263 74 8MB
English Pages 215 Year 1993
Author of the BestseIling 'STOCKS & HARE SIMPLIFIED'
Winning on the Stock Market
Winning on the Stock Market Low-risk and High-profit Strategies for Investors BRIAN J. MILLARD
JOHN WILEY & SONS Chichester' New York' Brisbane' Toronto' Singapore
Copyright © 1993 by John Wiley & Sons Ltd Baffins Lane, Chichester, West Sussex POI91UD, England All rights reserved.
No part of this book may be reproduced by any means, or transmitted or translated into a machine language without the written permission of the publisher. Other Wiley Editorial Offices
John Wiley & Sons, lnc., 605 Third Avenue, New York, NY 10158-0012, USA Jacaranda Wiley Lld, G.P.O. Box 859, Brisbane, Queensland 4001, Australia John Wiley & Sons (Canada) Ltd, 22 Worcester Road. Rexdale, Ontario M9W ILl, Canada John Wiley & Sons (SEA) Pte Ltd, 37 Jalan Pemimpin #05-04, Block B, Union Industrial Building, Singapore 2057
Library of Congress Cataloging-in-Publication Data
Millard, Brian J. Winning on the stock market: low-risk and high-profit strategies for investors / Brian J. Millard. p. cm. Includes bibliographical references. ISBN 0-471-93881-5 1. Stocks. 2. Investments. 3. Speculation. 1. Title. HG4661.M54 1993 332.63'22-dc20 93-14742 CIP British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library ISBN 0-471-93881-5 Produced from camera-ready copy supplied by the author. Printed and bound in Great Britain by Redwood Books Limited, Trowbridge, Wiltshire
Contents
1 Introduction
1
2 Brokers and Financial Information
13
3
22 セ。イォ・エiョヲゥオ」ウ
4 Trends in Share Prices
36
5 Insuring against Loss
55
6 Investment Tools
65
7 The Investment Climate
82
8 Picking Shares
110
9 Dominant Trends
121
10 Buying and Selling
140
11 Channel Analysis
160
12 Summary
189
Appendix
206
1 Introduction
It is a sad fact that for every investor who makes money out of the stock market, there are another two who do not. There are many reasons for failure, but foremost among these is inconsistency of approach, with impulse buying featuring at the top of the list. Naturally, a consistent approach is useless if it it lacks the wherewithal to make profits, but investors will find that the methods discussed in this book have stood the test of time, and have led over the long term to very useful profits for those who have followed them carefully. The prime objective of an investment in the stock market is to make money. It might be thought that this is so obvious that it needs no stating in this introduction, but this is not the case. Many investors have subconsciously changed their focus so that their objective is simply to hold shares. This is particularly true of new investors who have become involved in the stock market through the many privatisation issues They have no clear idea of what to do once they have received the share certificates. This is partly due to ignorance, not so much ignorance of the mechanics of selling shares, since the High Street banks are full of notices about how they can do this for you, but ignorance of how share prices rise and fall and the timescale over which they do this. MISPLACED LOYALTY Siren voices, especially emanating from the Government and the larger UK companies, constantly talk about investor loyalty and the advantages of holding shares for the long term. This is total poppycock on two counts. Firstly, as we will see later, the penalty for going for the long term can be to see the real value of a holding greatly reduced, not only due to the ravages of inflation but because of a fall in the share price.
2
Winning on the Stock Market
Secondly, loyalty should be a two-way process, but UK companies consider it to be nothing of the sort. If shareholders stay with their holdings for extended periods of time, then the costs of maintaining the share register are reduced, and the share price could stay firmer, While talking about difficult trading conditions, the directors often award themselves large and totally unjustifiable increases in salary or director's fees. The last thing such directors want is a stampede by investors to shed themselves of their shareholdings. The only loyalty required by an investor is to the value of his shareholdings. A share certificate has no emotional standing, and except for nostalgic certificates of long-gone Victorian railway companies and the like, has no intrinsic value as a work of art. The certificate is simply a means to an end, the end being to increase the value of an investor's shareholdings. If a particular share ceases to contribute to the upward momentum then it must be sold. TIlE OBSTACLES TO PROFIT Private investors have a major cross to bear in the form of dealing costs and the spread between quoted buying and selling prices of shares. In the absence of these costs, an investor who just buys and sells on a purely random basis should hope to break even over a long time period, although the aim of breaking even is not a very ambitious one! With these costs added in, the same investor must lose over the long run, since all profitable transactions must have these costs deducted from the profit, while all losing transactions have these costs added to the loss. An investor who just buys and sells at any time he feels inclined to do so is therefore almost certain to lose. Buying opportunities have to be chosen carefully so as to maximise the potential gain while reducing the potential loss, and the aim of this book is to discuss ways in which this can be done. Besides the fine detail of the timing decisions of share buying and selling, there is one simple requirement for investment success:
There shouldbemoreprofitable transactions thanunprofitable ones. The question which arises from this is, "How many more?" An investor who is correct 50% of the time cannot make a profit because of the effect of dealing costs. How much more of the time he should be correct in order to move into profit depends upon the level of these dealing costs and the average profit or loss per transaction. While we know the dealing costs, the average profit and loss depends upon the individual investor's decision making and the state of the market. In the absence of any other information we could hazard a guess that the investor needs to be correct 75% of the time, i.e. three transactions out of every four should be in profit. This is a very demanding requirement, and unlikely to be met by more than a handful of investors, and even then they would not be able to maintain this level of success for any extended period of time.
Introduction
3
4lIlI
3511
0
0
.....
1l1li
0
0
n[
511
nOD
11 11
m 211
o
0 411
611
811
1l1li
P.ro.nt... loss
F1aure 1.1 How the percentage gain required to offset a particular percentage loss rapidly reacheslevels which are impossible to achieve
Besides the negative influence of dealing costs on the overall profitability of stock market investment, there is also the unavoidable mathematics of profit and loss in a world where we look at everything in percentage terms. The percentage gain necessary toretrieve aloss isgreaterthan thepercentage which was lost. A simple example, which ignores dealing costs, serves to clarify this: Supposing an investor starts with £1000 and loses 20% of this on his first investment. He is then left with £800. Now to claw back the lost £200 and restore the capital back to £1000 a gain of 25% is required on the £800. The situation gets rapidly worse as the percentage loss increases. Suppose our investor loses 50% of the starting £1000. he is now down to £500, and requires a 100% gain to restore the original £1000. Figure 1.1 shows how rapidly the gain required to offset a loss rises as we move from modest to large losses. The gains are achievable when we are talking about small losses of say less than 25 to 30%, but by the time we reach large losses we require impossible gains to compensate. Thus it is vital to develop a method which wiU keep our losses down to the level of those at the left hand corner of the graph in Figure 1.1. As a spin-off, the method will also improve the situation so that the investor need not be correct such a high proportion of the time. The method is to place a restriction on the maximum.loss which can be made from the losing positions, while having no such restriction on
4
Winning on the Stock Market
the profitable positions. This is done via what is called a "stop- loss", and the mechanics of determining the stop-loss level and applying it to share price movements are discussed fuUy later in this book. THE FUNDAMENTAL AND TECHNICAL VIEWPOINTS There are two basic schools of thought in choosing shares to buy. The fundamentalists take the view that a thorough study of the fundamentals of a company will show whether its share price is too low, too high, or just about correct. They arrive at this decision by doing a large amount of work on the published accounts of the company, analysis of the markets in which it operates, the current and future competition, the quality of the top management, the quality of the products, the export potential of the product, and so on. It doesn't take much imagination to see that a vast amount of effort has to be expended in order to arrive at a conclusion that a share should be bought, sold or held. Just suppose that a private investor had the energy to put into such an effort, and came to the conclusion that the share was undervalued, and should be bought. His reasoning is that eventually the market will come to realise that the share is out of line, and the price will rise until it either correctly reflects the worth of the company, or has gone to the other extreme and overvalues the worth of the company. The major flaw in this reasoning is the belief that the market will come to realise that the share is undervalued. It may not do so, or may take such a long time to do so that the investor spends a long time in the situation where he has not even covered his dealing costs in buying the share. In practice, the situation is rather better than this. The professional fundamentalists publish their fmdings in the press and in brokers' circulars. Thus a large number of investors simultaneously read that the shares of company ABC are worth buying. We then have a self-fulfilling prophecy, since this excess of demand will make certain that the share price rises in the short term. The trouble is, the short term may be too short for a sensible profit to emerge once dealing costs are taken into acount, and it lies in the lap of the gods as to whether a longer term upward trend has been established. From the foregoing, the reader will perceive that I am not impressed by what we might call the result to effort ratio. An awful lot of effort has been put in by somebody, be it the investor or a professional analyst, for a result that may still be of questionable value as far as realised profit is concerned. The technical analysts take a rather different view of share price evaluation. They believe that the whole story is told by the price itself, and rather than go through the extensive analysis of the fundamentalists, it is only necessary to observe the movement of the share price. An increasing interest in a share, for whatever reason, will be reflected in a rise in the share price, and part of the job of the technical analysist is to determine whether the rise will be an extensive and therefore profitable one.
Introduction
5
Since this book is not about fundamental analysis, then on the broad definitions used above it falls into the technical analysis category. That being said, it should be pointed out that technical analysts form a very broad church indeed. The methods shown in the following chapters, while owing much to the subject at large, are also very much a Brian Millard view of how decisions can be arrived at from analysis of share price data. Although the effort involved in developing the skills to analyse share price movements in this way can be considerable, the resulting expertise can be applied to anyone share in a fraction of the time needed for a fundamental analysis of the same company. The advent of software packages to run on personal computers (see Appendix) takes a great deal of the drudgery out of technical analysis, and can even take the decision-making right out ofthe hands ofthe investor. The power of computers for storing of data, presentation of charts and the ability to perform extensive calculations in a few seconds is to be welcomed, and indeed this book would have taken twice as long to write and ten times as long for the research to have been carried out without a computer. Having said that, the black-box approach of entering the share prices and then obeying the instructions to buy or sell which the computer prints out is not one that I favour. A good analogy is that of driving a car. The driver who never cares what is under the bonnet but simply uses the vehicle to get from A to B is the driver who is more likely to suffer a breakdown on the road. The driver who is in sympathy with his car knows every noise and rattle, and will often begin to sense when something is starting to go wrong, taking appropriate action in plenty of time. The investor who understands the process behind the investment decision making will also be able to take appropriate action to limit the risk to his investments. THE SCIENCE OF INVESTMENT Investment has always been seen as an art rather than a science, and indeed a very famous investment book of yesteryear had the title "The Art of Investment". Art is almost totally subjective - a pile of bricks arranged in a neat cube in the Tate is one man's thing of beauty, and another man's pile of bricks. Artistic people can also be very suggestible, and many can be persuaded to see eventually that the pile of bricks is not just a pile of bricks but something of wonderment. Art is also subject to fashions, and new fashions rise while old ones decline. Investors also suffer from the same shortcomings, many of them being very suggestible to the shadier type of investment operator. Sectors of the stock market also become fashionable, such as the health sector over the last year, but as in the world of art, will suffer the inevitable decline. On the other hand, science is, or should be, highly objective. It is based on facts, and the development of theories from those facts. Until I became interested in the stock market, I spent some fifteen years as a university scientist in a branch of chemistry known as mass spectrometry. Although such a background may seem to be totally irrelevant to my later calling as an investment
6
Winning on the Stock Market
writer, nothing could be further from the truth. The scientific method imposes a discipline that is essential for success in the stock market. There is nothing magic about it; the scientific method is actually just an attitude of mind. Fortunately, this attitude can be developed by investors without having to take A levels in physics or chemistry, and reading and following the methods discussed in this book should suffice. Put simply, the stepwise scientific approach is as follows: 1. Make some observations and gather some facts. 2. Propose a theory to explain all the facts. 3. Make further observations. 4. If they are in line with theory, the theory is strengthened. If they are not, discard the old theory and go back to 2.
There have been many scientific scandals in the past, and there will continue to be more in the future, where a world-wide reputation has been made by scientists who have stopped at step 3 above. The scientist is so excited by his new theory that he is totally convinced that it is correct. He gets on a band waggon which carries him along, and new facts that are not in line with this theory are either ignored as being the result of experimental aberration or are deliberately suppressed. Investors are extremely prone to this failing. They come to some conclusion and then become blind to facts which subsequently negate the conclusion. While not damaging a world reputation, this will certainly damage their investments!
DEAliNG WITH RANDOMNESS The methods used in this book are based on the manipulation of numbers, the numbers being share prices. As will be shown later, the movements of share prices are partly random, and the degree of randomness in the movement changes from one period of time to another. Thus there willbe no such thing as a cast iron, 100% certain prediction of price movement, even if the prediction is limited to the simple question of whether the price will rise, fall or remain the same. The only way forward in such situations is to think in terms of the probability of a particular outcome. Thus we could determine the probability of the price of ICl shares rising on any particular day, or we could determine the probability of the price of ICI shares rising by exactly lOp on any given day. These two probabilities will be different, since quite obviously the probability of a specific rise of lOp is much less than the probability of a rise of any magnitude. It is quite simple to determine the probability of such outcomes in the stock market from past data. Taking a simple example, we could determine that over the last 100 days of ICI prices, there have been 20 occasions when the price has remained unchanged from that of the previous day, 43 days when it rose, and 37 days when it fell. This is irrespective of the actual amount by which the price rose or fell.
Introduction
7
With the proviso that the share price will continue in the same way into the future, we can say that there are 20 chances in 100 that it will remain unchanged tomorrow, 43 chances in 100 that it wiU rise and 37 chances that it will fall. These can be converted to percentages such that: Chance of unchanged price = 20% Chance of price rise = 43% Chance of price fall = 37% Thus there is a slightly better chance that the price will rise tomorrow than that it will fall, and a poor chance that it will remain unchanged. It should be pointed out that the probabilities shown above are not exact values, but estimates provided by the data from which they have been calculated. The estimates get better the more data we use to perform the calculations on, and get worse the less data we use. Thus the estimates obtained if we could be bothered to use, say, 10 years' worth of ICI data (over 2500 daily values) would be better than those above. They are also likely to be slightly different in actual value. Here we hit a theme which is constant throughout this book; a balance has to be struck between the amount of effort which should be employed and the results which can be achieved from that effort. It is a waste of time doing calculations on, say,25OO share prices if there is only a small improvement over the results obtained from 500 share prices. We will see this particularly with the question of using daily prices as opposed to weekly prices. It will become obvious later in this book that we do not improve our gains by a factor of five if we use daily prices (five per business week) as opposed to weekly prices. The choice is entirely up to the investor as to how much effort he is prepared to put into the analysis of share prices. That being said, there is certainly a minimum level below which the results will be disappointing to say the least. It is only if the chance of a rise is 100% that we can be absolutely certain that that is what will happen. If the chance is only 99.9%, then there is still a finite possibility that the price will not rise. Looked at in this way, probabilities are extremely simple, but extremely necessary, since, unless they are totally ignored, they prevent the investor from investing in a situation where the odds are totally against him. Looking at three probabilities above, an investor is bound to lose over the long run if he continues to bet on anyone of the three outcomes. Such are the vagaries of chance that he may well be correct 10 times in a row, but over a larger number, say 100 times, he will inevitably lose out. Thus, just as probabilities become more meaningful as we use more data in their calculation, so do future results get closer to the ones predicted by such probabilities as we amass more of them. This fact is so fundamental to the methods shown in this book that it must be clearly understood by the investor.
8
Winning on the Stock Market
It is only by consistent application of the techniques overan extended period of time that wecan be moreand morecertain thatthe expected results willbe achieved. A humorous example will serve to highlight both the workings of probability and the way in which individuals can fool themselves that probabilities always act in a certain way for them which may be quite different from normal expectation. We have all had the experience of a buttered slice of bread skidding off the plate and on to the carpet. We would all also say that more often than not, it lands so as to cause the maximum inconvenience, i.e. buttered side downwards. To the average person it would appear that the explanation is something to do with aerodynamics or the centre of gravity, causing the buttered side to end up underneath during the flight from the table to the floor. There is an excellent distinction here between what people think, and what actually happens. Two experiments serve to clarify both of these: 1. Carry out a poll of people. 2. Carry out an experiment with buttered bread.
When asked, "Do you think that if you accidentally drop a buttered slice of bread it will land face down on the carpet?", eighteen people out of twenty said "yes". In an experiment in which 1000 slices of bullered bread were flipped into the air from a machine specially designed to eliminate bias, the bread landed face down 505 times, and face up 495 times. These results are so close that there is virtually an equal chance of the bread landing either way up. There are two vital messages here for the investor, firstly that all preconceptions must be emptied from the mind, and secondly, that probability wins over the long term. Thus investors, having read this book and begun to implement the methods here, should not become disheartened if the first few investments go wrong. Their continued application will sooner bring the investor into profit. Conversely, the investor who finds that he makes a large profit from each of his first few investments should not think that he has found the Holy Grail. He is bound to make a few counterbalancing losses before long.
THE DARTBOARD INVESTOR Wherever possible in this book, the probabilities, or a rough estimate of the probabilities of certain outcomes are given. This enables the investor to take a much more informed view of the investment possibilities in certain situations. These probabilities are best expressed through the use of a mythical "dartboard investor". The dartboard investor is the rival of every investment author who puts forward any theory, method or system for making money out of the stock market. The dartboard investor operates by using a very simple technique. On one wall of his garage is fixed a large sheet of chipboard. On the chipboard he has pasted the lists ofUK shares which are found in the back pages of the Financial Times. Alongside
Introduction
9
these two pages he has an enlarged version of the Ff Actuaries table of market sectors. He has a dartboard in which all of the odd numbers have been marked by small red stickers bearing the word "sell", while all of the even numbers are accompanied by small green stickers bearing the word "buy".Finally, he also has pasted on the chipboard a total one-year calendar. The dartboard investor now has all of the tools available to enable him to take decisions. He can select a market sector by throwing a dart at the fT Actuaries Table, he can select shares by throwing a dart at the share pages of the Fl', he can decide about buying or selling by throwing a dart at the dartboard itself, while finally, he can decide on which day to do something by throwing at the calendar. In order to allow the laws of chance to operate freely, he should do this with his eyes shut. By this means we arrive at an investor whose decisions are arrived at purely by chance.
It is incumbenton every investmentauthorto show that, overthe long term; the dartboard investor willnot outperform him. If this turns out not to be the case, then the thousands of hours spent writing the book and the many more thousands of hours accumulated by readers of his book, have all been totally wasted. CHARTS
Producing a chart of a share price or index is the most logical way in which the movements can be shown over a period of time. The alternative is to show columns of figures, but very few individuals other than those with accountancy training are able to gain very much from this method of presentation. There are three types of chart which are in use today, linear, logarithmic and point and figure charts. The last have no place in this book and are not discussed further. Suffice it to say that they are based on price movements in a particular direction, and can only be used in the sense of drawing trendlines, support lines and the like on them as an aid to decision making. Both linear and logarithmic charts have two meaningful axes. The horizontal one is always taken to be the time axis, and is linearly divided according to the timescale being used. Short term price movements may be shown with daily divisions, longer ones with weekly or monthly divisions, while extended time periods will require yearly divisions. It is in the vertical, price scale, that linear charts differ from logarithmic ones. The logarithmic scale is one of constant ratios, so the distance between the divisions representing lOp and 20p is exactly the same as the distance between the divisions representing lOOp and 2OOp. An example of such a chart is shown in Figure 1.2. The advantage of such charts lies in the fact that a specific percentage change in a price gives the same apparent movement whatever the price. Thus the change from 800p to l200p represents a 50% change, as does the change from lOOOp to 15OOp, and as pointed out above, the distance on the chart is the same.
Winning on the Stock Market
10
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::1..........................................................................................................................··....·········..·····1-
:::::::::::::::::::::::::::::::::::.::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::11 ................................................................·····....······..·..·····....········....·····1-
1458 I .... 1:J5I IB
IZII 1281 1158 1188
· · . . . . ·.. ·.·.. · . . . ·.. "'11858
.......................................................................................................................................
............................................. •
8:l4l
................................................................................................................................................. 'lSl
..
............................................................................................................................................... 'lI8
..........
..
651 ........... -
68lI
Figure 1.2 A logarithmic chart. The vertical distance between say 800pand l200p is the same as that between l000p and 1500p. Thus the chart is based on constant ratios
The disadvantage of such logarithmic charts is in the loss of sensitivity to price movements at the upper part of the chart, i.e. at those points representing the higher prices. The chart appears to be flattened out. This also makes it difficult to determine the amplitude of cyclic movements. Since an analysis of price movements is the central theme of this book, all subsequent charts will be of the linear type, as shown in the example in Figure 1.3. Here a specific price difference rather than ratio gives the same apparent movement. Thus the distance from lOp to 20p will be the same as the distance from lOOp to HOp. For those who wish to construct price charts manually by plotting daily or weekly prices, a linear chart is far easier to maintain. TOOLS OF THE TRADE For the operations described in this book, the only requirements are graph paper for those who wish to construct their own charts, paper ruled with columns for maintaining price histories and simple calculations, and a calculator. The chart paper in the form of pads of A4 linear charts can be obtained from any good stationers, as can pads of paper ruled with squares. Better still would be an accounts book or accountant's ruled pad with columns already in place for calculations.
Introduction
11
Figure 1.3 A linearchart. The vertical distance between l000p and 1250p is the same as that between 1250p and 1500p. Thus the chart is basedon constantdifferences
The calculations are trivial, and require simple addition and division. As a short cut, printed values of many years of share price histories and linear share charts of the same shares can be obtained at modest cost (see Appendix). MICROCOMPUTERS The availability of relatively inexpensive microcomputers has taken much of the drudgery out of the manipulation of share price data. The ability to store share price data and plot a chart of this data is invaluable, but allied to this is the fact that they can perform calculations and display the results of those calculations in a fraction of the time it would take to do by hand. Unless the investor has the ability to program the computer, it is of little use without a program to do the required tasks. There are numerous software packages currently available for the investor, mostly running on IBM and IBM compatible computers. However, not all of them can carry out some of the apparently simple but nevertheless essential calculations and plotting shown in this book, such as presenting centred moving averages. The charts of share prices and indices displayed in this book have been produced using MICROVEST 3.1 (see Appendix).
12
Ulinning on the Stock Market
Since the plotting of a chart requires historical price data, the availability of extensive price histories on a large number of shares on floppy disk in a variety of program formats or even as printed values at modest cost simplifies the task of compiling a data base (see Appendix).
2
Brokers and Financial Information
Prior to "Big Bang", wheneverthere was a change in interest rates, or the balance of payments figures were horrendous, it was always the practice of the TV companies to send a camera crew to view the trading floor of the StockExchange. Viewers would see a scene of frantic activity with members rushing to and fro usually waving bits of paper at each other. The floor was the place wherejobbers, who dealt in shares, quoted prices to members of stockbroking firms who were actingto buy or sell shares on behalf of clients. Since Big Bang, the floor is deserted, because dealings are done via computer screens usingthe SEAQ system. In place of the jobber is the market maker, who decideson the buyingand sellingprices of the variousshares in whichthey make a market.The shares ofanyone companymaybe handled bymore than one market maker,and in fact for a share to be classed as an alpha share it is a StockExchange requirement that there must be at least two market makers for that share. The broker now staysback in his office, since all he has to do to quote a price to his client is to look at his SEAQ screen. The market maker updates the prices frequently, but has to deal at the price he is showing at anyone time. In the old days, if you requested a price from your broker, he would have to ask his representative on the trading floor to get a quotation. He would then call you backwith this price. You then decided that youwouldbuyor sell at this price, and by the time the request was sent back to the representative, the price may well have changed. To overcome this, it was necessary to give your broker some flexibility, usinga phrase such as "buy at the market", meaningbuywhateverprice wasbeing quoted. The advantage of having today's instant response is obvious. You knowas soon as you place your phone call what price you will be having to pay, or what price you are getting for your shares. You know your commitment instantly.
14
Winning on the Stock Market
There are very few newcomers to the stock market who did not participate in one of the privatisation issues, but the way in which these were sold to the public is quite different from the way in which the secondary, or after, market operates. The investor filled in his form, either cut out of the newspaper or obtained via a registration system, and filled in the appropriate number of shares requested and pinned a cheque for the correct amount to it. At some time in the future he then received an allocation of all, part, or even none of the shares requested. Although stockbrokers were involved in this process, they were acting behind the scenes, and not obvious to the investor. When the time came to sell the privatisation shares, the investor found that he needed the services of either an intermediary or a stockbroker. As far as buying shares is concerned, unlike the privatisation issues where the investor put the money up front, the secondary market in shares (the market in shares which have already been issued, the alphas, betas, etc.) is built on trust. Shares do not have to be paid for, and money for selling them is not received, until after the end of the Stock Exchange account in which the transaction is carried out. Stock Exchange accounts normally run for a fortnight, although there are the occasional threeweek accounts. Settlement days for transactions in an account are normally on the Tuesday eleven days after the last Friday of the account. Trust is therefore two-way. If you are selling, you expect to be paid, and if you are buying, the broker or intermediary expects you to pay him. Note that for dealing in gilts, the settlement is different. These have to paid for on the day following the transaction. You may buy gilts through the Post Office, paying for them at the time. For dealing in traded options, the settlement is again due the day after the transaction. Both brokers and intermediaries are members of regulatory bodies. Intermediaries and brokers are members of FIMBRA, the Financial Intermediaries, Managers and Brokers Regulatory Association, and brokers will also be members of the Stock Exchange. If you intend to deal with an intermediary, make sure that they are full, and not interim, members of FIMBRA. The most obvious intermediary for an investor to use will be his High Street Bank, presuming he has a satisfactory relationship with it. There are no worries about the financial standing of the bank, and other than the fact that they occasionally make mistakes, at settlement day your bank account will be credited with the proceeds of any shares sold or debited with the costs of any shares bought during the Stock Exchange account. Another route which may look attractive is one of the telephone dealing firms which usually advertise on Ceefax or Oracle. They appear to offer lower minimum commissions than banks or brokers, but this may be offset by a wider spread between the quoted buying and selling prices. There are also a few "share shops" which may be viewed in the same light. If the investor wishes eventually to deal in traded options, then it is better if the investor finds a broker rather than uses the bank. Since trust is involved, the broker
Brokers and Financial Information
15
will not take you as a client if you are totally unknown to him unless you provide bank details for a reference, or unless you are recommended by a well known existing client of the broker. As mentioned above, this is no problem if you have a normal relationship with your bank. If not, change banks before testing the water with a broker! In order to find a broker in your area, you can either look in Yellow Pages, ask a friend, ask your bank or write to the Stock Exchange in London (see Appendix). Obviously telephone calls are cheaper if your broker is in your local telephone area, but be prepared to go outside of this if you find a broker with a better rate of commission which will more than offset the extra telephone charges. BROKER SERVICES Brokers offer a number of services, for which they, like any other business, have to charge. The minimum possible, which, except for special circumstances is all that readers ofthis book require, is a "dealing only"service. Other levels of service include an advisory service and a full discretionary service.
Dealing OnlyService All you want from the broker is that he executes the buying or selling orders that you give him. You will receive confirmation of the order the next day in the form of a contract note, specifying exactly what you have bought or sold, the price you paid or received, and the amount of commission, etc. Advisory Service
Here you may wish to ask advice about a particular share, and your broker will also send you recommendations and lots of printed material from time to time.
Discretionary Service This is for the investor who wishes to do nothing but place his cash or portfolio in the hands of the broker. Hopefully any investor reading this book would not fall into this category. The broker buys and sells as he sees fit, sending you a monthly or perhaps quarterly report on the performance of your portfolio, or perhaps we should say his performance of your portfolio.
Traded Options As mentioned earlier. unlike the position with shares, traded options deals have to be paid for the next morning. An investor who wishes to deal in traded options must complete a set of forms which his broker will send him. These acknowledge that the investor recognises the risks inherent in traded options.
16
Winning on the Stock Market
The broker will not deal in traded options for you under credit. You have to deposit sufficient funds to cover any deals. In the case of cheques or drafts, these funds will have to be cleared before dealings commence. HOW MUCH MONEY IS NEEDED TO DEAL WITH A BROKER? Many investors have an inflated idea of how much money you should be investing before you can expect a broker to act for you, but by today's standards, not very much is needed. You may carry out a transaction in which only £500 of shares is bought or sold, but the dealing costs will be quite high as a percentage of this amount. For this reason, deals of less than that are not advised. There may be occasions with privatisation issues in which you have been allocated only a paltry number of shares. Many banks or building societies offer a good deal if these shares are used as the basis for a new account, and it is much better to liquidate such small holdings in this way. Dealing costs fall as a percentage as the value of the transaction rises, and it is best to view a deal of £10Q0 as about the minimum you would normally contemplate. For a discretionary account, a sum of the order of £25, OQO is probably the minimum that would interest most brokers. COMMISSION Commission is worked out as a percentage of the value of the deal. The value of the deal is termed the "consideration", and is simply the price obtained or paid for the shares multiplied by the number of shares sold or bought. Thus 2000 shares sold at 132p will give a consideration of £2640. The commission structure is complicated in two ways.Firstly there is a minimum commission, which varies from about £15 for a country broker to £25 to £30 for a big-city, and especially London, broker. Secondly there are often commission bands, so that the percentage falls once the consideration rises above a certain amount, say £5000 or £10, OQO. Most commission charges for deals of below a few thousand pounds are in the range 1.5 to 2%. The investor must shop around in order to obtain the best deal, Table 2.1 The increasing importanceof dealing costsas the sizeof deal decreases Commission % of consideration % including Consideration e£) e£) chargeof £3 5000 75 1.5 1.56 2000 30 1.5 1.65 1000 25 2.5 2.8 750 25 3.3 3.73 500 25 5.0 5.6
Brokers and Financial Information
17
comparing the quotes from several brokers. For an active investor, the difference may amount to several hundred pounds a year - well worth the cost of a few telephone calls at the outset. It is the minimum commission that is the killer for small deals. Thus if the minimum is £25 and the percentage rate is 1.5%, then the percentage of the consideration taken up by the commission rises rapidly as the value of the consideration falls, as can be seen in the third column in the Table 2.1. In addition to these there is a contract charge of £3 per deal, so this leads to the final column where total dealing costs are expressed as a percentage of the consideration. An investment must be considered to be a two-way transaction, since in order to realise the profit the security has to be sold again. The above percentages naturally apply to both sides of the transaction. In approximate terms, if you buy £1000 of shares and sell them later at £1000, you will be paying £28 twice, i.e. a total of £56. This represents 5.6% on the round trip, as it is called. Note that brokers make a concession if you buy and sell within the same account. You only have to pay one commission, although you still pay two contract charges. From this discussion of dealing costs, etc., the advantage of keeping deals comfortably above £1000 is obvious.
Spreads There is another source of costs which represent a considerable percentage to the investor, and that is due to the spread of prices. The closing prices quoted in the newspapers are so called "middle" prices. If a middle price is quoted as 250p, you will be disappointed to find that you will have to pay say 252p for the shares. Conversely, you will also be disappointed to fmd that when you sell them, you will only receive say 248p for them. The difference between these buying and selling prices at anyone time is called the spread. The above spread, of about 4p for a share around the 350p mark is typical of alpha shares. This spread is therefore building in an additional cost to the investor of about 1.5% on the round trip. Taking into account the dealing costs discussed above, then for a round trip in £1000 of alpha shares, the investor could pay as much as 5.6% + 1.5%, i.e. 7.2%. There is a great advantage in dealing with alpha shares because of the effect of spreads on dealing costs. Once you move outside of these shares and to other junior markets the spread can become extremely large in percentage terms. For those investors who have succumbed to the blandishments of telephone share pushers, it may be extremely difficult to sell such shares at other than knockdown prices. At the 7% level for alpha shares, the investor has a substantial deficit to overcome by a share price rise before moving into profit. It is very easy to get carried away when the share price rises by say 5% within a few days of buying it,
18
Winning on the Stock Market
but such a rise is only just getting to the point at which a profit will be produced. For the medium term investor, dealing costs will be offset to a great extent by dividends received. As will be seen shortly, these are normally a small percentage of the money invested in the share. CONTRACf NOTES AND SHARE CERTIFICATES Within a day or so of doing a deal via your stockbroker, you should receive a contract note with the date and time of the bargain, the settlement date, a contract reference and the details of the bargain, i.e. the quantity of shares and at what price. The consideration is calculated, and then in the case of a buying operation, the commission and contract charge will be added to the consideration, giving the total payable to the broker on settlement day. In the case of a selling operation, the commission and contract charges are deducted from the consideration to give the total payable to you. A day or so before settlement date, where you are in the net position of owing your broker, the broker gives you a gentle nudge towards paying by sending a statement with the amount payable. You should return any tear-off bit of this with your payment, and in any case never send a cheque without a note pinned to it listing the contract reference or references. The accounts staff of brokers, like the rest of us, can and do make mistakes from time to time, and having the correct number helps to reduce the chances of an error. If you are expecting payment and do not receive a cheque within a day or so of the settlement date, call up the accounts department of the broker to find out why. Besides the obvious mistake of a cheque lost in the mail or overlooked.dt sometimes happens that there is a delay in clearing a bargain. FINANCIAL INFORMATION The most accessible place for financial information is in the newspaper. The quality dailies carry listings of share prices, as do many provincial evening papers. None carry as complete a range of information as the Financial Times (the FT). For the serious investor, especially one who deals in traded options, access to price movements during the day is essential. The Teletext pages of BBC (both 1 and 2) and ITV (Channel 4) carry listings of many hundreds of share prices, updated frequently throughout the day, as well as announcements about changes in bank rate, company results and the like. A useful mine of financial information is the Investors Chronicle, published weekly, with plenty of company information being presented. Shares used to be marked in the FT with their category, alpha, beta etc. In the listings now they are not separated in this way, but listed together. There is now an indication of shares which are actively traded by means of the symbol •. The only separation is into "sectors" of the market, reflecting the markets in which the
Brokers and Financial Information
19
company operates. Thus there are banking, engineering, insurance, industrial and many other categories. New shares such as the recent electricity power generator shares are often kept for some time under a separate heading of "New Issues". In the FT the data is presented in columns. The first three entries at the time of writing under CHEMICALS are shown in Table 2.2. Table 1.1 A typical newspaper presentation of share priceinformation Notes .AkzoFl................ .Allied Colloids....
Price
+or-
£475/8 199
-2
1992 Mkt Yield High Low Cp£M Gr's £503/8 £393/8 1.188 4.1 2.7 220 186 518.5
PIE 12.7 'P
Notes: smallleUers with an explanation at the end of the list. For example: • actively traded g assumed dividend yield after rights issue xd ex-dividend xc ex-scrip issue 'P figures or report awaited Price:yesterday's middle closing price, sometimes with "xd" or ''xc'' or some other symbol whose meaning is clarified in the notessection. ''xd'' meansthat buyers of the sharedo not get the dividend. This goes instead to the previous ownerof the shares. "xc" means that the buyers of the share do not get the scrip issue. + or - : whether the price roseor fell from the previous day'sclose. 1991 highand low columns: theseare the highest and lowest prices reached during theyear so far. MKT Cp£M:thisisthe marketcapitalisation, the productof theshare price and the number of sharesin issue. It isgiven in unitsof £million. Yield Gr's: thisisthe gross yield, derived fromthe dividend expressed relative to the share price beforeapplying Advanced Corporation Thx at the appropriate rate (25%at the time of writing). PIE: thisis the price to earnings ratio,given by the currentshare price divided by the earnings per share, where earnings are the after-tax profits. PRICE TO EARNINGS (PIE) RATIOS These are worth a section of explanation since they are widely used by the investment community to attempt to value the shares of a company. If an investor knows the profits and the number of shares in that company issued, he can calculate the PIE figure, although the FT carries this out for you.
20
Winning on the Stock Market
The number of shares in issue, and the earnings of the company are obtained from the latest report of the company to its shareholders. By writing to the company secretary, you should be able to get hold of a copy of this even if you are not a shareholder. You could certainly telephone to find out these latest figures. In the figures given for the few shares in the Chemicals sector, only one had a PIE figure. That was Akzo with a value of 12.7.Byitself, this is a meaningless figure, because what the investor needs to be able to do is compare Akzo with other shares, and even better to compare it with other shares within the same sector. Without this comparison, all that one can say is that a high PIE ratio means that other investors think the prospects for the company are good, while a lowPIE ratio mean that investors do not think very much for the prospects of that company. The trouble with high PIE values is that unrealistic expectations may be built into the share price. A smart investor may be able to find a company with a low PIE ratio which has been overlooked by the market in general, where the prospects have the potential to become transformed. He can then expect that the share price will take off rapidly. Fortunately, the FT does provide us with a way of ranking one company against other companies in terms of their PIEs. This lies in the table of FT Actuaries Share Indices to be found on the inside pages. The great value of this is that details are kept for each sector, not only estimated PIEs for each sector, but special indices for each sector. By plotting the histories of these just as if they are shares, investors can spot sectors which are out of fashion, and which may provide an opportunity to get in before they become fashionable. From the PIE point of view, at the time of writing the PIE values ran from 6.89 for the metals/metal-forming sector up to 21.4 for insurance brokers. The number of companies in that sector is also stated, e.g. 8 for metals and metalforming and 8 for insurance brokers. Note the difference between electronics, with a PIE of 15.68 for 26 companies and electricals, With a PIE of 11.34 for 10 companies. Thus the market rates the earnings prospects of electronics much higher than electricals. Taking specific companies in these sectors, GEC, an electrical, had a PIE of 12, below the sector average, while Eurotherm in the electronics sector, had a PIE of 30.9, above the sector average. The market therefore judges that there is an excellent earnings potential for Eurotherm, as one would expect for a company operating in the high-tech electronics and control market. n the other hand the market has much lower expectations for GEC. Investment analysts spend a great deal of time analysing the PIE ratios for companies and analysing closely the companies within a sector that interests them. They pay visits to these companies and talk to higher management. The management goes along with this in the hope of a favourable recommendation from the analysis, which usually comes out as a broker's circular. A favourable report will elevate the share price, allowing the management to glow with
Brokers and Financial Information
21
satisfaction. Sometimes, to the fury of the management, analysts come away with a "sell" recommendation for that company's shares. Although the background information given in the financial press is useful as an aid towards understanding the overall investment climate, the methods discussed in this book focus simply on the price of the share, and the way in which this price has moved over the course of time. Thus the investor needs to do no more than update his share price histories from whatever source he uses, newspaper, Teletext, direct feed of prices from a databank via the telephone line, or prices on floppy disk. He should avoid shares which are not labelled as being active stocks, and stay mainly with the shares with the highest market capitalisations. By this means he will reduce the unfavourable spread of prices to the minimum.
3 セ。イォ・エャョヲオ」ウ
There is an important question to be asked about the stock market that will influence totally the way in which we approach investment. The question is this: Does the market act more like an indivisible whole, or more like a set of completely individual shares all doing their own thing? The focus of newspapers and TV on the FfSE100 Index as the barometer of share price movement may lead investors to the conclusion that it is the market itself which is of over-riding importance, and individual shares simply mirror the movement of the market itself. A moment's thought dispels this idea, since quite obviously an individual share must also reflect the view that investors have of that particular company. This may lead to the opposite view of the market, that it is simply a collection of unique shares, each behaving in complete isolation from every other share. The truth is, as is frequently the case, somewhere in between these two extremes, since there are several influences acting on an individual share price. The important influences can be summarised as in Figure 3.1. There are four of these, world markets, the UK market, the sector influence and the feeling about the company itself. Anyone who held shares at the time of the 1987 crash will appreciate the effect of world markets, since the crash started in Wall Street before reverberating around the world. The term "world markets" though is something of a misnomer, since there are just three markets of any importance, Wall Street, which is measured by the Dow-Jones Index, Tokyo, which is measured by the Nikkei-Dow Index, and London, measured primarily by the FfSE100 Index. Of these "Big Three", Wall Street is by far the most influential. There are a host of minor markets, such as Hong Kong and Sydney, but these exert very little influence in global terms. The UK market is conveniently divided into sectors, which are listed in Table 3.1 (p.27). The share pages of the Financial Times have now been radically altered so as to allocate shares more rationally into appropriate sectors. Just as the market as a whole is measured by the FfSE100 Index, then so are the sectors also
Market Influences
WORLD
COMPANY
MARKETS
I UK
23
-:
STANDING
FTSEIOO
2600
2400
False signals
2200 2100 2000
JFPI APlJ JA SON DJFPl APl JJA SON DJ FPlAPI JJ ASO "DJ FPlA PlJJ ASO
89 90
90 91
91 92
92 93
Figure7.13 False signals inthe 41-week averageduringthe periodJune to December 1991. Theseare the occasions when the averagechangeddirection for onlya weekor two before resuming its previous course
Winning on the Stock Market
108
In this chapter we have presented five methods of dealing with the problem of long term moving averages in order to emphasise how real the problems are and how essential it is to overcome them. The NN20 method was the standard against which other methods have to be judged in terms of their ability to reduce the delay in signalling the change in direction of the long term dominant trend in the FfSE100 Index and their ability to improve the problem of false signalling of the reversal of the dominant trend. The smoothing technique, while giving very impressive curvaceous trends, was actually inferior to the NN20 method in terms of the delay in signalling reversals in the dominant trend, although it did eliminate false signals. The graphical extrapolation method showed some improvement over the NN20 method in certain cases but not in all situations. The two most useful methods to give early warning that the current status of the dominant trend in the FfSE100 Index had reversed direction were the mini-market and the price analysis method. Although the individual shares ofthe mini-market are analysed by the NN20 method, and therefore individually are subject to the same drawbacks as when we apply this method to the FTSE100 Index, it is the spread of behaviour, with many shares anticipating a change in direction of the FTSE100 Index that gives it its power. Because of its advance warning property, and especially because of the excellent way it deals with the question of false signals, it is the method of choice for determining the status of the dominant trend. Having said that, it is useful to have confirmatory evidence 800 . . . - - - - - - - - - - - - - - - - - - - - - - - - - - - . , . , FTSElOO
700 600
soo 400 300 200
lOO 000
JFIl AMJ JA SON DJFM AI'! JJA SON DJ FMAPl JJ ASO NDJ FI'lA I'lJJ ASO
B9 90
90 91
91 92
92 93
Figure 7.14 Falsesignalsare removed by takinga second, 19-week centred averageof the
original 41-weekaverage
The Investment Climate
109
from another, different method. Thus, for the sake of the small amount of additional effort, it is well worth while to run the price analysis method to determine the status of the average as it was 10 weeks in the past.
8 Picking Shares
There are so many shares available for investment that the selection of just a handful for further evaluation is a formidable task. Even with a computer system, the time taken to analyse these thousands of shares makes an approach based upon such an analysis totally out of the question. Thus, the only way forward is to use some preliminary filtering system which can, rapidly and efficiently, reduce this large number to much more manageable proportions. One of the many criteria which can be applied is one of marketability. Since there are so many shares available, it makes no sense to include shares for which there is not a ready market. Even if such shares make a reasonable paper profit, it will be found that the eventual selling price will be profoundly disappointing. In the Financial Times at present, shares for which there is an active market through the SEAQ system are denoted by a small open square after their names. While there is still a large number of these, they account for about one-third of the market as a whole. Therefore the investor should not look outside of this group for investment opportunities. The potential for profit from a share depends obviously upon its volatility.The more volatile a share is, the larger is the difference between its low value and its high value. Naturally, the largest profit comes from those shares with the largest positive volatility, i.e, the volatility from a low to a high value. Any selection of shares based just on volatility will include shares with negative as well as positive volatilities, but negative volatilities will be weeded out by further analysis. Since we can only select shares by looking at their historical volatility, then it must be conclusively demonstrated that shares which have been volatile in the most recent past will continue to remain so in the recent future. If this turns out to be not the case, then any selection based upon volatility is a waste of time. A proper scientific approach to this problem would have to use a technique well established in medicine - the idea of a control group against which the results of the test group can be checked statistically. In theory, such a control group needs
Picking Shares
111
to be selected at random from the share pages of the Financial Times - another task for the dartboard investor. In practice, we have already in this book used a control group - the alpha share pool of about 120shares - for much of the research into investment methods. This alpha group has the advantage of being a good representation of the whole market in terms of the spread of these shares amongst the various market sectors, and although they have been selected as alpha shares because they have the highest market capitalisations, the spread amongst the sectors takes precedence as a criterion for using them as a control group. If we are unable to demonstrate that a selection procedure of shares based upon volatility does not give a group of shares with a higher profit potential than this group of alpha shares, then there is no point in going through such a procedure. The alpha shares can then be used as our primary pool of shares ready for additional analysis when the time for investment arrives. This time for investment will depend upon a change in the status of the dominant trend in the FfSE100 Index to a rising mode. The simplest way of selecting this volatile group of shares is to run down the column in the newspaper which lists the high and low values for the share over the previous year or part of a year. Early in the year, the period is split between the current year and the previous year, i.e. "1991/92 High Low",whereas later in the year the period will be restricted to that year, e.g. "1992 High Low". Which of these two forms appear is totally irrelevant, since all we are interested in is the ratio of the high to the low value. We aim to end up with a group of about 100 to 150 shares which on the basis of the ratio of their high to low values over this previous period are the most volatile in the market. As a first scan, it is convenient to set a cut-off point of two for the ratio. This makes it easy to do without calculator. It is fairly easily to decide mentally whether the high value is twice or more times the low value. If it is, just put a tick alongside the share name. At the end of the scan down the share pages, if it is found that there are too few shares, then the ratio can be reduced somewhat to say 1.75 and the scan carried out again. It might be necessary to use a calculator for such an odd value, but bear in mind that extreme accuracy is not required. Even if a few shares are missed by the procedure, the validity of the method will not be affected. To show the effectiveness or otherwise of the selection procedure, some 115 shares were selected in this way on the 22nd February 1991.This date was chosen because it was the point at which the percentage of rising 41-week averages in the mini-market (see last chapter) rose above 60%, signifying a positive change in direction of the dominant trend in the market. The shares are presented in decreasing order of volatility. The highest ratio of high to low was 19.79, and the lowest 1.9. The average volatility was 3.65 for this group. The list is given in Table 8.1. By comparison, the group of 120 alpha shares, whose volatility was calculated on the same date, had a highest value of 17.64 and a lowest value of 1.23. The
112
Winning on the Stock Market
Table 8.1 The most volatile shares on the basis of the ratio of 1991 high to low price, chosen on the 22nd February 1991 Brent Walker WPPGroup Next Berisfrd Speyhawk Rosehaugh Anglia Sec Erskine House Cookson Wickes Securiguard TIP Europe Misys Sketchley Thlfos Clayform Props Gold Greenlees lVS Entertain Barren (Henry) Goldsmiths Group Cityvision Howden Group Business 'Iech Lawrence (W) BET Burton Group 600 Group Robinson (Thos) Firth (G.M.) AB Electronics Countryside Barrat Dev HlV Etam Crest Nichol P&P Wheway Shandwick Markheath Emess Phoenix limber Norcros Utd Scientific Standard Chartered Mountleigh Christies InU Cannon St Inv
HIGH
LOW
RATIO
376
19 42
19.79 17.02 16.62 10.79 9.42
715
108
6.5
151
14
35 476 132 202
59 17
8
8.07
7.76 7.21
306
28 48
209 327 212 377 374
33 52 34 62 65
228 239
44
5.18
5.09
216
47 43
184 280
38 58
4.84
145 149
32 34 39.5 54
6.38
6.33 6.29 6.24
6.08
5.75 5.02 4.83 4.53 4.38
67
4.30 4.20 4.17 4.15
226 115
55 28
4.11 4.11
109 87
4.04
299 231
27 22 76 60
3.85
216
57
3.79
136
36
3.78
196 217 220 140
52 59
3.77
39
3.59
148 86
43 26
3.44
98
30
3.27
110 248 85 616 163
34
3.24
77 28 205 55
3.22 3.00 2.96
142 58
2.86
170
227 96 278
413 166
23
60
3.95 3.93
3.68 3.67 3.31
3.04
2.91 (continues)
Picking Shares
113
Table 8.1 (con/d)
Eurocopy Blacks Leisure Budgens Bunzl Stakis Saville Gordon Carlton Comms Canning(W) MoreO'Ferr Midland Ransomes Granada FKI Trafalgar House Unitech BodyShop PenUand Group Logica Norex Bridon LexService Ratners Aegis group Sirdar Devenish Lovell (YJ.) Avon Rubber Aran Energy Young (H) Premier Cons BridgendGroup FiveOaks Invs Reuters Cresta Holdings McLeod Russel Eurotunnel NMW Computers Adwest Group APV
Finlay (lames) INVESCOMIM Resort Hotels CRTGroup CapitalRadio Wellcome Thames TV Quadrant group Renold
HIGH
LOW
256 82 113 118 98 60 810 239 433 404 175 357 93 382 399 326
90
96
358 218 209 297 275 280 90 286
256 530 69 88
110 62 71 1318 61 143 689 39 176 157 125 131 112.5 110 258 795 555 193 104
29 40 42 35 22 298
RATIO
2.84 2.83 2.83 2.81 2.80 2.73 2.72
88
2.72
160 150 65 133 35 145 152 125 37 139
2.71 2.69 2.69 2.68 2.66 2.63 2.63 2.61 2.59 2.58 2.53 2.52 2.52 2.50 2.50 2.50 2.49 2.49
86
83 118 110 112 36 115 103 214 28 36 46 26 30 557.5 26 61 295 17 77
69 55 58 50 49 115 360 253 88
48
2.48
2.46 2.44 2.39 2.38 2.37 2.36 2.35 2.34 2.34 2.29 2.29 2.28 2.27 2.26 2.25 2.24 2.24 2.21 2.19 2.19 2.17 (continues)
114
Winning on the Stock Market
Table 8.1 (comd) Calor Group Siebe Costain Stanley Leisure Peek Kalamazoo HazIewood Frogmore Ests Cray Electronic SeniorEng Suter Bullough Evered Ibstock Johnson Assoc Brit Ports Berkeley Group Clarkson CH) Hewden-Stuart VauxGroup
HIGH
LOW
RATIO
420 530 318 206.5 91 39
280 268
194 245 148 97 43 18.5 117 223 33 32 83 71 75 76 167 142 138
124 329
174
2.16 2.16 2.15 2.13 2.12 2.11 2.09 2.06 2.06 2.05 2.02 2.01 2.01 2.00 1.99 1.97 1.94 1.94 1.89
244 460
68 65.5 168 143 151 152 333
64
average volatility for this group was 2.04. Quite clearly, in terms of average volatility, the test group has a higher value than the control group of alpha shares, as one would expect since we are considering that the control group is somewhere around the market average. At the end of the year, on 20th December 1991, the mini-market showed signs that the dominant trend was turning negative, and so this is a good point to assess whether the test group showed highest volatility over the period since February 1991 than did the control group. The volatilities are taken as the ratio of the price at 20th December 1991 to that at 22nd February 1991 if the first is higher than the second, and vice versa if the price had fallen over the period. Table 8.2. shows the test group of shares, now sorted in terms of decreasing volatility over this period. The group of 115 shares had a highest value of 9.71, with an average volatiJityof 1.85. In contrast, the control group of alpha shares had a highest value of 4.54 and an average volatility of 1.33. In statistical terms the difference between these two sets of results is significant, showing that shares which are more volatile than the average based upon the historical price movements over a certain period of time will remain more volatile than the average over a similar period into the future. Although there is not room to present vast amounts of supporting data here, similar results are obtained at other points in the past where the dominant trend made a positive change in direction (Stocksand SharesSimplified, see Appendix). It is not possible to gain an appreciation of the percentage gains and losses made from the data in Table 8.2 since the shares were ranked in order of volatility. High volatility may mean a large gain, but it may also mean a large loss. The percentage changes between February and December 1991 for these shares are shown in
Picking Shares
115
Table 8.3. The highest gain in the alpha shares was 105.5%, while the highest gain in the test group was 115.4%. The largest loss in the alpha group was 78% while in the test group is was 89.7%. These are not significantly different, but what is different is if the investor is able to capture the best 30 performances in each group. In the alpha group the best 30, running from 105.5% to 1.3%, averaged 29.5%, but in the test group the gains ran from 115.4% to 15.8%, averaging 49.7%. Thus quite clearly, the investor who is able to maximise his performance has the potential to achieve considerably higher gains from the more volatile group of shares. Since the thrust of this book is to enable the investor to maximise his performance, then it is obvious that this prior selection of the most volatile shares is an essential part of this process. Table 8.2 The shares from Thble8.1 ranked on the basis of their volatilitybetween the 22nd February 1991 and the 20th December 1991
lVS Entertain Rosehaugh Lovell (YJ.) Speyhawk Ratners Mountleigh Cannon St Inv Brent Walker Costain Robinson (Thos) Business 'Iech Ransomes P&P Crest Nichol Burton Group Quadrant group Wheway Blacks Leisure Misys AranEnergy Next Gold Greenlees Premier Cons Clayform Props Sirdar Etam Wellcome WPPGroup 'Iratalgar House Barrett (Henry) Anglia Sec Ibstock Johnson Phoenix Timber Evered
HIGH 184 476 256 358 275 163 166 376 318 109 227 175 220 217 226 193 140 82 377 69 108 216 110 239 90 196 795 715 382 280 132 152 110 151
LOW RATIO 22-2-91 20-12-91 RATIO 38 4.84 68 7 9.71 59 8.07 74 8 9.25 103 2.49 150 17 8.82 38 9.42 83 10 8.30 110 2.50 153 25.5 6.00 55 2.96 68 12 5.67 4.93 58 2.86 69 14 19 9.79 49 12.75 3.84 148 2.15 204 58 3.52 27 4.04 33 10.5 3.14 4.20 88 29 3.03 54 2.69 2.81 65 87 31 60 3.67 109 40 2.73 37 59 3.68 94 2.54 90 35.5 2.54 4.11 55 88 2.19 108 46 2.35 66 29 2.28 39 3.59 2.83 29 52 112 2.15 62 6.08 94 200 2.13 28 2.46 34 16 2.13 6.5 16.62 27.2 56 2.06 5.02 100 198 1.98 43 47 46 2.39 24 1.96 47 5.09 62 32 1.94 89 36 2.50 46 1.93 52 3.77 115 221 1.92 360 2.21 497 950 1.91 42 17.02 1.85 61 33 145 2.63 124.5 1.84 229 58 4.83 80 44 1.82 7.76 26 1.81 17 47 76 2.00 110 65 1.69 34 3.24 35 21 1.67 75 2.01 100 60 1.67 (continues)
116
Ulinning on the Stock Market
Table 8.1 (contd)
Berisfrd Assoc Brit Ports CRTGroup Countryside Christies Inll Sketchley Thames1V Body Shop Stanley Leisure Emess Barrat Dev Standard Chartered AB Electronics FIve Oaks Invs Senior Eng Stakis Resort Hotels Erskine House Lawrence (W) Thlfos Eurotunnel Markheath Siebe Penlland Group Lex Service 600 Group Firth (G.M.) Shandwick Budgens Devenish BET Cresta Holdings Bridgend Group INVESCOMIM Young (H) Saville Gordon Reuters Cookson Kalamazoo More O'Ferr Avon Rubber Canning(W) Carlton Comms Clarkson (H) Howden Group Midland Renold Berkeley Group
mGH
LOW
151 333 110 231 413 374 555 326 206.5 98 216 616 299
14 167 49 60 142 65 253 125 97 30 57 205 76 30 32 35 50 28 23 44 295 26 245 37 118 28 22 43 40 115 67
71
65.5 98 112.5 202 96
228 689 86
530 96
297 115 87 148 113 286
278 61 62 131 88 60 1318 306 39 433 530 239 810 268
170 404 104 280
26 26
58 36 22 557.5 48 18.5 160 214 88 298 138 39.5 150 48
142
RATIO 22-2-91 20-12-91 RATIO
10.79 1.99 2.24 3.85 2.91 5.75 2.19 2.61 2.13 3.27 3.79 3.00 3.93 2.37 2.05 2.80 2.25 7.21 4.17 5.18 2.34 3.31 2.16 2.59 2.52 4.11 3.95 3.44 2.83 2.49 4.15 2.35 2.38 2.26 2.44 2.73 2.36 6.38 2.11 2.71 2.48 2.72 2.72 1.94 4.30 2.69 2.17 1.97
26
208 76 132 201 74 274 213 119 38 80 273 112 32 43 48 67 53 33 85 473 29 378 78 161 42 26
57 43 159 139 36 36 89 49 35 789 120 38.5 189 265 124 393 158 58 175 64
280
16 333 121 83 127 117 176 331 184 25.5 54 402 77 22 61.5 34 94 74 24 113 358 22 494 100 206 33 33 45 34 198 173 29 29 72 60 29 949 100 45.5 160 313 146 460
183 51 198 57 250
1.63 1.60 1.59 1.59 1.58 1.58 1.56 1.55 1.55 1.49 1.48 1.47 1.45 1.45 1.43 1.41 1.40 1.40 1.38 1.33 1.32 1.32 1.31 1.28 1.28 1.27 1.27 1.27 1.26 1.25 1.24 1.24 1.24 1.24 1.22 1.21 1.20 1.20 1.18 1.18 1.18 1.18 1.17 1.16 1.14 1.13 1.12 1.12
(continues)
PickingShares
117
Table 8.2 (con/d)
HIGH Beazer Bridon APV Unitech Adwest Group Goldsmiths Group Suter 'huxGroup NMW Computers Capital Radio Aegis group H1V McLeod Russel Peek Norcros TIP Europe Securiguard Hewden-Stuart Calor Group Bunzl Cray Electronic Utd Scientific Frogmore Ests FKI Bullough
Finlay (James) Hazlewood Norex Granada Logica Eurocopy Cityvision Wickes
178 209
157 399 176 145 168 329 39 258 280 136 143 91 248 212 327 124 420 118 68 85 460
93 143 125 244 218 357 358 256 149 209
LOW 81 83 69 152 775 32 83 174 17 115 112 36 61 43 77 34 52
64 194 42 335 28 223 35 71 55 117 86 133 139 90 34 33
RATIO 22-2-91 20-12-91 RATIO 1.11 2.20 128 115 2.52 99 1.11 110 2.28 104 115 1.11 2.63 217 197 1.10 2.29 94 103 1.10 4.53 36 33 1.09 2.02 122 1.09 112 1.89 220 202 1.09 2.29 34 37 1.09 2.24 169 183 1.08 125 2.50 135 1.08 3.78 54 1.08 50 84 1.08 2.34 78 2.12 57 53 1.08 1.07 3.22 111 104 1.07 6.24 57 53.5 98 1.05 6.29 93 1.05 1.94 84 80 2.16 1.05 225 215 2.81 80 83 1.04 64 1.03 2.06 62 42 43 1.02 3.04 2.06 317 324 1.02 2.66 47 48 1.02 2.01 1.02 112 114 2.27 63 1.02 62 2.09 179 181 1.01 2.53 131 130 1.01 2.68 1.01 186 187 1.01 2.58 193 194 2.84 90 90 1.00 4.38 48 1.00 48 6.33 1.00 56 56
Table 8.3. The shares from Thble 8.1 ranked on the basis of their percentage gains or losses between 22nd February 1991 and the 20th December 1991.
HIGH Blacks Leisure Misys Next Gold Greenlees Sirdar Etam Wellcome Assoc Brit Ports
82 377 108 216 90 196 795 333
LOW RATIO 22-2-91 20-12-91 %GAlN 29 2.83 52 112 115.38 62 94 200 112.77 6.08 16.62 6.5 27.25 56 105.50 43 100 198 98.00 5.02 36 46 89 93.48 2.50 52 115 221 92.17 3.77 497 950 91.15 360 2.21 208 333 60.10 167 1.99 (continues)
118
Winning on the Stock Market
Table 8.3 (coned) CRTGroup Sketchley Body Shop Stanley Leisure Standard Chartered Senior Eng Resort Hotels Erskine House Thlfos Siebe PentIand Group Lex Service Firth (G.M.) Devenish BET Young (H) Reuters Kalamazoo Avon Rubber Canning(W) Carlton Comms Clarkson (H) Midland APV Adwest Group Suter NMW Computers Capital Radio HTV McLeod Russel Securiguard Bunzl Cray Electronic Utd Scientific Frogmore Ests
FKI Bullough Finlay (lames) Hazlewood Granada Logica Wickes Eurocopy Cityvision Norex Calor Group Hewden-Stuart TIP Europe
HIGH 110 374 326 206.5 616 65.5 112.5 202 228 530 96
297 87 286 278 88 1318 39 530 239 810 268 404 157 176 168 39 258 136 143 327 118 68 85 460 93 143 125 244 357 358 209 256 149 218 420 124 212
LOW RATIO 22-2-91 20-12-91 %GAlN 49 2.24 76 121 59.21 65 5.75 74 117 58.11 125 2.61 213 331 55.40 97 2.13 119 184 54.62 205 3.00 273 47.25 402 32 2.05 43 43.02 61.5 50 2.25 67 94 40.30 28 7.21 39.62 53 74 44 5.18 85 113 32.94 245 378 2.16 494 30.69 37 78 2.59 100 28.21 118 2.52 161 27.95 206 22 3.95 26 26.92 33 115 159 2.49 198 24.53 67 4.15 139 173 24.46 36 2.44 49 60 22.45 57.5 789 2.36 949 20.28 18.5 2.11 38.5 45.5 18.18 214 2.48 265 313 18.11 88 17.74 2.72 124 146 17.05 298 2.72 393 460 138 1.94 158 15.82 183 150 175 2.69 198 13.14 69 2.28 104 115 10.58 77 2.29 94 9.57 103 83 2.02 112 122 8.93 17 2.29 34 37 8.82 115 169 2.24 183 8.28 36 3.78 50 54 8.00 61 2.34 78 84 7.69 52 6.29 93 5.38 98 42 2.81 80 83 3.75 33 2.06 62 64 3.23 42 28 3.04 43 2.38 223 317 2.06 324 2.21 47 35 2.66 48 2.13 71 2.01 112 114 1.79 55 2.27 62 63 1.61 117 2.09 179 181 1.12 133 2.68 186 187 0.54 139 2.58 193 194 0.52 33 6.33 56 56 0.00 90 0.00 2.84 90 90 34 4.38 48 48 0.00 86 2.53 131 -0.76 130 194 -4.44 2.16 225 215 64 -4.76 1.94 84 80 34 6.24 57 -6.14 53.5 (continues)
PickingShares
119
Table 8.3 (contd)
Norcros Peek Aegis group VauxGroup Goldsmiths Group Unitech Bridon Beazer Berkeley Group Renold Howden Group More O'Ferr Cookson SavilleGardon INVESCOMIM Bridgend Group Cresta Holdings Budgens Shandwick 600 Group Markheath Eurotunnel Lawrence (W) Stakis Five Oaks Invs AB Electronics Barrat Dev Emess ThameslV Christies InU Countryside UlIey Berisfrd Phoenix Timber Evered Ibstock Johnson Anglia Sec Barrett (Henry) 'Irafalgar House WPPGroup Clayform Props Premier Cons Aran Energy Wheway Quadrant group Burton Group Crest Nichol P&P
HIGH 248 91 280 329 145 399 209 178 280 104 170 433 306 60 131 62 61 113 148 115 86
689 96
98 71 299 216 98 555 413 231 66.75 151 110 151 152 132 280 382 715 239 110 69 140 193 226 217 220
LOW
RATIO 22-2-91 20-12-91 %GAIN 3.22 111 104 -6.31 43 2.12 57 -7.02 53 112 2.50 135 -7.41 125 174 1.89 220 202 -8.18 32 4.53 36 33 -8.33 152 2.63 217 197 -9.22 83 2.52 110 99 -10.00 81 2.20 128 115 ·10.16 .10.71 1.97 142 280 250 48 2.17 64 57 -10.94 4.30 39.5 58 51 ·12.07 160 2.71 189 -15.34 160 48 6.38 120 100 -16.67 22 2.73 35 29 -17.14 -19.10 58 2.26 89 72 26 2.38 36 29 -19.44 26 2.35 36 29 ·19.44 43 34 -20.93 40 2.83 43 57 -21.05 3.44 45 28 4.11 42 33 -21.43 26 29 -24.14 3.31 22 2.34 473 -24.31 295 358 4.17 -27.27 23 33 24 35 2.80 48 ·29.17 34 30 2.37 32 -31.25 22 76 77 -31.25 3.93 112 80 54 -32.50 57 3.79 -32.89 30 3.27 38 25.5 274 253 2.19 176 ·35.77 -36.82 142 2.91 201 127 -37.12 60 3.85 132 83 39 45.5 28.5 1.71 ·37.36 -38.46 14 10.79 26 16 -40.00 34 3.24 35 21 75 -40.00 2.01 100 60 -40.91 76 2.00 110 65 17 7.76 47 26 -44.68 80 44 -45.00 58 4.83 -45.63 145 2.63 229 124.5 42 17.02 61 33 -45.90 47 5.09 -48.39 62 32 46 2.39 47 24 -48.94 28 2.46 34 16 ·52.94 39 3.59 66 29 -56.06 88 2.19 108 46 ·57.41 -60.56 55 4.11 90 35.5 -60.64 59 3.68 94 37 -63.30 60 3.67 109 40
77
(continues)
120
Winning on the Stock Market
Table 8.3 (contd)
HIGH Ransomes Business tech Robinson (Thos) Costain Brent \\alker Cannon St Inv Mountleigh Ratners Speyhawk Lovell (Y J.) Rosehaugh lVS Entertain
175 227 109 318 376 166 163 275 358 256 476 184
LOW 65 54 27 148 19 58 55 110 38 103 59 38
RATIO 22-2-91 20-12-91 %GAlN 2.69 87 -64.37 31 4.20 88 29 -67.05 4.04 10.5 -68.18 33 2.15 204 58 -71.57 19.79 49 12.75 -73.98 2.86 69 14 -79.71 2.96 68 12 -82.35 -83.33 2.50 153 25.5 9.42 83 10 -87.95 -88.67 2.49 150 17 8.07 74 8 -89.19 4.84 68 7 -89.71
9
Dominant Trends
In the last chapter we saw how, at the time when the dominant trend in the market turned positive, we could select a group of shares which had a high probability of maintaining a higher level of volatility than the market in general. We also saw that volatilitycan work in favour of the investor, or it can work against him, depending upon whether the share price rises or falls. We saw that the average performance of the volatiles was very similar to that of the control group, averaging a loss of 7% against a loss of 6% for the control group between February and December 1991. Thus the dartboard investor would expect a small loss from either group of shares. There is a small probability (less than 1%) that the dartboard investor might have chosen the top performers of either group, in which case he would have made a gain of 105.5% in the case of the alpha shares (from Next) or 115% from the volatiles (from Blacks Leisure). On the other hand there is a similar small probability that he might have made a loss of 78% in the case of the alpha shares (from ASDA) or 90% in the case of the volatiles (from TVS Entertainment). Looking at the two groups in another way, out of the volatiles, 49 out of 115, i.e. 42.6%, made a profit, while out of the alpha shares, 47 out of 120, i.e. 39.2%, made a profit. Thus the dartboard investor has roughly a 40% chance of selecting a winning share out of these two groups of shares. All that the selection of the group of volatile shares has done for us so far is to give us a potential for greater profits or greater losses. We now have to develop a further selection procedure that keeps us away from the losing shares and points us in the direction of the winning shares. Since no method of selection is going to be perfect, we have to accept the fact that we will end up with some losers, but as long as we are strict about applying the stop-loss method discussed in an earlier chapter, we will not suffer the large losses which appeared in the latter half of Table 8.3. In Chapter 8 we adopted the rule that we should never invest if the dominant trend in the market is running adversely. The best method of determining the
122
Winning on the Stock Market
status of the dominant trend in the market was to look at the direction of the 41-week average of the shares in the mini-market of alpha shares. As long as 60% or more of these were rising, then it was acceptable to invest in a share that passed all the further criteria for investment. In this chapter we come to one of these further criteria: the dominant trend in the share itself must be rising at the time it is proposed to invest. We saw the importance of a rising dominant trend in a share price when we discussed the potential gain under various conditions. The maximum gain was obtained when both the long term and the short term trends were rising. Since we have to accept that we will make a certain number of losing investments, it is vital that we maximise the gains from those decisions which are correct, otherwise our overall performance may well be disappointing. MOVEMENT OF DOMINANT TRENDS OVER A PERIOD For the volatile shares as listed in Table 8.3 in the last chapter, 42.6% of these made a profit if the buying point was taken to be the 22nd February 1991 and the selling point the 20th December 1991. These dates are somewhat artificial from the investment point of view, since we are not suggesting that the shares were actually bought and sold on these particular dates. The importance of these dates lies in the facts that the February date was the point at which the dominant trend in the market turned up, and that the December date represented a sensible period over which to check the volatility of the share during what was a favourable time for investment. Thus it is important to understand that it was the period between these two dates that was favourable for investment. Only if other criteria were satisfied would any share be bought on the 22nd February and sold on the 20th December. There are four possible situations for the dominant trend in a particular share on these two dates: 1. The dominant trend is falling on the 22nd February but rising by the 20th December. 2. The dominant trend is falling on the 22nd February and still falling on the 20th December. 3. The dominant trend is rising on the 22nd February and still rising on the 20th December. 4. The dominant trend is rising on the 22nd February but falling on the 20th December. Of these four situations, the second will probably give rise to the largest loss and the third probably the largest gain. In situations 1 and 4 we may achieve a gain or a loss, depending upon where between the two extreme dates the change in direction of the dominant trend occurs. Thus if at the point when we think we have been given a signal to buy the dominant trend in the share price is falling, it is in the lap of the gods as to whether
Dominant Trends
123
it might change direction sooner rather than later. Although a short term uptrend at the time the investment is made may take the share price up slightly, this initial gain will soon be eroded by the falling dominant trend. Although a stop-loss will have a limiting effect, nonetheless it is inevitable that a loss will be made if the dominant trend continues to fall. Thus it is a pure gamble to invest at such a time just in the hope that the adverse trend will change for the better. The only sensible rule to obey is not to invest unless the dominant trend is rising. The opposite rule is obviously that we can invest when a buying signal is given provided the dominant trend is rising. There should be a qualification to this rule: if the dominant trend has been rising for a considerable time, say for more than about six months, then the probability is building up that this rise will come to an end. Under such circumstances the investor is better off looking for an alternative investment where the upward dominant trend is still young. The situations above are best illustrated by examples taken from the list of volatile shares in Table 8.3. However, there was no share in the list that followed situation 1, and therefore an example from the list of alpha shares, viz. Bass, is used as an illustration. Note that in these examples the 41-week average is plotted as a centred average, i.e. it is a good representation of the dominant trend itself. This is done more than a year after the period shown in the charts so that the position of the centred average is known across this whole period. Remember that at the time we consider an investment we can only estimate the current status of the dominant trend in the share - see later. Situation 1: Dominant trend falling on 22nd February, rising OD 20th December
Example: Bass (Figure 9.1) The reversal of the falling long term trend occurred in mid-August 1991, and the net effect of the price of the downtrend followed by an uptrend was to leave it virtually the same at the end of the period as at the beginning. The price was 501p on the 22nd February and 496.5p on the 20th December. Situation 2: Dominant trend faUing on 22nd February, falling on 20th December
Example: Bwton Group (Figure 9.2) Here we can see the drastic effect of a continuously falling long term trend. There
isjust a sign that the fall may have ended by December 1991, but the shares showed a fall of 60.6% over the period.
"Winning on the Stock Market
124
. セ
BASS
525
500
450
....•.............................................................
425
,
F
91
, ...............•............•.............
11
A
11
J
J
A
s
o
N
o
Figure 9.1 The Bassshare pricewithdominant trend from February to December 1991. The trend is falling in February but rising in December Situation 3: Dominant trend rising on nnd February, rising on 20th December
Example: Wellcome (Figure 9.3) Here we can see exactly the opposite effect to the adverse trend in Burton. The uptrend in Wellcome took the price up by 91.2% Situation 4: Dominant trend rising on 22nd February, falling on 20th December
Example (a): Trafalgar House (Figure 9.4) The dominant trend was rising from late January 1991, but reversed direction in July so as to be falling on the 20th December. The fall was more severe than the rise, thus causing the share price to undergo a loss of 45.6% between the two extreme dates.
Example (b):BET (Figure 9.5) As with Trafalgar House, the dominant trend changed direction during the period, but this time rather later, in early September. Besides the point that the period of fall was therefore only four months, whereas the period of rise was six months, it
125
Dominant Trends ............ セAャ[iG
120
.
110
100 90 BD
70 60
..
40 JO
.......•.•.•.•...•........•...•......•.•.•••..•••.•••••.•.•.•••••• F
91
"
j:I
"
J
J
j:I
s
o
Pt
D
Figure9.2 The Burton share pricewith dominanttrend from Februaryto December 1991. The trend is fallingcontinuously from Februarythrough December
LElLCO£
F
91
"
j:I
"
J
J
j:I
s
o
Pt
o
Figure 9.3 The Wellcome share price with dominant trend from February to December 1991. The trend is risingduring the whole period
Winning on the Stock Market
126 300
•......•.........•....•.•............•.....•....•.....•.•.•..•.... TRAF'
. セ
200
'"
.
1'0
.
.
125
F 91
"
Fl
"
J
J
Fl
s
o
N
D
Figure 9.4 The Trafalgar House share pricewithdominanttrend fromFebruaryto December 1991. The trend is rising in February but falling in December
can be seen that the rate of fall was slightly less than the rate of rise. Because of these two factors, BET still finished at a higher level in December than it started in February, giving a gain of 24.5%. Now we have a clear picture of the directional movement of dominant trends in these examples over the period in 1991 when the market trend was favourable, it is of interest to see how our estimates of the trends at the time a buying signal was given would affect the final decision to buy or not to buy. The way in which the buying signals were given by short term trends is discussed in the next chapter. The interest in this chapter is in the qualification of the buying signal by estimates of the current status of the long term trend. As we saw in the last chapter, there are several ways in which the current status can be estimated. The NN20 method, in which the calculated value of the 41-week average at a point in time is taken as the estimate of the current status of the trend, was seen to be the least accurate, but had the virtue of simplicity, since only the one calculation is needed. The price analysis method required more effort, but gave a much better picture in most cases of the current direction of the trend, and whether the 41-week average might have changed direction during the intervening 2O-week gap. Although these two methods were compared for the FTSEl00 Index, it is necessary to compare these two methods in the case of particular shares in order to demonstrate conclusively the merits of the price analysis case.
Dominant Trends ZlO
127
............................•.•....•.•.•..............•.•....•.•.. BET
200
.
150
. .
125
..
.
..
F
'ill
.
.
"
A
"
J
J
A
s
o
N
o
Figure 9.5 The BET share price with dominant trend from February to December 1991
APPLYING THE NN20 METHOD Bass
There were three buying signals given by short term trends in the Bass share price during 1991 - one on the 26th April (price 477.5p), one on the 26th July (price 460p) and one on the 16th August (price 463p). Calculations of the 41-week average on those dates showed that the 41-week average was still falling on the first two occasions; only on the 21st August was the average rising. Bass could then have been bought for 463p, and it ended at 496.5p on the 20th December for an accumulated profit at that point of 7.2%. Buying on the other two dates would have given profits of 4% and 7.9%. Burton
There was just one buying signal given in Burton on the 9th September with the price at 36.5p. The 41-week average calculation showed it was falling, and so the share would not be bought. If it had been bought at 36.5p, then the price of 35.5p on the 20th December would have given a loss of 2.7%.
128
'Winning on the Stock Market
Wellcome There were two buyingsignalsin Wellcome,one on the 25thMarch withthe price at 503p, and one on the 29th November with the price at 837p.Only at the latter date was the 41-week average rising. With the price rising to 950p on the 20th December, buyingin November would have givena profit of 13.5%. If it had been bought in March at 503p,then the gainwouldhavebeen a massive 88.8%. Trafalgar House
There were two buyingsignalsgivenfor Trafalgar House - on 26th April at 254p. and on the 16th August at 237p. The 41-week average was falling on the first occasion but rising on the second one. If the share had been bought at that point, then with the price at 124.5p on the 20th December a loss of 47.4% would have been sustained. If the first buyingpoint had been followed, a loss of 50.9%would have followed. BET As withTrafalgar House there were twobuyingsignals givenfor BET, the firston
the 1st March with the price at 140p and the second on the 14th June with the price at 181p.The 41-weekaverage was fallingon both occasions. The price on the 20th December was 174p,so that buyingon the first occasion would have given a profit of 24.2%,.while buyingon the second occasionwould have given a loss of 3.8%. In summary, out of the ten buying signals given for these five shares, only on three occasions were the 41-week averages moving favourably. The buying operations that would have been carried out on these two occasionsgavegainsof 7.2% and 13.5%.in four months and three weeksrespectively, and a lossof 47.4% in four months. Out of the other seven occasions when the shares would not have been bought, the gains would have been 4%, 7.9%, -2.7%, 88.8%,-50.9%, 24.2%and -3.8%. Thus the method has protected us against the large loss of 50.9% in Trafalgar House, did not protect us againsta lossof47.4%inTrafalgar House and prevented us taking advantage of the large gain in Wellcomeof 88% and a good gain in BET of 24%. Notwithstanding the point that in practice we may well have sold the shares before or after the 20th December accordingto the sellingmethod discussedlater, and thus made better profits, it has to be said that the performance of this method is extremelypoor, virtuallyguaranteeing the investor a loss.
Dominant Trends
129
TIlE PRICE ANALYSIS METHOD The method is familiar from the discussion in the last chapter. If the method is to be an improvement on the NN20 method, it has to be able to allow us to take advantage of the large rise of 88% in WeUcome, denied to us by the NN20 method. It would also be useful if it would allow us to ride the good rise of 24.2% in BET. As well as taking part in these lost profitable rises, the method should keep us out of the other losing position which the NN20 method failed to do - the loss of 47.4% in Trafalgar House. It should also keep us out of the other three losing positions. This is a rather tall order, requiring us to be almost 100% successful. However, the estimation of the current status of the dominant trends in Wellcome, Trafalgar House and BET in the following examples shows that the method is extremely successful in achieving these goals. WeUcome
The position is shown at the time of the first buying signal, on the 25th March 1991, in Figure 9.6. The extreme right hand axis is at the point corresponding to the 22nd March 1991 (the point NOW), while the extreme left hand axis is the point 40 weeksback (NOW - 40), 15th June 1990. Thus the chart contains 41 data points, and the value of the 41-week average calculated from these points is the [mal plotted point in the 41-week centred average. This point appears at a time on the axis corresponding to the 2nd November 1990. To fill the 2O-weekgap between 2nd November 1990 and 22nd March 1991 with the estimated path of the average we adopt the procedure used in the last chapter. We assume that for the next twentyweeks the price will remain at its last value of 506p. As plotted at present, the drop-point for the next calculation of the average, i.e. on the 29th March is the point 40 weeks back from the present latest point, i.e. (NOW - 40), the point on 15thJune 1990 on the extreme left hand axis. The value ofthis point is 662p. Thus nextweek's price would have to rise above this level of662p if the 41-week average is to rise. The probability of this happening is so low as to be virtually negligible. Thus we can estimate the next point on the average (NOW - 19) as being much lower than the last true point. Looking at the left hand part of Figure 9.6, it can be seen that it is about 10weeks before the price falls below the value at NOW, i.e. below 506p. Thus as we step along from NOW - 40, comparing each point in turn with the price at NOW, we will estimate each successive point of the average to be lower than the previous value, i.e. for the average to fall. Thus we can fill in the next nine points from NOW - 19 (we have already estimated the first at NOW - 20) with a falling average. This takes us to the estimated position of the average at 11th January (NOW -10). It is then that we come to the situation that the next drop-point (on 24th August) at 423p is lower than the value of 506p at NOW. We can therefore
Winning on the Stock Market
130 ;100
•..........•.•...........................•......•.................
ilEl.LCO£
............. , .
600
nowセRッM
'
.
..
400
A
s
o
D
J 90 91
M
Figure 9.6 The position in Wellcome on the 25th March 1991. The chart covers 41 weeks from the 15th June 1990. The last calculated position of the centred 41·week average is on the 2nd November 1990, marked as NOW - 20
estimate that the average will rise. From that position onwards we see that the price does not rise above 506p, and therefore we can conclude that the average will keep on rising, and that the estimated status of the dominant trend at NOW is a rising one. Thus, with a buying signal being given by the short term trend, all is set fair for us to invest in Wellcome on the 22nd March 1991 with the price at 506p. This should be compared with the performance of the NN20 method, which kept us out of Wellcome until the 29th November, when the price had already risen to 837p. It simply remains now to confirm the accuracy of this prediction by showing the actual position of the average as calculated twenty weeks later. This is shown in Figure 9.7. We can see that the average turned up in the middle of December 1990, some four weeks earlier than we predicted. Our prediction is therefore astonishingly accurate, since we have to remember that we are predicting what the actual position will be 20 weeks ahead of the latest true value of the average. It would be about the middle of May, about eight weeks further into the future, before the NN20 method would tell us that the trend had changed direction.
131
Dominant Trends ;IlIO
•.•.•..•.•.•.•.•.•......•.•.•.•....•.....•.•....•...•.•.•.••••.•.•
WEl..LCCJ£
400
.
. セ
J 90
s
o
o
F
PI
FJaure 9.7 The actual course of the 41-week averageover this period can be determined by calculations on the necessary data that is available 20 weeksafter the end of the period
Trafalgar House
Thisisthe opposite case to Wellcome,since the NN20method on the 16thAugust gave us an estimate of the current status as rising,leading us to make an incorrect investment which lost 47.4%. The chart of Trafalgar House, with the centred 41-week average, is shown in F"JgUI'e 9.8.The chart covers 41 weeks, from 9th November 1990 (NOW - 40) at the extremeleft hand side to 16th August 1991(NOW) at the extreme right hand side. The last plotted position of the average is at the 29th March 1991 (NOW-20). The price at NOW is 237p. It takes 14 weeks from NOW - 40 before the price rises above this level of 237p. Thus we can estimate that for 14 weeks from NOW - 20 the averagewillcontinue to rise. This takes it to NOW - 6. At that point the price reaches 239p,fallsback to 229pbefore risingto 264p.It stayswellabove 237p untilthe 28thJune 1991. Thus the averagewillfallat NOW- 5, rise at NOW 4, and then continue to fall for at least these next four weeks. Thus we estimated that the average peaked out around NOW - 4, Le about the 19th July, and is still headed south on the 16th August. Therefore we would ignore the buying signal given by short term trends at this time. The actual path of the average for this period could be determined on or after the beginningof January 1992, and this is shownin Figure 9.9. Once again,we can see how astonishingly accurate the estimation was, with the average actually
Winning on the Stock Market
132 300
.
TRAF'
V5
.
1'-'-'-':"-'----"/"\' . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . • . . . • . • . . . . . . .
1;>5
...
.
"
90
,
J 90 91
F'
'NOW' :...'20'
"
.
"
J
J
Figure 9.8. The position in ltafalgar House on the 16th August 1991. The chart covers 41 weeks from the 9th November 1990. The last calculated position of the centred 41-week average is on the 29th March 1991, marked as NOW - 20
peaking out in late June, some four weeks prior to the estimated peak. Remember it would not be for twenty weeks after this actual peak, in November 1991that the NN20 method would give us this same information. BET
The position in BET on the 1st March 1991 is shown in Figure 9.10. The price shows a consistent fall across almost the whole of the 41-week period. Only during late January/early February did the price fall below the latest price of 140p. Thus the only prediction that can be made is for a fall in the 41-week average right up to the 1st March. The plot of the average made 20 weeks later (Figure 9.11) shows that the prediction was 100% accurate. The average did indeed fall at a consistent rate until the 1st March. It turned up the following week, but that was due to an unexpected rise in the price after the 1st March, which can be seen from the chart shown in Figure 9.12. The price rose from 140p on the 1st to 154p on the 8th and 167p on the 15th March. Obviously, these price rises could not be predicted in advance from the position shown in Figure 9.10.Note that this particular dominant trend was short-lived, petering out in the autumn of 1991. Without a stop-loss in operation the investor would have taken a nasty loss.
Dominant Trends
133
:soo
. TRAF'
:vs
.
N 'ilO
F"
J 'ilO 91
A"
J
A
J
Figure 9.9 The actual course of the 41-week averageover this period can be determined bycalculations on the necessary data that is available 20 weeksafter the end of the period
BET
300
.
:vs
.
225
200 175
ISO
.........NOW 20
125
.
100
. J 'ilO
A
s
o
"
D
so
J
91
F
.
"
Figure 9.10 The positionin BET on the 1st March 1991.The chart covers41 weeks.The last calculated positionof the centred 41-week averageison the 12thOctober 1990,marked as NOW-20
134
"Winning on the Stock Market BET
300 275
.........•.•.............................•...................•....
Figure 9.11 The actual course of the 41-weekaveragein BET over thisperiodcan be determined bycalculationson the necessarydata whichis available 20weeksafter the end of the period
BET 300
V5
225 200 Qセ
130 125
100
Figure 9.12 The rise in the dominant trend after 1st Marchcould not have been predicted in advance.The large price riseson the 8th and 15th Marchwere responsible for the risein the trend
Dominant Trends
135
THE REASON FOR FAILURE OF THE NN20 METHOD The NN20 method is only satisfactory as an estimator of the current status of the dominant trend at periods when the trend has not changed direction during the gap between the last true plotted point and the last data point. Thus, if a 41-week average is used for highlighting the dominant trend, the dominant trend should not have changed direction during the last 20 weeks. If it has done so, then the NN20 estimate is 100% incorrect. This can be stated in another way: the NN20 method will always give an incorrect estimate of the current status of the dominant trend during the first 20 weeks of the existence of this trend. This shows us that although 2OO-day and 4O-week averages are widely used in investment systems, they have this basic flaw that makes their application certain to lead to losses except for those periods 100 days or 20 weeks after the long term trend has changed direction. Since the most profitable time to enter an uptrend is as close to its beginning as possible, we are faced with two diametrically opposed requirements. In order to be correct about the direction of the dominant trend, we have to hold off an investment until we have just seen a turning point in the 41-week average. Then the new trend will have already been in existence for 20 weeks.The investment will be profitable as long as the dominant trend lasts for say30 weeks or so. If it does not, and changes direction again before such a time haselapsed, then we will be forced to exit the investment by virtue of the triggering of the stop-loss which we are applying. We saw in the recent behaviour of the FrSE100 Index in the last chapter that we can have periods in time, such as late 1991 to early 1992, in which the dominant trend lasts for less than 20 weeks, and evenfor as little as two to five weeks. Thus any estimation of the current dominant trend using the NN20 method can be hopelessly wrong. This point can be illustrated quite clearly by using the short-lived dominant trend in BET as an example. In Figure 9.13 we show how this trend moved through the whole of 1991, the values being calculated at a point 20 weeks into 1992. The turning point on the 8th March 1991 would not have been reflected in the calculated 41-week average until 21 weeks after this. During the first 20 weeks of this 21-week period the NN20 method would be estimating the dominant trend as fallingcontinuously. Thus the estimate would be in error until the 26th July 1991 when the latest calculation of the average showed it was higher than the previous calculation. From that point onwards the NN20 method would be estimating the dominant trend as rising, and would continue to do so for a further 20 weeks until after the next turning point. The next turning point in retrospect can be seen to be the 30th August. Thus the NN20 method is only correct for the period of time from the 26th July, when it finally showed the trend had begun to rise, until the 30th August when it changed direction again, Le. for four weeks. The calculation would not reflect this latest turning point until 21 weeks after the 23rd August, i.e. on the 24th January 1992. Thus the NN20 method was incorrect for the rest of 1991 from
Winning on the Stock Market
136 ZlO
......................•.................................•......... BET
200
......
1'5
150
.
.
125
.
J
91
F
A
"
J
J
A
s
o
"
D
Figure 9.13 The dominant trendin BET through 1991. Thelast 20 weeks arecalculated from BET prices inthe first halfof 1992
the 30thAugust onwards.These periods of correct and incorrect estimationof the dominant trend using the NN20 method for 1991 are shownin Figure 9.14. The estimate of the direction of the dominant trend is correct from the 4th January 1991 until the 1st March 1991, incorrect from the 8th March untilthe 19th July, correct from the 26th July until the 30th August, incorrect from the 7th September until the end of 1991. The net result is that out of the 52 weeks in 1991, the NN20 estimate of the direction of the dominant trend was correct for only14weeks. The methods of producing buying signals which will be discussed in the next chapter will produce several such signals over the lifetime of a favourable dominant trend if the latter lasts for sixmonths or longer.The first of such signals will almost certainly lie within this 2O-week window of error (as far as NN20 is concerned) at the start of the dominant uptrend. Thus we either have the option ofignoring the initialbuyingsignals, or employingthe price analysis method toget a better estimate of the current status of the dominant trend at this crucial period close to its directional change. If we think about it, there is no contest. If the NN20 estimate is always wrong in this period of time, then anyother method must be an improvement! A good test of the price analysis method is to compare its estimates of the dominant trend in BET through the whole of 1991 with the poor estimatesshown above for the NN20 method. The key points where the trend can be estimatedas having changed direction were on the 7th June 1991 (Figure 9.15) and the 20th
137
Dominant Trends
. セ
BET
225
.
200
1"5
-
125
J
91
F
A
J
J
A
.
5
o
"
o
Figure 9.14 The dominant trend in BET is represented by the centred 41-week average, and theestimateprovidebythe NN20method isrepresented by the unlagged 41-week aver-
age. Only when these twoare moving in the same direction(heavydouble arrowed lines)is the NN20 estimateof the trend directioncorrect December 1991 (Figure 9.16). In the first case it can be seen that the price at NOW
is higher than the price a few weeks after NOW - 40, heralding a change in direction. In the second case the price at NOW is lower than the price a few weeks after NOW - 40. Thus with the price analysis method, the estimate of the direction of the dominant trend is correct from the 4th January 1991 until the 1st March 1991, incorrect from the 8th March until the 31st May, correct from the 7th June until the 30th August, incorrect from the 7th September until the 20th December and correct on the 27th December. The net result is that out of the 52 weeks"in 1991, the price analysis estimate of the direction of the dominant trend was correct for 22 weeks. These periods are shown in Figure 9.17. Thus, in this particularly difficult case, the price analysis method was 50% more successful than the NN20 method in estimating the current status of the dominant trend. Occasionally with certain types of price movement, the price analysis method will not be able to predict clearly that the average should turn up before the point is reached at which the true calculated average turns up. More often than not though, the price analysis method is able to predict the turn up in the 41-week average about five to six weeks before the actual event. Since the share price will have risen, perhaps by a considerable amount during this period, the five to six
138
Winning on the Stock Market
2'll
. BET
225
...........•.....•..........•......•.•.•........•...•.....•......•
200
1i'5
.
150
125
.
..•.......................•...•...
o
A 90
N
• .•....................•......•
J
D
J
A
/'I
90 91
Figure 9.15 The dominant trend can be estimated as having changed direction by analysis of the position in BET on the 7th June 1991
. セ
BET
225
150
..
•.....................•..................•.•.................•.•..
/'I
91
"
J
J
A
s
o
/'I
o
Figure 9.16 The dominant trend can be estimated as having changed direction by analysis of the position in BET on the 20th December 1991
Dominant Trends m
139
.
BET
225
.
DOMINANT mEND
......... ·t·
セi
iセ
.
....
125
.
J 1'1
F
.
1'1
A
1'1
J
J
s
o
"
o
Figure9.17 The periodswhenthe priceanalysis estimateof the directionof the dominant trendare correct are shownas heavy double-headed arrows
weeks advance warning of the turn up in the average allows the investor to capture, on average an extra 4.6% gain. This is an extremely useful additional profit, which will offset the dealing costs involved in the transaction. In some cases the advantage can be extremely valuable, for example an extra 30% in Barclays, 21% in BAT, 19% in ICI, 18% in Hanson and General Accident, and 17% in BTR, Cable & Wireless and Wellcome. The investor should develop this skill by constant practice on a few charts of share prices, checking constantly on a weekly basis the estimate he made 20 weeks ago against the current calculation of the 41-week average.
10 Buying and Selling
We have come to the conclusion that a share should be bought at a time when we have estimated that the dominant trend is rising. Now we have to decide on what is the optimum time to buy during this period in which the dominant trend is headed north. INVESTMENT TRENDS If we look at a section of price history for a share where the dominant trend has reversed direction and is rising, then the shorter term fluctuations in the price provide several opportunities for buying. In Figure 10.1 we show the position for the Boots share price. The dominant trend as highlighted by the centred 41-week average is shown as well as the price movement itself. The dominant trend can be considered to have started with the low price of 199p on the 9th September 1988 and ended with the high price of315p on the 8th September 1989,lasting one year and resulting in a gain of over 50%. During this period there are several obvious buying points. We can use a criterion used previously - a valid buying point is one following which the price does not fall below this level until the ultimate selling point. In other words, the investor is never in a losing situation. Naturally this is only clear from the historical perspective after the event, and in practice we are never going to be able to sustain such a perfect performance. However, identifying such points after the event may lead us to develop methods which have a high probability of avoiding other, more problematical buying points. Using this criterion, there are five such points from which the investor is guaranteed a profit. Each of these points represents the start of a short term trend which takes the price up for a period of a few weeks at a much higher rate than that of the underlying dominant trend. These are the investment trends, and a
Buyingand Selling
141
m
. BOOTS
300
ZI5
.
.....
250
200
..
......
JJ
AS
OND
JFI'I
AI1J
JAS
OND
JFI'I
A
ss B8 B9 99 90 Figure10.1 The Boots share price witha centred 41-week averagesuperimposed. The points labelled 1 to 5 are obvious buying points
method has to be found of identifying them so that the timing of the investment can be fme-tuned during the upward phase of the dominant trend. Since the investment trends can only be seen to have started on their upward climbs the week after their low points in the case of weekly data such as this, or the day after the low point in the case of daily data, these gaps of one week or one day represent the earliest point at which any signal can get us into the share. This might lead us to think that the most efficient method is the one that generates a buying signal with this minimum of delay of one week or one day. However, this is an incorrect perception of the meaning of the word efficiency. The most efficient method is the one that gets us into the share at the next best price after the low points labelled in Figure 10.1. This mayor may not be the point one week or one day after the trough - it depends upon the exact nature of the price movement in the weeks or days after the trough. The distance apart of these troughs in the share price are 13, 12,6 and 12 weeks. As a reasonable approximation they represent cycles with these wavelengths. If we are to use a moving average to isolate short term cycles, we must use a span of less than the shortest cycle, i.e. less than six weeks. Thus in this particular case a 5-week average will still allow through these troughs in the price, so that such an average can be employed to generate a buying signal. A 5-week average, and its 25-day equivalent, are widely used to generate buying signals from weekly and daily data, although we shall see that they can be used in
Winning on the Stock Market
142
a variety of ways. Whichever span of average we use, and however we analyse the average data, we still need to satisfy two prime requirements: 1. The buying signal should be as dose as possible to the best price available after the trough in the share price. 2. The number of false signals, Le. those where the price fails to rise sufficiently to make a profit, should be minimal. Criterion 1 will lead us to the maximum profits from correct buying signals, while criterion 2 will reduce the number of incorrect buying signals. It should be realised that because of the partially random nature of share price movement, we can never reduce the number of false signals to zero. Since the data in Figure 10.1. is weekly data, it is appropriate to apply a 5-week moving average. Since we are interested only in whether the 5-week average has changed direction or not at its latest calculation, rather than using it as an estimate of the current status of the short term trend, we do not need to plot it as a centred average. Thus in Figure 10.2we have plotted a centred 41-week average to remind us of the current status of the dominant trend, and the 5-week average with no lag, so as to give us the exact date at which the calculation of the average showed it to have changed direction. We can see that troughs 1, 2, 4 and 5 from Figure 10.1 are reflected by troughs in the 5-week average, but that the picture for trough 3 is not quite as clear. The average hesitates after point 3 rather than falls to an obvious trough, so that if we base buying signals on troughs in the average we will see four rather than five signals. J2:l r-r..-.-.-.-.-. .-.-.-.-. .-.-.-.-..-.-.-.-.-. .-.-.-.-..-.-.-,-.-. .-.-.-.-..-.-.-.-..-.-.-.-.-. .-.-.-.-. .-.-.-.-. .-.-.-... ' BaJl'S
JOO
275
..
.....
250
.
225
..
200
,
JJ Bll
AS
ONO
JF'I'I Bll B9
AI'IJ
,
JAS
0"0
JF'I'I 89 90
.
A
Figure 10.2 The chart from Figure 10.1 now has an unlagged 5-week average superimposed.The troughs in these signal the start of short term uptrends
Buyingand Selling
143
In order to compare the buying signals generated by the change in direction of theS-weekaverage with the starting points in the trends themselves, it is important to compare like with like. The start of an uptrend is only confirmed when the price rises on the next occasion from its low point. Thus a 100% perfect trend-calling system would put the investor in not at the trough, but at the next price point. The same argument must be used with turning points in moving averages. Not until the average rises fromits low point is the signal generated. In the comparisons made in the following tables,we will call the price point following the price trough the trend buying point. We can see from Table 10.1. that out of the four signals, there was no delay in one,so that no profit was missed. Out of the other three signals, each of which was delayed by two weeks, there was a disadvantage to the investor in two of them, but in the third the investor would have bought at a better price than that at the trend buyingpoint. Overall, the price had risen by 1.9% from the trend buying point by the time the buying signal was given, a very small price to pay for the additional certainty given by using the average. Before moving to a discussion of other ways in which moving averages can be used to generate buying signals, it is instructive to show whether an investor would use the signals in Table 10.1 in the light of the behaviour of the dominant trend. At the time of the first buying signal on the 30th September 1988, there is no indication by either the NN20 method or the price analysis method that the dominant trend was moving other than downwards. Hence this signal would not be acted upon. This is a general point that can be commented on. The signal that is given on or just after the time when then dominant trend changes direction (when viewed historically) can never be acted upon, because the estimate of the direction of the dominant trend will be that it is still falling. We have seen that it requires anything from ten to twenty weeks before an estimate changes to a view that the dominant trend is now rising. Thus any buying signals occurring in this region will have to be ignored. This is the penalty we pay for waiting until the direction of the dominant trend is much more obvious. By the time the second signal is given on the 30th December 1988, the picture is very positive. Historically we can see that at that point we are now some four Table 10.1 Comparison of dates and pricesat the trend buying pointwiththosegiven by the 5-week average turningpoint of the Boots share price 'Irend buying point Date Price 16-09-88 16-12-88 21-04-89 14-07-89
213 217 277 287
Date
Average turningbuying point Price delay % missed
30-09-88 30-12-88 21-04-89 27-07-89
210 2 -1.4 231.5 2 6.7 277 0 0 294 2 2.4 Average % IriiSSed 1.9
144
Winning on the Stock Market
months past the actual turning point in the dominant trend, and a price analysis carried out at the time, as shown in Figure 10.3, estimates the dominant trend to be rising. This is because for almost all of the price region for twenty weeks from the left hand side of the chart, the price is below the level of the 30th December at the right hand side of the chart. This should be reflected in a rising 41-week average and hence a rising dominant trend. The position at the time of the third signal on the 21st Aprll1989 is against the background of an extremely positive estimate for the dominant trend (Figure lOA). Thus the share should be bought at this point. The only slight cautionary note is that the length of time for which the dominant trend has been rising - some nine months. The reversal in direction of the trend may occur sooner rather than later. By the time the signal is given on the 28th July 1989, the dominant trend has already been rising for a year. Thus the probability that it will change direction soon has increased significantly since the beginning of 1989. Although price analysis at that time shows that there may still be a few more weeks of a gently rising dominant trend, discretion should cause the investor to look for other opportunities where the dominant trend is much younger.
n---------------------------:--Tl セッ
BCDTS
245 240 235 230 225
220
215 210 205
A 88
J
J
s
o
N
o
Figure 10.3 The chart shows the price history and 41-week average of Boots as at 30th December 1988. The current estimate of the dominant trend is rising, since the latest price is above most of the price levels to the left of NOW - 20
Buyingand Selling :lOO
145
..•....•.•.•.....•......•.•.•.•....•.•................•.•........• BaJrS
Z'5
.
NOW-20
................................. j
.
. .............................•......•.•.............
200
. J
5
0"
0 Nセ
J
F
"
A
Figure 10.4 The chart shows the price history and 41·week average of Boots as at the 30th December 1988. The current estimate of the dominant trend is rising, since the latest price is wellabove the price levels to the left of NOW - 20
PRICE PENETRATION OF THE AVERAGE The use of a turning point in an average such as the 5-week used here is intellectually the most satisfying because the turning point in the average is a reflection of the turning point in the trend, with contributions due to very short term fluctuations almost totally removed. However, another method of using averages has become widely used in investment research. In this alternative method the signal is given by the price rising above the average in the case of buying signals, and falling below it in the case of selling signals. The average is plotted as in the previous figures with no lag. In Figure 10.5 the places are marked where the price rises above the average so as to generate a buying signal. These signals are compared with those generated by the turning method in Table 10.2. Two of the signals occurred at the same time as the turning method, one occurred earlier and one later. Thus on balance there is no great difference between the two methods. Over a large of group of shares there is also no significant difference in performance between the two. However, there is a difference in the number of signals generated. In this particular example there is an extra signal on the 23rd June 1989, where there is no equivalent turning point in the 5-week average. The difficulty is that most such signals are false. Thus the price of 297p on the 23rd June turned out to be the highest price for six weeks, putting the investor in a situation where he failed to
146
Winning on the Stock Market
J:l'
. IlCDI'S
JOO
.
..
275
250
..
225
JJ 88
AS
ONO
JF''' 8889
A"J
JAS
ONO
bYセ
JF'"
A
Figure 10.5 Buying signals in the Boots share price generated by the price rising above the 5-week average Table 10.2 Comparison of the turning and penetration methods using 5-week averages of the Boots share price Turning method Date 30-09-88 30-12-88 21-04-89 28-07-89 28-07-89
Penetration method Price 210 231.5 277 294 297
Date 16-09-88 23-12-88 21-04-89 28-07-89
Price 213 226 277 294
profit for this length of time. His money could have been more usefully employed elsewhere in a situation which showed an immediate profit from the time of investment. Not only that, hut if the investor is operating a stop-loss selling system, then the fall to 271p on the 7th July would trigger any stop-loss up to 8%, giving a nasty loss of over 13% in just two weeks if dealing costs are taken into account. Since a prime aim of any investment strategy is to reduce to a minimum the number of false buying signals, it is of interest to see why the price penetration method gives a signal when there is no turning point in the average. The basic reason that investors use this method is due to a misunderstanding of the simple arithmetic involved in the moving average calculation. It might be thought that if the price rises above the average price, which is the situation at the time the penetration signal is given, then the average must subsequently rise at the
147
Buying and Selling
next calculation. This totally ignores the three components of the average calculation- the running total, the next price point, and the drop-point price. Only if the latter is less than the value of the next point will the average rise. The fact that the current point is above the value of the average does not even guarantee that next week's price is above this week's price, let alone that it is above the drop-point price. Thus there is no relationship whatsoever between what will happen at the next calculation and the fact that the price is currently above the average price. It is this lack of any fundamental relationship between the two that is the cause of so many erroneous signals given by the price crossing method. THE 25·DAY AVERAGE It would seem logical to expect that similar results would be obtained for a 25-day average operating on daily data as for a 5-week average operating on weekly data. This is certainly not the case, and for a very good reason. The sampling of data on a weekly basis from what is fundamentally daily data is in itself a smoothing operation. Thus a 5-week average of weekly data would be expected to be smoother than a 25-day average of daily data. That this proves to be the case can be seen from Figure 10.6, where both the daily data and the 25-day average are shown. The figure should be compared with Figure 10.1. It should be obvious that the weekly data is smoother than the daily data, and the weekly average is smoother than the daily one.
. セ
300
..
MDJFPl"'"JJ"'SOMDJFPl"'"JJ"'S セ
Nセ
セ
Figure 10.6 Daily data and the 25-day average of the Boots share price
148
Winning on the Stock Market
The consequence of this is that we will get many more buying and selling signals from the 25-day average than from the 5-week average, irrespective of whether we use the turn or price penetration methods. The problem is well illustrated in Figure 10.7, where the chart of the Boots daily price from April 1989 to September 1989 is shown, along with the 25-day average. This was a time period over which we saw just two signals for the 5-week turning or price crossing methods. In Figure 10.7 we have marked six signals from each of the average turning and price penetration methods. Only two of these six can be said to be profitable buying points, the other four put the investor into a losing position within a few days of buying. In order to outweigh the highly negative effect of these false signals, the true signals, i.e. those nearest to the four weekly signals produced in the weekly data shown previously in Tables 10.1 and 10.2, should produce considerably better buying prices. Whether this is true or not can be established from the data in Table 10.3,where the results for the signals in the daily data are given. In the daily data it can be seen that the price penetration method appears to be slightly better than the average turning method, getting in at the same price on two occasions and at better prices on two occasions. The more interesting comparison is with the weekly data in Table 10.2. The 5-week average method is better than the 25-day average penetration method on one occasion, the same on another occasion, and not as good on two occasions. Where it is better, it gets the investor 310
.
300 29S 290 295 290 VS 270 26S
T
"
8'1
J
J
A
Figure 10.7 In this expanded chart of the 25-dayaverage,the excessive number of signals generated by daily averages is illustrated by the sixcrossing signals (X) and sixturningsignals (T) shown here. Only two of each type were profitable
Buyingand Selling
149
Table 10.3 Comparison of the turning and penetration methods using 25-day averages of
the Boots share price Thrningmethod Date 04-10-88 23-12-88 19..Q4-89 03.{)8-89
Penetration method Price 215 226 277 299
Date RoセMX
20-12-88 21-04-89 27-07-89
Price 215 223.5 277 286
in 5p lower, while when it is not as good it gets the investor in at 8p higher. As an average over the four signals, the 25-day average gets the investor in ara price only 1% better than the 5-week average method. This small difference may seem surprising to investors brought up on the virtues of daily data. It has to be said that for the extra work entailed in maintaining a daily data base as opposed to a weekly one, the additional gain is really not worth the extra effort. When all the additional negative signals are taken into account, the use of daily data may be considered to be an actual disadvantage. Naturally, as we have maintained throughout this book, it is dangerous to draw conclusions from a limited set of data. When the pool of alpha shares is examined over a period of time from the beginning of 1988 to mid-1992, the conclusion drawn from the study of Boots was fully supported. The number of signals generated by using a 25-day average penetration method is about twice as many as those generated by the 5-week average turning method. SELLING SIGNALS To show the profits (and losses) which can be made from buying on the turn of the 5-week average, ten shares were chosen so as to represent relatively involatile shares, moderately volatile shares and very volatile shares. Only signals which occurred while the dominant trends in the both the market (using the FrSE100 Index) and the share itself were estimated to be favourable were taken as genuine buying signals. Price ConlroUed Selling Signals
Although there are many ways in which selling signals can be generated, the stop-loss method we discussed earlier has the virtues of simplicity and the ability to lock in a considerable proportion of the profit from a correct buying decision. Although a 5% stop-loss is a good general purpose value, this percentage can be relaxed somewhat if we have used buying techniques which reduce the overall risk in the investment. The strategy of relying on a favourable dominant trend in the share and the market is such a technique, and being a dominant trend, should last for a reasonable length of time. The advantage of relaxing the stop-loss to say 7.5%
150
Winning on the Stock Market
means that we should be able to stay in the share for longer periods than if a 5% stop-loss is used, thereby taking advantage of such longer trends. The selling points used in Table 10.4 are generated by triggering of a 7.5% stop-loss. The buying points listed are those between the autumn of 1988, when the estimate of the 41-week average of the FfSE100 Index was that it had started to rise, and J one 1992, at the time of writing. Buying points are not listed for occasions when the 41-week average of either the FrSE100 Index or the share was estimated to be falling. There are 67 buying and corresponding selling signals for the ten shares, and the average gain of these is 13.8%. This average of seven transactions per share in
Thble 10.4 The transactions generated in 10 shares by the 5-week average turning method. Signals given when the dominant trend is rising in the market or the share itself are not listed Buy date
Price
Sell date
Price
%gain
Wksinvstd
268.5 355 346 378 445 436 454 454 587
19-05-89 29-09-89 10-08-90 05-06-92
350 341 419 665
30.4 -3.9 21.1 75.9 49.4 52.5 46.5 46.5 13.3
19 10 40 68 50 42 39
231.5 277 329 374 420
02-06-89 02-06-89 28-06-91 28-06-91 20-03-92
281 281 359 359 432
20.9 1.4 9.1 -5.8 2.8
22 6 21 4 21
398 407 487 436 687 715 530 539
16-09-88 16-09-88 25-11-88 15-09-89 15-09-89 15-09-89 26-04-91 26-04-91
455 455 443 648 648 648 600 600
14.3 11.8 -9.0 48.6 -5.7 -9.3 13.2 11.3
14 12 7 36 10 2 18 14
203.5 253 253.5 258
19-01-90 11-10-91 11-10-91 15-05-92
218 258.5 258.5 250.5
7.1 2.1 1.9 -2.9
9 25 19 5
BM 06-01-89 21-07-89 03-11-89 15-02-91 21-06-91 16-08-91 06-09-91 15-11-91 10-04-92 Boots 30-12-88 21-04-89 01-02-91 31-05-91 25-10-91 British Aerospace 10-06-88 24-06-88 07-10-88 06-01-89 07-07-89 01-09-89 21-01-90 18-01-91 British Gas 17-11-89 19-04-91 31-05-91 10-04-92
29
8
(continues)
151
Buyingand Selling 'l8ble 10.4 (con/d)
Buy date British Telecom 30-09-88 30-12-88 31'{)3-89
19-05-89
Price
sell date
Price
%gain
242.5 256 275 276 277.5 257 255 259.5 265 292 378
21-04-89 21-04-89 02-06-89 02-06-89 30-06-89 27-10-89 27-10-89 27-10-89 09-03-90 28-06-91 28-06-91
263 263 246.5 246.5 252 250 250 250 274.5 348 348
9.0 2.7 -10.3 -10.6 -9.2 -1.9 -1.2 -2.9 3.6 19.2 -5.3
29 16 9 2 3 16 12 9 15 21 5
02-06-89 02-06-89 28-06-91 22-11-91
290 290 461 505
14.1 7.0 19.7 2.6
20 8 22 19
02-06-89 02-06-89 02-06-89 02-06-89 22-11-91 22-11-91
445.2 445.2 445.2 445.2 650 650
18.8 21.5 20.8 -0.9 22.7 -2.4
34 28 22 6 43 6
13-10-89 13-10-89 13-10-89 13-10-89 13-10-89 13-10-89 16.{)2-9O 29-03-91
169 169 169 169 169 169 194 188
19.7 20.0 19.0 8.3 8.9 8.6 8.9 14.6
54 44 40 28 18 14 14 9
27-10-89 27-10-89 27-10-89 27-04-90 28-06-91 15·11-91 05-06-92 05-06-92
891 891 891 1044 765 903 1158 1158
85.2 24.9 13.8 13.6 -10.3 10.8 22.0 -0.2
45 28 15 21 4 15 24 7
21-04-89 10-05-91 08-11-91 05-06-92 05-06-92
454 414 498 718 718
15.8 18.6 14.4 41.6 20.9
16 12 22 25 14 10
09-06-89 07.{)7-89 04-08-89 25-08-89 24-11-89 01-02-91 24'{)5·91 kingfISher 13'{)1-89 254 07-04-89 271 18'{)1-91 385 12'{)7-91 492 Reckitt & COlman 07-10-88 374.8 18-11-88 366.4 30-12-88 368.4 21-04-89 449.2 08-02-91 529.6 25-10-91 666 IOY81 Bank or Scotland 30.{)9-88 141.2 09-12-88 140.8 06-01-89 142 31'{)3-89 156 09.{)6-89 155.2 07.{)7-89 155.6 10-11-89 178 25'{)1-91 164
WIts invstd
Reuters 16-12-88 14-04-89
lW7-89
481 713 783 919 843 815 949 1160
01-12-89 31'{)5-91 02-08-91 20-12-91 17-04-92 Siebe 30-12-88 392 15-02-91 349 07-06-91 435 13-12-91 507 28-02-92 594 Average oU7 signals
13.8
152
Winning on the Stock Market
just four years is far too high, and ways have to be found of reducing it and possibly increasing the average gain. Reducing the Number oI'False Signals False signals are those that produce no profit to the investor over a reasonable time span, say six to eight weeks, so that the investor is at best standing still at a time when he could have been more profitably invested. At worst he suffers a considerable loss during this period. Quite obviously, an investor needs a method of reducing these to the very minimum, it being recognised that the random nature of the stock market makes it impossible to eliminate those instances where everything looks perfect for an investment, only to turn sour just a few days or few weeks later. There are several sensible and logical rules which we can apply to do this. The first one which will remove a large number of transactions is to treat as void any buying signal which is given before the selling signal from the previous buying operation has been generated. In other words the investor would be buying into a share which he would still be holding from the previous buying signal. The risk involved in taking two bites at the same short term trend must be considered to be relatively high. Examples would be the transactions from the 21st June 1991 to the 10th Apri11992 in BAA before the buying operation on the 15th February 1991 is completed. By adopting this approach we reduce the number of transactions to 36. The second type of signal which should be ignored is that where at the point that the latest calculation of the average shows that it has turned up, the price has fallen from the previous week. It should be an inviolate rule that an investor never buys on a price fall. The reason is simple: the start of an uptrend, be it a very short term one-or otherwise, is always from a price rise. Thus the investor must wait for a rise to be sure that the uptrend has begun. If the price has fallen from that of the previous week, there is no guarantee that it will not fall the next week. An example of such a rise in the average on a fall in the price is shown for Reuters in May 1991 (Figure 10.8). The subsequent price action confums that the buying signal was a false one, and the transaction would have lost the investor 10.3%. Although naturally there are many examples of the reverse behaviour, where although the price has fallen, it then recovers and continues upwards, there are many more where the buying signal turns out to be false. On balance the risk is too high to take positive action in such circumstances. The third type of signal is a development of this second type. In this case we ignore a signal when the trend which is even shorter than the one isolated by the 5-week average is falling even when the 5-week average has turned up. These shorter term trends would be isolated by a 3-week average, but for our purposes it is not necessary to calculate such an average. If we think in terms of drop-points, the 3-week average turns up when the price at NOW which is added in to the
Buyingand Selling 875
153
•.........................•.•.•.................•.•.•........•.•.. REUTERS
24th MAY
850
..
5-WEEK AVER 800
................................................ AGE .....
....
SIGNAL GIVEN 31st MAY
rn
..
oaos
91
1005
91
1406
91
Figure 10.8 A false buying signal in the Reuters share price on the 31st May 1991. This should be ignored, since, although the 5-week average has risen, this is accompanied by a fall in the share price from the previous week
3-week running total is higher than the price at the drop-point at NOW - 3. Thus all we need to do when we get a signal that the 5-week average has turned up is to check that the latest price is higher than the price three weeks ago. On about 90% of the occasions it will be. If it is not, then very short term trends are running adversely, and the signal should be ignored. An example of such a negative signal is shown in Figure 10.9 for the buying signal in British Telecom on the 31st March 1989. The investor who ignored this negative signal would have lost over 10% from the complete transaction. These last two methods of eliminating false signals would remove six signals from the 40. These false signals are shown in Table 10.5. Although they do not all lead to losses, where they do not, they certainly do not give anything other than very small gains which would be wiped out by dealing costs. The net result of removing these is to leave us with 36 buying and selling transactions from the 10 shares, i.e. somewhere between three and four per share over the four years. These transactions are shown in Table 10.6. The average gain is now improved to 15.5%, and the average length of time for which the investor would be invested is 21 weeks. This represents a rate of gain of 0.73% per week, before dealing costs are taken into consideration. If we allow 5% for these, then the gain becomes 10.5% over 21 weeks, i.e. 0.5% per week, or 26% per annum, an impressive performance. Although space does not warrant lengthy tables of such data for all of the alpha shares, similar results are obtained on all but the most
154
Winning on the Stock Market
:l'O
0
7.5
o.
0
••••
0
•••••
0
•••
0
••
0
•
0
•••••
0
•
0
0
0
0
0
•
0
•
0
0
0
•••••
0
0
0
••••••
0
•••••••••••••••
••
IIlITIEl.E
:NQW セNMZ
285
0
0
•
0
•
..
2.5 290
V5 2.5
S -H A R ·Eo PRICE
270 7.5
0
265
0
•••
0
0
••••••
0
•••••
••••
0
•
••••••••••
0
•
•
0
•••
0
••••••••••••••
••••••
0
•••
0
••••
0
•
0
•
0
0
••••
0
•
0
••••••••••
0
0
•
•
••••••••••
0
•
0
•
•
•
0
•
•
•
•
0
••
0
•••
0
•
000.
•
•
•
•
•
•
0
••
0
0
•
000
•
0
•••••
•
••
0
0
0
••
••••
0
•••
••
0
•••
••••
0
•
0
NOW 2.5
0
0
••••
0
•
0
0
••
セPR
Figure 10.9 A different type of false buying signal in the British 'Ielecom share price in March 1989. The very short term trend which would be shown by a 3-week average is falling. It is not necessary to calculate this since a rising 3·week average will have the price at NOW higher than the price at NOW - 3. In this example the price at NOW is lower, so the 3-week average is falling and the share should not be bought Table 10.5 The false signals eliminated (a) because the price has fallen (b) because the price at NOW is lower than the price at NOW - 3 at the time the 5-week average turning signal is given Buy date
Price
Sell date
Price
%gain
Wksinvstd
420
20-03-92
432
2.8
21
275 257 265
02-06-89 27-10-89 09-03-90
246.5 250 274.5
-10.3 -1.9 3.6
9 16 15
843
28-06-91
765
-10.3
4
Boots
(a) 25-10-91 British Telecom (b) 31-03-89 (b) 07-07-89 (a) 24-11-89 Reuters (a) 31-05-91
involatile shares, and as wehaveshownin a previouschapter, the investorwill find it a simple matter to eliminate such shares from consideration for investment.
Buyingand Selling
155
TIme Controlled Selling Signals The results obtained using a stop-loss of say 7.5% were impressive, but it should be pointed out that any method that depends upon the price movement itself to generate a selling signal has a blind spot. It is possible to stay invested for long time periods over which the share price makes no significant advance. Theoretically the price could see-saw up and down within 2 or 3% above and below a horizontal long term trend, with the falls from the peaks being too small to trigger the stop-loss. Such a position is shown in Figure 10.10. Thus the investor is locked into a situation which is producing no profit. It is of little compensation that he is also apparently not losing money while this is happening. Any time spent not making profits must be considered to be wasted time where money could be better employed elsewhere. Thus we have to have a last-ditch method to prevent such situations building up. Since a price controlled system is the type which fails, we need a back-up system which is independent of price. Since the only two variables in the stock market are price and time, we need to use a back-up system based on time. Thus our selling signal will be given when a certain amount of time has elapsed with no selling signal having been generated by the stop-loss or other selling signal that we are using. The only decision to be made is on the length of time that should be used. We can get some help with this by looking at the average length of time for which we stayed invested in the ten shares used as examples in Table 10.6. This was 21
セBイ
.
セ
.....
\ 'lOO
.......
HO
9:'
J
9:' 99
F'
H
III
11
J
J
AS
0
Figure 10.10 It is possible to remain invested for a long unprofitable period without a sellingsignalifthe priceoscillateswithina narrow horizontal band. The fallsfrom peak to trough (double-headed arrows) are less than 7.5% hence the stop-loss is not triggered
Winning on the Stock Market
156
weeks for the average return of 15.5%. Rounding this off to 20 weeks, we could adopt a method in which we sell the share after 20 weeks if a selling signal has not been given by the 7.5% stop-loss within that time. In Table 10.6 there were 16 shares that had not given a selling signal within this period of time. It is useful to compare the selling price and gain achieved after 20weeks with the ultimate selling price and the additional amount of gain achieved during the period after this 20 weeks. This shows us whether the rate of gain was higher during the initial 20 weeks than during the subsequent period. If this turns out to be the case, then there is no point in remaining invested for longer than 20 weeks in the same share; much better returns could be obtained by switching to another share that was just reaching a buying point. The data from this comparison are given in Table 10.7.
Table 10.6 The transactions which remain after elimination of all false signals Buy date
Sell date
Price
%gain
Wksinvstd
268.5 355 346 378
19-05-89 29-09-89 10-08-90 05-06-92
350 341 419 665
30.4 -3.9 21.1 75.9
19 10 40 68
231.5 329 420
02-06-89 28-06-91 20-03-92
281 359 432
20.9 9.1 2.8
22 6 21
398 487 436 530
16-09-88 25-11-88 15-09-89 26-04-91
455 443 648 600
14.3 -9.0 48.6 13.2
14 7 36 18
203.5 253 258
19-01-90 11-10-91 15-05-92
218 258.5 250.5
7.1 2.1 -2.9
9 25 5
242.5 275 277.5 257 265 292
21-04-89 02-06-89 30-06-89 27-10-89 09-03-90 28-06-91
263 246.5 252 250 274.5 348
9.0 -10.3 -9.2 -1.9 3.6 19.2
29 9 3 16 15 21
254 385 492
02-06-89 28-06-91 22-11-91
290 461 505
14.1 19.7 2.6
22 19
Price
BAA 06-01-89 21-07-89 03-11-89 15-02-91 Boots 30-12-88 01-02-91 25-10-91 British Aerospace 10-06-88 07-10-88 06-01-89 21-12-90 British Gas 17-11-89 19-04-91 10-04-92 British Telecom 30-09-88 31-03-89 09-06-89 07-07-89 24-11-89 01-02-91 Kingfisher 13-01-89 18-01-91 12-07-91
20
(continues)
157
Buying and Selling Table 10.6 (contd)
Buy date
Price
Rec:kitt & Col man 07-10-88 374.8 529.6 08-02-91 ROyal Bank of SCotland 30-09-88 141.2 10-11-89 178 25-01-91 164 Reuters 16-12-88 481 31-05-91 843 02-08-91 815 20-12-91 949 Siebe 30-12-88 392 349 15-02-91 07-06-91 435 13-12-91 507
Sell date
Price
%gain
Wksinvst
02-06-89 22-11-91
445.2 650
18.8 22.7
34 43
13-10-89 16-02-90 29-03-91
169 194 188
19.7 8.9 14.6
54 14 9
27-10-89 28-06-91 15-11-91 05-06-92
891 765 903 1158
85.2 -10.3 10.8 22.0
45 4 15 24
21-04-89 10-05-91 08-11-91 05-06-92
454 414 498 718
15.8 18.6 14.4 41.6
16 12 22 25
The advantage of selling at 20 weeks is obvious. In only one case, BAA, was the rate of return per week over the subsequent period after 20 weeks higher than that during the initial 20 weeks. In 10 out of the 16 cases, the subsequent rate was negative, i.e. the investor's profit was declining from its position at the end of the initial 20 weeks. Taking an average, the investor was achieving a rate of return of 1.2% per week during the first 20 weeks, and then losing at a rate of 1.2% per week for every week he stayed invested after that. Thus there is no contest. The investor must sell after 20 weeks if a selling signal has not been given by the stop-loss method. By doing this, the gains and number of weeks invested that were shown in Table 10.6 are now changed to those shown in Table 10.8. The average gain is now 14.0%, but reached over a shorter average timescale of 15.5weeks. The rate of gain per week is 0.9%, and if dealing costs of 5% are taken into consideration, the net gain of 9% in 15.5 weeks is equivalent to a net rate of gain of 0.58%. This is equivalent to a net annual gain of 30%. Since dealing costs have been totally taken into account, the investor will also have the benefit of dividends from the various shareholdings.
158
U'inning on the Stock Market
Table 10.7 The buying prices, prices at 20 weeks and ultimate selling prices for the longer investment trends. Also given are the rates of gainper weekfor the first 20 weeks and the
periodsubsequent to this Date Buy price
Pnceat 20 weeks
Pnce at signal
Gamperweek at 20 weeks after 20 weeks
BAA 03-11-89 346 15-02-91 378 Boots 30-12-88 231.5 25-10-91 420 British Aerospace 06-01-89 436 British Gas 19-04-91 253 British Telecom 30-09-88 242.5 01-02-91 292 Kingfisher 18-01-91 385 Reckitt & Colman 07-10-88 374.8 08-02-91 529.6 Royal Bank or Scotland 30-09-88 141.2 Reuters 16-12-88 481 20-12-91 949 Siebe 07-06-91 435 13-12-91 507 Average
399 438
419 665
0.765 0.79
0.29 1.25
307 448
281 432
1.63 0.335
-5.85 -3.9
682
648
2.82
-0.49
273.5
258.5
0.405
-1.2
275 359
263 348
0.67 1.145
-0.44 -3.5
495
461
1.425
-4.4
425.6 616
445.2 650
0.675 0.815
0.38 0.28
160.5
169
0.685
0.18
766 1234
891 1158
2.96 1.5
1.04 -2.0
525 693
498 718
1.035 1.83 1.218
-3.15 1.0 -1.28
159
Buyingand Selling Table 10.8 The transactions which remain after application of20-week limit Buy date
Price
BM 06-01-89 268.5 21"()7-89 355 03-11-89 346 378 15-02-91 Boots 30-12-88 231.5 01-02-91 329 420 25-10-91 British Aerospace 10-06-88 398 07-10-88 487 436 06-01-89 21-12-90 530 British Gas 203.5 17-11-89 19-04-91 253 258 10-04-92 British Telecom 30-09-88 242.5 31-03-89 275 277.5 09-06-89 07-07-89 257 24-11-89 265 01-02-91 292 Kingfisher 13-01-89 254 18-01-91 385 12-07-91 492 Reckitt & Colman 07-10-88 374.8 08-02-91 529.6 Royal Bank of Scotland 30-09-88 141.2 10-11-89 178 25-01-91 164 Reuters 16-12-88 481 31-05-91 843 02-08-91 815 20-12-91 949 Slebe 30-12-88 392 15-02-91 349 07-06-91 435 507 13-12-91 Averages
Sell date
Price
%gain
Wksinvstd
19-05-89 29-09-89 23-03-90 05-07-91
350 341 399 438
30.4 -3.9 15.3 15.8
19 10 20 20
19-05-89 28-06-91 13-03-92
307 359 448
32.6 9.1 6.7
20 6 20
16-09-88 25-11-88 26-05-89 26-04-91
455 443 682 600
14.3 -9.0 56.4 13.2
14 7 20 18
19-01-90 06-09-91 15-05-92
218 273.5 250.5
7.1 8.1 -2.9
9 20 5
17-02-89 02-06-89 30-06-89 27-10-89 09-03-90 21-06-91
275 246.5 252 250 274.5 359
13.4 -10.3 -9.2 -1.9 3.6 22.9
20 9 3 16 15 20
02-06-89 07-06-91 22-11-91
290 495 505
14.1 28.5 2.6
20 20 19
24-02-89 28-06-91
425.6 616
13.5 16.3
20 20
17-02-89 16-02-90 29-03-91
160.5 194 188
13.7 8.9 14.6
20 14 9
05-05-89 28-06-91 15-11-91 08-05-92
766 765 903 1234
59.2 -10.3 10.8 30.0
20 4 15 20
21-04-89 10-05-91 25-10-91 01-05-92
454 414 525 693
15.8 18.6 20.7 36.6 14.0
16 12 20 20 15
11 Channel Analysis
The strategy we have developed so far has simply required us to buy in at the first indication that a rising investment trend had started, against the background of a rising dominant trend in both the share in question and the market itself. The indicator used to determine the start of the investment trend was the turning point in the 5-week moving average. No attempt was made to analyse in detail the price movement of the 5-week average in the way in which the price movement of the 41-week average was analysed, since good results were obtained simply by eliminating two types of false signal. Implicit in this profitable strategy is that the dominant and investment trends are two unrelated entities. We now show that this simple picture is not true, and that the dominant trend puts limits on the levels to which the short term trends can rise or fall. Taking a familiar example, Boots, we show in Figure 11.1 a chart upon which we have superimposed the dominant trend as signified by the 41-week average. Looking at the extent to which the shorter term trends oscillate around this average, it becomes obvious that there is a limit to how far vertically above and below the average these short term trends take the price. It is possible to draw a duplicate of the average curve at an equal distance vertically above and below the original position of the average such that the majority of the price movement is contained within these upper and lower levels. This is shown in Figure 11.2.The upper and lower boundaries constitute a channel for the shorter term trends. Such channels can be drawn by computer, as shown here, or by eye, simply by drawing smooth curves through the high and low extremities of the price. It is important that the vertical depth of the channel remains constant, Le. a fixed number of points in the case of the FTSE100 Index, or a fixed number of pence for a share. Use of percentage envelopes invalidates the channel procedure. The more peaks and troughs that lie on a channel, the more valid is the channel. A small number, say two to four, overshoots of the channel boundaries by the price is permissible,
Channel Analysis
161
Figure 11.1 The Boots share price with the dominant trend as shown by the centred 41week average
Figure 11.2 The shorter term trends in the Boots share price can be contained within a channel constructed byduplicating the centred 41-week average above and below the original position
Winning on the Stock Market
162
for example the unavoidable overshoots at the time of the 1987 crash, but these should be kept to a minimum. Note that in Figure 11.2 there are at least 16 peaks and troughs that lie almost directly on the channel boundaries. On the four occasions in which overshoots occurred in 1987, the amount of overshoot is still quite small. The important implication of these channel boundaries is that there is a limit (in pence or points) to the extent to which the short term trend can normally take the price (or Index) away from the centre of the channel. Having reached this limit the price rebounds back towards the centre. In this example of Boots, we have set the limit at 32p above and below the centre of the channel. Thus the price has a very high probability of bouncing back towards the centre of the channel once this limit is reached. The price is therefore oscillating within a channel which is 64 pence deep throughout its history. On the few occasions during the period of the crash in 1987 this limit was exceeded, but only by a few pence. An idea of the way in which the price moves within a channel can be seen in Figure 11.3. The distance of each price point from the average is plotted in units of O.5p, and the number of occasions which the price spent at each level over the time period shown in Figure 11.2 is shown on the vertical axis. As we would expect from Figure 11.2, the number of occasions when the price is at a large distance,
Boot!;
CHANNEL CENTRE
" o
• 18 + - - - - - - - - - - - , r - n - - I - - H - I f - - - - - - - - - - - - - - i
o
I
gc セ
セ
•5
SETat32p
SET at 32p
.
•
UPPER BOUNDARY
LOWERBOUNDARY
5
+------+------+
i セ
a
J
a
セ
i セ
6
5
8
5
6
セ
I
セ
I
セ
Figure 11.3 The movement of points within the Boots channel. The horizontal axis shows the distance of the price from the position of the moving average, negative values being below and positive above the average. The price levels are divided into units of O.5pand the vertical axis shows the number of points which lie at each position over the period 1983 to present
Channel Analysis
163
e.g, 30p or more from the average, is quite small. The extremes were SOp above the average and 49p below on just one occasion each. The price spent most of its time within a band 15p either side of the average. The channel boundaries at 32p above and below are also shown, so that the very few occasions on which the price reached or penetrated these boundaries can be seen clearly. A better view of the underlying theory of channel analysis can be obtained from Figure 11.4. Here the vertical axis expresses a different quantity. The centre point represents the position of the average itself, and to the left of this is the percentage of points that lie further below the average than the particular value on the horizontal axis. For the right hand side, the percentage refers to those points which are further above the average than the value on the horizontal axis. Taking particular points on the graph, we can see that 20% of the points below the average lie more than 17p below it, while 20% of the points above the average line lie more than 20p above. The graph is not quite symmetrical because of the effect due to the gradual rise in the Boots share price from 1983 to the present. The boundary positions of 32p are marked on the chart, and now we can see that a very small percentage (3.5%) of the points lie outside either the lower or upper boundaries. In fact, the boundary position was chosen by the computer program so as to achieve this level of overlap. We can now see that the position of the boundaries is not terribly critical, since moving it in to, say, 25p above and below still leaves only about 5% of the points outside of them. This is why it is also quite simple to get good results if the channels are drawn by eye, since the position tolerance is fairly high. If we think of the vertical axis in terms of probability then we can see why channel analysis works. Once a closing price lies at or very near the boundary, the probability of the next closing price, be it weekly or daily, staying there is only 3.5% in the case of this particular example. The probability is high that it will move in the direction of the centre line of the channel. Having said that, the extent of its movement towards the centre of the channel will depend upon the usual weekly or daily price movement. Thus it would be unrealistic to expect the Boots price to move by 32p to the centre of the channel in one day! A value of 1 to 5p is much more probable. It is the directional probability that we are really concerned with. It is important to use this picture of channels to reinforce the fundamental significance of centred moving averages and the fundamental insignificance of the unlagged averages so beloved of chartists. By plotting the average and channel as unlagged, the chart shown in Figure 11.5 was obtained. We can see quite clearly that all we have is a graphic mess from which virtually no useful information can be obtained. To emphasise the validity of this approach, the same channel is shown expanded over the shorter period from March 1989 to June 1992. There are no fewer than six peaks or troughs which lie almost exactly on the channel boundaries. If we extrapolate the 20 week gap to the end of the data, we can see that we expect another four points to lie on the extrapolated channel. Thus, in this 3-year period,
164
Winning on the Stock Market Boas
......
j
+
\T
f
.f LOWER BOUNDARY SETat32p j
"\
I
UPPER BOUNDAR SETat32p
:f 2li1
+-----+---4'-----+------"'tII""...----f-----l
CHANNEL CENTRE
i i a セ
I
i セ
セ
,
5
5
liI
6
セ
a
i セ
セ
= , I
Figure 11.4 For any point on the horizontal axis, the vertical axis gives the percentageof points that liefurther from the position of the average. The left hand of the graph represents points lying below, and the right hand side pointslying above the average. At a channel boundary of 32p, only3.5% of pointslie outside the channel
BalTS
450
400
.
セッ
300
.
250
200
150
84
86
87
88
81'
90
1'1
1'2
Figure 11.5 If the channelsare based on unlagged rather than centred averages then no sense can be made of them
ChannelAnalysis
165
we will have encountered 10 occasions on which the price fails to move significantly beyond the boundary. No such result would be obtained if prices moved totally randomly, or if there was no relationship between the dominant trend and the shorter term trends. It is important to study, in Figure 11.6, the price movement that occurs once the price retreats from a boundary. The price always appears to retreat to the centre line within a few weeks, but may not continue on to the opposite boundary. In fact, the numbers of half-channel and full-channel movements are about the same. Since our aim is to be invested only when the dominant trend is rising, it is of interest to analyse the price movements during the upward leg of the channel, i.e from early 1990. Although the distance from the channel boundary to the centre, measured vertically, is 32p, the price changes following a bounce from the lower boundary to the centre will be more than this. This is because the channel itself has risen after the bounce, and this rise in the channel must be added to the normal half-channel depth of 32p. Because of this additive effect, useful profits can be made from half-channel rises, although full-channel rises obviously provide the optimum gains. Note that there is no apparent way in which we can predict whether a rise will be a half-channel or full-channel rise at the time the price rebounds from the lower boundary. Taking some examples of the ultimate gains from such lower boundary bounces, the half-channel bounce from 276p on the 21st September 1990 took the price to 323p on the 19th October, Le. a movement of 17% in just four weeks. The
__ __ __
:lOOM M セ M M セ M M M ⦅ Z ⦅ M M M ョ
セ
セ
LN ⦅Z セ
BOOTS
475 450
.
425
400
375
..
350
..
325
300 275 250
1'1 A" セ
J J AS 0 N 0 J F 1'1 A 1'1 J J A SO" 0 J F " A 1'1 J J A SON 0 J F " A" J J AS ONO セ
セ
セ
Figure 11.6 The price retreats towards the centre of the channel once it bounces off a boundary. It mayor may not continue to the opposite boundary
166
Winning on the Stock Market
full-channel bounce from 359p on the 28th June 1991 took the price to 439p on the 23rd August 1991, Le. a movement of 22.2%. Obviously, the earlier in the development of the upward dominant trend the bounce occurs, the greater will be the gain expressed as a percentage, since it is starting from a lower base. This is why the full channel bounce on the 28th June 1991 does not give about twice the gain of the earlier half-channel bounce. The price movements above are simply the extent to which the price changes from the channel boundary. Naturally the limiting prices are not the ones at which the investor could have bought or sold. The investor can only buy when the price has bounced back from the boundary, i.e, has formed a trough. Thus the earliest the investor can buy is at the price immediately after the trough, i.e. one week later in the case of weekly data, or one day later in the case of daily data. The buying and selling signals for the above two price movements would therefore be: Buy on the 28th September 1990 @ 284p, sell on the 26th October @ 313p. Buy on the 5th July 1991 @ 378p, sell on the 30th August @ 429p. This gives gains of 10.2% and 13.5% respectively. NOT QUITE THE HOLY GRAIL Since we have noted that once the price reaches the lower boundary, it always reacts back to the centre, giving real profits of 10% or more in a short time, we appeared to have found a completely foolproof method of profitable investment. This would be true if we knew exactly the position of the channels at the point NOW, Le. how they have moved, in the case of channels based on a 41-week average, between NOW - 20 and NOW. The problem is exactly the same as that addressed earlier using the price analysis method. Then the approach used was to assume that the price remained constant for 20 weeks into the future and calculate the resulting average, thus filling in the gap between NOW - 20 and NOW. In the case of channel analysis, we have to have an even better method, since we need an accurate estimate of the current position of the average, and hence the channel boundaries, since the latter are a constant distance below and above the average. The position of the boundaries up to NOW - 20 has been determined by the more extreme of the peaks and troughs. The boundaries are either drawn by eye so as to pass through these extremes, or calculated by computer so as to allow only a certain, small percentage of points, normally about 3.5%, to lie outside the resulting channel. Thus any method to extrapolate the boundaries to NOW should adopt this same procedure. If there are peaks and/or troughs in this 2O-weekgap that lie close to the estimated position of the boundaries, then the boundaries can be bent upwards or downwards so as to pass through these peaks or troughs. Naturally there must be a limit imposed on the amount of bending towards the centre of the channel which is tolerable. If an unusual bend is required, then the
Channel Analysis
167
peak or trough in question should be accepted as one which does not lie on the boundary, but some distance inside of it. Bending away from the centre of the channel is a different matter. This will be required in those circumstances where the price lies outside the estimated boundary position. Since we allow only a very small proportion of points to lie outside the boundary, and even then not too far outside, then we must bend the channel, however extreme this bend may be, in order to keep the peak or trough close to this limit. This will only happen under the very occasional circumstance where the price has moved much more than the typical range of daily or weekly movement. Before proceeding to such an estimate carried out in this way, it is instructive to see what is a permissible error in the estimate of the boundary position. The effect of error will be different, according to whether we are approaching a lower or upper boundary. Error in Lower Boundary
If an estimate in the current position of the lower boundary is lower than the correct position, then the price will never reach it. The practical effect to the investor is that the share will never be bought. If the estimate of the current position is higher than the correct position, then the price will pass down through this incorrect boundary within a few days or weeks. Thus at the time the price is at the estimated boundary, we may take a premature buying decision. Error In Upper Boundary
If the estimate in the current position of the upper boundary is lower than the correct position, then the investor may sell prematurely at the apparent boundary, only to see the price re-establish its upward path to the correct higher boundary. If the estimate in the upper boundary is too high, then an investor relying on the fact of reaching the boundary as the selling signal will never receive such a signal. Thus a loss will inevitably occur if a stop-loss method is not used. In practice it turns out that the degree of tolerance in boundary position is quite high. Thus in Figure 11.7 we show boundaries that are 20p above and below the centre line rather than the correct value of 32p. If we take the lower boundary in Figure 11.7 to be that which we might have estimated incorrectly, then we can see that the effect of premature buying which this causes is not to give us an eventual loss,but to put us into a position where it is a few weeks before we move into profit. Our annual gain will be reduced by the amount by which our buying price stands above the buying price we would enter at when the price bounced back from the real channel boundary. These problems with inaccurate estimates of the channel boundaries can be reduced by two methods. Firstly, we noted that half of the bounces of price from
168 'IlO
Winning on the Stock Market •.................................................•...............
IllDTS 475
,
.
425
400
375 350 325
250 225
...........•...... ,
".,
.
M A" J J A SON 0 J F 1'1 A M J J A SO" DJ F " A 1'1 J J A SON 0 J F" A" J J AS 0 NO so 91 91 92 B9 B9 so
Figure 11.7 This shows the position if the boundaries are estimated incorrectly. The channel depth is set at 4Op, instead of 64p
the lower boundary do not take the price to the upper boundary, but only to the approximate midpoint. It is imperative to operate a stop-loss methods to insure against either this incorrect estimate of the boundary, or the premature retreat of the share price because of random influences. Secondly, since peaks and troughs lie on the boundary, then where a peak or trough lies close to an estimated position of the boundary, the boundary can be finely adjusted to touch these peaks or troughs. If we are fortunate, then there are a number of such peaks and troughs in the gap between NOW and NOW - 20. Thus we can continually fine-tune the channel boundaries as we proceed to interpolate across the gap from the last true position of the 41-week average, taking care to keep the vertical depth constant. The thought processes that occur during an estimate of the channel position can best be illustrated by means of three shares.
1. Boots On the 11th January 1991
This particular point is taken as an example because the price has just risen from a point which at first sight would appear to be at the lower channel boundary (Figure 11.8).
ChannelAnalysis
169
The example shows that great care must be taken, otherwise erroneous decisions can be taken. In Figure 11.8, we see the last calculated position of the centre line and channels 20 weeks ago from the 11th January. We proceed to move to the right from this point, viewing each peak or trough that we come across to see whether the channel should be adjust to pass through each feature as it occurs. The first trough we see is on the 21st September, and a simple extrapolation of the lower channel boundary at its current rate of curvature would bring it very close to this trough. Therefore we adjust the boundary so as to go exactly through the trough. Remembering that the channel depth is 64p, we then know the position of the upper boundary as being 64p higher than this trough. We also know that the centre line, Le. the average itself, is 32p higher than the trough. The next feature is the peak on the 19th October. The upper boundary extrapolated from the newly estimated position on the 21st September will pass a great deal higher than this peak. If the boundary is bent so as to pass exactly through this peak, then the degree of bending would be excessive. Not only that, but if the channel was bent in this way, the next peak on the 26th October would violate the upper boundary by a considerable extent. Thus we accept that the peak on the 19th October is one of those that does not approach the boundary even remotely. The same arguments apply to the trough on the 26th October. It is too near the middle of the channel to be allowed to affect the channel position. The next major peak on the 26th November is a different proposition. The upper boundary as extrapolated from the last estimated position on the 21st September
3':l0
.......•..............................................•...........
BaJrS
325
225
.
.................................................................•
A 90
"
J
J
s
o
"
Figure 11.8 This shows the position in Boots on the 11th January 1991
D
J 90 91
Winning on the Stock Market
170
would pass very close to it. Thus it is permissible to move the boundary slightly so as to pass directly through this point. Now we come to the trough on the 4th January which is led us to believe at first glance that the position on the 11th January was a buying one. We would have to bend the channel upwards at a sudden unacceptable rate from its curvature prior to the last estimated point on the 26th November. Thus we come to the conclusion that this trough was not close to the boundary (Figure 11.9), and that the share should not be bought on the 11th January. Once another few weeks have passed, the position shown in Figure 11.10 is reached. Although the trough just formed is almost at the same price point as the one on the 4th January, the upward trend of the channel in the intervening period means that this trough does lie close to the boundary. Therefore we can adjust the boundary to go through it, and can take the position on the 1st February at 329p as being a good quality buying point. The validity of this buying point can be seen from the subsequent rise in price as shown in Figure 11.11. The price runs up quite rapidly to the upper boundary of the channel. If, as suggested in the last chapter, a 7.5% stop-loss is used, then the investor would sell on the 28th June 1991 at a price of 359p for a gain of 9.1% in six weeks. Note that this buying point is the same as that signalled by the 5-week moving average method (see Table 10.6).
800TS
Z2S
......
250
225
.
I't 90
J
J
o
D
J 90 91
f
Figure 11.9 This shows the extrapolated channels for Boots on the 11th January 1991. Note that to pass through the trough on the 4th January would require an excessivebend upwards from the position of the channels on the 23rd /30th November
171
Channel Analysis l50
n-----------------------:;:r-------,-
225
22S
...
"
90
J
J
o
o
J 90 91
F'
Figure 11.10 Bythe time the next trough occurs in late January onlya slightly greater increasein the slopeof the channelis necessary to enable it to pass through this trough. The risefrom the boundarycan now be taken as a valid buying signal
If the investor runs a second selling method in parallel to the stop-loss method, where bounce-back off the estimated position of the upper channel is the criterion used, then a better selling point at a price of 382p on the 21st June is obtained.
British Telecom During 1989
This is chosen as an example because it was a very difficult share to make money from by using the 5-week moving average. In Table 10.6 in the last chapter, it was shown that, out of six buying/selling transactions, three of them during 1989 made a loss, and a fourth, also in 1989, made a very small profit of 3.6% before dealing costs. The British Telecom share price with the 41-week channel isshown in Figure 11.12. Now we can see the reason for the difficulty in 1989. The price does not touch the channel boundaries throughout this period. Thus channel analysis would not generate any buying signals during 1989, and the investor would not therefore suffer the poor performance shown in Table 10.6.
172 4'0
Winning on the Stock Market
...------------------------------n IllDTS
425
400
3/5
350
325
J 90 91
M
F
J
S
J
Figure 11.11 Once bought on the 1st February, the Boots share price makes a useful gain by rising to the upper boundary ofthe 41-week channel in seven months
4'0 セ M M M M M M M M M M M M M M M ョ BRITTELE
SI1DJFI1AMJ JASONDJFM AI1J JA SONDJFI1AM JJASOI1DJFI'IAMJJ
98
98 89
89 90
90 91
91 92
Figure 11.12 The 41-week channel of the British 'Ielecorn share price. The price did not touch the lower boundary during 1989, and therefore the share would not be bought during this period
173
Channel Analysis DuringAprill990
Following the barren period of 1989, the first point at which the British Telecom share price did touch the lower boundary was in April 1990. The position at the end of April when the price had rebounded from the lower channel is shown in Figure 11.13.We can obviouslyextrapolate the upper boundary so as to go through the two peaks in December 1989 and January 1990 by maintaining the previous rate of curvature, Le.no bending other than the most minor adjustment is required. By continuing the same rate of curvature, the lower boundary will pass through the trough which has just been formed in April. Since the direction of the channels gives us the direction of the dominant trend, this trend is downwards at the time of the proposed buying operation in April. Thus the share would not be bought. This conclusion as to the direction of the dominant trend is of course the same as we arrived at in Chapter 10 by using the price analysis method.
On the 1fxhNovember 1990 This example (Figure 11.14) is presented to illustrate one of the main rules of investment - when in doubt, don't invest. Although the price has moved up from a trough which is considerably lower than the upper boundary peak in early July 1990, it is not possible to take any . セッ
. セRU
225
,
5
o
,
N
'
o
.
J
89 90
F
M
Figure 11.13 Extrapolation of the boundaries at the end of April shows that the dominant trend is falling. British 'Ielecorn would not be bought at this time
11
Winningon the Stock Market
174
decision. This is because, other than the peak in July which locates the position of the upper boundary early in the extrapolation, there are no troughs formed during July and August which can be viewed as being even remotely close to the most sensible line for the lower boundary. Thus we cannot say that the lower boundary is passing through the trough formed in early November, because this would require us to bend up the boundary in order to do this. There is no obvious reason why the downward curve in the channel should be reversed in this way. Thus the probability that the trough was on the lower channel boundary is very low, and the share cannot be bought. In retrospect, it turns out that the channel did reverse direction at this point, and that buying in November would have been profitable. This can be seen from the relevant section of the channel in Figure 11.12. However, the investor must only invest on those occasions when the probability of success is high, and this was not the case on the available evidence in November 1990.
On the 1&hJanuary 1991 The position here is shown in Figure 11.15. At the middle of the chart it is obvious that the downward curve of the channel has come to an end and that the channel is poised to rise. There are no peaks from September to January that lie anywhere near the most sensible line for the upper boundary, so the only features we have to locate the channel are the two troughs, in November and January. The gentle 350
. llRITTELE
325
275
..
......
225
,.,
PI
90
A
PI
J
J
A
.
s
o
N
Figure 11.14 The 41-weelc channel of the British 'Ielecorn share price on the 16th November 1990. It is difficult to decide how to extrapolate the channel, and therefore the share should not be bought
175
Channel Analysis
upward curve that we estimate for the channel would need to be bent upwards rather more than is comfortable in order to put the trough in November on the channel boundary. By the time the trough in January is reached, the amount of bending required is minimal, and therefore we can say that there is a high probability that the boundary does pass through this trough. The channels extrapolated to this point are shown in Figure 11.16. Since we now assume that the boundary does pass through this trough, then the point after the trough is a valid buying point. This puts us into the share at a price of 277.5p. Looking at this section of the price movement in Figure 11.17 shows that the price made a cross-channel rise to the 350-375p price band within a few months, givinga very useful profit. U sing channel analysis as the selling trigger the investor would have noted that by the 12th April, the price at 365p was at the extrapolated upper boundary. The fall to 361p the following Friday would be the indication to sell. Since the investor would have bought at 277p on the 18th January, this gives a gain of 88p, or 31.7% in a matter of three months. British Gas The chart of British Gas in November 1989 is shown in Figure 11.18.The fact that there is a group of three peaks all at the 210p level during August and September is significant. This resistance level implies that the upper channel boundary at this . セッ
BRITIEl.E
325
,.,.,
, .. , . , . , . ,
.
300
..................................................
225
,
1'1
90
J
J
,
,
s
.
o
N
D
J
90 91
Figure 11.15 The 41-week channel of the British 'Ielecom share price on the 18th January 1991
Winning on the Stock Market
176 3'0
.
BRITIEl..E
22S
" .. ,
J
1'1 90
J
s
A
.
o
J 90 セi
D
"
Figure 11.16 The channels have now been extrapolated to January 1991, making the rise from the trough in January a valid buying point
2!lO
,
,
, .. ,
.
225
N 90
J
90 91
"
A
1'1
J
J
A
Figure 11.17 The price makes a good cross-channel rise following the bounce from the lower channel boundary in January 1991
Channel Analysis
177
275
.
BRIT6AS
250
225
.
......
200
..
150
.
F' 89
M
A
"
J
J
A
s
o
"
Figure 11.18 The 41-weekchannel in British Gas in November 1989
point is horizontal and passing through these peaks. Only a modest amount of downward bend from the middle of the chart is necessary to allow the upper boundary to do this. Thus it is highly probable that the dominant trend is not rising at the time there appears to be an investment opportunity in November. Thus channel analysis tells us not to make an investment. The subsequent movement of the share price (Figure 11.19) shows that the channel had changed direction in December, so that a useful rise was made by the price. Even so, on the evidence available in November, the correct decision was made. USING DAILY DATA Whereas the application of moving average to daily data in order to generate buying signals offered no real advantage over the use of weekly data, the position is rather different with channel analysis. This is because the buying signal is based on a rise from a trough which can be estimated to lie on the lower channel boundary. The trough can be much nearer to the start of the uptrend than is the turning point in the 5-week average. In fact, the most powerful aspect of channel analysis is that the trough is more often than not the exact start of the short term uptrend. Thus using daily data the investor can buy into the share at the earliest point that is theoretically possible. Although there will be a few occasions when the price obtained from weekly data is a better one that that from daily data, this
Winning on the Stock Market
178 300
..................................................•............... BRIT6AS
175
.,
2M
.....
.
1'5
,.,
S
89
D
,
J
8990
F
'"
.
A
'"
J
Figure 11.19 The subsequent movement of the British Gas share price after November 1989
only happens when the price falls back slightlyfrom the first point after the trough by the time of the Friday close. On balance, the price at the start of the uptrend wiU be the best price obtainable. It is not necessary to keep lengthy daily histories of share prices in order to take advantage of the daily movement at or near the channel boundaries, although such histories are available (see Appendix). The simplest approach is to continue to maintain weekly data, and then when it appears that the price is approaching a lower channel boundary, the price can be monitored daily. Thus it will only be necessary to keep daily data from a point just prior to the approaching buying point to the point at which the share is sold. This reduces the work necessary to an absolute minimum. To show that there is more often than not an advantage in using daily data in this way, the results obtained can be compared with those obtained using weekly data as shown in the previous examples. Boots
The position in Boots on the 3rd January 1991, with the price at 312p, is shown in Figure 11.20. The price the day before had been at 308p, and an extrapolation of the lower boundary of the channel would pass a considerable way below this trough. Thus we do not yet have a buying signal. By the end of January (Figure 11.21), the price rose from a trough of 315p on the 29th to 327p on the 30th. On
ChannelAnalysis 400
179
.......................................•..........•.......•..•.... IlOJI'
J75
.
350
.
"
J
J
A
s
o
"
D
J 90 '1
Figure 11.20 The position in Boots on the 3rd January 1991 with daily data
this occasion, it is not necessary to have to bend the lower boundary upwards in order for it to pass through this trough. Thus we can take this point on the 30th as being a valid buying point. Thus we have bought at 327p and with a selling signal being given by a bounce-back off the upper boundary on the 16th April at 367p for a gain of 4Op, i.e. 12.2%. In contrast, usingweeklydata as discussed earlier, we would have bought at 329p and sold at 382p for a gain of 53p, i.e, 16.1%. Thus we in this case have reduced the gain by using daily data to fine-tune the exact buying and selling points by being given a premature selling point. British Telecom
On the lst May 1990 In Figure 11.22is shown the position in British Telecom on the 1st May 1990.The price had risen to 249p from its close the previous day of 245p. The extrapolated boundary would appear to go through this trough. In spite of this, no investment would have been made because the best estimate of the channel at this time is that it is falling.
180
Winning on the Stock Market
400
. BCDl'
:PS
.
350
.
225
..
"
90
J
J
A
s
o
N
o
Figure 11.21 The position in Boots on the 30th January 1991 with daily data
. セッ
BT
325
300
225
.........•........................................•.......•.......
lL__- - - - - - - -
....•.....................................................•..•....
A 89
s
o
o
J
B9 90
F'
1'1
A
Figure 11.22 The position in British 'Ielecom on the Ist May 1990 with daily data
1'1
181
ChannelAnalysis British Telecom
On the 6th November1990
The situation here is shownin Figure 11.23. The price made a lowthe day before at 255p, then rose to 260p. A reasonable extrapolationof the boundary wouldput it at this 255plevel,but the rise to 260p cannot be taken as a buyingsignal,since the estimateddirection of the channel is down.This is the same conclusionas was reached withweekly data.
. セ
BT
325
................................................•............•.•..
225
.............................................................•.•..
F'