Читать книгу Millard on Channel Analysis - Brian Millard - Страница 8

The alternative to buying shares and holding them for long periods of time is to buy and sell them over shorter time periods. We still have to satisfy the above criterion, i.e. that the price rise over the shorter timescale will be more than sufficient to offset the buying and selling costs. Whether we will still have the advantage of any dividends will depend upon the time period over which we hold the shares. If we are lucky, then holding the shares for just one day could capture a dividend.

It is interesting to view Grand Met shares in terms of their yearly performance since the beginning of 1978, i.e. look at the gain or loss that occurred over each calendar year since that time. These annual changes are shown in Table 1.1.

Table 1.1 Yearly starting values, ending values and gains/losses made in the Grand Metropolitan share price from 1978 to 1995

We can see that out of these 18 yearly changes, seven were either losses, or gains of only 2.2%. Thus, there is no question that if we had not been invested in Grand Met during those seven years, but had found a better home in the money market for our funds, then even taking into account the dealing costs involved in selling and buying back a year later we would have made a much better return over the whole 18-year period.

Ignoring the price movement in the shares themselves over any particular time period, the major disadvantages of such a strategy of buying and selling frequently would appear to be:

 We have to carry buying costs of 2.5% and selling costs of 1.5% with each buying and selling transaction. Switching from one share to another therefore is an expensive operation.

 We have to spend time managing our portfolio.

The second of these disadvantages should be ignored by any serious investor. If the reward becomes high enough through the application of a successful investment strategy, then the time spent is worthwhile. The only negative aspect therefore is the high dealing cost of carrying out a strategy of buying and selling at frequent intervals.

The success of such a strategy now depends upon the answers to just three questions:

1 Do prices rise sufficiently over the investment term to offset the transaction costs and generate profit?

2 Since profits will be compounded, and bearing in mind the transaction costs, what is the shortest practicable time period over which to hold a share in order to maximise profits?

3 How difficult is it to capture good price rises and avoid bad price falls in a given time period?

In this chapter we will look at these first two questions. The objective of this book as a whole is to answer the third question.

Before any transaction can generate a profit, the price rise over the period of that transaction has to be considerably in excess of 5% in order to comfortably clear the dealing costs. A logical approach to this question of multiple transactions is to investigate shorter and shorter time periods over which the shares are held in order to decide at what point the average gain per transaction falls below that necessary to make a profit, i.e. falls below say 5%. There will come a point at which no profit will be made, and so we can say that shortening the timescale would appear to be working against us as far as the level of profit is concerned.

On the other hand, as we fit more and more transactions into a certain time period and reinvest the total proceeds of one transaction into the next, the compounding effect will move in our favour, increasing profits dramatically. Thus we expect that:

1 Reducing the transaction time reduces the real gain per transaction.

2 Reducing the transaction time increases the compound gain of multiple transactions.

The exercise therefore comes down to an investigation of the combined effect of these two factors which are acting in opposite directions. A staged approach to this question is valuable in helping us to gain an insight into the relationship between these two factors for a typical share such as Grand Metropolitan.

There is a third point to be made here, and that concerns the rate of gain. Thus although the reduced transaction time should reduce the gain made per transaction, since this occurs over a shorter timescale the rate of gain expressed, say, as a rate per week may be much better than that made from an investment with a longer transaction period. This is apart from any advantage to be obtained by compounding successive gains.

Figure 1.2 Medium-term trends in the Grand Metropolitan share price since 1978. These are represented by a centred 41-week moving average

Looking at the Grand Met share price in Figure 1.1 again, we can see some upward surges in share price which make good gains over time periods of up to about two years. These trends are displayed in Figure 1.2, and have been isolated by using a centred 41-week moving average as discussed in Chapter 3. Taking the rising part of the trends only, there are 13 such uptrends in the figure. In this chapter we are concerned only with the share prices at the time the trends started and when they finished. The beginning and end of a trend is signified by the changes in direction of the average.

The share prices at the turning points in these 13 uptrends are given in Table 1.2.

Table 1.2 Starting values, ending values and gains made in 13 upward trends in the Grand Metropolitan share price from 1978 to 1995

The timescale of these trends varies from 16 weeks up to 84 weeks with an average time of 45 weeks, i.e. nearly one year from the beginning to the end of the average trend. Note that the longest trend, 84 weeks, gave the largest gain, but one of the two shortest trends of 16 weeks did not give the smallest gain. In other words there is no obvious direct relationship between the length of time of an uptrend and the rise that occurs during it. The average gain for these 13 trends was 43.6%, corresponding to a gain factor of 1.436. We are, in the present exercise, trying to compare the performance of an investor who bought and held Grand Met shares for 18 years with one who took advantage of these 13 trends, buying at the start of each trend and selling at the end of that trend. In order to do this we have to adjust the performance of each investor to the same time period.

This can be done in several ways:

1 Adjust the gain from the seven transactions over 45 weeks to a gain over 18 years, i.e. 936 weeks.

2 Adjust the gain from the one transaction over 936 weeks down to a gain over 45 weeks.

3 Adjust both gains to some other common time period, e.g. one week.

Although at this point option 3 gives us twice as much work to do as options 1 or 2, adjusting the gains from both types of investment to an equivalent gain over one week will have the advantage that we will be able to compare other transactions over other time periods to this same common standard, giving a more realistic comparison between them.

Taking the long-term investor first, the gain factor of 8.92 over 18 years (936 weeks) will reduce down to an annual gain of the 18th root of 8.92, which is a gain factor of 1.129271 per annum. This is a gain in percentage terms of 12.9% per annum.

Brought to a weekly basis, the weekly gain is the 936th root of 8.92, which is 1.002341 over one week. In percentage terms, this is equivalent to 0.2341% per week. This is the gain that, if reinvested each year, would compound to a gain factor of 8.92, or 792%, over 936 weeks.

Carrying out the same calculation for the investor who buys and sells with the 13 trends with an average gain of 1.436 over 45 weeks, we have to raise 1.436 to the power (52/45) which gives an annual gain factor of 1.519. In percentage terms this is equivalent to 51.9% per annum. This is the gain that if reinvested each year for 18 years will give an ultimate gain of 1.436, i.e. 43.6%.

Compared on an annual basis, therefore, and with the important proviso that dealing costs are ignored, the gain from taking advantage of 13 upward surges in the share price over the 18-year period rather than one such surge lasting 18 years improves the gain on an annual basis from 12.9% to 51.9%, i.e. by a factor of about four.

Because dealing costs are high, they will have a considerable influence on profit as we increase the number of transactions that take place in a given time period, and therefore we have to take them into account when computing the various possibilities which we wish to compare. The buying and selling prices given in Table 1.1 have to be adjusted for these costs if we are to get a realistic idea of the gains which would be made from these seven transactions. To do this we adjust the buying price upwards by the typical amount of a buying cost, say 2.5%, and adjust the selling price downwards by the amount of these selling costs, say 1.5%. These details are given in Table 1.3. The effect of these costs is to reduce the average gain per transaction from 43.6% down to 38.05%.

Table 1.3 Buying prices, selling prices and gains in the 13 major trends in the Grand Metropolitan share price adjusted for dealing costs

Taking the long-term investor first, the adjusted gain factor of 8.57 over 18 years (936 weeks) will reduce down to an annual gain of the 18th root of 8.57, which is a gain factor of 1.126762 per annum. This is a gain in percentage terms of 12.67% per annum.

Brought to a weekly basis, the weekly gain is the 936th root of 8.57, which is 1.002298 over one week. In percentage terms, this is equivalent to 0.2298% per week. This is the gain that, if reinvested each year, would compound to a gain factor of 8.57, or 757%, over 936 weeks.

Carrying out the same calculation for the investor who buys and sells with the 13 trends with an average adjusted gain of 1.381 over 45 weeks, we have to raise 1.381 to the power (52/45) which gives an annual gain factor of 1.452. In percentage terms this is equivalent to 45.2% per annum. This is the gain that if reinvested each year for 18 years will give an ultimate gain of 1.381, i.e. 38.1%.

Compared on an annual basis, therefore, and with dealing costs now being taken into account, the gain from taking advantage of 13 upward surges in the share price over the 18-year period rather than one such surge lasting 18 years improves the gain on an annual basis from 12.67% to 45.2%, i.e. by a factor of about three and a half.

The clear message so far is that the theoretical annual rate of gain made from shorter-term transactions is vastly superior to the rate of gain made by buying and holding.

The reason we use the word theoretical is because we have made the assumption that we buy at the exact beginning of a trend and sell at the exact end. We shall take a more realistic view of where an investor might have got on board a rising trend, and where he would get off it, later in this chapter. At the moment we are simply trying to evaluate the theoretical effect of increasing the number of transactions over a certain time period. The reason, of course, that these 13 transactions give a superior gain is because the perfect timing of our theoretical investor takes him out of the market while the price is falling, whereas the buy and hold investor has to cope with the ups and downs of the 18-year period.

Figure 1.3 Short-term trends in the Grand Metropolitan share price since 1978. These are represented by a centred five-week moving average

Looking at Figure 1.1 more clearly, we can see that as well as the medium-term trends we have been analysing so far, there are trends of a shorter timescale. These trends are isolated by means of a five-week average, shown in Figure 1.3. The share prices at the turning points can be extracted just as in Figure 1.2 in order to analyse the price changes caused by these short-term trends. There are 41 such short-term uptrends, and the price data for these are given in Table 1.4. These trends lasted for an average of 12 weeks, as opposed to the 45 weeks of the longer-term trends. The average rise of each of these 12-week trends was 22.7%.

Table 1.4 Gains made in short-term trends in the Grand Metropolitan share price

Table 1.5 Buying prices, selling prices and gains in short-term trends in the Grand Metropolitan share price adjusted for dealing costs

Just as in the case of the 13 longer-term trends, we have to adjust the buying and selling points of the trends to allow for buying and selling costs. This is done in Table 1.5. We find that the average gain per transaction now falls to 17.94%. In order to compare this gain over a 12-week period with the previous values for 18 years and 45 weeks, we have to recalculate the gain as if it occurred over one year. We find that the gain factor of 1.1794 over 12 weeks is equivalent to a gain factor of 2.044 per annum, i.e. 104.4% per annum. This value supports our view that the rate of gain increases as we shorten the transaction time, even though of course the gain per transaction is less.

Figure 1.4 Very short-term trends in the Grand Metropolitan share price between July 1984 and September 1987. These are represented by the share prices themselves

Since we have this rate of gain moving so positively in our favour, the natural next step is to look for even shorter uptrends to take advantage of in this way. In Figure 1.4 we show an expanded portion of the Grand Met chart between July 1984 and September 1987. The very short-term movements which could not be seen clearly in Figure 1.1 can now be seen easily. In this time period there are 34 such trends. The actual price movements for these 34 trends are given in Table 1.6. Many of these trends last for only one week, and the longest for eight weeks. The average length of time for which these very short-term trends persist is 2.6 weeks. The average gain of these 34 transactions is 7.9% compared with the 22.7% in Table 1.4. We now appear to be coming to the shortest possible trends which will give us a profit, since we still have to adjust these for the dealing costs.

Table 1.6 Gains made in very short-term trends in the Grand Metropolitan share price

This is done in Table 1.7, once again by increasing the buying prices by 2.5% and decreasing the selling prices by 1.5%. Now we can see that the average gain per transaction has fallen to 3.7%. Once again, in order to compare with the previous calculations, we have to express this gain as if it occurred over one year.

Table 1.7 Buying prices, selling prices and gains in very short-term trends in the Grand Metropolitan share price adjusted for dealing costs

As before, we have to upgrade this gain to the equivalent gain over a one-year period, and this works out as a gain factor of 2.068, or 106.8% per annum. Since this is only marginally higher than the rate of 104.4% per annum obtained with the 41 short-term trends of Tables 1.4 and 1.5, it would appear that we are at about the optimum number of trades over the 18-year period in terms of rate of gain per week. However, bearing in mind the additional effort required for these very short-term transactions, we can consider that using the short-term trends rather than the very short-term trends represents the optimum, and its annual gain of 104.4% is a vast improvement over the annual gain of 12.67% made by the buy and hold investor.

The four situations we have examined so far are summarised in Table 1.8. Dividends have been omitted from each of the transactions in order to simplify the comparison.

Table 1.8 Length of trend, percentage gain and annual rate of gain for transactions in Grand Metropolitan shares

Two major points are illustrated by Table 1.8. The first of these is that as we take advantage of trends of shorter and shorter timescale, the gain made during the course of the trend falls lower and lower. This is a direct consequence of the properties of cyclical movements, and we shall see quite clearly later in this book that the longer the period of the cycle, the larger is the gain from the trough to the peak. Conversely, of course, very short-term cycles make small gains. The second important point is that the rate of gain, expressed as an annual gain for comparison purposes, increases as we move from one very long-term transaction of 18 years’ duration to 13 transactions of lesser duration. As active investors it is this rate of gain that we have to maximise, since we will be continually ploughing gains back into subsequent investments. The rate of gain increases again as we move to transactions of a shorter timescale, averaging 12 weeks per transaction, but then only marginally improves as we move to even shorter time periods of 2.6 weeks.

The reason for this is the effect of the dealing costs which really start to bite once we are down to lower gains per transaction. Thus there is a critical value of gain and a critical time period over which this gain is made, below which there appears to be no advantage to the investor. This time period lies between 12 weeks’ and 2.6 weeks’ duration for Grand Metropolitan shares. For other shares, the investor can determine this time period by going through the same exercise that we have in this chapter, but the results should be broadly comparable to those in Table 1.8.

Table 1.9 The percentage gain per investment needed to double the original investment assuming proceeeds are reinvested

Figure 1.5 The percentage gain per investment required to double the starting capital for various numbers of consecutive investments