Technical Analysis Explained

by Martin J Pring

  Momentum

  A simpler method is to calculate a momentum series and smooth it by an appropriate MA, as determined by trial and error. This approach will bring out the underlying rhythm in the price movement, just as a deviation-from-trend calculation does. It is doubtful whether the momentum approach alone can be successfully used for cycle identification, but it can prove to be an invaluable confirmation of cyclical reliability when used in conjunction with the technique of simple price observation discussed later.

  The position of a momentum indicator can also be useful in warning of potential cyclic inversions, i.e., when a projected cyclic low might turn out to be a cyclic high, and vice versa. For example, a cyclic inversion may occur when the observed data project that a cyclic low is likely to develop around a specific date, while the momentum indicator used in conjunction with this study is at, or coming down from, an overbought level. A good example is shown in Chart 24.5 featuring the S&P Composite. In 1987, the 9.2-year cycle appeared to reach its peak, as measured by the 55-month rate of change (ROC) at exactly the time when a low should have been recorded.

  Simple Observation

  Chart 25.1 shows the Philadelphia Gold and Silver Share Index. The solid line represents an 82-week cycle of lows and the dashed line a 126-week cycle of highs. The ROC has a time span of 41 weeks, i.e., half the 82-week cycle. Neither of these cycles is perfect but they do, for the most part, explain most of the turning points in the period under consideration. These two cycles were isolated on a trial-and-error basis using the cycle line tool in the MetaStock program.

  CHART 25.1 Philadelphia Gold and Silver Share Index and a 41-Week ROC, 1995–2002

  If you do not have access to a package such as this and wish to accomplish this task manually, the easiest method of identifying cycles is to start by observing on a price chart two or three major lows that appear to be relatively equidistant. The next step is to pencil in the projections for that particular cycle. If a substantial proportion of those projections result in either highs or lows, it is a good idea to mark them with a colored pencil. If most projections result in failure, the cycle should be abandoned and a new one sought. A cycle high occurring at any of these points should be treated as a successful projection, since the first objective of cycle analysis is to determine potential turning points. Once a reliable cycle has been established, the analyst should look at all the important cycle lows that are “unexplained” by the first cycle and try to “explain” them by discovering another cycle. The chances are that the second cycle will not only fit some of the unexplained lows, but will also occur at or near some of the cycle lows previously established. This is very important because a basic principle of cycle analysis is that the greater the number of cycles making a low around a certain time, the stronger the ensuing move is likely to be. Such knowledge must be used in conjunction with other technical evidence, but if that, too, offers a green light, the odds that a significant up wave will occur are increased.

  The next step in the method is discussed in the following section.

  Combining Cycle Highs and Lows

  The vertical lines in Chart 25.1 point up a fairly reliable pattern for both cycle highs and lows. One of the most important points that comes out of this exercise is the fact that the various turning points derive their significance from the direction of the main or primary trend. In this respect, the arrows on the chart flag the various bull and bear market environments. Note how the cycle tops tend to have greater magnitude in a bear market, such as the 1987 and 1990 tops. Conversely, the 1986 and late-1992 lows developed in a bull market and experienced far greater magnitude than the 1997 and 1999 signals, which developed in a bear market.

  One of the advantages of combining high and low cycles is that this approach makes it possible to obtain some idea of how long a rally or reaction might last. This arises from the proximity of the high and low. For example, the late-1992 low developed just after the high. The ensuing decline was quite brief. The reverse was true at the end of 1999, where the low was very close to the early 1990 high. The rally in this instance was short. The position of the ROC can often provide a clue as to whether a particular cyclical turning point will “work.” For example, the strong peaks in 1987, 1990, and late 1999 all developed when the ROC was at or close to an overbought condition. Similarly, the 1986 and 1988 lows were associated with moderately oversold conditions that failed to materialize.

  Not all examples work out quite as accurately as that shown in Chart 25.1. Readers are cautioned not to try to make a cycle “work.” If it does not fit naturally and easily, the chances are that it either does not exist or is likely to be highly unreliable and therefore should not be used. In any event, such analysis should always be used in conjunction with other indicators.

  Summary

  1. Recurring cycles, both of low points and high points, can be observed from charts of financial markets.

  2. A cycle turning point is significant both for the time interval between cycles and for the number of cycles that are turning at the same time.

  3. Cyclic analysis should always be used in conjunction with other indicators.

  4. Suspected cycles that do not easily fall into a consistently recurring pattern should not be made to “work” and should be discarded.

  26 VOLUME II: VOLUME INDICATORS

  Chapter 7 outlined some basic principles for interpreting volume. Now we will explore several volume indicators that can be applied to any security, finishing up the discussion with volume indicators that are more suitable for analyzing markets as a whole.

  Rate of Change (ROC) of Volume

  Normally, volume is displayed as a histogram underneath the price. A quick glance at any chart often reveals a noticeable increase in the size of the volume bars that are associated with breakouts, selling climaxes, and so forth. This is all well and good, but occasionally there are subtle shifts in the level of activity that are not easily detectable by this method. By massaging the volume data with an ROC calculation, it is possible to observe some new insights into the dynamics of volume interpretation.

  Chart 26.1 shows a 10-day ROC of volume together with a regular volume histogram for Northern Trust. I am using a 10-day ROC here, but, of course, it is possible to use any time span you wish. The price peak at A looks quite normal under the histogram method, but the ROC technique indicates a dramatic surge commensurate with an exhaustion move.

  CHART 26.1 Northern Trust, 2000–2001

  Peaks in the ROC indicator can often signal exhaustion that is not readily apparent with the histogram. In Chart 26.2 featuring T. Rowe Price, the initial peak at A is well signaled by the ROC, but not by the histogram in the bottom panel. The peak at B is indicated by an expansion in the histogram levels, but the rise is nowhere near that of the volume ROC, which was close to a record for the period covered in the chart. At C, the histogram rallies to a record level, but this is not picked up by the ROC. In this instance, the high volume followed the initial peak by 2 days and appears to be an isolated affair totally unimportant from an analytical point of view. Finally, at D, both series flag the selling climax at the first bottom but the comparison is far more dramatic for the ROC.

  CHART 26.2 T. Rowe Price, 1999–2001

  The ROC sometimes flags divergences. In Chart 26.3 featuring Stanley Works, for instance, the two dashed arrows point up the series of declining momentum peaks and rising price highs. This indicated serious potential weakness, and sure enough, the price did experience a decline once it had confirmed by violating its up trendline.

  CHART 26.3 Stanley Works, 1999–2000 and a Volume ROC

  In Chart 26.4 we see a situation at A where it is possible to observe a down trendline break in volume, which is also confirmed by a similar break in the price. Later on at B, the volume ROC traces out a small inverse head-and-shoulders pattern, which indicated a short-term trend of higher activity. This time, though, the price violated a trendline on the downside. This indicated an expansion
of volume on the downside, which is a bearish sign. Sometimes, short-term trades can take advantage of overbought readings since they often flag short-term turning points.

  CHART 26.4 Stanley Works, 1999–2000 and a Volume ROC

  It is fairly evident by now that a simple ROC of volume can be a pretty jagged indicator suitable only for pointing up exhaustion moves, divergences, and on a limited basis, constructing trendlines. Consequently, it makes sense to run a moving average through a volume rate of change because it smooths out the jagged nature of the raw data.

  In Chart 26.5 this has been taken one step further by calculating a 10-day moving average of a 25-day volume ROC.

  CHART 26.5 Snap-on Inc, 1992–1994 and a Volume ROC

  This series is a lot smoother than those we have been looking at so far. It is fairly evident that this approach lends itself more readily to price pattern and trendline analysis. First note that the overbought and oversold lines are not drawn on an equidistant basis. This is because the calculation treats the ROC as a percentage. Volume can expand, even on a smoothed basis, by 200 percent or 300 percent quite easily, yet can only fall by 100 percent. This means that downside action is far more limited than upside potential. Later on we will look at some alternative form of calculating a smoothed ROC of volume. There are two principal events on the chart. The first is a head-and-shoulders top. The head represented a buying crescendo, which indicated a change of trend. Once the neckline had been violated, this merely represented a confirmation that the volume trend was down. Since the price also violated an up trendline, both price and volume were now in gear on the downside, and it was reasonable to expect an extended correction.

  The end of the sell-off was signaled as both the price and volume violated important downtrend lines. Since they were now in gear on the upside, this provided a nice confirmation that the uptrend was healthy. The down trendline for volume actually represents the neckline of a reverse head-and-shoulders pattern. I did not display the H and S letters on the chart because the left shoulder was also the right shoulder of the previous upward-sloping head-and-shoulders top, and this would have complicated things.

  Volume ROC: The Percent Calculation

  We discovered earlier that the volume ROC indicator, when calculated with a percent method, does not lend itself to pointing up oversold volume conditions very well. A solution is to calculate the ROC using a subtraction method, as I have done in the lower panel of Chart 26.6. An oversold condition is signaled at A for the subtraction method, but not for the division calculation. The disadvantage of the subtraction calculation is that volume momentum cannot be compared over long periods of time if the security being monitored experiences a substantial increase in average daily volume. For charts that span less than 2 years of data, the subtraction technique is probably a better approach, though it’s important to remember that because the volume level for individual stocks and markets can vary tremendously, the overbought and oversold lines will have to be adjusted accordingly. In Chart 26.6 the upper dashed horizontal lines represent an overbought reading in both volume indicators. Since the price had been declining, the high level at B signaled a selling climax. The subtraction-based oscillator was deeply oversold at A, the next slightly lower bottom. This indicated a classic double-bottom characteristic. It is also apparent, if it was not before now, that a high overextended reading in a volume oscillator does not necessarily mean that the price is overbought—merely that volume is overextended. A high reading in volume oscillators can mean a top or a bottom, depending on the previous price action. I’ll have more to say on that point a little later.

  CHART 26.6 Snap-on Inc, 1994–1995 and Two Volume ROC Calculations

  Major Technical Principle A high overextended reading in a volume indicator does not necessarily mean that the price is overbought—merely that volume is overextended. A high reading can mean a top or a bottom, depending on the direction of the previous trend. A reversal from a high reading may also indicate a change as opposed to a reversal in trend.

  Primary-Trend Volume ROC

  Annual (12-month) ROCs are a useful way of measuring primary-trend price momentum, but since volume trends are more random, such indicators tend to be quite jagged in nature. Chart 26.7 shows the annual ROC price for the Standard & Poor’s (S&P) Composite and New York Stock Exchange (NYSE) Volume. However, in order to overcome the jagged nature of the volume data, both series have been smoothed with a 6-month moving average (MA). When used together, the two series can be extremely instructive.

  CHART 26.7 Price versus Volume Momentum, 1969–1990

  Several observations can be made:

  1. The volume curve has an almost consistent tendency to peak out ahead of price during both bull and bear phases.

  2. In most instances, fairly reliable indications of a potential trend reversal can be obtained when the volume momentum crosses price.

  3. When the price index is above its zero reference line and is falling but volume is rising (e.g., 1953, early and late 1976, 1981, and late 1987; the experience in early 1964 appeared to be an exception), the expanding activity represents distribution and should be interpreted as a very bearish factor once the rally has terminated.

  4. A reversal in volume at a market bottom should be confirmed by a reversal in price momentum.

  5. Very high readings in the volume indicator are usually followed by strong bull markets.

  6. When volume crosses below zero, it is normally, but not always, a negative sign. The most bearish situations seem to arise when the price indicator is well above zero. In 1988, for instance, price momentum was well below zero when volume moved into negative territory but the market rallied. On the other hand, in 1969, 1973, and 1977 volume crossed below zero just after price momentum had started to roll over from an overbought level, and this was followed by a major decline.

  7. During the initial stages of a bull move, volume momentum is always above price. (The 1988–1989 rally represents the only exception.)

  Chart 26.7 shows that a reversal in one curve unaccompanied by a reversal in the other at market bottoms tends to give a premature signal. For example, volume turned up ahead of price at the end of 1973 and 1977. Consequently, it is wiser to await a signal from both, even though it may occur at a slightly higher price level.

  Volume Oscillator

  The volume oscillator is an alternative method of presenting volume in a momentum format. It is calculated by dividing a short-term measure of the volume trend, usually a moving average, by a longer-term one and plotting the result as an oscillator. It is also possible to divide the close by a measure of trend as well. In effect, the calculation is identical as that for price under the trend deviation section in Chapter 14. The only difference is that volume is substituted for price. An example is shown in Figure 26.1 The resulting indicator is an oscillator that revolves around a zero reference line. A zero reading in the indicator occurs when both MAs are at identical levels. Positive readings develop when the shorter-term (10-day in the figure) MA is above its longer-term (25-day) counterpart, and vice versa. At points A and B the two moving averages cross and the indicator also moves through zero. On the other hand, C and D represent overbought and oversold readings, respectively, where the distance between the two averages is greatest. Ten- and 25-day time spans have been used in the example, but any widely separated spans could be used instead.

  FIGURE 26.1 Calculating a Volume Oscillator

  An example is shown in Chart 26.8 featuring Humana. Here we can see that the volume oscillator breaks above a small down trendline in March 1999. This upside breakout merely indicated that the trend of volume was likely to be up. It said nothing about the course of prices. In this instance, the price violated a horizontal trendline, which indicated a bearish combination of declining prices and expanding volume. This was then followed by an overbought reading in the oscillator, a sign of a selling climax. That could have proved to be the final bottom. However, the oscillator broke above a trendlin
e in July 2000 and the price violated one on the downside, indicating an almost exact replay of the

  CHART 26.8 Humana, 1998–1999, and a Volume Oscillator

  Chart 26.9 features Advent Claymore, where we can see that oscillator breakouts at A and B signified a new trend of rising volume. We can also see a selling and buying climax in October 2011 and 2012, respectively. A joint trendline break in July 2012 also signaled a nice rally. The volume breakout indicated that the advance would take place under a background of rising activity, which clearly added strength to its validity until the bearish October buying climax.

  CHART 26.9 Advent Claymore, 2011–2012, and a 10/25 Volume Oscillator

  Chart 26.10 features a 15/45 volume oscillator for Columbia Energy. This longer-term interrelationship offers more deliberate and less jagged price action. The most obvious thing to note is the late-1999 summer reverse head and shoulders in the oscillator. Once again, an increase in volume was indicated, but this time the price broke to the upside. We see another example later on in November. It is rare when these oscillators are able to trace out price patterns, but when they do, the signals tend to be very reliable. The chart also shows a buying climax in early November, which was later confirmed by a trend break in the price itself, then a selling climax in the following March.