Journal of Technical Analysis (JOTA). Issue 22 (1985, November)

MARKET TECHNICIANS ASSOCIATION

JOUR-N.AL Issue 22

November 1985

MARKET TECHNICIANS ASSOCIATION JOURNAL Issue 22 November 1985

Editor: Henry 0. Pruden, Ph.D. Adjunct Professor Golden Gate University San Francisco, CA 94105

Manuscript Reviewers: Arthur T. Dietz, Ph.D. Professor of Finance Graduate School of Business Administration, Emory University Atlanta, Georgia

Frederick Dickson Portfolio Manager Millburn Corporation New York, New York

Richard Orr, Ph.D. Vice President for Research John Gutman Investment Corporation New Britian, Connecticut

David Upshaw, C.F.A. Director of Portfolio Strategy Research Waddell and Reed Investment Management Kansas City, Missouri

Anthony W. Tabell Technical Analyst Delafield, Harvey, Tabell Princeton, New Jersey

Robert A. Wood, Ph.D. Associate Professor of Finance Pennsylvania State-University State College, Pennsylvania

Printer: Golden Gate University 536 Mission Street San Francisco, CA 94105

Publisher: Market Technicians Association 70 Pine Street New York,. New York 10005

MTA Journal/November 1985 1

MTA JOURNAL--NOVEMBER 1985

TABLE OF CONTENTS

FROM THE EDITOR: A SPECIAL CALL FOR PAPERS.................

MTA OFFICERS AND COMMITTEE CHAIRPERSONS.....................

MEMBERSHIP AND SUBSCRIPTION INFORMATION. . . . . . . . . . . . . . . . . . . . .

STYLE SHEET FOR SUBMISSION OF ARTICLES......................

LOOKING BACK/LOOKING AHEAD Robert J. Farrell.......................................

WE'VE LASTED FOR 2500 YEARS. WILL WE SURVIVE THE NEXT TEN? David D. Halt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

A SIMPLE APPROACH TO VOLATILITY Richard C. Orr, Ph.D....................................

THE DOUBLE POWER SCALE John R. McGinley, Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7

14

23

32

CONSTRUCTING ADVANCE/DECLINE AND SYNTHETIC PRICE LINES FOR DIVERGENCE ANALYSIS OF THE CAPITAL SECTOR OF THE TREASURY MARKET

R. Bruce McCurtain . L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

MOVING AVERAGES AND THEIR VARIATIONS John Carder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48


MTA JOURNAL

SPECIAL CALL FOR PAPERS

Sentiment: Concepts, Indicators, Contrary Opinion

Has the whole concept of sentiment slipped away from the grasp of technicians and become lost in the shadows? Should the traditional indicators of sentiment based upon short-selling behavior be relegated to a dark corner of the technical "war room"? Is Contrary Opinion now so popular that it is no more than a cliche?

The MTA JOURNAL wishes to devote an entire issue to exploring the problems, and the possible solutions to the problems, besetting technical concepts of sentiment, technical indicators of sentiment, and the uses and abuses of Contrary Opinion. The JOURNAL seeks a wide variety of articles and notes shedding light into the sentiment area. Outstanding, full-length articles are sought, but so too are one or two page notes and book reviews. Moreover, since market sentiment is the lively, amusing corner of technical analysis, humorous incidents, anecdotes and cartoons are often illuminating and so shall be welcomed.

Please direct your manuscripts or your inquires to the Editor, MTA JOURNAL, Box 1348, Ross, CA 94957, or telephone (415) 459- 1319. If enough materials are on hand by next summer, then the November, 1986 issue of the MTA JOURNAL could become our Special Edition.

Thank you,

Pruden, Ph.D. Editor, MTA JOURNAL


1985-86 MARKET TECHNICIANS ASSOCIATION

OFFICERS

PRESIDENT SECRETARY Gail Dudack John Murphy Pershing/Div. DLJ JJM Technical Advisors 212-312-3322 212/724-6982

VICE PRESIDENT Cheryl Stafford Wellington Managment 617/227-9500

TREASURER David Krell New York Stock Exchange

VICE PRESIDENT (Seminar) Robert Simpkins Delafield, Harvey, Tabell 609/924-9660

COMMITTEECHAIRPERSONS

PROGRAMS Robert Colby 212/399-6002

ETHICS & STANDARDS/PUBLIC RELATIONS Tony Tabell 609/924-9660

NEWSLETTER Robert Prechter 404/536-0309

PLACEMENT John Brooks 404/266-6262

JOURNAL Henry Pruden 415/459-1319

- . EDUCATION

Fred Dickson 212/398-8489

ACCREDITATION COMPUTER SPECIAL INTEREST GROUP Charles Comer & John Brooks John McGinley 212/825-4367 404/266-6262 203/762-0229

MEMBERSHIP Phil Roth 212/742-6535

FUTURES SPECIAL INTEREST GROUP William Byers 212/952-6651

LIBRARY Ralph Acampora 212/510-3750

MTA Journal/November 1985

MARKET TECHNICIANS ASSOCIATION MEMBERSHIP AND SUBSCRIBER INFORMATION

ELIGIBILITY: REGULAR MEMBERSHIP is available to applicants --- "whose total professional efforts are spent practicing financial technical analysis which results in an identifiable research product that is either made available to the investing public or becomes a primary input into an active portfolio management process." (From revised Constitution)

ASSOCIATE MEMBER status is ---7 "reserved for professional users of technical analysis (i.e. money managers, traders, brokers, floor specialists, etc.) who are not engaged primarily in technical research, but for whom technical analysis is the basis of their decision-making process." (From revised Constitution)

SUBSCRIBER category is available to individuals who are 7--------y interested in keeping abreast of the field of technical analysis, but who don't fully meet the requirements for regular or associate membership. Privileges are noted below.

Application Fees: A one-time application fee of $10.00 should accompany all applications for regular and associate members, but not for subscribers.

Dues: Dues for regular members, associate members and subscribers are $100.00 per year and are payable upon receipt of dues notice in September each year. ----------------_------------------------------------------------

Regular Associate

Invitation to Monthly MTA Educational Meetings

Members Members Subscribers

Yes Yes Yes

Receive Monthly MTA Newsletter Yes Yes Yes

Receive Tri-Annual MTA Journal (Nov-Feb-May) Yes Yes Yes

Use of MTA Library Yes Yes Yes

Participate on Various Committees Yes Yes Yes (Exceptional membership)

Eligible to Chair a Committee Yes No No

Eligible to Vote Yes No No

Fee Discount - MTA Annual Seminar (May) Yes Yes Yes

Annual Subscription to the MTA Journal ONLY -- $35.00 per three issues. Single Issue of MTA Journal (including back issues) -- $15.00 each.


STYLE SHEET FOR THE SUBMISSION OF ARTICLES

MTA Editorial Policy

The Market Technicians Association Journal is published by the Market Technicians Association, 70 Pine Street, New York, New York 10005 to promote the investigation and analysis of price and volume activities of the world's financial markets. The MTA Journal 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 Association and registered with the Library of Congress. All rights are reserved. Publication dates are February, May, and November.

Style for the MTA Journal

All papers submitted to the MTA Journal are requested to have the following items as prerequisites to consideration for publication.

Short (one paragraph) biographical presentation for inclusion at the end of the accepted article upon publication. Name and affiliation will be shown under the title.

All charts should be provided in camera ready form and be properly labeled for text reference.

All tables should be properly labeled and in camera ready form. I ,

Paper should be submitted typewritten, double-spaced in completed form on.8 l/2 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 references should be put at the end of the article.

Greek characters should be avoided in the text and in all formulae.

Two submission copies are necessary.

Manuscripts of any style will be received and examined, but upon acceptance, they should be prepared in accordance with the above policies.


LOOKING BACK/LOOKING AHEAD

Robert J. Farrell

It is a pleasure and privilege to be the introductory speaker at this 10th MTA Seminar. I am proud of the progress the MTA has achieved since a group of us got together under our equivalent of the Buttonwood tree in 1972 to found a professional union of technicians. From that beginning nucleus of 20 the MTA has grown to 160. From a more limited geographical concentration in New York and Boston, we have gone national and now international. This is progress. But we have more than size and impressive seminars to be proud of. We have fostered professionalism. We have introduced a code of ethics. We have increased understanding through dialogue and communications including our monthly newsletter and the prestigious quarterly MTA journal. And thanks to Ralph Acampora and others, we have an excellent library of technical books and manuscripts.

The MTA, of course, means different things to different people. To Bostonians, the MTA is the Metropolitan Transit Authority. To Transylvanians, the MTA is the Morticians and Taxidermists Association. To perennial bears the MTA stands for more trouble ahead; to those who have been frustrated by a trading range market it stands for my tortuous art. But for those of us who believe in the usefulness of technical analysis it might best stand for, "the market tells all."

We have accomplished much of what we set out to achieve 13 years ago. Most of you may not remember that technicians were second class citizens in Wall Street for many years. Merrill Lynch once ran an ad back in the 1960s headlined "OK Charlie put those charts away". Fortunately, my name was not Charlie. Technical Analysis was relegated to the back of the bus. Today we still have to deal with academic criticism and references from the press about crystalballgazing and fortune telling but we have achieved a significant degree of professional respect and recognition. It is important that we preserve this professional standing by answering our critics and not condoning shoddy work from members of our organization. If we forget the mistakes of the past, we are condemned to repeat them, is a line we all know well. We have to look ahead to the new challenges and constantly reassess our role in the investment process so that we may be more, rather than less effective. Organizations and people arelike markets . . . they thrive and grow until they reach maturity and then decay of themselves. To prevent the process of aging it is important to focus on the changing environment and the organization's objectives. This is what I would like to spend some time on this evening. But first, I would like to relate an appropriate story --

A Southern Congressman received a letter from a constituent demanding to know, how do you stand on whiskey?

MTA Journal/November 1985 7 c

Knowing that his constituency was divided on this volatile subject, the Congressman formed his reply as follows:

"I had not intended to discuss this controversial matter at this time. However, I have never shunned a controversy and I will take a stand on whiskey. Here is how I stand:"

"If, by whiskey, you mean the devil's brew, the poison scourge, the bloody monster that defiles innocence, dethrones reason, destroys the home, creates misery and poverty -- yea, literally takes bread from the mouths of little children -- if you mean the evil drink that topples godfearing men and women from the pinnacles of righteous and gracious living into the bottomless pit of degradation, despair, shame, helplessness, and hopelessness -- then I am against it with all my power!

"But," the letter went on, "If when you say whiskey, you mean the oil of conversation, the philosophic wine, the elixir of life, the ale that is consumed when good fellows get together, that puts a song in their hearts and laughter on their lips, and the warm glow of contentment in their eyes -- if you mean the stimulating sip that puts a little spring into the step of an elderly gentleman on a frosty morning -- if you mean the drink that enables man to magnify his joy and happiness, and to forget, if only for a little while, life's great tragedies, and heartbreaks and sorrows -- if you mean that drink, the sale of which pours into our treasuries untold millions of dollars that provide tender care for our little crippled children, our blind, our deaf, our dumb, our pitiful aged and infirm -- to build highways, and hospitals, and schools -- then certainly I am in favor of it. This is my stand."

"And I will not compromise."-

We all have learned tohedgelikethe good Senatorbutour future in this business is going tobe determined by our ability to help our clients make correct money making decisions. Confusing rhetoric is just as counterproductive as overstated claims of accuracy and infallibility. We have to look at the changing environment and decide what our contribution is going to be.

The major change in the analysis of markets is the expansion in the number of financial instruments available to investors and speculators. This is revolutionizing our business by increasing the variety of ways to hedge positions and directing interest to short term forecasting. Many of our traditional indicators are no longer dependable. Our new indicators need seasoning to be trusted. We are at an important crossroad. In adapting to changing markets and the reality of the new financial instrument world, it is necessary to reassess.our value in the investment process.

The focus point of most financial future's trading is short term. As indicators of short term sentiment, future's premiums and


option's put call ratios are gaining in value. But the risk is that we will become too short-term oriented in an investment world needing long-term guidance. My impression of the non- technical world's view of those of us who are practicing technicians is that our time horizon is measured by days and weeks rather than months, years and long-term cycles. The new emphasis on financial instruments only shortens the days and weeks to hours and minutes. We must not let our potential important contributions to long-term theme investing be overshadowed by this overemphasis on short term instruments. I find it amusing when an institutional client or research associate will refer to our work as mostly short term and the economist or fundamentalist as offering the truly long term insight. In fact, over the years what success I have had, has been in recognizing major long term trend changes leading to new trends that lasted many years. Yet many cannot get away from the stereotype of the technical analyst as only a short term trader ready to change his opinion with the next tick. I have been very conscious of a major overlaying inflationary cycle for at least the last six years and it permeates my whole approach to the market.

Yet, the supposedly long term oriented economists and fundamentalists change their forecasts regularly with each new quarterly GNP forecast. The period ahead should be especially challenging in this regard.

The one long term trend of which I a.m.convinced is that the public is again going to become more involved in the market through increasing ownership of equities. It will be involuntary and indirect at first as lower interest rates force money out of money market funds, CDs and savings accounts.

There is over a trillion dollars that has been shifted around among short term cash equivalents to take advantage of the high returns. The historical odds are that these high returns will not continue and the public will gradually shift to longer term financial asset alternatives. The tax laws which encourage savings through IRA and 401K programs also motivate the individual investor in this direction. Initially, the equity alternative of preference will be third party management particularly mutual funds because the public is confused and wants to avoid risk. In time, however, today's successful investor will become tomorrows speculator. If we can help the individual investor make the right early choices again by stressing the long term themes and trends, we could do a great service.

To illustrate how low public interest is in the equity market, we have only to look at what kind of business the average retail broker is doing today. In our firm today most of the brokers say equities account for no more than 25% of their production. We have a whole new generation of young brokers who have little knowledge of equities at all. They are selling bonds and other


products because their customers also have little interest in stocks. As an example of how low the interest level has sunken, I attended a conference for 400 of the mostsuccessfulbrokers in our company in San Francisco recently. In addition to the prevailing consensus that most did much less equity business than in the past, I ran into three different brokers at a social function, introduced myself as Bob Farrell from N.Y. and they responded by asking me what I did. I do not have a big ego but it told me something of the level of excitement about the market. For you Contrarians this has to have bullish implications for stocks.

It is also apparent that we have a tremendous educational job ahead of us. We have a whole new generation that has to learn that you need patience to make money consistently in the stock market. And unfortunately, it is almost a sure thing that those staying away from equities today are destined to be excited buyers atmuchhigherprices a few years from now.

Despite this, when I look at all the potential demand for equities in coming years from the cash rich public, pension funds and foreigners, and add in the reduction in supply arising from record corporate share repurchases ($84 billion in 1984 and an estimated $30 to 40 billion in 1985) plus increased popularity of indexing among pension sponsors (an estimated $100 billion is indexed) which locks away stock, the case for another big market rise in the next few years is quite persuasive. I could even make a scarcity case for stocks but hesitate because that only happens at tops. Valuation parameters and low equity ownership figures for the public and institutions reinforce this view. The pendu lum will swing from disinterest to excitement once again.

The important point for us all is to have a sense of history and to see where we fit into the picture. It is possible to go through a 10 or 20 year bull markets and not make money if we only buy when the market is exciting and going up and stay out when it is correcting and going down. It is possibletolose money if we are always buying sectors that are most popular (yesterday's stocks) and avoiding the out of favor stocks. We have to be awareofthe long-term trends and usethemtoouradvantage. Our whole focus is being directed to short-term trading due to financial futures, shortened capital gains holding period, a possible flat tax which will tax short-terms gains similar to long-term gains; performance pressures in the institutional area and reduced transaction costs. One correct long-term theme or sector decision in my opinion is worth a dozen correct short-term decisions. Yet, we are being lured into the dissection of the markets short-term trading swings by all these factors. In the end, high turnover is a self-defeating process.

Large pools of'money are managed successfully by recognizing a theme or making big sector bets like energy in the late 1970's or consumer growth in recent years. The crowd wants a number (what level is the DJIA going to) and a date (when will it get there).

MTA Journal/November 1985 c

10

The old saw says give them a number or give them a datebutnever both. I say neither is relevant to investment success. -The world is grey far more often than it is black or white. In the grey world having the rightsectors,groups or stocks is far more important than the next intermediate peak or trough level of the DJIA. Another thing, there are very few people in the Forbes list of 400 richest people in the U.S. that got there because of their stock investments. The ones that did, like John Templeton and Warren Buffet got there because they were long-term theme or value buyers. Technical analysis is amenable to this same long- term concept of investing. It is up to us to recognize this and help our clients to be better long-term investors, or they too will only buy when stocks are advancing and exciting and wind up not making money.

Are there industry trends that are going to affect the way we do business? The buy side has eliminated many technical positions for one reason or another. On the sell side, there are still ample technical positions but the number could decline as firms consolidate or disappear. The trend in recent years has been for technicians to go off into independent advisory positions on their own or to join the ranks of those managing money. With third party research payments now under question as pension sponsors seek to recapture some of their commission flow, the institutional advisory business is likely to be contracting. The independent market letter writers may always have a market but live in a precarious cyclical world dependent on the success of their last call and the stage of the market cycle. It seems to me that while certain nitches will always have to be filled, there is going to be an increasing number of technicians going from advisory work to money management. The greatest need for public investors is in the area of informed counselling on asset allocation and money management. The advisory business is helpful but in the old saying, you can lead a horse to water but you can't make him float on his back. The public needs knowledgeable money management and asset guidance and the opportunities are unlimited for market technicians with their knowledge of markets to fill this demand.

In the 1976-77 MTA Seminar my theme was flexible analysis was going to be more valuable than mechanical technical models due to the changing environment. To quote what I said at that time, "At Merrill Lynch we do not consider market analysis a science. In fact, I have found the more exact I try to become or the more minutely I examine date, the more the results are closer to chance. Our greatest problem is integrating the discipline of mechanical approaches of analysis effectively with the subjective assessment of what will be different in the present bull or bear cycle. The latter involves the study of the long-term cycles or trends in equity popularity".

The indicators of course do not always tell a story and some are more useful than others. .People are always asking which is the best indicator. All of us have biases on what we consider most


important, but we allknowthatno one thing works all the time. There is in effect no prettiest girl. Or, the prettiest girl last time is not likely tobetheprettiestgirlthis time. When we all clue in on the one sure thing approach, then it is sure not to work or new parameters will develop. This is the nature of the market place." This has proven correct as so many indicators have been distorted or changed in the intervening period.

For those with economic models, we saw behavior change as we went from an inflation world to a disinflation world. All the models based on the inflationary trends of the late 1960's and 1970's lost much of their relevance in the disinflationary 1980's. For those with technical models the appearance of financial instruments such as options and futures, has brought a new dawn. Odd lot figures have been obsolessed as trading has become predominantly round lot; the short interest has been distorted by arbitraging, interest sharing and option hedging; margin debt has been altered with the inclusion of bond debt; specialist short sale figures have been distorted by option related activity; the Merrill Lynch margin account sample has been affected by CMA accounts shifting cash accounts to marginable accounts; volume has been distorted by institutional dominance that has led to even higher average size of trades and futures programs that buy and sell baskets of stocks only to capture premium excess; institutional cash samples have been biased by lengthening maturities of cash equivalents and promise to be further changed as futures hedging takes the place of raising cash. The list of indicators with changing parameters is almost endless and if it were not for the appearance on the scene of many new indicators particularly by those related to options and futures, one might despair of being able to make confident strategy decisions and forecasts at all. It is a touch of irony as well that the markets are undergoing these changes that make mechanical modeling most hazardous just at a time computers are beginning to dominate the field of market analysis. Computers are important tools but they can give a false sense of certainty in an increasingly uncertain world. It is beginning to appear that technical analysis is becoming too popular as computer models proliferate and stress short-term results. We should not promise more than we can deliver. We should remember that technical analysis in most forms is not the prime mover of stocks. Someone with knowledge has to take the steps to make the footprints that we try to follow.

In the long run, the more complex the market becomes the better off we will be keeping it simple. Breadth, rate of change analysis, relative strength, supply demand studies, monetary analysis, and common sense intuitive thinking will survive as the analyst's basic tools. Long-term trend analysis will be equally crucial. While many whipsaws develop over the short run as the crowd responds to sudden news or chart breakouts, long-term tops and bottoms look no different than they did 50 years ago. Keeping our analysis simple and having a correct sense of long-


term trends and sector themes may be the key to our effectiveness in this accelerating future shock world.

The intuitive aspect of our analysis gains in importance as quantitative analysis becomes more endorsed by the crowd. Bennett Goodspeed made a valuable contribution to understanding the role of right brained intuitive thinking in the investment process in his book entitled The Tao Jones Averages -- A Whole Brained Approach to Investing. -Hxmessage was thatspotting chanae is essentianv visual and falls within the domain of the d

right brain, the intuitive side. He believed that each person has the potential for greater wisdom if he treats intuition as an equal partner with logic. Edward Johnson, (Mr. Johnson), of Fidelity likewise said, "I know this is no science. It is an art form. It is personal intuition, sensing patterns of behavior." I have to endorse that conclusion. Going back to my 1977 conclusion, we cannot afford the luxury of being mechanical in a constantly changing non-mechanical world.

The purpose of this 10th Annual Seminar is both a retrospective of past trends and an attempt to gain some perspective of the future. It is a well planned program that should cause all of us to think a little deeper into our roles as technicians, market analysts and investors. We should try to relate to each other and understand the problems of change without becoming inbred non-thinking crowd. Enjoy these next few days and have an open mind. Experience has taught me when I stop learning I stop enjoying one of the most fascinating and rewarding professions in the world -- the study of markets.

This article is based upon the keynote address to the 10th Annual MTA Conference, Hilton Head, North Carolina, May, 1985.

Mr. Robert J. Farrell is Vice President-Chief Market Analyst, Market Analysis Department, Merrill Lynch.


WE'VE LASTED FOR 2500 YEARS. WILL WE SURVIVE THE NEXT TEN?

David D. Holt

In approximately 500 B.C., Confucius said, "Study the past if you would divine the future."

As far as I have beenabletodetermine, this makes Confucius the first market technician, and, by far, the best known.

During the intervening 2,500 years the practice of studying the past to determine the mathematical probabilities of future market direction has been successfully employed by technical analysts all around the world. And yet, their methodology has experienced more changes than perhaps any other analytical process having to do with the securities markets.

From the first recorded results of barter trading (the origin of the auction and dealer marketplaces of today), through the buying and selling of options in the days of the Roman Empire, 16th century tea trading in Japan, the tulip-bulb boom and bust in Europe to the "big crash" in the United States securities market, the market technician has been the epitome of survival by displaying a remarkable ability to adapt to the environment within which he must operate.

He has been especially successful in surviving what was undoubtedly the most evolutionary decade of his lifetime-- the last ten years.

I _

And yet, I cannot help but wonder after all the battles he has won, will he be alive and well ten years from now. As dramatically efficient as the evolutionary process has been in recent years, it stands to pale by comparison to the potential changes rushing towards us as we meet here today.

Let's review the immediate past so we can solidify the changes that have already occurred, why they changed, what were the results, as well as the remedies. We can then compare them with the potential change coming to see just how we, as market technicians, will be affected.

In other words, study the past to divine the future.

More V and V ----- (Increased Volatility and Volume)

It is difficult to imagine anyone involved in our markets who doesn't recognize the tremendous increase in both volatility and volume in most of the marketplaces technicians follow. Contrary

M!l?A Journal/November 1985 14

--

to popular belief by the non-believers, statistics don't lie. They can be distorted, bent, misused, and even misinterpreted. But, unless they are materially altered, which, of course, removes them from the category of statistics, they tell their own true story in a very clear and understandable way.

Take volume on the NEW YORK STOCK EXCHANGE as an example. Perhaps the most infamous day of all was "Black Tuesday". .As dramatic as that one-day comparison is, a much clearer picture of the explosion in volume can be seen when you compare recent levels to those in existence just ten years ago, on a trend basis.

After the most disastrous six-year bear market in history (1968/1974), NYSE Volume was averaging (10% exponential) around 23 million shares in December of 1974. From this primary low, prices appreciated in a steady uptrend until July, 1975 at which point the volume trend has increased to almost 25 million shares. Except for a two-month penetration during February and March of 1976, the 25 million level held until the second quarter of 1978. The 50 million barrier was exceeded in early 1980, and both 75 and 100 were surpassed during 1982. The high for the NYSE 10% volume trend was established on February 7, 1985 at 130.3, and is currently slightly over 106 million shares.

As we see it, there are four principal reasons for the recent volume explosion:

1. Performance "now" syndrome of institutions

2. Repositioning of asset dollars among market segments and industries (part of Number 1)

3. Massive arbitrage and hedging programs

4. Offshore investment capital

The first contribution to volume expansion is a direct result of the lesson learned during the early 70's by portfolio managers that one-decision stocks are both prudent and desirable -- even if it's only for three weeks. Decisions may still be based on fundamentals which are projected into the future, but today's manager, with his low cost base, sophisticated analysis input, and keen sense for survival will not sit on a position turning sour. Where it used to be, "wait for it to curdle", it's now, "when it starts to smell, throw it out".

Repositioning of assets is prevalent today not only because technicians have shown managers how to keep track of the segmentized inflow and outflow of investment capital, but because there is no viable alternate market that can absorb their billions of dollars. So, instead of leaving the securities markets, they merely move it from one segment to another. Yet another form of answer to the question, "What have you done for


me lately?

The massive arbitrage and hedging programs of the institutions have been well documented during the past several years. So, I won't belabor you with yesterday's news except to state my opinion that as long as the number of dollars in circulation continues to increase in almost exponential proportions, so will the size and frequency of institutional arbitrage and hedging programs, which, of course, guarantees the continued expansion of volume. Because of the political clout of those who control these pools of investment capital, don't look for this newest game in town to be legislated out of business.

As far as offshore capital coming into our equity market is concerned, it also is a self-fulfilling prophecy. The more capital there is that has to be placed in the most advantaged place, the more it will be forced into U.S. securities, which is the last marketplace in the world with the liquidity to absorb such massive amounts of capital. The doom-and-gloomers get a lot of ink when they give their scenarios about how our federal deficits will cause a total collapse of our government, which, in turn, will destroy western civilization as we know it. And yet, the people who control large pools of capital know that the United States will be the last domino to fall, and logically conclude that it, therefore, shouldn't enter their day-to-day decisions. They make their search through the total spectrum of alternatives and invariably end up with the U.S. securities markets as their most viable and liquid source for investment.

When you are looking for a cause for the dramatic increase in volatility, you end up with basically the same four reasons that produced the increase in volume. The cause and effect are obviously intertwined to the point where they become almost identical.

The only additional observation I will make on volatility is that the arbitrage and hedging by institutions are the culprits for the increase of short-term volatility only. The "quick-hook" and “move-i t-around" policies of the institutions have produced the intermediate-term increases.

Numbers Lie (Distortions in statistics from historical norms)

It is a generally accepted fact that some of our oldest and previously most reliable indicators are no longer working as well as they used to. The statistical base is still the same, but the number-crunching results are different. And, as is always the case, when you have a "first time" result, you don't have anything to compare it with. This makes your conclusion suspect at the least, and invalid at the most.

I would like to touch on several areas of statistics that appear to have fallen victim to evolutionary change.


Odd Lots --

There appear to be a multitude of reasons why these statistics are not as useful as they were in the past.

Perhaps the most widely accepted theory is that the traditional odd-lotter has diverted his small capital base to the derivative markets. Instead of buying 10 shares of a $40 stock, he buys two call contracts on the stock for the same initial cashflow of $400. If not stock options, it's the cash settlement options and futures that attract, and, in many cases, consume his capital.

Several other reasons for the demise in odd-lot statistical performance are also fairly well recognized by technicians.

You now have the rather common practice of member firms matching and melding their customer's orders rather than passing them on to odd-lot brokers to process.

Another change that has materially altered the odd-lot picture is the tendency for the participants in automatic investment plans, investment pools (clubs), IRA's, etc. to go either into mutual funds or derivative securities.

Short-sales

Another old standby of technicians that is falling on hard times is the various statistical categories of short-sales.

One of the primary changes in public short-sale activities is the adoption of options as a viable alternative to short-sales for bearish strategies. It isn't difficult to build up a rather imposing list of advantages of options over short-sales all of which are well known to you. The inconsistency in this logic appears to be a lack of acknowledgement of the corresponding list of disadvantages. Be that as it may, put options on individual stocks, although not as nearly liquid as calls, are moderately popular as evidence by the current open interest for stock put options, which has exceeded the 1.5 million contract level. If you multiply that by the normal size of 100 shares per contract, you get a better idea of just how much participation it has attracted.

While we are on the subject, we might as wellsuggestto you that the recent surge in option volume by not being reflected in open interest is proof positive of just how large the arbitrage and hedging activities have really become. It is not at all uncommon to find less than 25% of all option volume being converted to an increase in open interest.

There are a number of option indicators that had great promise at one time or another that are no longer viable because of the ability of the players to totally box out their positions, and,

MTA Journal/November 1985 L 17

in many cases, totally eliminate their risk exposure, the number- one drawback of what has become perhaps the most efficient market place of them all.

Incest (Uncontrolled proliferation)

Traditional boundaries of the various marketplaces are coming down everywhere. Even though the natural instinct to protect ones "turf" still runs high, the interbreeding is almost out of control.

An example of the blurring betwe.en the auction markets of the exchanges and the dealer network of the OTC market is a recent decision by the SEC allowing each exchange to trade up to 25 OTC securities. It's true this is a test, but so was the 25 stocks the CBOE were allowed to trade options on in 1973.

Don't think for a minute that the problem of listing fees will keep the exchanges from trading OTC stocks. Internal competition as well as from foreign marketplaces virtually guarantees that the exchanges will not forego this opportunity to generate new business.

The logical question then becomes "will authorized off-floor trading not be too far behind?" The SEC has already gone on record that they "haven't forgotten" the relationship between off-board trading restrictions and unlisted trading privileges.

NASD members already have been approved to make markets in roughly 900 NYSE stocks even though they account for only 10% of the volume in these 1973 stocks. The logical conclusion as we see it is that if the exchanges capture a significant market share of OTC securities, the Intermarket Trading System (ITIS), which is currently the exclusive property of the exchanges, will merely be expanded to include NASDAQ. The inter-marriage will then be complete between the equity specialist/market maker marketplaces.

The SEC has also sanctioned a one-year pilot program of side-by- side trading in options and their underlying stocks. This is another controversial but "door-opening" form of incest, scheduled to begin on January 20, 1986. It isn't restricted to the five options exchanges, since it also applies to NASD and other regional exchanges as well.

Perhaps not earth shattering in relative magnitude, but the recent entrance by NASD into the cash settlement options market is still another traditional barrier that has been removed.

Inter-marriages are not restricted to just the equity/options markets. The CBOE has already started proceedings that will allow them to become a futures exchange.


Of course, with futures on OTC stock indexes already trading, it's getting to the point where soon there won't be any barriers at all. Equity markets will be options markets and futures markets. Option markets will be equity and futures markets, etc., etc. It will then be a battle of supermarkets on a monstrous scale.

The Future is Now --- (Electronics come into their own)

As numerous and as radical as the changes have been in recent years, our securities markets have done an outstanding job of adapting existing technology to smooth out the stress points.

Both institutional and private investors have swarmed to computers and electronic telecommunications like bees to honey.

Not only is instantaneous transmission of data a common occurrence these days but instantaneous analysis, both fundamental and technical, is available, even if you are sailing in San Francisco Bay. Almost all of the space-age technologies are used to transmit data. Everything from radio bands to satellites are constantly sending current market information all over the world. Because of some rather dramatic break-throughs in both hardware and software, the cost has been reduced to affordable levels. Now an individual investor in the comfort of his home office car, boat, or airplane can get the same data at the same time as institutional investors, 'who control hundreds of millions of investment capital dollars. He can even punch a key on his personal computer and instigate a trade in his account without another human being's involvement in the equation.

Even though it is a trial balloon at this point, 24-hour trading of equities is already a reality. We know of a brokerage firm in Los Angeles that offers the service, and we have been told it will soon be available in New York and Chicago.

There is almost 24-hour global trading also available for certain commodities with networks springing to life nearly monthly.

The American and Toronto exchanges began electronically linked trading last month. Switzerland's three main stock exchanges have revealed joint information, settling, and trading systems scheduled to start next year. The computerized linkups are intended to ease the exchanges' adjustment to the coming era of 24-hour trading. Within two years they will be able to link up their systems anywhere in the world.

An official of a .major U.S. exchange recently predicted that international stock trading would surpass current estimates of $227 billion in U.S. and foreign international stock activity by ' 1988. He went on to poin.t out that U.S. firms now have over 250 branches in 30 foreign countries, while close to 100 foreign firms have branches in the U.S. The following statement reveals,


at least in our opinion, how perceptive his insights on the matter really are. "The experience of foreign branches indicates that, more and more, sophisticated customers want global asset management. When customers think in these terms, firms and markets respond."

And you thought the extra half-hour in the morning was a big deal!

A group of equity dealers in London are organizing a new international equity traders association similar to NASD. Closer to home, our own MARKET TECHNICIANS ASSOCIATION has provided the leadership necessary for the International Federation of Technical Analysis to become a reality.

If you don't believe that global 24-hour trading is going to become a reality within the next ten years, encompassing all of our major marketplaces for securities, you are more of a wishful thinker than a realist.

Other major phases in the evolutionary process of our markets that we expect to seebecome realities in the next decade are:

A centralized market (as mandated by Congress years ago)

Computerized trading floors (adios specialists!)

Electronic money managers (Mac to Lisa, sold 100 shares @ 48 l/4)

Whatimpactwillthese changes,bothpastand future,have on the professional technical security analyst? How will he adapt to this potentially hostile environment? What new tools will he be required to obtain, and where will he get them?

These are questions I have been asking myself for sometime. Because the answers have been so elusive, they became the motivation behind my selection of the theme for today's presentation.

If I weren't convinced the technicians of the world would successfully accept and conquer this challenge, I would not be here today. I would be looking for another field of endeavor, since I most certainly hope to be a productive component of society in the year 1995 and beyond.

In all candor, though I must say it will be an extremely difficult time for us in the years immediately ahead, we are going to be required to develop new indicators to replace old ones that no longer are efficient.. We are going to be pushed more and more into the forefront of the newest generation of computer hardware and software. We will, in all probability, have to considerably shorten our timing constants as well as our means of communicating to our clients and customers. We will


experience more frustration and disappointments than we think we can handle. We will be working with data and materials that will be unprecedented in that they did not previously exist.

As an example, how will we handle global 24-hour trading? Will we consider a time unit to be from noon-to-noon or midnight-to- midnight? Will we break a 24-hour unit into three 8-hour "shifts", and, if so, will it be a 9 to 5 sub-cycle or another start/finish sequence? When the central market, as mandated by Congress, is in full operation, what time zone will we use? Will New York City get the nod on a "grandfather clause", or will we use whichever time zone we operate in? Will we use a volume coefficient to smooth out the new trading patterns as they evolve? How about breaking out submarket activities by continents or countries?

The single subject of global 24-hour trading markets alone creates more questions that I have remaining time to present, let alone broach the subjects of computerized trading floors, central markets, and electronic money managers. But, I am sure you can already see where I'm coming from. We are at the right place at the right time. We are on the cutting edge of an industry that is going to be going through its parabolic evolutionary phase during the next decade. We will be in position to make an indelible imprint on history, and we will. I know it! I can sense it in this room. The vibes are there, the clock is running, and I have never felt more fortunate to be a market technician than I do right now. I hope you have the same feeling.

Thank you for having me, and may I leave you with a glimpse of our competition in the years ahead. Just last month a Massachusetts based company introduced an artificial intelligence computer system for personal financial planners, with analytical and reasoning powers almost beyond belief. It has the ability to reason, analyze and make judgments. It has a knowledge base containing more than 6,000 decision-making rules, along with a data base of more than 125 financial vehicles available to investors. Those included are securities, real estate, shelters, partnerships, tangible assets, and fixed income assets. It also contains federal and state tax laws, which will be updated quarterly.

If it's true that when the going gets tough, the tough get going, I suggest we leave here today in compound low, but be in overdrive by Monday.


After all, we have a 2,500 year tradition to uphold, and there's no way in hell we're going to be the weak link.

This article was presented as a keynote address to the Technical Securities Analysts Association of San Francisco, October, 1985.

Mr. David D. Holt is a member of the Market Technicians Association and President of T.L.Communications, Inc., publisher of the Trade Levels Report.

-


A SIMPLE APPROACH TO VOLATILITY

RICHARD C. ORR, Ph.D.

INTRODUCTION.

One of the more elusive concepts in technical analysis is that, of volatility. The dictionary defines volatility as the quality of being changeable or transient, that is: passing quickly or soon. The problem one faces in trying to utilize volatility in forecasting the market is in trying to construct a volatility measure, which in itself is not too volatile. A measure of volatility that gives a low reading one day and a high reading the next would be useless. At the same time, simply constructing a volatility function that evolves slowly from high values to low values accomplishes nothing if it does not accurately reflect the likelihood of the future movement of the market.

In what follows we will propose an easily calculated measure of volatility, give the reader a feel for its behavior over more than a ten year period, and suggest a property that volatile markets seem to have, which could be exploited in market timing work. We will also discuss multiplicative filters generally, and look at a 5 percent multiplicative filter on the S&P 500 index to see how volatile the market has been at intermediate-term (about 2 months) tops and bottoms in the market over the past ten years.

A CANDIDATE FOR A VOLATILITY MEASURE.

Definitions.

A classic mathematical definition of volatility might be that of variance. calculates

For a given set of values {xl, x2, . . . . x,) one its mean or average, Z, and then proceeds to sum the

squares of the differences between each value and the mean value, finally dividing this sum by n-l (one less than the number of terms addedjl, Algebraically stated, variance is given by:

VAR = -& 2 (Xi - ii)* i=l

lThe question as to when we divide by n-l or by n depends on whether we are dealing with a whole population or a sample of that population, but will not be important for our discussion.


Notice that if all the values {xl, x , t

. . . . x,1 are equal to some value, a, then the mean is also equa to a, and the variance would be 0. There is in that case no variability of the values around a. In all other cases variance would be positive, with values far below the mean contributing just as much to the variance as values far above the mean. This latter feature is important since with volatility the changeability of the market, but not the direction of such a change, is being measured.

'One of the drawbacks of using variance as the measure of volatility is that it may be less sensitive than it should be to the extreme values, which are of the utmost importance within a given time interval. A much simpler measure of volatility, which we now consider is the following: For some fixed period, say 25 days2, calculate the largest and smallest values of the data within that period. The measure of volatility we will use will have the characteristics of the ratio of the largest piece of data to the smallest piece of data, but to make the function easier to work with we define it as follows: Let MAX be the largest of 25 consecutive daily closing values of the S&P 500 index, while MIN represents the smallest value of the same set. Our volatility function is then defined as follows:

VOLATILITY = 100 x%-1)

All this definition does is to simplify the interpretation of its values, e.g., if VOLATILITY = 20 then MAX is 20 percent larger than MIN.

Notation.

In the tables that follow we will use a calendar notation YYMMDD, which is common to many databases. This is simply a six digit representation of a date where the first two digits represent the last two digits of the year, the second two digits represent the month and the last two digits represent the date. For example, 791008 represents October 8, 1979, while 800327 represents March 27, 1980.

MAIN RESDLTS.

Having defined volatility to be the percentage that the largest value in the 25 day range is above the smallest value in the same

2Any number could be used here. The larger the number of days considered the longer term this measure of volatility would be. Units are daily, but could be hourly, weekly, monthly, or anything else.


range, let us look at the distribution of its values from 750206 through 850916. Table I on the next page displays a frequency distribution in half-percent increments for the 2679 days in question. For example, there were 224 days in which the volatility value was at least 5.5 percent but less than 6.0 percent, while the total number of days with volatility less than 6.0 percent was 1597.

The first question we deal with is that of the transience of our volatility measure. By the way we have defined volatility, we have guaranteed that successive values will be highly correlated. The data from which the volatility values are calculated overlaps until 25 days have elapsed. We consider the autocorrelation of the time series of volatility values with lag 25. This simply means that we look at the correlation between each pair of values separated by exactly 24 trading days. We would like to see a positive correlation between these two values of volatility even though 25 days have passed. The process begins on 750206, the 25th trading day of the year, and 750314, the 50th trading day of the year. The volatility values for all pairs of days that are 25 days apart are compared from this point until 850819 and 850924, the last pair considered. For these 2660 pairs of days we have a correlation coefficient r=+.178, which is very signif.icantly positive. Thus we have shown that over the ten-plus year period in question volatility levels tend to maintain some stability into the future.

Next, it is our intention to show that periods of high volatility are characterized by a certain pattern of market action. We claim that the following schematic describes the typical market behavior during periods of high volatility and beyond.

S&P 500 PRICES

SUBSEQUENT TOP

BEGIN PERIOD OF HIGH VOLATILITY

OF LOW VOLATILITY

SUBSEQUENT BOTTOM


At Least But Less Than Count 0.0 0.5 0 0.5 1.0 0 1.0 1.5 0 1.5 2.0 2 2.0 2.5 39 2.5 3.0 105 3.0 3.5 211 3.5 4.0 307 4.0 4.5 293 4.5 5.0 215 5.0 5.5 201 5.5 6.0 224 6.0 6.5 154 6.5 7 .o 156 7.0 7.5 141 7.5 8.0 100 8.0 8.5 107 8.5 9.0 72 9.0 9.5 42 9.5 10.0 40

10.0 10.5 56 10.5 11.0 23 11.0 11.5 42 11.5 12.0 19 12 .o 12.5 15 12.5 13.0 1-l 13.0 13.5 27 13.5 14.0 16 14.0 14.5 2 14.5 15.0 6 15.0 15.5 6 15.5 16 .O 19 16.0 16.5 3 16.5 17 .o 6 17.0 17.5 5 17.5 18.0 2 18.0 18.5 0 18.5 19.0 1 19 .o 19.5 1 19.5 20.0 7 20.0 20.5 1 20.5 21 .o 0 21 .o 21.5 2

Cumulative 0 0 0 2

41 146 357 664 957

1172 1373 1597 1751 1907 2048 2148 2255 23 27 2369 2409 2465 2488 2530 2549 2564 257 5 2602 2618 2620 26 26 2632 2651 2654 2660 2665 2667 2667 2668 2669 2676 2677 2677 2679

Table I. Frequency Distribution of Volatility Values. 750206 through 850916


For our purposes, high volatility will be any value of volatility that is at the 75th percentile or higher, while low volatility will be any value that is at the 25th percentile or lower. From the frequency distribution (Table I> we find these values to be VOLATILITY 17.5 and VOLATILITY < 4 respectivelg. We then study all intervals that begin with the first day that VOLATILITY is at least 7.5 and end with the first day that VOLATILITY is less than 4.0. Within each interval there is a largest price that we refer to as the TOP and a smallest price that we will refer to as the BOTTOM. What the above schematic asserts is that most of the time the BOTTOM will precede the TOP and that the starting day will have a lower price than that of the ending day. Table II on the next page demonstrates this characteristic for the period 750206 through 850916.

A MULTIPLICATIVE FILTER

While for very short term-market moves a standard additive filter may be fine, for most moves we feel that a multiplicative filter is superior. With an additive filter, moves up equal moves down: a move to 100 from 50 would have as its counterpart a move to 0 from 50 (50+50 and 50-50). A similar multiplicative filter would pair a move from 50 to loo-with a move from 50 to 25 (50*(Z) and 50*(1/Z)). Given that prices are bounded below by 0 but unbounded above, we feel that.a multiplicative filter handles this situation more appropriately.

For the period 750206 through 850924 consider a 5 percent multiplicative filter on the S&P 500 index. Under this filter a TOP is any largest price in an interval that is 1.05 times (larger than) some previous price and subsequent price in that interval. Similarly, a BOTTOM is any smallest price in an interval that is l/1.05 times (smaller than) some previous price and subsequent price in that interval. Table III, displayed on the two pages following Table II, lists all the extrema for a 5 percent multiplicative filter on the S&P 500 index. Also listed are the volatility values as well as their percentile values on the day of each TOP or BOTTOM. It is worth noting that on the average BOTTOMS tend to be more volatile than TOPS. Perhaps even more interesting is that both TOPS and BOTTOMS tend to.occur in more volatile times. Of the 53 extrema

listed in TABLE III, 20 of 27 TOPS and 21 of 26 BOTTOMS were above the 50th percentile in volatility.

3For those who are interested, the distribution of volatility has a mean of 6.12 and a standard deviation of 2.92, but is not normal (in fact, it is skewed to the right), which is why we are using the frequency distribution to find the percentiles.


First Date

750206 78.56 750210 78.36 BOT 750715 760106 93.53 760106 93.53 BOT 760220 761022 99.96 761110 98.81 BOT 761215 780424 95.77 780424 95.77 BOT 760517 780802 102.92 780802 102.92 BOT 780906 781020 97.95 781114 92.49 BOT 790126 791015 103.36 791107 99.87 BOT 791217 800123 113.44 800327 98.22 BOT 800903 801015 133.70 801030 126.29 BOT 801128 810325 137.11 810325 137.11 TOP 800414 810827 123.51 810925 112.77 BOT 811130 820111 116.78 820129 120.40 TOP 820308 820604 110.09 820611 111.24 TOP 820707 820806 103.71 820812 102.42 BOT 830304 830428 162.95 830519 161.99 BOT 830525 840208 155.85 830223 154.29 BOT 830227 840529 150.29 840608 155.17 TOP 840615 840803 162.35 840808 161.75 BOT 840821 850121 175.23 850121 175.23 BOT 850213

S&P 500 First Extremum Date S&P 500

Second Extremum Date S&P 500

95.61 TOP 751111 89.87 102.10 TOP 760227 99.71 105.14 TOP 761223 104.84

99.60 TOP 780526 96.58 105.38 TOP 780906 105.38 101.86 TOP 790307 98.44 109.33 TOP 791231 107.94 126.12 TOP 800910 124.81 140.52 TOP 810225 128.52 132.68 BOT 810415 134.17 126.35 TOP 811228 122.27 107.34 BOT 820511 119.42 107.22 BOT 820709 108.83 153.67 TOP 830330 153.39 166.21 TOP 830531 162.39 159.30 TOP 840312 156.34 149.03 BOT 840716 151.60 167.83 TOP 840907 164.37 183.35 TOP 850227 180.71

Last Date S&P 500

Table II. Order of Precedence of Extreme Values During Roves from High to Low Volatility. 750206 through 850916


INTERMEDIATE TOP DATE PRICE VOLAT

750317 86.01 9.76

750715 95.61 6.14

751117 91.46 3.83 32

760921 107.83 6.48 71

770103 107.00 4.80 48

770719 101.79 3.09 15

771125 96.69 6.59 72

780606 100.32 4.61 45

780912 106.99 3.58 27

790126 101.86 7.55 81

790410 103.34 4.98 51

791005 111.27 4.58 45

800213 118.44 7.78 82

PERC 91

67

INTERMEDIATE BOTTOM DATE PRICE VOLAT

750407 80.35 7.04

750916 82.09 6.13

751205 86.82 5.34

761110 98.81 4.79

770531 96.12 4.35

771102 90.71 6.65

780306 86.90 4.52

780705 94.27 6.42

781114 92.49 13.95

790227 96.13 5.96

790514 98.06 5.38

791107 99.87 11.41

800327 98.22 17.12

Table III, Part I. Volatility Values for Extreme Values (5 Percent Filtef). 750206 through 800630

v?A Journal/November 1985 29

PERC

77

67

57

48

41

73

44

70

98

65

58

95

100

INTERMEDIATE TOP DATE PRICE VOLAT

800922 130.40 6.82

801015 133.70 8.22 85

801128 140.52 11.27 95

810106 138.12 8.45 87

810325 137.11 8.32 86

810811 133.85 5.29 56

811130 126.35 6.93 76

820507 119.47 4.13 38

821109 143.02 13.54 98

821207 142.72 7.59

830622 170.99 i.97

831010 172.65 5.03

841106 170.41 5.41 58

850717 195.65 5.57 60

Average for TOP 6.53 64 Average for BOTTOM 7.70 73

PERC 75

81

65

52

INTERMEDIATE BOTTOM DATE PRICE VOLAT PERC

800929 123.54 6.82 75

801030 126.29 8.22 85

801211 127.36 10.33 93

810220 126.58 6.47 71

810722 127.13 4.89 50

810925 112.77 14.60 98

820308 107.34 9.94 92

820812 102.42 8.90 88

821123 132.93 7.59 81

821215 135.24 7.36 79

830808 159.18 7.02 77

840724 147.82 4.75 69

841213 161.81 4.25 40

Table III, Part II. Volatility Values for Extreme Values (5 Percent Filter). 800701 through 850924

-


Summary

We have attempted in this paper to define volatility in a manner that is both readily understood and easily computed. We have then shown that, under our definition, volatility is a stable enough measure to have value in forecasting. We have also shown that there is a prevalent market pattern during periods of high volatility and, finally, that while intermediate-term BOTTOMS tend to be more volatile than do intermediate-term TOPS, that both are considerably more volatile than average. An interested reader might wish to explore further the consequences of changing the parameters we have used: e.g., 65 days instead of 25 days or hourly as opposed to daily data. For homework, determine what is so unusual about the value of VOLATILITY for 850910.

Dr. Richard C. Orr is a member of the Market Technicians Association and Vice President for Research, John Gutman Investment Corporation.

31 c


THE DOUBLE POWER SCALE

By John R. McGinley, Jr.

One of the impediments to good charting - and good chart reading - has always been, for lack of a better term, noise. When finally your eagerly sought, very important subtle signal appears, sometimes it can hardly be distinguished from all the other points on the chart. What has always been needed is a way to make such readings literally jump off the page at you. Raising the wheat of meaning above the chaff of noise is the worthy objective of all who draw charts.

Arthur Merrill has made his reputation on the fact that Technical Trends charts are meticulously drawn to make reading his charts as simple - and as clear - as possible. He has always carefully chosen his scales to emphasize the information content and to reduce the "noise". How often his charts are reproduced is a measure of his success.

As a part of our on-going efforts in this endeavor, I have developed a new scaling method, which we call the Double Power Scale. It highlights important points in oscillating data with a clarity not previously possible with older scaling methods.

The Double Power Scale uses a simple concept: from an indicator's central tendency point, the scale - both up and down - begins to broaden out the farther away from center you move. In other words, each scale marker up or down from the oscillator's midpoint is placed NOT equally distant from its neighbor, but at a point which is the square (or other power) of the distance from the midpoint.

For example, if the median is 1.0 and the next scale graduation is one mm. away (marked l.O), the next few graduations (arithmetically marked 2.0, 3.0, 4.0, 5.0, etc.) would be drawn at2mm, 4mm, 8mm, and 16mm, etc., using the power of two. As you can see from the example below, each successive point on the axis gets progressively further away from its predecessor, both up and down.

Our example is Dick Arms' Trading Index (TRIN, STKS, or MKDS on some machines). The Double Power Scale serves particularly well in quieting most of the noise normally found in the daily or weekly readings of this very useful index. Ratios by definition center on unity. This indicator, however, keeps unity at Arm's (sic.) length. The historical median of the Arms Index tests at approximately 0.89, or 1.12 on our chart (scale reversed to make upward readings bullish). Minor variations about this point (roughly 1.20 to 1.00) have little meaning. However, excursions far above or below the immediate central range are often particularly important.


1P

AVER. VOLUME UP S70CKS/OOWN STOCKS 1.4 I

ft I.3

1.2

I.1

1.91 1.5 Power Scale

I I I

AVER. VOLUME UP STOCKS/DOWN STOCKS l.4[p+&+J

I984

When notable points (such as the second week in September 1983 on the chart) are all but lost amid the noise on the normal chart, the 1.5 power scale pops it out for immediate inspection. In this case, a 2.0 scale would have been even more dramatic. Nevertheless, one can easily see how greatly emphasized the important deviations are. And the more important they are, the more highlighted they become.

The charts of many other oscillators can be effectively improved with this scaling method. We have converted several TECHNICAL TRENDS charts to advantage. Now more than ever, you'll find a single glance is usually all you need.

This helpful innovation can easily be instituted in your own charts with the help of the following computer program, written in a very basic BASIC by Arthur Merrill. It will calculate where on the Y-axis to place your scale markers. Enter the program into your machine exactly as written, neatness counts! The up caret (tn(t is the symbol for "to the power of". It varies on some machines, but most use this symbol. (6)-(3) is six cubed. Use the symbol appropriate for your machine.

The entries , when you run the program will be:

Median Height of chart, in inches Top scale point Lowest scale point Increase, one scale point

to the next.

Power

*Scale below

The centraltendency point From top border to bottom

On the Arms chart, 1.5 From the chart, 0.8" As below, 0.1. For minor

points, it would have been 0.01. The amount of spreading wanted; 1 is normal 3 a lot. Typically 1, 1.5, 1.7, 2, & 2.2 are useful.

0.8 drawn by hand to 0.74


A Program to Calculate Double Power Scale Markings by Arthur A. Merrill

10' 20' 30' 40 Input "Enter Median"; Al 50 Input "Enter Height of Chart in Inches"; S9 60 Input "Enter Top Scale Point"; P9 70 Input "Enter Lowest Scale Point"; P8 80 Input 'Enter Increase from One Major Scale Point to Next"; T9 90 Input "Enter Power"; P5

100 D2 = P9 - Al:D3 l= Al - P8 110 Kl = S9/D2-PS+D3-P5):S3 = Kl * D3-P5 120 CLS: Print "Median ="; Al 130 Print "Height of Chart in Inches ="; S9 140 Print "Top Scale Point ="; S9 150 Print "Lowest Scale Point ="; P8 160 Print "Increase One Point to Next ="; T9 170 Print "Power ="; P5 180 Print:Print "Location of Scale Markers": Print 190 Print "Scale Marker "; P8, " 0 Inches" 200 N = 0 210 N = N + 1:P7 = P8 + (N * T9) 220 IF P7 > Al GOT0 270 230 D5 = Al - P7:SS = Kl * DS*PS 240 S6 = S3 - S5 250 IF S6 > S9 GOT0 290 260 Print "Scale Marker "; P7, S6; "1nches":GOTO 210 270 D5 = P7 - Al ' 280 S5 = Kl * DSAP5:S6 = S3 + S5:GOTO 250 290 Print "Completed": END

A sample run: run Median =1.12 Height of Chart in Inches = 6 Top Scale Point = 1.5 Lowest Scale Point = .8 Increase One Point to Next = .l Power = 1.5

Location of Scale Markers Scale Marker .8 0 Inches Scale Marker .9 1.124533 Inches Scale Marker 1 2.01485 Inches Scale Marker' 1.1 2.574597 Inches Scale Marker 1.2 2.942397 Inches Scale Marker 1.3 3.718862 Inches Scale Marker 1.4 4.75619 Inches Scale Marker 1.5 6 Inches COMPLETED Ok

Mr. John R. McGinley, Jr. is a member of the Market Technicians Association and is associated with Merrill Analysis, Inc.


CONSTRUCTING ADVANCE/DECLINE AND SYNTHETIC PRICE LINES FOR

DIVERGENCE ANALYSIS OF THE CAPITAL SECTOR OF THE TREASURY

MARKET

R. Bruce McCurtain

Divergence Analysis, 2 Classic Tool

Since technical analysts were first attracted to the rattlings of the stock market ticker, one of the most helpful indicators has been the venerable advance-decline line. Many bull and bear markets have been preceded by the disparate action between this breadth measure and a price index. An advance decline is a leading indicator, or as Harvey A. Krow explained the concept of divergence: II . ..future deviations from the direction of present movement of prices is detectable first in subtle changes in the relativity of issues rising to issues falling," (Stock Market Behavior p. 64).

As with the "classic" A-D line for stocks, divergent actions by treasury A-D line prior to a move by government bond prices is desirable - it provides a potential early warning of a change in market direction.

Unfortunately, the lack of readily available statistical information for the treasury market has made it difficult to produce a similar technical barometer for the study of government debt instruments. With a little extra work, however, an analyst can construct an advanced-decline series for the capital sector of the treasury market.

Constructing Two Cumulative Flow Lines

To conduct a divergence analyses of the U.S. Government Market, two cumulative flow lines are needed.

The first line is the U.S. Government Advance/Decline Line. -- Fis the simpledifferential between the number of issues that advanced for the previous day less the number that declined.

Each day the leading financial journals publish a list of the prices for the previous day's action in treasury bonds, notes and bills (a) and (b). The final bid (or closing price) and the net change from the previous day are quoted. Except for the yield, no other data are revealed. To construct an advance/decline line the analyst must count the number of issues advancing, the number declining, the number unchanged, and the total number of issues traded (c). What the analyst


should include in the count are notes (out to ten years) and bonds (over ten years and out to 30 years).

First the analyst must add up all of the pluses and minuses for the advance-decline for that day. The result is then cumulated into a flow line over days so that divergences can be identified.

Next, the analyst must sum all of the "net" price changes for the advances and the declines. This results in the cumulative total of the amount each issue advances or -- declines. This cumulation total price line is the second line or the -- issues.

Etyncic" U.S. Government Pricei% of capital --

There are several advantages to constructing these two cumulative flow lines. Although there are published indexes which will mirror price action in the extant long broad market (e.g., Merrill Lynch, Ryan, Shearson, etc.), there could be a problem with the retrieval of historical data for calculations. In addition, there is no promise of a continuation of those published indexes. Moreover, the advantage of the price flow line is that it, unlike a cash bond or a futures contract, will never expire even though some of the components in the listing are "retired" and new issues are included.

Demonstrating Divergences Between the Treasury A-D Line -- and the Treasury Price Line --

Over the past several years (1983-85), whenever there has been a significant divergence between the U.S. Government A/D line and the U.S. Government Price Line, an important change in the direction of interest rates (i.e., bond prices) has developed. In those instances lacking preliminary disparate action between the A/D and price lines, reversals could still be caught by the confirmation of one line's reversal movement by a similar reversal movement in the other line.

In 1983 top formation in the debt markets took several months to reach completion. The Treasury A-D line peaked in January 1983, while it took the Treasury Price Line until May, 1983 to reach its high. After the January peak in the A/D Line, the price highs in February, 1983 and May, 1983 were not confirmed by action in the A-D Line. These upside non- confirmations presaged the ultimate bear trend in treasuries even though the May peak in the A-D Line did exceed its February high point.

Once the bear decline had c,ommenced, no significant divergences between A-D and the Price Line developed until the period from August 1983 through December 1983. A rally followed from the joint lows in August 1983. After the initial burst, the two indexes parted ways with repeated


-

upside non-confirmations by the U.S. Government A-D Line. This prolonged divergent action again forecasted price weakness. A point (8) the A-D broke the new lows. Soon, the Price Line followed and the market began another major decline.

During the second week of January 1984, some optimistis were beginning to suggest an end to the bear market as prices rallied past an earlier short-term peak. The A/D did not confirm this price action, however. The A-D indicator simply slipped to a new low. The major decline was on again!

At points (11) and (12) the bear market finally did end, but with a rare twist in the divergence relationship. While the A/D Line settled to new lows in June, 1984, the Price Line refused to confirm on the downside. While it could be argued that the A-D was again leading the Price Line toward new lows, the divergence nevertheless held out a warning flag that something could change. Soon thereafter, at point 13, both lines confirmed each other's action by making new short- term highs. So an upside reversal following a divergence was suggesting a rally of at least intermediate term consequences. As it turned out, the bear market was over.

The two indexes climbed in tandem throughout the rest of 1984. Some lessening in strength by the A-D Line created an inharmonious relationship (13) which preceded the intermediate correction into March, 1985. At point (19) the two cumulative flow lines again crossed to the upside for a bullish signal.

Points (20) through (22) highlight another series of divergences which lead to the mid-June, 1985 through July, 1985 contraction in price. At points (23) and (24) a severe breakdown in the Price Line was not confirmed by action in the A-D Line. When coupled with the oversold level in the trading oscillator, there was a strong suggestion of "externalV price weakness greater than the "internal" market dynamics warranted.

Conclusion

In the "wasteland" of debt instrument analysis, where there is a dearth of statistical information, it is nonetheless possible to construct a "classic" Advance/Decline Line and the Price Flow Line. These were done for the capital sector of the Treasury Market. Daily price data were used to construct cumulative A-D Lines and Price Lines.

Although this is still a new frontier, an intial study of data between 1983-85 demonstrated divergent action between the U.S. Government A-D Line and a Price Line of U.S. Governments can be a useful tool for anticipating trend reversals in debt instruments.


NOTES :

(a) Prices are quoted as of 3 pm even though cash bond trading does notend until 5 pm. The earlier closing bids are released to the Associated Press and ultimately the financial journals by the federal reserve. According to the government agency, 3 pm prices are more "accurate" than 5 pm level when trading is thinner.

(b) While an advance/decline series for treasury bills can also be constructed, this study deals only with intermediate to long- term debt instruments.

(cl Recently, the total issues traded has been near 180. When I began the series in 1981, the number of outstanding issues was 135. The increase reflects the amplified federal debt and the supply which has kept a damper on bond prices over the past several years.

(d) The oscillator at the bottom of each chart is a simple lo- day ratio of advances divided by declines. Movement above and below 1.00 can provide buy and sell points.

W The consolidation from August, 1983 through December, 1983 proved to be a "bear flag" with a downside "measured move" which culminated in the June-July, 1984 bear market lows.

(f) This exception was probably a function of the flat yield curve and extant weakness at the short-end of the capital sector. While net price amounts of all issues declined little on the retest of the June lows, *the net of the advances and declines brough the A-D precariously near its previous lows--thus the inverted divergence. If the yield curve had been "normal", it is likely the divergence between the A-D and price lines would have shown a classic disparity.

(9) A longer term overview of market price action (Chart #9) can be compared to the "compressed" A-D line to get a visual "feel" for the bias of movement in the lines. Since March, 1985, despite confirming action on the upside by prices and the A-D line, the overall "mood" of the internal indicator has less dramatic than external pricing.

Mr. R. Bruce McCurtain is Chief Technical Analyst, Reid, Thunberg & Co., Inc.



*

CHART #2 (DAILY) U.S. Government price line (TOP)

(4) with Advance/Decline line and trading oscillator

I ‘\E\ ,’ 1’

(5 )

I I I I I I I I I ‘ I I 1 - I I


I CHART $13 (DAILY) U.S. Government price line (TOP) with Advance/Decline line and trading oscillator

h7m-m . Tha ontire consolidation from August, 1983 through L hnr 198?. nroved to be the mid-way point in DecemY--, ____

the bear marketrihi% began in May, 1983:

n

W 10/3 11/7

I t I I 1 I 8 I I I I I I


CHART 84 (This point was a new short-term high in

It was not confirmed by the A/D.)

.

New low (10 1 *a_

\ -

\ ; \ l

\ J \ .

\

\ . .

. . . . k. - . . .

(DAILY) U.S. Government price line (TOP) with Advance/Decline line and trading oscillator

, . \

, - \ ’

1 .

\ l

\ ’

- \

.

\ - .

’ .

\4

\ l


CHART 115 (DAILY) U.S. Government price line (TOP) with Advance/Decline line and

I .

. .

.

BUY

913 I I I 1 I


CKART DAILY) U.S. Government price line (TOP) with Advance/Decline line and‘trading oscillator

J . . . I

‘*, l .*

(15)

1

+

/\I/ / .* ‘.

,,*.p

‘J (15)

i’ . . . . . . . .

. . .

.

. . . . . . \vJ . -. . - .

BUY

\/ SELL V

‘V ‘V 913184 10/l 11/5 1213 l/7/85

I I I I I I I I I I I I I 1 1 1 I I 1


(DAILY) U.S. Government price line (TOP) with Advance/Decline line and trading oscillator

411


CHART #8 (DAILY) U.S. Government price line (TOP) with Advance/Decline line

ng oscillator

(23)


-

CHART #9

(WEEKLY) U.S. Government price line (TOP) with Advance/Decline Line

Notice the relative weakness of the A/D

, line since June, 1984.

I

I I I I

516183 l/6/84 6/l/84 l/4/85 6/7/85 c


MOVING AVERAGES AND THEIR VARIATIONS

John Carder

When markets are discussed, one of the most frequently used words is "trend". Dow Theory uses the concepts of Primary Trend, Secondary Trend and Minor Trend. Elliot Wave Theory (as practiced by Robert Prechter) mentions waves of degrees ranging from Grand Supercycle to Subminuette. Usually, the first step in

t analyzing a market is to identify the primary trend. That is, the trend that is of sufficiently large degree to be in effect for the expected duration of your trade. Unfortunately, prices don't move in smooth curves. They are upone momentanddownthe next. Moving averages were developed in an attempt to tame the zigs and zags on a chart into a smooth curve. This paper will discuss some of the refinements to moving averages and suggest a new one.

Conventional Moving Averages (MA's):

Conventional moving averages are a simple idea. An "n period MA" is simply the average (mean) of the last n values. That is, take the sum of the last n values and divide by n. One of the most common examples is the 200-day moving average of the Dow-Jones Industrial Average (see Figure 1). This turns the daily gyrations of the stock market into a gentle curve. The most common usage of the indicator is to be long stocks when the DJIA is above its 200-day MA and out or short when the DJIA is below its 200-day MA. This example uses days for the periods. Any time period is possible for-this and the other moving averages mentioned in this paper. I have seen periods ranging from 5 minute bar charts of futures prices to yearly charts. Whatever the period, the curve can be smoothed by using a moving average.

Exponential Moving Averages (EMA's):

The first variation is the exponential moving average. There is no standard way of defining an EMA. Martin Pring suggests the following formula in his book Technical Analysis Explained: To calculate an "n period EMA" for a range of values, you first calculate the "n period MA" for the first n values. Then, take the difference of the n+lst value and the "n period MA". Multiply the difference by t e "exponent", 2/n, and add the product to the previous 1 value. An example will probably make more sense than this explanation. We will take a four period EMA of the following values: Since n=4, the exponent is 2/4 or 0.5.


-

Table 1:

1 2 3 4 5 6 Previous Difference Difference EMA (co1.3+

Period Value EMA (co1.2-~01.3) times Exponent co1.5) ===========================================================================

1 10 2 11 3 12 4 11 5 12 11 1 0.5 11.5 6 13 11.5 1.5 0.75 12.25 7 14.25 12.25 2 1.0 13.25 8 12 13.25 -1 -0.625 12.625

------------------------------------------------------===================== -_-------__-_----___----------------------------------

The previous EMA (column 3) of the fifth period is the average of the first four periods. Figure lalso shows the 200-Day EMA of the Dow-Jones Industrial Average.

Two advantages are usually claimed for the EMA over the MA. The first is that it is easier to compute than the MA. Since you only use two values (the previous day's EMA and today's value) as opposed to all n values in an MA, it is supposed to be faster. However, you don't need to add the last n values to compute an MA. Most of the numbers in that sum are the same as in the prior period's calculation. You add one new value and subtract one old value. Again, a table is useful. We'll compute a 4 period MA with the same values.

Table 2: 1 2 3 4 5 6 7

Previous Value four Difference MA (~01.3 Period Value MA periods ago co1.2-~01.4 col.5/n +co1.6) ===========================================================================

1 10 2 11 3 12 4 11 5 12 11 10 2 0.5 11.5 6 13 11.5 11 2 0.5 12 7 14.25 12 12 2.25 0.5625 12.5625 8 12 12.5625 11 1 0.25 12.8125

-------__--_-_______-------------------- -__------_--_-----_---------------- --------____--______-------------------------------------------------------

The point of my discussion is that using the above formula, a conventional MA should take about as long to compute as an EMA.

The second advantage claimed for an EMA is that it is front - weighted. All of the last n values are equally weighted in a

conventional MA. The most recent value is the most heavily


weighted in an EMA. In fact, an EMA is always moving toward the most recent unsmoothed value.

Figure 2 contrasts the weig 9 ting for a10 PeriodEMAwiththatof

a conventional 10 period MA. As you can see, the four most recent periods are more heavily weighted in the EMA than in the MA, giving it its front weighting. Unfortunately, an EMA is also "tail weighted". Over ten percent of the weight is assigned to values that are more than ten periods old. The situation gets worse for longer term EMA's, as you can see in Table 3.

Table 3: DAY EMA: TAIL WEIGHT ===================== . . 10 : 10.7% : : 15 : 11.7% : . . 20 : 12.2% : : 30 : 12.6% : : 50 : 13.0% : . . 75 : 13.2% : . . 100 13.3% : . . 200 r 13.4% : ======================

This means that while an EMA is always trending toward the most recent unsmoothed value, it is being held back by a substantial amount of stale values. When you have a large move in prices, the EMA is held back by its "tail weighting".

Front Weighted Moving Averages (FWMA's)

In order to avoid the problem of "tail weighting", you must use a real front weight moving average (FWMA). The most common way to construct an "n period FWMA" follows. This kind of smoothing has been called the Coppock Guide. Multiply the most recent value by n, the next most recent by n-l, the third most recent by n-2, an so on. Sum the products. Divide by the sum n+(n-l)+(n- 2)+...+2+1. (That sum equals n*(n+l)/2.) The result is the most common FWMA. Tables 4 and 5 illustrate the calculation for a four period FWMA.


Table 4: Weights

Day 1 2 3 4 ==================================================== Weight 4 3 2 1 Sum of weights=4*(4+1)/2=10

Table 5: 1 2 3 4 5 6 7 8

4 times 3*last 2*2nd 1*1st Sum of FWMA Period Value most recent value value value products co1.7/10 ----------------------------------------------------------------========== ----------------____--------------------------------------------

1 10 2 11 3 12 4 11 44 36 22 10 112 11.2 5 12 48 33 24 11 116 11.6 6 13 52 36 22 12 122 12.2 7 14.25 57 39 24 11 131 13.1 8 12 48 42.75 26 12 128.75 12.875

=====================================================================------ ------

Figure 3 contrasts the weighting of a lo-day EMA and a lo-day FWMA. As you can see, the FWMA loses the tail weighting of the EMA. The FWMA is truly front weighted and is not "tail weighted". Its main disadvantage is that it involves a lot of computation. That has been overcom,e by the advances in computers. We can eliminate the step of dividing each sum of products by dividing weight by that sum before we begin. That' is, we would multiply by 0.4 instead of 4 in column 3 of table 5. This means that an n period FWMA requires n multiplications and n-l additions. The EMA required one subtraction, one multiplication and one addition. The MA required one subtraction, one division and one addition. For long period FWMA's, a computer becomes essential.

Figure 1 also illustrates the 200-day FWMA of the DJIA. As in the cases of the MA and EMA, shorter period FWMA's are less smooth than longer period FWMA's.

A problem arises with all three of these moving averages when markets make large sudden shifts, as the gold market did in early 1983. All those old numbers will tend to draw the average away from current values for some time after the shift.

A shorter FWMA will lose the old values more quickly, but it won't be as smooth as a longer term FWMA. I have devised a simple system to vary the front weighting of an FWMA continuously from one extreme of a simple MA to the other extreme of the most recent value.

Variable FWMA's:

Consider the function Y=X raised to the power P, with X ranging


from 0 to 1.3 Figure 4 shows this function for several values of P. To construct an n period FWMA, you begin by constructing the weights. You will need n of them. Begin with the fractions l/n, 2/n, . . . I (n-U/n, n/n. Raise each of them to the power P. These are your unadjusted weights. Unfortunately, they do not add up to one (unless n=l), so we do one last thing to obtain our weights. Sum them and divide each of them by the sum. Now you have your weights. Let's construct the weights for a 3 period FWMA with P=2 as an example:

Table 6: Weight: 1 2 3 -_--_------____-__--------------------------- --------------------------------------------- l/3 l/3 213 1 >>>>>>>> the fractions squared 1/g 4/g 1 >>>>>>>> the unadjusted summed = 14/9 = (l/9+4/9+1) weights --------------------------------------------- -_--------------_---------------------------- The weights l/14 4/14 9/14 >>>>>>>> the unadjusted

weights divided by the sum

To determine the n-period FWMA for a given period, you multiply the last n values by the corresponding weights and then sum their products. Table 7 provides an example:

Table 7: 1 2 3 4 5 6

9/14 times most 4/14 times l/14 times FWMA (~01's Period Value recent value 2nd value 3rd value 3+4+5) -------------_-----_------------------------------------------------------- ------^---_--_--____------------------------------------------------------- _

1 10 2 11 3 12 108/14 ' 44/14 lOj14 162/14 (11.57) 4 11 99/14 48/14 11/14 158/14 (11.29) 5 12 108/14 44/14 12/14 164/14 (11.71) 6 13 117/14 48/14 11/14 176/14 (12.57) _

---------------------------------------------------------================== ----------_--_--_--_-------------------------------------

Figure 5 shows some adjusted weighting curves for various values of P. If you compare Figure 5 with Figures 2 and 3 you will notice that P=O yields the weighting curve of a conventional moving average and that P=l yields the weighting curve of the Coppock Guide described above. Values of P between 0 and 1 yield FWMA's that are less front weighted than the Coppock Guide, while values greater than 1 yield FWMA's that are more front weighted than the Coppock Guide. Also, there is none of the tail weighting of the EMA.

Why bother with variable FWMA's? Conventional MA's are slow, and a sudden shift in underlying values makes too sudden shifts in the MA; one when it enters the sum and one when it departs the sum. EMA's are tail weighted, and I don't think that that is what we want in a smoothing. Probably the most compelling reason is that the ability to vary P gives us another degree of freedom -


in attempting to find the "right" moving average for a given application. The second part of this paper shows how that degree of freedom is used to improve a valued indicator.

An Example:

In his excellent Winter 1980/81 Chartbook, a tool to illustrate the four year cycle.

4Ian McAvity introduced He started with the

monthly average of the daily closing level of the S&P 500. He took the 12 month percentage change of these values. He then used a lo-month Coppock Guide (10 month FWMA with P=l) to smooth the % change. A buy signal was generated whenever this smoothed value was falling and negative and then turned up. Figures 6, 7, and 8 illustrate this tool. To quote from his chartbook: "Yes, it's always LATE on the upturn (but sure)". I want to thank Mr. McAvity for his generous assistance in the work that follows. He was kind enough to send me the underlying data, along with a worksheet explaining the Coppock Guide and how it is constructed. The refinements which follow are not intended as a criticism of his insightful work, rather they are an attempt to "fine tune" an already excellent tool.

First, I varied the percent change from 10 to 14 month changes. Then for each of these I computed 42 FWMA's on the data. P varied over 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4. Also, I computed 9, 10, 11, 12, 13 and 14 FWMA's for each of the P's. I will use the terms fast and slow in the following discussion. One FWMA is faster than another it is shorter in period or has a greater P. I firstnoticedthatthe faster FWMA's were whipsawed by bear market rallies, and the slower ones were even later in their signals. Figure 9 illustrates this effect. How do we optimize this tool? Which combination is best? Is there a best combination? The problem here is that by using a faster --m-m combination to get an earlier (more profitable) buy signal, we risk being whipsawed. In order to compare the various values, I calculated the average gain of the signals generated by each FWMA. I assumed a 28 month holding period, and if a subsequent signal was generated during those 28 months, the holding period was extended to 28 months from the subsequent buy signal, and it was considered one long signal. Table 8 shows the result of averaging these calculations. As you can see, the lo-month FWMA with P=l that Mr. McAvity used produced an excellent average gain (34.46%). The best average gain (36.92%) came from a 13 month FWMA with P=1.3 of the ll-month percent change. Figures 10, 11 and 12 illustrate it and its signals. I

There is another way of looking at the signals. Cumulative gains might yield different results. Table 9 shows the results of compounding your gains at each signal, that is, as if you had invested $1.00 at the first signal, sold it as above, and invested the proceeds at the next signal, etc. The major difference in the results is in the very slow signals. -- They generated fewer signals and thus had lower cumulative returns. The best result was once again a 13-month FWMA with P=1.3 of the


11-month percent change.

There were three other combinations that should be mentioned, since they had good results n both average and cumulative gains. They are 12-month FWMA with P=O.9 and 1.0 of 11-month percentage change and 12-month FWMA with P=1.3 of 12-month percentage change. Tables 10,11,12 and 13 show the signals generated by these three combinations and the best one. None of the four combinations generated a losing signal. The fourth one was odd, though. It was generally late and generated two extra signals. It does illustrate the problem of relying solely on gains. I would describe it as a lucky anomaly and would place little faith in it as a technical tool. The remaining three smoothings all were of ll-month percentage change. The 13 month FWMA with P=O.9 was generally slower, but August 1938 was a lower place to buy than July 1938. This kept it in the running. Obviously, none of them is best for all purposes. The 13-month FWMA with P=1.3 has yielded the best results.

The difference in gains does not seem particularly significant to me. Most of the improvement is due to its having given a signal a month earlier here and there. A shorter conventional FWMA would have signalled earlier too, but would also have been easier to whipsaw. The Coppock Guide was trapped by bear rallies in March 1948 and February 1958, while the 13-month FWMA with P=1.3 did not generate signals in either bear market rally. The additional degree of freedom has allowed us to find an FWMA t= generates more rel=ble signals, and it q

--- ---- enerates them sooner. --.-

It could be argued that this has been merely an exercise in curve-fitting. If you examine Tables 8 and 9 in the area of this FWMA, you will seethatany of the FWMA's near it would have also produced good results. . I think that this continuity of the results indicates that we haven't found an anomaly. Why should we expect this to continue to be effective? Mr. McAvity introduced it in 1980. We have had two signals since then, in 1982 and 1984. At this point, they appear to me to be good signals, confirming the indicator's predictive value.

Figures 13 and 14 show one more refinement. I have taken the 13- month FWMA with P=1.3 of ll-month percent change and converted it to daily figures. As you can see, the original Coppock Guide and the three smoothings above all gave buy signals in January 1985. There is no guarantee, but it indicates that the bull market is going to be alive and well for some time. No indicator is perfect. It is still late, but it is not quite as late. It still can be whipsawed, but if markets are similar to those of the last 60 years, this is not likely to happen. That is the benefit that the additional degree of freedom provides. It gives one more way to refine good tools.

Moving averages are the technician's tools for smoothing. Each has its advantages and disadvantages. A conventional MA is simple, easy to explain and easy to use. It is the slowest of

-

MTA Journal/November 1985 54 c

the n-period MA's that I've covered. The EMA is faster and easy to calculate. It does have that "tail weighting". The main disadvantage of the FWMA is its computation time. Luckily, computers are overcoming that disadvantage. By adding the FWMA with parameter P to your collection of smoothing tools, I hope that you will be able to refine your indicators.


FOOTNOTES:

1) Often, EMA's are described not as an n period EMA, but as an m% EMA, where m is the exponent viewed as a percent. The conversion formulas follow:

An n period EMA is a (200/n)% EMA. An m% EMA is a (200/m) period EMA.

One advantage of the percent nomenclature is that it allows a wider range of values for the exponent. On the other hand, I have seen no standard way to start an EMA with an exponent equivalent to a non-integral number of days.

2) Where does the tail weighting come from? It is inherent in the construction of the EMA. First note that in the construction of the EMA described in Table 1, we multiplied the difference (column 4) by the exponent and added that to the previous EMA. That is equivalent to multiplying the Value (column 2) by the exponent, then multiplying the Previous EMA (column 3) by (l- exponent) and finally adding the two products. We will use this equivalent construction in the example below. An example shows how an EMA gets its tail weighting: Assume a lo-day EMA. Its expoent n equals 2/10=0.2. So 20% of Tuesday's unsmoothed value is added to 80% of Monday's EMA. So Tuesday's unsmoothed value has a weighting of 20% in Tuesday's EMA. On Wednesday, we repeat the process, adding 20% of Wednesday's unsmoothed value to 80% of Tuesday's EMA. In so doing we have reduced the weighting of Tuesday's unsmoothed value from 20% to 16% (80% of 20%). On Thursday, we repeat the process, adding 20% of Thursday's unsmoothed value to 80% -of-Wednesday's EMA. The weighting of Tuesday's unsmoothed value is thereby reduced from 16% to 12.8% (80%of 16%). As each day goes by, the weighting of Tuesday's unsmoothed value is reduced to 80% of what it was the day before. While repeatedly reducing the weight to 80% of its previous value does shrink its value, it NEVER goes to zero. So any day's unsmoothed value will be reflected in the EMA, albeit to a small degree.

In general for an EMA with exponent n, the weight of the dth period's value in the EMA is given by:

(d-1) n(l-n)

where the most recent period is considered the first period, the previous period is the second period, etc.

3) If the reader is unfamiliar with exponents, the following explanation maybehelpful. Inallt.hatfollows assumethatxis a positive number. And if P is a natural number (i.e. 1,2,3,4 , . . . 1 t then X to the power P is the product of P X's. If P=O then X to the power P is 1 for any positive X. This definition .has been expanded to cover any number P (even infinite


cardinals). If you are writing a program in BASIC you only need type XAP and the language will perform the calculations for you. Some calculators offer LOG and/or LN functions, they usually also offer their inverses. If you insist on doing it by hand, LOG and LN tables can be found in the Handbook of Chemistry and Physics --- published annually by the Chemical Rubber Co. 18901 Cranwood Parkway, Cleveland, Ohio, 44128. To calculate X to the power P, first take the LOG (or LN) of X, then multiply the result by P and apply the inverse function to the product.

4) Ian McAvity edits DELIBERATIONS, published twice monthly by Deliberations Research, Inc., P>O> Box 182, Adelaide Street Station, Toronto, Ontario, Canda M5C 251


REFERENCES :

The Dow Theory Today, Richard Russell, Edition1981 a division of:

Fraser Publishing Company

Fraser Management Associates Box 494 Burlington, Vermont 05402 Library of Congress Catalog Card Number: 81-68858 ISBN: 0-87034-061-l .

The Elliott Wave Principle, -- Robert Rougelot Prechter, Jr. and Alfred John Frost .

Second Edition 1981 New Classics Library, Inc. P.O. Box 1618 Gainesville, Georgia 30503 Library of Congress Catalog Card Number: 81-80170 ISBN: o-932750-02-8

Technical Analysis Explained, Martin J. Pring, McGraw-Hill Book Company 1980 Edition

ISBN: O-07-050871-2

.

-


BIOGRAPHY

John Carder received a B.A. in Mathematics from the University of Colorado at Boulder. He currently acts as an advisor to several of his family's trusts. He is the author of the programs used to draw the charts appearing in this paper.


I

I

I-

t I-

I

t, .

1 .

i- I J- ) l-

-


I

l-

3-- I I I I I 1 ’ ’ ’

m . Ei

H

I3 C 7J m

N

. .

v CI

ul

HEIGHT

MTA Journa l/November 1985 61

4 I

-1 ,L

I

. 1-

a

I-

Jl-

I H

c) c xl rl

W

I -1 x\

l- I

‘L +

+ I 9

P +

‘-1

9 I- - Yl

I I I

L-- 1 -l I,---

I- i

MTA Journa l/November 1985

WEIGHT

62

FIGURE 4

0.75

0.50

0.25

0.25 0.50 0.75

Y=XP for various P’s


FIGURE 5

1.75 /

1.50

1.25

1.00

0.75

0.50

0.25

0.00 0 0.75

Adjusted weighting curves for various F’s


.

I I 1 I I 8 8 1 I 1 8 ‘ I 1 I I 1 I I I I I

f4

30-

S&P 500

Honthly Average of the Daily Closing Level

FIGURE 6 2~

.

12 month % change (monthly)

- 10 month FHMA with P-l

V -40%.


S&P 500 flonthly Average of the

40%-

I V



80. SW 500

. flonthly Average of the 6 0. Daily Closing Level

40.

FIGURE 8

40%.

12 month % change

10 month FWMA with P-1

30%


P =

4

6

a > 38 G

1 iJ rl0 0 >

12

14

*

/

+

. 2

/

*

. 4

/

/

*

.b

/

/

*

. 8 1 1.2

/ /

/ /

/

if the March 1935 bottom was s

later than May 1935

x 1.4

x

/ /

/ /

/

gnall ed

if was trapped by'the 1957 bear market ral

if was trapped by a 1974 bear market rally

(FWMA's of 12 month percent change on

FIGURE 9

Y)

Y


Monthly Rverage of the Daily Closing Level

FIGURE 10

30

20

10


b 50%,

n b- 0%,

13 month Wlfl with P-1.3 13 month Wlfl with P-1.3 -40%. -40%.


S&P 500 Honthly Average of the Dally Closing Level

FIGURE 11

40%

11 month % change

13 month FWMA with P-1.3 3 0%.

20%.

10%

V ‘J V 0%

-10%.

. +


.-

Monthly Rverage of the Dally Closing Level

FIGURE 12 150-

125-

100-

75-

40%-

20%-

(monthly)

v 0%-

-20%-

13 month FHMR with P-l .3 T -20%-


FIGURE 13

I I drawn daily from 2


1982 1983 1984 1985 I

DAY MOVING AVERAGE

180

140

-Jan 1978 to 3 Ott 1985 I MTA Journal/November 1985

FIGURE 14

1978 1979 1980 1981

1 I I I I I I I 1 I 1 1

231 DAY % CHRNGE OF THE 21 DRY MR OF THE

drawn daily from 2 2.:


1983 1984 1985

1 I I 1 1 I I 1 I I I 1

S&P 500 WITH ITS 273 DRY FWMR USING Pd.3

40

20

0

I

an 1978 to 3 Ott 1985


-I-FtBl-E 8 FIUERF~GE GfiI N

10 MONTH CHANGE WITH MONTH P FWMA .8 .9 1.0 1.1 1.2 1.3 1.4 f=======~===============Z=========================================================

9 i0.39% 30.39% 35.48% 35.51% 34.93% 34.01% 33.78% 10 31.17% 31.17% 31.97% 30.64% 35.57% 35 . 25% 35.22% 11 30.11% 30.43% 31.17% 31.58% 31.58% 30.64% 30.64% 12 30.29% 30.62% 30.70% 30.43% 31.58% 31.58% 31.58% 13 27.53% 27.53% 27.81% 28.15% 30.43% 31.58% 31 .S8%

14 29.40% 27.53% 27.53% 27.53% 23.05% 28.15% 27.96% =I==D===I=I=PI==PE====IPI=IPID==5==II=P=========================================

11 MONTH CHANGE WITH MONTH P FWMA .8 ‘9 1.0 1.1 1.2 1.3 1.4 ====================P=====P=========================================================

9 32.89% 32.51% 31.41% 30.12% 35.33% 35.33% 35.33%

10 31.82% 31.82% 33.93% 32.09% 31.46% 31.46% 30.93% 11 33.46% 3i. 15% 33.15% 33 . 93% 33.55% 32.02% 31.78%

12 34.48% 36. a32 36.87% 32.54% 34.35% 33.13% 33.13% 13 33.29% 32.99% 34.56% 33.92% 34 . 39% 36.92% 32.54% 14 31.613% 33.66% 33.08% 32.95% 33.15% 33.92% 34 * 797 ,e

======p====e==o=========-==P= I==fP=I==t=PI========IPP===Pt==E=======P===========

12 PlONTH CHANGE WITH MONTH P FWMA .8 .9 1.0 1.1 1 .- 3 1.3 1.4 ===========================I=========== I===3P=L===I===I=I=*==IP=-90===P===T===I===

9 34.a1’/. 34.86% 30.66% 30.25% 30.82% 28.55% 28.58% 10 35.31% ’ 34.86% 34~46% 34.86% 34.86% i0 . 777 e 30 .77%

11 34.93% 35.75% 35.75% 34.86% 34.86% 34.86% 30.05%

12 35.61% 35.08% 34.60% 3S.75% 35.75% 36.31% 35.56%

13 33.42% 33.21% 33.75% 34.60% 34.84% 35.41% 35.75% 14 3 0 . 2 s % 31.08% 31.08% 32.89% 33.40% 33.40% 34 ~ 842

===P=l=zI======f ==Pf======X3=P=PI=P==I===E=0==I==P==9=-=========~===========---- ----

13 MONTH CHANGE WITH MONTH P FWMA .a .9 1.0 1.1 1.2 1.3 1.4 =I=p===p==I==r=I==pI-==-=p=I=====pI==I==========================================

9 27.20% 29 .22. Y 28.93% 29150% 29.33% 29.41% 29.67%

10 29.33% 23.9ax 28.74% 27.79% 28.34% 28.93% 29.50%

I1 32.13% 31.73% 29.15% 28.76% 28.62% 29.29% 28.71%

12 31.51% 31.70% 31.70% 27.43% 27.40% 27.40% 27.07%

13 29.87% 30.83% 31.06% 31.64% 27.00% 27.61% 27.40%

14 28.67% 30.28% 30.28% 30.77% 31.13% 30.88% 27.18% ==0==0=1I1=DE1Er==I==IIZP==S-------3lf-========================================= _------

14 MONTH CHANGE WITH MONTH P FWMA .8 .9 1.0 1.1 1.2 1.3 1.4 ==I=PIP51P=tll=PP===I=P==P=O-==PPllltl=I===~===============~= ==I=pP===o=P=IP===r

9 31.21% 31.34% 32.08% 32.08% 32.08% 32.93% 32.93%

10 30.28% 31.21% 30 98% 26.90% ’

31.28% 31.34% 31.29% 31.88%

11 26.90% 27.92% 27.42% 27.60% 31.73% 31.73%

12 26.08% 25.88% 25.41% 26.71% 27.06% 28.71% 28.99%

13 24.79% 25.96% 21.42% 21.23% 25.41% 25.41% 27.14%

14 24.81% 25.96% 2S.96% 21.42% 20.95% 20.42% 21.22% I=PP==Ilf==PI=====P=I-IPI=0==-P=II=IP==I===========~============================


-I-f=+E3BLE 9 CUMULATIVE GFIIN

10 MONTH CHANGE WITH MONTH P FWMA .8 .9 1.0 1.1 1.2 1.3 1.4 =4==============================================================================

9 14.7 14.7 16.1 16.2 15.5 14.7 14.4 10 16.1 16.1 17.1 15.0 16.1 15.8 15.8 11 14.6 14.9 16.1 16.7 16.7 15.0 15.0 12 14.6 15.2 15.3 14.9 16.7 16.7 16.7 13 14.2 14.2 12.1 12.6 14.9 16.7 16.7 14 21.2 14.2 14.2 14.2 15.1 12.6 12.3

======================P===I===P==IL==Pr=========================================

11 MONTH CH3NGE LJITH MONTH P FWMA .9 .9 1.a 1.1 I.2 1.3 1.4 ----------------================================================================ -_--em----------

9 19.3 18.7 17.1 15.2 17.3 17.3 17.3 10 17.6 17.6 2 1. .5 18.1 17.2 17.2 16.4 11 20.5 20.1 20.1 21.5 20.9 18.1 17.9 12 30.5 36.6 36 . 3 18.6 21.7 20.2 20.2 13 34.2 33 . 0 30.6 28.9 30.0 36.8 18.8 14 29.7 35.4 33.5 33.2 33.9 28.9 31.3

====================fP=I============================================================

12 MONTH CHANGE !JITH . MONTH P

FWMA .8 .9 1.0 1.1 1 .2 1.3 1.4 ==I====-------------- --------------===========================================================

9 29.1 31.6 16.9 15.3 17.3 13.4 13.5 10 32.4 31.2. 30.3 31.6 31.6 17.2 17.2 11 31.5 33.6 33.6 31.2 31 . 2 31.6 16.3 12 33.7 32.2 30.7 33.6 33.6 35.8 33.6 13 33.4 32.7 35.8 30.7 31.2 3:. 8 33.6 14 24.5 26.5 26.5 31.7 34.6 34.6 31.2

===p==o======pp======-==3=I=2=zp=D====9=========~===============================

13 MONTH CHANGE WITH MONTH P FWMA .8 .9 1.0 1.1 1 .2 1.3 1.4 =====================f=PE=E=======P--=I=========================================

9 14.1 17.1 16.5 17.1 16.9 17.2 18.4 10 17.2 16.6 16.3 14.8 15.7 16.5 17.1 11 30.5 29.3 16.9 16.3 16.1 17.0 16.3 12 28.6 29.3 29.3 14.2 14.1 14.1 13.6 13 22.9 25.0 26.5 29.1 13.6 14.4 14.1 14 19.6 23.7 23.7 25.6 26.7 26.1 13.8

============211====opDsp====I==IIpII==’=====~=====~====~~=========~=========~==~

14 MONTH CHANGE LJITH MONTH P FWMA .8 .9 1.0 1.1 1.2 1.3 1.4 ====P===IE=I==II=rPE============================================================

9 16.1 16.5 17.6 17.6 17.6 19.0 19.0 10 14.8 16.1 15.8 16.4 16.5 16.5 17.3 11 14.2 13.4 14.5 14.0 14.2 17.0 17.0 12 12.6 12.4 11.2 13.1 13.6 15.6 16.2 13 11.2 12.9 6.7 6.3 11.2 11.2 13.7 14 11.2 12.9 12.9 6.7 6.1 5.7 6.4

PP==PE=====P=======IPI=r===-=PI==PI==5======================================~===


TFSBI-E 18 11 month %change smoothed by 12 month FWMA with P= .9

==================r====D-=3P=Pr=

1*+*++28**APR 1924 *** 54.90% 2**+*12B*+AUG 1932 *** 12.11% 3**+*160**APR 1935 l ** 71.69% 4****200++AUG 1938 +*+ 14.81% S+*+*ZlS**MAR 1940 *** 14.81% 6,,++“32*+#‘R 1941 L l ** 14.81% 7*+**249**SEP 1942 l *+ 14.81% 8**+*308+*AUG 1947 l ** 57.70% 9****334*+OCT 1949 l +* 57.70%

l0***+3BS+*FEB 1954 +** 74.14% 11+*+*439*+JUL 1958 l ** 16.29% lZ++**d69*+JAN 1361 a** 36.79% 13*+**494**FEB 1963 *+* 36.79% 14****542*+FEB 1967 *+* 10.37% 15****585**6EP 1970 *** 40.33% 16*+**638*+FEB 1975 +*+ 18.51% 17****676**APR 1978 l ** 26.78% l$j++++728+*AlJG 1982 l ** 0,Q0%

19*+**757*+JAN 1985 l ** 0.00%

=~z==z I========I===I====t==I=-I=P


-

78 -

1-ABLE 11 11 month %change smoothed by 12 month FWMt4 with P= 1

===t================I=PS=i==P==l

1*+***28**APR 1924 l * * 54.90%

2**++128**AUG 1932 l * * 12.11%

3****160+*APR i935 l * * 71.69%

4++**199**JUL 1938 +** 9.59%

5+++*219*+MAR 1940 +** 9.59% 6++++232**APR 1941 l * * 9.59%

7****249**SEP 1952 .*** 9.59%

8****Z08*+AUG 1947 *** 57.70%

9**++334**OCT 1949 l ** 57.70%

10****?86*-FE9 1954 l *+ 74.14%

11****439**JUL 1958 ii* 16.29X

1:+***4S3++OEC 1550 *** 42.40% 13*+*+434++FE3 is63 +** 42.40% 14*++*542**FEB 136-i *+* 10.87%

15*+**5SS**SEP !970 *** 40.33% 16*+*+638*+FEB 1,375 *** 18.51% 17****676**APR 1978 ++* 26.78X

18**++728**AUG 1332 *** 0.00%

19+*++757**JAN 1335 *it* 0.(30x ============t===================


-

-I-FIBI-E 12 12 month Xchange smoothed by 12 month FWMA with P= 1.3

I~=============II==PLI-============ lttttt2gttMAy 1924 *** 54.35% 2***+128**AUG 1932 **+ 12.11% 3****161**MAY 1935 l ** 42.00% 4****199**JUL 1938 *+* 13.24% 5****220**APR 1940 l ** 13.24% 6****233**MAY 1941 *** 13.24% 7****250**OCT 1942 l t* 13.24%

8**+*308**AUG 1947 *** 60.62% 9****315**MAR 1948 l ** 60.62%

10****333+*SEP 1959 l ** 60.62% 11****386*+FEB 1954 *** 74.14% 12****434**FEB 195d ttt 27 . 59"

13****438**JUN 1358 l t* 27.59;

14****469*+JAN 1361 l ** 43.61% 15****493**JAN 1363 *** 43.61% 16****542**FEB 1367 *** 10.87% 17****585**SEP 1370 l * * 40.33%

18****638**FEB 1975 l ** 18.51% 19**+*676**APR 1978 ++* 26.78% 20****729**SEP 1932 *** 0.00:: 21****757**JAN 1385 +** 0.00%

=======rp===============9r511========


-

80

TABLE 13 11 month %chanQe smoothed by

13 month FWMA with P= 1.3 ================================

1+*+**29**f1PR 1924 l * * 54.90%

2****129**AUG lS37 l +* 12.11% ~++++~6~++ApR 19;j it+ 71.69%

4****199+*JUL 1938 *+* 9.59%

5****21g**MAR 1940 l * * 9.59%

6***+232**A?R 1941 l * * 9.59%

7*+**24gr*SEp 1942 *+* 9.59%

9**+*307**JUL !947 l * * j3.622

9+*++33l+*OCT 1949 l ** 53.62% ~~**t~~g~**~~~ 1954 ++* 74.14;:

11****439*+JUL 1958 *** 15.29% 1~****4~g*+DEC 19E8 ++* 4?.40T! !3++**4g4+rF~~ 1g-;; l ** 42.40% l~*f+*j4~*+~~~ 1957 *+* 10.37T! 15++**5~5**~~p 1972 *** 4a.33x 16****637**JAN 1975 l * * 23.30%

17+***S7~+*ApR 1378 *+* 26.78% l~*+**723**hUG lS92 l * * 0.00%

19****75?*+JAN 1925 l * * 0.00Z!

PS=f======P==r=I==I===r==I=PIP=I


Journal of Technical Analysis (JOTA). Issue 22 (1985, November)

Economy & Finance

mta journal mta journalnovembe

mta journal special

market sentiment

mta officers

papers sentiment

sentiment area

concept of sentiment

new york richard orr