 5.00

I4.50

14.00

13.50

I3.00

I2.50

: 2.00

: I .50

1985 ’ 1986 ’ 1987 ’ 1988 ’ 1989 1990 ’ 1991 I 1992 ’ 1993 ’ ! .oo

- JAPANESE YEN PER U.S. DOLLAR - U.S. - JAPANESE YIELD SPREAD (RIGHT HAND SCALE)

Source: DATASTREAM

40 MTA JOURNAL ! WINTER 1993 - SPRING 1994

1.00 _

0.90 :

0.80 :

0.60 :

0.50 :

0.40 :

FIGURE 5 1 1 -3

12

I 1

LO

I-1

I-2

7

1-4

l-5

I-C

r 0.30 1

’ 198.3 ’ 1984 ’ 1985 ‘-

-

1977 ’ 1978 ’ 1979 I l?W ’ 1981 ’ I ? P, 2 - BRITISH FOUNDS PER U.S. DOUAR - U.S. - BRITISH YIELD SPREAD (RIGHT HAND SCALE) Source: DATASTREAM

some divergence, with the yield spread rising from 1.2% to 2.8% while the dollar has moved sideways. One possible explanation for this is that the rally in the yield spread has not been strong enough to lift the dollar.

However, in early 1993 the yield spread peaked and turned downwards. Simultaneously the dollar changed its trend from sideways to downwards. Since the yield spread and the dollar are now trending downwards in unison, the outlook for the dollar against the yen would appear to be strongly bearish.

3. U.S.-British government bond yield spread The U.S.-British government bond yield spread

is depicted in Figures 5 and 6. Before the currency markets were liberalized in the early 1970’s the exchange rate was approximately f0.40 to the dollar. The British pound subsequently weakened to f0.60 per dollar in 1976 before starting to strengthen.

Between 1977 and 1979 the U.S.-British yield spread was approximately -5% (i.e. British yields were significantly higher than U.S. yields) and the trend was sideways.

As shown in Figure 5, from 1980 to mid-1984 the yield spread reversed from -5% to +2% in a major uptrend which displayed the classic pattern of ascending peaks and troughs. The start of the uptrend in the yield spread, it could be argued, anticipated by

one year the start of the uptrend in the dollar. This period was characterized by the following sequence of events (also identified in Figure 1 earlier with regard to the dollar-Deutschmark): the yield spread increased, the dollar strengthened, the yield spread closed somewhat, the dollar weakened, and the yield spread then increased to a new high.

In mid-1984 the yield spread peaked at 2% and then fell to and broke its upward trendline which started in 1980. When the yield spread broke its trendline in late 1984 there was a clear divergence with the dollar whose upward trend was not only intact but accelerating. However the dollar peaked in early 1985, three months after the yield spread provided a clear warning by breaking its trendline. Subsequently the yield spread and the dollar trended sharply lower in unison.

As shown in Figure 6, in late 1986 the yield spread halted its downtrend by rallying from -3% to 0% (i.e. the yield spread was completely closed). Over the next two years the yield spread produced what turned out to be an ascending triangle which might have anticipated the subsequent sideways movement in the dollar.

In early 1989 the yield spread broke the lower line of its triangle. This corroborated the bearish view presented by the dollar’s rising wedge formation. Indeed, nine months after the yield spread broke the lower line of its triangle the dollar broke


0.80 _ FIGURE 6 -0.50

U.S. _ British yield spread 10

L-O.50

:- 1 .oo

‘0.65 :

:- 1.50

0.60 :

c-2.00

o’45-i 1985 ’ 1986 ’ 1937 ’ 1988 ’ t

1989 ’ 1990 ’ 1991 ’ 1992 ’ 1933 --3.50

- BRITISH POUNDS PER U.S. DOLLAR - U.S. - BRITISH YIELD SPREAD (RIGHT FL4ND SCALE) Source: DATASTREAM

the lower line of its wedge formation. Subsequently the yield spread and the dollar trended lower in unison.

In mid-1990 the yield spread rallied and this anticipated by six months an end to the dollar’s decline. However, although the yield spread rallied from -3.3% to 1.2% between mid-1990 and mid-1992

this only coincided with a sideways movement, rather than a rally, in the dollar.

Since late 1992 there has been some divergence between the yield spread and the dollar. More specifically the yield spread has moved sideways but the dollar has rallied. It may be recalled that in September 1992 the British pound left the European

: .45 _ FIGURE 7

_ 0.50

1.40 I

1.35 I

1.30 I

1.25 I

1.20 I

1.15 1

1.10: U.S. - Can&ii yield spread

:O

L-0.50

I- 1 .oo

1-1.50

r-2.00

~-2.50

I- 3.00

r-3.50

’ IY77 I 1978 r 1979’ II I I II-0 Is31 ’ 1982 ’ 19% ’ 1984 ’ 1985 :- 4.51:

- CANADIAN DOLLARS PER U.S. DOLLAR - U.S. . CANADLAN YIELD SPREAD (RIGHT HAND SCALE)

Source: DATASTREAM


Monetary System. Moreover, the earlier quotation may be recalled:

“Dramatic one-off economic or political events.

have a major, albeit often transient, influence on exchange rates’?

Consequently, it appears that the dramatic economic and political events of late 1992 may have disrupted the normal relationship between the U.S.- British yield spread and the dollar. However, on the assumption that this disruption is only temporary, Figure 6 would suggest that the current rally in the dollar is probably not sustainable.

4. U.S.-Canadian government bond yield spread The U.S.-Canadian government bond yield spread

is depicted in Figures 7 and 8. It is apparent from Figures 7 and 8 that the U.S.-

Canadian government bond yield spread has at best a weak relationship to the U.S.-Canadian dollar. The principal problem would appear to be that since 1977 the yield spread has predominantly remained within a band of -1% to -2.5% (i.e. its amplitude is typically 1.5%). In contrast, the U.S.-German yield spread typically has an amplitude of 4% and the U.S.-Japa- nese and U.S.-British yield spreads typically have an amplitude of 5%.

It is possible that Canadian government policy and central bank intervention are responsible for the relatively stable U.S.-Canadian government bond yield spread. However, this stability obviates the yield spread in this instance as a predictive tool in intermarket analysis.

Conclusion From the above examination of four bond and

currency markets, one may conclude that the intermarket analysis of bond yield spreads provides an important strategic overview of the currency markets because:

l Government bond yield spreads frequently trend in the same direction as the relevant currencies.

l Government bond yield spreads frequently lead turns in the relevant currency markets.

It should, however, be acknowledged that the data do not always support these hypotheses, but at least do so frequently enough to make this a worth- while tool for helping to understand currency markets. More specifically:

l Dramatic and unique economic or political events can disrupt the hypothesized relationships, stated above, between bond yield spreads and the relevant currencies.

1.45 _

1.40:

1.35 _

1.30 _

1.25 _

FIGURE 8

U.S. _ Canadian yield spread

Canadian dollars per U.

,O

_- 0.50

_- 1 .oo

_- I .5c

e-2.50

1.10 4 - 3 IJj:! 1985 ’ 1986 ’ 1987 ’ 1984 ’ 13P9 ’ 19’jO ’ 1991 ’ 1992 ’ 1993

- CANADIAN DOLLARS PER US. WLIAR - U.S. - CANADIAN YIELD SPREAD (RIGHT HAND SCALE) Source: DATASTREAM


l Where the bond yield spread is relatively stable over a protracted period of time the hypothesized relationships, stated above, between bond yield spreads and the relevant currencies may be considerably weakened.

Nevertheless it would appear that the government bond yield spread captures many of the myriad of factors which precipitate currency movements.

REFERENCES

1. John J. Murphy, Zntermarket Technical Analysis-Trading Strategies for the Global Stock, Bond, Commodity and Currency Markets (1991), John Wiley & Sons, p. 274.

2. John J. Murphy, “The Impact of Commodity Prices on Bonds and Stocks-An Intermarket View”, MTA Journal, Spring 1992, Issue 39, p. 39.

3. John J. Murphy (1991) op.cit. p. 255.

4. Howard Flight and Bonita Lee-Swan, All You Need to Know About Exchange Rates, (1988), Sidgwick & Jackson, p. 55.

5. Christopher Dunis & Michael Feeny (Eds), Exchange Rate Fore- casting, (1989), Woodhead Faulkner, p. 22.

R. John Slatter M.Phil.(Oxonl C.Z?A. M.S.TA. is an analyst and investment manager with Post& Investment Management Limited in London,


The Derivative Oscillator: A New Approach for an Old Problem by Connie Brown

Introduction Traditional momentum indicators frequently offer accurate confirmation to the completion of textbook Elliott Wave Principle patterns. However, complex wave patterns remain a challenge for analysts. Com- plex Elliott Wave patterns develop when a simple three wave correction becomes the first wave within a larger corrective pattern. Other complex patterns develop when extensions unfold. These Elliott Wave patterns frequently cause incorrect or premature buy/sell signals from traditional momentum indi- caters such as Momentum, RSI, Stochastics, and MACD. The need to minimize these incorrect momentum indicator signals, while maintaining the original additive value of these studies with the Wave Principle, has led to the development of a derivative oscillator.

The oscillator is a triple smoothed derivative of RSI plotted as a histogram. The histogram has the added benefit of frequently displaying precise equality relationships between the amplitude of crest highs and trough lows. This mathematical equality adds visual clarity to the identification of high risk pivot levels for a market without the usual lag time limitation. The objective of this paper is to explain the use of this indicator, demonstrate how it is currently being used in conjunction with the Elliott Wave Principle, and identify potential areas for further research. After three years of development, testing, and real-time performance evaluation, this indicator is being offered to the Technical Analysis community for the first time.

Elliott sequence. As an example, a fifth wave when compared to a third wave is frequently accompanied by momentum divergence as the fifth wave is usually less dynamic. Traditional momentum studies perform well in such text book pattern scenarios.

More challenging Elliott Wave patterns tend to cause multiple entry/exit signals from traditional momentum indicators. This can be demonstrated when an impulsive fifth wave extends into a nine or thirteen wave structure. Extensions temporarily renew dwindling volume and yield multiple diverg- ing momentum signals which can be misleading to the analyst and costly for the trader. Complex corrective Elliott Wave patterns also cause timing dif- ficulties for momentum indicators. The simplest three wave correction, labeled a-b-c, would ideally be followed by a distinct, impulsive five wave structure. However, should the market then develop a choppy, nearly indiscernible wave structure uncharacteristic of an impulsive five wave pattern, the choppy wave personality would warn the analyst that the correction is incomplete. The first a-b-c structure likely completed wave A within a larger A-B-C corrective pattern. Wave B can then unfold with several different admissible counts. Incorrect momentum signals, in conjunction with several equally viable wave interpretations, can make it extremely difficult to correctly identify the completion of wave B. The momentum indicator problem will be illustrated and described in detail when a new indicator is compared to RSI in Chart 7 and MACD in Chart 8.

Pros and Cons of Momentum Indicators with The Elliott Wave Principle

Robert Prechter and A.J. Frost write in the Elliott Wave Principle, “As waves are in the process of unfolding, there are times when several different wave counts are perfectly admissible under all known Elliott rules“ (p.68). They further add, “it is at these junctures that a knowledge of wave personality can be invaluable.” Supporting technical indicators, such as momentum, help the analyst identify the personality of specific waves in the

Introducing the Derivative Oscillator The Elliott Wave Principle provides the analyst

with a sequence which serves to map the current position of a market. The Elliott Wave practitioner knows when a pattern is complete but needs a warning when a more extensive pattern should be expected. In these situations fewer signals generated by a momentum indicators would be of greater value than numerous signals which serve to warn that a market is in the process of forming a top or bottom. Charts la, lb, and lc illustrate the DJIA in a daily time interval. The oscillator directly below the bar

MTA JOURNAL / WINTER 1993 SPRING 1994 45

CHART 1 a

chart is a Derivative Oscillator plotted as a histogram. The histogram is labeled CMB Momentum- Hist throughout charts la-lc. In Chart la, a traditional Momentum study is plotted by TradeStation

6-?--5-.-ii- I b--E ” -iJ r-r-r--z-i-z----

CHART 1 b

at the bottom of the page. First visual comparison of these two histograms shows the objective of developing a momentum indicator with fewer signals has been accomplished. The histogram directly below the bar chart shows very distinct and clean crest peaks and trough lows when compared to Momentum. Chart lb illustrates this similar comparison to the Commodity Channel Index. Chart lc demonstrates Rate of Change. In terms of visual clarity, the Derivative Oscillator shows promise.

The DJIA illustrates an upward trending market during October 1992 to February 1994 time frame of

CHART Ic

Charts la, lb, and lc. An indicator frequently demonstrates strength during a trending market or a sideways consolidation, but rarely offers significant value in both market conditions. ChartsBa, 2b, and2c illustrate a daily Cash S&P bar chart with the same indicators displayed in Charts la to lc. The Cash S&P market demonstrates a general upward trend inter- rupted by two large contracting triangles. The triangles have been marked on the charts. More traditional momentum studies give multiple signals in complex corrective patterns, therefore it was important to have an indicator which could offer value during consolidations and perform in strong trending markets where extensions are expected. The rough visual comparison seems to show promise in both market conditions. A more rigorous evaluation on the


profitability potential of the Derivative Oscillator was then indicated.

The Derivative Oscillator Formula The Derivative Oscillator is a triple smoothed

RSI. The formula incorporates two exponential moving averages and a simple moving average.

STEP 1: An exponential average of RSI is calculated.

STEP 2: The result in Step 1 is used to produce a new exponential average using a shorter period than that used in Step 1.

STEP 3: A simple moving average is then obtained from the result in Step 2.

CHART 2a

STEP 4: Obtain the difference between the results obtained in Step 2 and 3. The result is then charted as a histogram.

Omega TradeStation and SuperCharts both use the symbol ‘XAverage’ to denote an exponential moving average. A custom formula can be created for both these products in the following manner: (XAverage(XAverage((RSI(Close,length)),Periodl),Period2)) -

(Average(XAverage(XAverage((RSI(Close,14)),Periodl), Period2),Period3))

Using “input names” for the exponential and

CHART 2b

simple averages will allow easy adjustment to these variables. The DJIA and S&P markets should both use a 14 period RSI and the following periods for hourly and daily analysis:

CHART 2c


r Study InputName Default Value RSI Length 14 XAverage Period1 5 XAverage Period2 3 Average Period3 9

Telerate’s TeleTrac and the Snap module within CompuTrac would be setup in the following manner:

TYPE NAME DEFINITION COEF rsicoef 14 COEF period1 5 COEF period2 3 COEF period3 9 STUDY rsi RSI(last,rsicoef) STUDY exp-ma Exp-ma(rsi,periodl) STUDY exp-ma2 EXP_ma(exp_ma,period2) STUDY mov-avg Mov-avg(exp-ma2,period3) STUDY deriv-os Oscl(exp-ma2,mov-avg) (Plot the last line as a histogram.)

the current Derivative Oscillator was greater than the prior period, and the Derivative Oscillator calculation two periods back was greater then the prior period. This allowed a short position to be reversed to a long position when ever the oscillator began to advance changing a former down trend.

SELL: Exact logic used for buy signals but reversed.

ASSUMPTIONS: Always in the market.

STOPS: Neither trailing, percentage loss, nor profit stop losses were used to record actual gains and losses between oscillator signals. Trailing profit pro- tection stops and incorporating percentage stop losses would have significantly increased profits according to further test evaluation.

A Profitability Evaluation of the Derivative Oscillator for the DJIA

The preliminary visual examination of the Deriv- ative Oscillator showed distinct crests and troughs. A more thorough evaluation was indicated to test the actual profitability of the oscillator signals. Table 1 lists all the daily buy and sell signals generated by the Derivative Oscillator for the DJIA over the period of July 22,1992, to February 4,1994. The full evaluation covered a 20 year look back period in the DJIA, and a six year look back for the Cash S&P and OEX (S&P 100) markets. Table 1 is a detailed listing showing a portion of the test results.

A summary of the test results from 7122192 to 2/4/94 yielded a cumulative profit in DJIA of +269.49 points. The objective of this test was to determine if the oscillator showed some merit on its own before divergence analysis and other observations could be applied which would increase the indicators’ performance. The test results from the DJIA, S&P Cash, and OEX were viewed promising and justified further evaluation of the indicator.

The criteria used to generate the buy/sell signals follows:

BUY Enter a long position when the current period of the Derivative Oscillator is greater than the prior period of the Oscillator. More simply stated, go long when ever the oscillator changes direction and begins to advance. Instead of using the next day’s open to enter a long position, use the next day’s close to provide a worse case scenario to cover slippage, com- mission, and intraday confirmation after the signal has been triggered. No criteria was established to require an oscillator bottom within a specified range. A turn upwards in the oscillator at the - 10 level was given equal weighting as a signal generated at + .02. Once the system was long it could not add multiple positions when secondary signals where generated. The objective is to find an indicator which does not give several premature signals, therefore this restric- tion was incorporated. Indicator whipsaw had to be evaluated. This was accomplished by permitting short positions to be reversed to a long position if

The test results for the time period illustrated in Table 1 can be summarized:

Total Net Profit: +269.49 DJIA points. Gross Profit: 1162.72 DJIA points. Gross loss: -893.23 DJIA points.

The total number of winning trades: 31 out of a total of 71 signals.

The largest winning position: 141.29 DJIA points. The largest losing position: -57.02 DJIA points.

The average winning trade: 37.51 DJIA points. The average losing trade: -22.33 DJIA points.

The win/loss ratio is 1.68. (Remember there are no stops.)

The average winning position was 9 days in duration, losers averaged 3 days duration.

The complete test results from the period 7122192 to 214194 follow in Table 1 and then accompanying charts mark when each buy/sell signal occurred.

In the chart showing buy/sell signals from August 1992 to February 1993, the letters ‘A’ and ‘B’ have been marked on this chart. The very distinct sell signal at ‘A’ and then the buy signal at ‘B’ are not captured by the logic used to generate the black box test signals. One of the objectives of the profitability test was to test premature signals. The logic therefore does not permit multiple entry/exit signals. A few premature signals will be seen in the charts for Table 1, they are limited and final test


r CMB Momtm/3x3 DJIA.D-Dal ly 05/12/92 - 02/09/94 Date 'rime 'bw Cnts Price Signal Name Entry P/L Cumulative 07/22/92 Lwy 1 327-I. 61 07/27/92 LExit 1 3202.20 5 4.59 s 4.s9 07/21/92 Sell 1 3282.20 07/28/92 SExit 1 3334.0’7 s -51.07 $ -41.28 07/28/92 Buy I 3334.07 00/05/92 LExit 1 3365.14 5 31.07 $ -16.21 00/05/92 Sell 1 3365.14 00/10/92 SExit 1 3329.4Q 5 35.66 $ 19.45 08/10/92 uuy I 3329.40 on/19/92 LExit 1 3307.06 5 -22.42 $ -2.97 00/1.9/92 se11 1 3307 06 08/28/92 SExit 1 326’7 .61 5 39.45 $ 36.48 00/28/92 uuy 1 3267.61 09/08/92 LExit 1 3260.59 5 -7.02 5 . s29.46 09/08/92 Sell 1 3260.59 09/10/92 SExit 1 3305.16 s -44.57 5 -15.11 09/10/92 Buy I 3305.16 09/17/92 LExit 1 3315.70 5 10.54 $ -4.57 09/17/92 SC11 1 3315.70 10/12/92 SflXlL 1 3 114 .4 I s 141.29 s 136.72 10/12/92 n1ry 1 3ITi.41 10/19/92 LExlC 1 31110.45 s 14.04 5 150.‘76 10/19/92 Sell 1 3100.45 10/21/92 SExit 1 3107.10 5 1.35 $ 152.11 10/21/92 Buy 1 3107.10 10/30/92 LExit 1 3226.20 5 39.18 $ 191.29 10/30/92 Sell 1 3226.28 11/02/92 SExit 1 3262.21 5 -35.93 $ 155.36 11/02/92 Buy 1 3262.21 11/04/92 LExit 1 3223.04 5 -39.17 $ 116.19 11/04/92 Sell 1 3223.04 11/11/92 SExit 1 3240.33 $ -17.29 $ 90.90 11/11/92 my 1 3240.33 11./12/92 LIzxit 1 3239.79 5 -0.54 $ 98.36 11/12/92 Sell 1 3239.79 11/20/92 SEXit 1 3227.36 5 12.43 5 110.79 11/20/92 Buy 1 3227.36 12/03/92 LExit 1 3276.53 5 49.17 $ 159.96 12/03/92 Sell 1 3216.53 12/07/92 SExit 1 3307.33 5 -30.80 $ 129.16 12/07/92 Buy 1 3307.33 12/11/92 LExit 1 3304.08 $ -3.25 $ 125.91 12/11/92 Sell 1 3304.08 12/18/92 SExit 1 3313.27 5 -9.19 $ 116.72 12/18/92 Buy 1 3313.27 12/29/92 LExit 1 3310.04 5 -2.43 $ 114.29 12/29/92 Sell 1 3310.04 12/30/92 SExit 1 3321.10 $ -10.26 5 104.03 12/30/92 Buy 1 3321.10 12131192 LExit 1 3301.11 $ -19.99 5 84.04 12/31/92 Sell 1 3301.11 01/14/93 SExit 1 3267 .f30 5 33.23 $ 117.27 01/14/93 Buy 1 3267.88 01/19/93 LExit 1 3255.99 5 -11.89 $ 105.38 01/19/93 Sell 1 3255.99 01/25/93 SExit 1 3292.20 $ -36.21 $ 69.17 01/25/93 Buy 1 3292.20 02/11/93 LExit 1 3422.69 s 130.49 $ 199.66 02/11/93 Sell 1 3422.69 02122193 SExit 1 3342.99 5 79.10 5 279.36 02/22/93 my 1 3342.99 03/12/93 LExit 1 3427.82 5 84.83 $ 364.19 03/12/93 se11 1 3427 .02 03/10/93 SExit 1 3465.64 s -37.82 5 326.37

TABLE 1


CHB Homtm/3x3 DJIA.D-Daily 'X/12/92 - Date

02/09/94 Time Type Cnts

03/18/93 Price Signal Name

Buy 1 Entry P/L Cumulative

3465.64 03/24/93 LWtlt 1 3445.30 03/24/93 5 -20.26 Sell 1 S 306.11

3445.38 03/30/93 SExit 1 3451.27 03/30/93 s -11.89

Buy 1 s 294.22

3457.27 03/31/93 LExlt 1 3435.11 03/31/93 s -22.16 se11 1 $ 272.06

3435.11 04/12/93 SExit 1 3428.09 04/12/93 s 7.02

Buy 1 S 279.08

3428.09 04/20/93 LExit 1 3443.49 04/20/93 S Sell 1 15.40

3443.49 $ 294.48

04/29/93 SExit 1 3425.12 04/29/93 S 18.37 Buy

1 $ 312.85 3425.12

05/07/93 LExit 1 3437.19 05/07/93 s 12.07 se11 1 3437.19

$ 324.92

05/11/93 sExit 1 3468.75 OS/11193

s -31.56 my 1

$ 293.36 3460.15

05/14/93 LExit 1 3443.01 05/14/93 s -25.14 se11 1 3443.01

$ 267.62

05/19/93 SExit 1 3500.03 05/19/93 s -57.02 my

1 $ 210.60 3500.03

06/07/93 LExit 1 3532.13 06/07/93 s 32.10 se11 1 $ 242.70 3532.13

06/16/93 SExit 1 3511.65 06/16/93 S 20.48 Buy 1 $ 263.18 3511.65

06/18/93 LExit 1 3494.77 06/18/93 S -16.88 se11 1 $ 246.30 3494.77

06/21/93 SExit 1 3510.82 06/21/93 5 -16.05 Buy 1 $ 230.25 3510.82

06/22/93 LWtit 1 3497.53 06/22/93 s -13.29 Sell 1 $ 216.96 3497.53

06/28/93 SExit 1 3530.20 06/28/93 S -32.67 Buy 1 $ 184.29 3530.20

w/02/93 LExit 1 3483.97 07/02/93 S -46.23 Sell 1 $ 138.06 3483.97

07/08/93 SExit 1 3514.42 07/08/93 s -30.45 Buy 1 s 107.61 3514.42

07/22/93 LExit 1 3525.22 m/22/93

s 10.80 Sell 1

s 118.41 3525.22

07/23/93 SExit 1 3546.74 07/23/93 s -21.52 Buy

1 $ 96.89 3546.74

07/30/93 LExlt 1 3539.41 07/30/93 S -1.27 Sell 1 $ 89.62 3539.47

08/06/93 sExit 1 3560.99 08/06/93 5 -21.52 Buy 1 $ 68.10 3560.99

08/13/93 LExlt 1 3569.65 08/13/93 s 8.66 Sell 1 $ 16.76 3569.65

08/16/93 SBxlt.1 3579.71 00/16/93 S -10.06 Buy 1 $ 66.70 3579.71

09/01/93 LExit 1 3645.10 09/01/93 s 65.39 Sell 1 $ 132.09 3645.10

09/13/93 SExit 1 3634.20 09/13/93 s 10.90 $ 142.99 Buy 1 3634.20

09/17/93 LExit 1 3613.25 09/17/93 5 -20.95 se11 1 $ 122.04 3613.25

09/21/93 sExit 3567.70 09/27/93 S 45.55 BUY 1 s 167.59 3567.70

11/03/93 LExit 1 3661.87 11/03/93 S 94.17 se11 1 $ 261.76 3661.87

11/10/93 sEwit 3663.55 11/10/93 s -1.68 $ 260.08 Buy 1 3663.55

11/22/93 LExit 1 3670.25 s 11/22/93

6.70 S 266.78 Sell 1 3670.25

11/26/93 sBxit1 3683.95 S -13.70 $ 253.08

S3iB tdbmtm/3X3 LNIA.D-Daily 05/12/92 - Date

02/09/94 Time Type cnts

11/26/93 Price Signal Name

my 1 3683.95 Entry P/L Cumulative

11/29/93 LExit 1 3677.80 11/29/93 S Sell 1 -6.15 3677.80

$ 246.93

11/30/93 sExit 3683.95 11/30/93 5 Buy 1 -6.15 $ 240.70 3683.95

12/X/93 LExit 1 3116.92 12/15/93 S se11 1 32.97 3716.92

$ 273.75

12/17/93 sExit 1 3751.57 12/17/93 s my

-34.65 1 3751.57

$ 239.10

12/31/93 LExit 1 3754.09 12/31/93 s Sell 1 2.52 3754.09

5 241.62

01/05/94 sEx1t 1 3798.82 01/05/94 S -44.13 Buy

1 3798.82

$ 196.89

02/04/94 LExit 1 3871.42 Q2/Q4/94

s 72.60 Sell 1

$ 269.49 3071.42

50 MTA JOURNAL ! WINTER 1993 . SPRING 1994

,,...,.I

results show a positive return even without multiple entry signals.

The test results are not conclusive but are viewed a preliminary and important step taken during the development phase of the Derivative Oscillator. The effect of using different periods for the RSI and moving averages within the formula will be discussed at a later time in this paper.

TABLE 1: ~uylSell Signals lrom hn ‘93 to Aug ‘93 (signals are clmied nexl day on mark.9 close)

How to Interpret the Derivative Oscillator The profitability test results justified continued

development of the oscillator and observations within a real-time context of the oscillator suggest the following guidelines for its use and application.

Amplitude Equality The first observation of the Derivative Oscilla-

tor is the number of relatively few oscillator crests and troughs. Upon a more careful evaluation of these peaks and lows, an equal and opposite reaction can be expected. Let’s take a close look at Chart 3. In this chart a strong rally developed into the momentum peak labeled ‘C’. A peak extreme in the DJIA is considered any momentum value which exceeds + 10. When such occurrences develop, an analyst can expect a momentum low of equal magnitude in both hourly and daily charts. The momentum peak at ‘C’ has a peak high of + 11.89. A momentum low marked ‘c’ then attains an extreme value of -12.93 with


momentum lows on either side of the trough low near equality at -11.86. The momentum low does not have to immediately follow the momentum high as demonstrated between ‘C’ and ‘c’. However, after the extreme momentum low has been attained, the analyst can expect the next momentum high to be equal to the momentum high preceding the extremes. In Chart 3 the momentum low at ‘B’ has a value of -7.19 and the momentum high at ‘b’ has a value at the peak of +7.17. Again displaying near equality. The observer will find this attribute of the indicator occurs several times within the same chart. Ampli- tude equality is attained between momentum crests and troughs labeled at points ‘A’ and ‘a’, ‘D’ and ‘d’, and again between ‘E’ and ‘e’. The equality attribute provides a distinct objective for an analyst to mark a high risk pivot level for the market.

-_.--. ,iiaEs. --rziBziiiiiiiizFi, /_v..“,s_-< _

observer align important trend reversals in the oscillator with price. At these pivot levels, the oscillator reversed before a momentum high or low developed on the opposite side of the zero line. In each case marked ‘M’, the former near term price trend resumed with conviction. When the oscillator velocity begins to show no net gain/loss, or when the oscillator actually reverses its trend, a significant short term buy/sell signal can be realized as early as the next day in the market. The momentum peaks labeled 5,6, and 7, do not confirm the price highs and bearish divergence develops between oscillator and price. When a series of new price highs are accompanied by diminishing momentum peaks descending toward the zero line, a sharp price breakdown can be expected before the oscillator crosses the zero line. Each oscillator pivot between the peaks labeled 5,

In cF+ zx=im,B:R’i”;“I,,....,, ,-izii:;.iij--~---~--

I--

The characteristic of equal and opposite momen- 6, and 7, helped to warn that the developing market turn amplitudes is attributed to the market’s cycles. top was not in place. The peak at point 7 would be Further study is indicated. given a high probability of marking a top preceding

In addition to frequent momentum equality, a decline because the oscillator peaked marginally other analytic techniques such as divergence analy- above +2.00 and then proceeded to decline further sis should be applied to the Derivative Oscillator. In towards the zero line while prices stalled for two Chart 4, the momentum lows labeled ‘A’ and ‘B’ show days. In this example the price high associated with bullish divergence with the corresponding price lows the momentum peak at point 7 had completed an marked ‘A’ and ‘B’. A substantial rally then unfolded Elliott Wave pattern which fit the profile suggested in the DJIA. by the oscillator. Peaks 5,6, and 7 also display bear-

Three dashed lines are marked ‘M’ to help the ish divergence with the momentum peaks 1 and 4.


The momentum low in November marked point 3 has an equality relationship with the October momentum high at point 1. An analyst can suspect point 3 will be a high probability pivot level for the market based on the discussion for Chart 3.

Three small momentum lows are labeled ‘C’ in December 1993 to January of 1994. Shallow oscillator declines which have turned upwards frequently mark the end of a correction within a upwards trending market. Visa versa in a downwards trending market. In July of 1993, the price low on the far left of Chart 4 is accompanied by a shallow momentum low labeled ‘X’. Bullish divergence at ‘X’ occurs with the momentum low in June 1993 and the secondary momentum pivot in late June of 1993. This combination of a shallow oscillator decline and bullish divergence should be interpreted as a significant nearby market bottom.

The three momentum lows labeled ‘C’ also suggest horizontal support and resistance levels can develop. This is more fully discussed with Chart 5.

Chart 5 simply demonstrates that momentum peaks can form horizontal resistance levels. The momentum peaks at points 1,4, and 5 are repelled from the same horizontal resistance line. Point 3 shows a marginal breach of the resistance level, while point 6 failed just beneath it. When a momentum peak develops well above the horizontal resis-

CHART 5

tance level, an analyst should expect the market to at least develop a momentum peak again which tests resistance, as is seen in point 3 after the extreme momentum high in January. Secondary horizontal resistance levels will form which are closer to the zero line. Momentum lows can also develop horizontal support levels.

Double bottoms and double tops have been wit- nessed. However, diagonal trend lines drawn from peak to peak or from trough to trough provide little information of value. The single exception is when a consolidation is unfolding in the market. Chart 6 offers an opportunity to discuss this exception.

CHART 6

In Chart 6 the daily Cash S&P is displayed. A price consolidation causes the momentum peaks and troughs at points 1 through 6 to develop. These momentum points appear to form a distinct contracting triangle pattern. A contracting triangle pattern does not begin to develop in price until point 3. From point 3 to point 4, the Elliott Wave Principle suggests wave a down developed. The contracting triangle pattern observed in the Cash S&P price is considered complete at point ‘B’, where wave e forms a small contracting triangle itself. The Derivative Oscillator displays a series of contracting peaks and troughs which yield shorter trough amplitudes than the corresponding momentum peak amplitudes. The momentum low at point 2 is slightly shorter than that for peak

MTA JOURNAL /WINTER 1993 SPRING 1994 53

1, the same can be seen for the low at point 4 with the peak at 3, and similarly point 6 is shorter than peak 5. When a momentum consolidation develops in this formation, the analyst can expect the market to break in the direction of the longer momentum amplitudes, in this situation upwards.

The series of contracting momentum highs and lows can show an apex for the triangle which corresponds with the momentum trend reversal that occurs at point ‘B’. Point ‘A’ shows the first breakout through the upper trendline connecting the momen- turn peaks. Further research is suggested, but this breakout appeared to be a valuable early buy signal. At point ‘B’ the oscillator illustrates a trend reversal discussed in Chart 4. This trend reversal in the oscillator triggers a buy signal at ‘B’, and again at ‘C’. Point ‘B’ is the lowest risk entry for a long position. It is also the only point at which the momentum oscillator offers confirmation to a completed Elliott Wave pattern. A more thorough demonstration of how the Derivative Oscillator is used in conjunction with the Wave Principle will be offered shortly.

Similar momentum patterns have been observed with bearish resolutions. In these scenarios, the momentum peaks became progressively shorter than the amplitudes of the reaction momentum lows within the contracting triangle formation. A strong thrust downwards then followed just beyond the apex of the consolidation formation.

Chart 7 displays the daily Cash S&P with the Derivative Oscillator and a 14 period RSI.

At point ‘1’ the RSI displays bearish divergence with price. The difficulty caused by this divergence is that a five wave advance can be considered complete from the February 1993 low into the price high one bar preceding point 1. This premature sell signal would prove costly later as the market then rallied six S&P points toward the March high. In contrast to the RSI at point 1, the up trend displayed in the Derivative Oscillator had not begun to roll over. This strongly suggested that an extension to the simple five wave structure should be suspected and that the rally was therefore incomplete. The next period in the oscillator does rollover, but the internal wave structure in price is corrective. The potential for the oscillator to form the descending peak formation towards the zero line would be given consideration at that time.

At point 2 in price the Derivative Oscillator is forming a flat plateau below the momentum peak. This is interpreted as rapid market decay and viewed as a sell signal when scaling into size positions. Histogram plateaus do not develop in market consolidations. When the oscillator first

CHART 7

declines below the histogram plateau a fully leveraged short position should be established. The dashed line marked ‘S’ denotes the first pullback in the oscillator breaking the plateau in the histogram. This period in the histogram corresponds with the price high. The price high com- pletes a nine wave progression from the February low which satisfies the completion of a five wave advance with an extension. Histogram plateaus rarely lead to the descending peak pattern described earlier in Chart 6, or an ascending peak progression in a bullish formation. A plateau can be interpreted as a rapid loss of market velocity which will lead to a trend reversal shortly after the plateau formation is broken.

The RSI at point 2 does not display divergence until the next day. A sell signal can then be viewed as accurate in this indicator.

The RSI at point 3 is caught on the wrong side of the market when a six point decline unfolds in the cash S&P RSI divergence cannot be viewed until after point 3 and after the six point decline occurs. A costly late signal if point 2 was not acted upon after the premature signal at point 1 in the RSI.

This chart demonstrates the precise problem created by the RSI with the Wave Principle and why an alternative indicator was developed to offer fewer signals. The Derivative Oscillator does not replace


the RSI, but can be used to enhance the timing and warn of premature RSI signals.

Chart 8 displays the daily DJIA with the De- rivative Oscillator and MACD. The MACD differential shows some similarity to the Derivative Oscil- lator at first glance. There are several very important differences between the signals suggested by these two indicators.

At point ‘A’ in the Derivative Oscillator, the bullish divergence was described in Chart 4. The same time interval marked ‘B’ in the MACD does not display any divergence with price. The MACD signal to buy in early October is regarded as too late for use with the Wave Principle.

Point ‘C’ in the Derivative Oscillator shows an important price bottom in place. The MACD displays a shallow differential trough in its histogram, which could be viewed bullish, but the signal at point ‘X’ in the MACD is confusing and longer horizon investors would view the second crossover in MACD in late November as a confirmed intermediate sell signal. In this scenario establishing a short position at the signal marked ‘sell?’ would lead to a position on the wrong side of the market. A sell signal could be interpreted at the same point in the Derivative Oscillator as the histogram does roll over. The position is covered near the same price as the position was established when the histogram rolls back up. The profit test in Table 1

I”

i

CHART 0

shows this signal represented a loss of - 1.68 DJIA points, no loss occurs if the position had been reversed on market open rather than on market close.

Point ‘E’ in the Derivative Oscillator has already been described as short term bullish signals forming a horizontal trendline in the histogram. Point ‘F’ in the MACD is bullish by nature of a positive differential displayed, but does not suggest levels to add to existing positions, or at which levels to reestablish a long position if stopped out during the minor corrective declines.

Point ‘G’ in the Derivative Oscillator displays bearish divergence, while point ‘H’ in the MACD is considered confusing by this analyst. The MACD sell signal in early February lags the sharp market break which occurs in the fourth price bar from the far right hand side of the chart.

The MACD lag is viewed as a major obstacle for an indicator used in conjunction with the Wave Principle.

The Derivative Oscillator Applied Real Time

The original objective for developing an indicator with limited and timely signals was to compliment and support an analyst during complex Elliott Wave patterns, or to help warn the analyst when alternate patterns should be given a higher probability. A demonstration follows to show how the Derivative Oscillator is being used currently with hourly data of the DJIA and March S&I?

A market report is transmitted to subscribers each evening by Fax or through Reuters, Bloomberg, Telerate, and Knight-Ridder electronic quote screens. Five market reports follow which were transmitted to subscribers. The reports cover two important market turns for short term traders, the time horizon of interest for institutional subscribers to the World Stock Markets Outloolz”“, or Futures Outlook”“. The hourly charts for both the DJIA and March S&P are included with notations to help the reader identify where the market closed when each report was prepared.

The Derivative Oscillator is plotted below the bar chart for each market.

In the right hand margin, beside the market report, are analysis notes to guide the reader to important relationships between price and oscillator which had an influence to this analyst’s market opinion at that time. The combination of wave structure and oscillator was an invaluable combination for this analyst during this market transition and continues to be of value.


124th 127th list 13rdilth 19thillth 116th 122nd 124th 127th list 13rd l7thl9thlllth 116th 122nd

;2 Feb 94 09:53

, -. -. - -. , [nerlttance, 473.4,. 474.16. *75..59. 476.46. 477.01. i4??.?~-.92).48o.M).(485.~~ 851 ,Suppor,, 468.41. (456.92-467.02~ (461 91..99l. (458.04- 20). (45:.1’S23)

Analysis Notes For Monday’s Report,

Globally Transmitted: 2136 GMT, Friday 2/l 1194:

DJIA Polnf ‘1’ I” the asctllator rolled upwards

when the puce low at ‘1’ occurred.

The 3870.7 larger was a Fibonacct

oblectrve to complete an Elliotl Wave partern. Recording an actual price low of 3867.5 near the objectwe. and wth a

completed Elliott Wave internal sub- structure. rxreased the probability that a

shon term low was made on Frtday.

March S&P Powlt ‘1’ in the oscillator shows an upwards trend was developing when Ihe price low at ‘1’ occurred. Bullish divergence is present between the oscillalor low at point ‘1’ and the prior oscillator low from Feb. 7th compared to

corresponding price lows.

The momentum low corresponds wth a pmr mome.ylum low from the 24th.

The oscillator low at point ‘1’ has an equality

relauonship wtth the peak high on the 7th.

L


I 124th !27th list l3rdl7thl9thlllth 116th 122nd I 124th 127th list l3rd i7thl9thillth !16th 122nd J

22 Feb 94 09:53

Anaiysis Notes For Wednesday’s Report, Globaliy Transmitted:

CaJ:lO GMT, Wednesday 2/l&34:

DJIA The uptrend I” fhe osc~llafor has turned

down ar Point ‘2’ suggesting the current ra!ly IS only corrective I” character.

The Eliion Wave internal sub-struc:ure from point ‘1’ 10 ‘2’ is incomplete.

Therefore the momentum oscillator was suspected to be forming three dwerglng peaks as the osctllalor approached the

zero line.

March S&P 01stmct bearish divergence at osctilator points ‘2’ and the oscillator high on the

1 lth with corresponding prices warn that the rally will likely fall.


I 124th 127th list 13rd l7thl9thlllth 116th 122nd I I 124th 127th list l3rd l7thl9thlllth 116th 122nd I

FA..,‘nOoLmI*OluNE: ‘~(M~~J-GG”7-1*OURS:?1:00-Jc:00CL~. ANY rnOoLEMS AFTER TIICSC llouns SJ,O”LD nc DImcTED ‘ia: I.400-534.OF80

THE WORLD STOCK MARKETS OUTLOOK SM

Analysis Notes For Thursday’s Report. Globally Transmitted: 23:09 GMT, Wednesday Z/16/94:

DJIA AT PM! ‘3’ Ihe asc:llaror :s !ornu,g the

fhrd pew am con:mues !3 show a decllmng pean xagiess~an towards the zera I1ne. The 31cr: Wave Pnnc:ple req”mS One rcre qn !a cxmlele ;he pattern from me kw at pcinc ‘1’. A sell

stgnal IS then ::ans-mea for c!ienfs :n rhe firsr morning .ccale af:er !he nen

new hqn above 3955 is :ecsrbed

March S&P

hnf ‘2’ Snows The oscllafcr has alreaw crossed lhe zem iine. One mere raliy is expec:ea la ml mo as an apposne momentum hl9n fcrrrs with a s”oner ampliruae than lhat neasljred at pc~nt ‘3’.

The E!lion Wava Pnnwle suggesled cna more hlgn lo asfaclisn shon pcsn~cns. Wave c UD from point ‘i’ lormed a

diagonal wangle panem wh wave 5 of c

UP enaiog ar :he upper trend line.


124th 127th list 13rd l7thl9thlllth 116th 122nd

22 Feb 94 09:53

Analysis Notes For Friday’s Report, Globally Transmitted:

23:39 GMT, Thursday 2/M/94:

DJIA

At point ‘4’. the oscillator was s:~ll trending downwards and had nor begun

to roll up, supponing further losses.

March S&P Potnt ‘4’ in the oscillator shows the down trend is still un place.

Point ‘4’ in the osctllator has an exact

equality relatlonshlp with the momentum peak recorded on the 27th. While :he

oscillator has not turned upwards. the equality relatlonshtp warns thaf an

oscillator prvot could be near and the expectations for a five wave decline in price (from the 161h) could be incorrect.

Therefore, the osclllalor warned that a larger corrective pattern could be

unfolding and the next decline would end wave b down at an ideal target of 465.25 for this panem.

The nefl day 465.30 was the actual low.


r

124th 127th list 13rd 17th 19th 111th 116th 122nd

22 Feb 94 09:s)

Analysis Notes For Tuesday’s Report,

Globalfy Transmitted:

23:52 GMT, Friday 2/l&% (U.S. markets were closed Monday.)

DJIA

Point ‘5’ in the oscillator is equal IO pomt

‘1’ suggewng a momentum double bottom IS I” place.

In addltlon. the last two hours form a double bonom at precise momentum

lows suggesting an extreme oversold scenario for prices.

The 3879 objeclrve was precisety met suggesting the market was tracking lhe

proposed E!l~ott Wave pattern.

March S&P

Point ‘5’ shows momentum is clearly IrendIng upwards from point ‘4’. (Point

‘4’ had satisfied an equalify relalionship

wth the peak on the 27th.) Because of the relatlonship of equality the nexl obfec!rve for the oscIIIator would be near

the momenrum high just preceding point ‘2’, which is equal in amplitude to the momentum peak preceding the last

momentum extreme high. (See the first peak on the left hand side of the chart before the 24th.)

The 465.25 objec!ive was met with a significant retacement, supporting the scenario that wave b down developed.

60 MTA JOURNAL /WINTER 1993 SPRING 1994

Suggested Areas for Further Research

All developmental work has been directed to only three markets which are all correlated; the DJIA, the S&P Cash and Future markets, and the OEX. Only daily and hourly time intervals have been tested.

A quick visual glance when the indicator is applied to currencies and bonds shows different periods are required within the oscillator’s formula. The differences are likely cycle dependent. The relationship of period intervals and cycle interval should be given a more through evaluation. As an example, George Lane recommends using a period one half the market cycle when setting up Stochastics. This has pro- ven to be an accurate guideline. A concise interval relationship between the exponential averages and simple average period used within the Derivative Oscillator formula needs to be researched. In addition different RSI period intervals should be tested in conjunction with the oscillator. The RSI period was selected by performing an independent optimiza- tion for the markets and time intervals in question using very elementary buy/sell logic. Using a 14 period interval may not be the ideal period,

intervals such as weekly and monthly horizons should be thoroughly tested to support investment objectives rather than trading applications.

Wave International, located in GainesLsille, Georgia. She

intraday updates for institutional clients cia Reuters,

The Derivative Oscillator raises a very interest- ing area of research with the amplitude equality characteristic of histogram peaks and troughs. Pre- liminary research has revealed that amplitude can be more valuable than the number of periods sep- arating two signals with RSI also. Signal amplitude is a topic where little work has been recorded for the Technical Analysis community.

Conclusion The Derivative Oscillator shows promise as an

indicator to assist the Elliott Wave practitioner when complex wave patterns develop. This oscillator offers fewer signals than most common momentum oscil- lators used currently and it displays distinct, clear visual signals, which do not favor a buy or sell signal. The value of this indicator is apparent in both trending and sideways market action. The signals generated from daily and hourly time intervals can significantly aid the aggressive short horizon trader.

This paper shares with the reader attributes to help interpret the oscillator and offers a demonstration of how it is currently being used with profitable success. The oscillator attributes described in this work are offered as a solid foundation for further study. Performance results to date will certainly justify any time spent to add to the information shared at this time. However, additional research is strongly recommended before applying the oscillator to markets other than US. stock indices. Longer time


NOTES


journal of technical analysis (jota). issue 42 (1993, winter)

Economy & Finance

mta members

mta journal winter

mta library

mta journd

single issue of mta

monthly mta newsletter

market technlclans association

financial technical