Technical Analysis Explained

by Martin J Pring

  One of the drawbacks of the method used in the construction of the DJIA is that if a stock increases in price and is not split, its influence on the average will become substantially greater, especially if many of the other Dow stocks are growing and splitting at the same time. In spite of this and other drawbacks, however, the Dow has, over the years, acted fairly consistently with many of the more widely capitalized market averages. The ETF that represents the DJIA is the SPDR Dow Jones Industrial Average ETF (symbol DIA).

  The Standard & Poor’s (S&P) Composite, which comprises 500 stocks representing well over 90 percent of the NYSE market value, is another widely followed bellwether average. Its ETF is the SPDR S&P 500 (symbol SPY). The index is calculated by multiplying the price of each share by the number outstanding, totaling the value of each company, and reducing the answer to an index number.

  Over the years, the S&P 500 has become the benchmark against which professional money managers are judged. It is also the most widely traded equity futures contract.

  Most of the time, the DJIA and S&P 500 move in the same direction, but there are times when a new high or low is achieved in one index but not the other. Generally speaking, the greater the divergence, the greater the next move in the opposite direction is likely to be. Chart 21.1 shows that in late 1968 the S&P 500 reached a new all-time high, unlike the DJIA, which was not able to surpass its 1966 peak. This development helped to signal a bear market that wiped nearly 40 percent off the value of both averages. On the other hand, the 1973–1974 bear market was completed with a double bottom. In the case of the DJIA, the second bottom in December 1974 was lower than the October one, yet the S&P 500 failed to confirm the new low in the DJIA. In the space of the next two years, the DJIA rose by some 80 percent. This is also shown in Chart 21.1.

  CHART 21.1 Key Market Averages, 1964–1978

  Chart 21.2 compares the DJIA to the S&P Composite for the turn of the century. For most of the 1990s, both series were in gear. However, the DJIA made its peak in January 2000, and the S&P topped out in March and September of that year. This indicated that both averages were out of gear with each other. The confirmation of a bear trend came later in the year when both averages violated important up trendlines. Do not get the impression that the lack of a divergence means a healthy market, because there was no discrepancy at the next bull market peak in 2007, and this was followed by one of the worst bear markets in history.

  CHART 21.2 The DJIA versus the S&P Composite, 1998–2001 and Divergences

  The NASDAQ Composite is a capitalization-weighted index consisting of all the stocks listed on the NASDAQ. Since it contains most of the technology heavyweights, it is very much a technology-driven index. However, when it comes to ETFs, the NASDAQ 100 (symbol QQQ) is the preferred vehicle. This index/ETF is constructed from the 100 NASDAQ issues with the largest capitalization.

  The NYSE compiles an all-encompassing index called the NYSE Composite. In a sense, it represents the ideal average, since its value is based on the capitalization of all shares on the exchange. Its movements are very similar to those of the DJIA and the S&P 500. Nevertheless, divergences between the trends of these three averages offer additional confirmation of changes in the overall technical structure.

  The most comprehensive indicator of all is the Wilshire 5000 Equity Index, which represents the value-weighted total, in billions of dollars, of all actively traded common stocks in the United States. Conceptually, this is the indicator that should be used for monitoring trends of the overall market, but because of the lethargy of the investment community and the obvious vested interest of the sponsors of the other popular averages, it has not received the widespread recognition that it justly deserves.

  Value Line has published the Value Line Arithmetic, an equally weighted price index that reflects the broad market. Since its construction emphasizes smaller stocks, it occasionally differs in its trajectory with, say, the Wilshire 5000 in a significant way. Chart 21.3 compares the two. Note how there was a huge discrepancy between them just after the 2000 stock market peak. That was undoubtedly due to the unwinding of the tech bubble since tech stocks had fought their way to huge weightings in the cap-weighted indexes. Generally speaking, any form of discrepancy whichever index leads is a sign of weakness in the prevailing trend that, when confirmed by price, usually results in a worthwhile reversal. You can see a small divergence at the 2007 peak, which was confirmed by two trendline breaks.

  CHART 21.3 Wilshire 5,000 versus the Value Line Arithmetic 1998–2012 and Divergences

  The Market Averages and MAs

  When experimenting with a moving average (MA) from the point of view of trend determination, it is first necessary to assess the type of cycle to be considered. The 4-year stock market cycle has corresponded to the U.S. business cycle for many decades. Since the stock market is greatly influenced by business cycle developments, this 4-year (or, to place it more exactly, 41-month) cycle is of great significance in trend determination. Consequently, the choice of an MA to detect such swings is limited to anything less than the full period, i.e., 41 months, since an MA covering this whole time span would smooth out the complete cycle and theoretically become a straight line. In practice, the MA does fluctuate, since the cycle is rarely limited exactly to its average 41 months and varies in magnitude of price change. Through computer research1 it has been found that a 12-month MA for the S&P Composite was the most reliable between 1910 and the early 1990s. Between then and 2012, there were only four whipsaw signals.

  Major Technical Principle When choosing a time span for a moving average, go for consistency over a number of securities rather than perfection.

  In his book The Stock Market Indicators (Investors Press, 1968), William Gordon calculated that a 40-week crossover gave 29 buy and sell signals for the DJIA between 1897 and 1967. The average gain for all bull signals (i.e., between the buy and sell signals) was 27 percent, and the average change from sell signals was 4 percent. For investors using the buy signals to purchase stocks, nine resulted in losses, although none greater than 7 percent, while gains were significantly higher. This approach has worked reasonably well since 1967, though it is important to note that 40-week MA crossovers of the S&P Composite resulted in many whipsaws in the late 1970s. As so often happens after a number of whipsaws, the 1982 buy signal was superb. It captured most of the initial advance of the 1982–1987 bull market, while the second, in late 1984, would have kept investors in the market until the Friday before the 1987 crash.

  The arrows in Chart 21.4 show whipsaw 40-week MA crossovers for the S&P Composite between 1996 and 2012. They may look plentiful on the chart, but developed, on average, about once every 18 months or so. There were also quite a lot of occasions when the average acted as a support or resistance area by turning back advances and declines. When combined with the valid crossover signals, it can be argued that the 40-week MA acts as a fairly reliable benchmark, but as always, keep the following principle in mind.

  CHART 21.4 S&P Composite, 1996–2012 versus a 1/40 Price Oscillator

  Major Technical Principle Moving averages should always be used in conjunction with other indicators to obtain a weight-of-the-evidence approach.

  For intermediate swings, crossovers of 13- and 10-week (50-day) averages have proved to be useful benchmarks, but naturally, an MA covering such a brief time span can result in many misleading whipsaws and is, therefore, less reliable than the 40-week average. For even shorter swings, a 30-day (6-week) MA works well, although some technicians prefer a 25-day average.

  The Major Averages and ROCs

  There are many ways in which the techniques described in previous chapters can be adapted to the major averages. In Charts 21.5 and 21.6, for instance, the S&P Composite is featured with a 9-month rate of change (ROC).

  CHART 21.5 S&P Composite, 1900–1950 versus a 9-Month ROC

  CHART 21.6 S&P Composite, 1950–2001 versus a 9-Month ROC

  It seems that an excellent signal of an interme
diate to primary trend bottom develops when the ROC either recrosses above its oversold line at –20 percent or touches the –20 percent level and then reverses. Alternatively, a recrossing of the +20 percent level appears to be a reasonably reliable intermediate peak or bear market signal. Obviously, this is not a perfect indicator, but for the most part, it works with a high degree of statistical reliability. Some of the most glaring errors are flagged by the ellipses. The first, in 1929–1930, which was obviously premature, and the second in the late 1990s, where several signals of weakness completely failed. Chart 21.7 shows the same exercise for more recent times, but in this instance, the S&P has been adjusted by the Consumer Price Index (CPI). Note that in a secular bull market there is a slight tendency for the sell signals to fail, whereas the opposite is true in the secular bear between 2000 and 2012.

  CHART 21.7 CPI Adjusted S&P Composite, 1979–2012 versus a 9-Month ROC

  Another technique is to construct an up trendline joining the bear market low with the first intermediate bottom. This is then combined with a 12-month ROC where a similar trendline is constructed or a price pattern, if available, flagged. That is the idea behind most of the trendlines drawn in Charts 21.8 and 21.9. Sometimes it is not possible to construct such lines and we are left with the alternative of a secondary trendline. When both are violated, this is usually a good sign that the bull move is over. Most of the time, the signals come fairly close to the bull market peak.

  CHART 21.8 S&P Composite, 1966–1983 and Trendlines

  CHART 21.9 S&P Composite, 1989–2012 and Trendlines

  Major Technical Principle In cases where it is obvious that trendlines are going to be violated well after the turning point, it is usually best to disregard them and rely on other evidence.

  Chart 21.10 shows the trendline break technique with a 65-week ROC of the Dow Industrials. With a time span of this length, signals tend to be few in number. Sometimes the ROC offers a useful trendline break, such as at A and B, but the price trend is so sharp that it is not possible to construct a meaningful trendline. That, unfortunately, is a fact of life and is better ignored than forced, i.e., drawing a sharp trendline just to make the data fit.

  CHART 21.10 DJIA, 2000–1012 and Trendlines

  A simpler technique for identifying intermediate trends is to use reversals in the trend of a 13-week ROC of a market average, such as the S&P Composite, in conjunction with a reversal in the trend in the level of the average itself. The technique used in Chart 21.11 involves the drawing of trendlines for both the weekly closing price of the DJIA and its 13-week momentum. When a break in one index is confirmed by the other, a reversal in the prevailing trend usually takes place. Such signals are illustrated in the chart by the arrows. This type of analysis should be supported where appropriate with price pattern analysis for the S&P, and with other techniques utilizing the momentum principles described in Chapter 13. This method does not always give a signal, but whenever there are clearly definable violations of trendlines that have been touched three or more times, the conclusions drawn are usually extremely reliable.

  CHART 21.11 DJIA, 1970–1975 and a 13-Week ROC

  In Chart 21.12 the letters A–E indicate where overbought/oversold crossovers are not confirmed by the price. In the cases of B and D, these were preliminary signals, where a subsequent oversold decline was eventually confirmed. A, C, and E were never confirmed.

  CHART 21.12 DJIA, 2001–2011 and a 13-Week ROC

  The Dow Jones Transportation Average

  In the last part of the nineteenth century and the early part of the twentieth century, rail was the dominant form of transportation and, therefore, an average composed solely of rails represented a good proxy for transportation stocks. In 1970, the Rail Average was expanded to embrace other transportation segments, and the index was renamed the Dow Jones Transportation Average.

  The Transportation Average is basically affected by two factors: volume of business and changes in interest rates. First, when a business recovery gets under way, inventories are low and raw materials are needed to initiate production. Transportation volume picks up, and investors, anticipating such a trend, bid up the price of transportation shares. At business cycle peaks, companies typically overbuild their stocks; the result is that when sales start to fall, their requirements for raw materials are reduced. Transportation volume then falls sharply, and the stocks react accordingly. Second, transport companies tend to be more heavily financed with debt than industrials. Because of the leverage of this heavy debt structure, their earnings are also more sensitive to changes in interest rates and business conditions than those of most industrial companies. As a result, the Transportation Average quite often leads the Industrial Average at important juncture points. Indeed, recently conducted research data since the mid-1950s confirmed that the Transportation sector outperformed the market during the early primary bull phase and underperformed during the latter stages of the cycle. (Note: See my book The Investor’s Guide to Active Asset Allocation.)

  The significance of the Dow theory rule requiring confirmation of both the Industrials and Transportations should now be more obvious, since a move by the producer stocks (the Industrials) really has to be associated with an increased volume of transportation, which should be reflected by a similar move in the Transportation stocks. In a similar vein, increased business for the Transportation stocks is likely to be of temporary significance if the industrial companies fail to follow through with a rise in sales and production levels. The longer-term cycles of the Transportation Average and the Industrial Average are more or less the same as a result of their close association with business conditions. The techniques and choice of time spans for MAs, ROCs, etc., are, therefore, similar to those described earlier for the Industrials.

  One principle that is not normally used for the Industrials but that can be applied to the Transportations is that of relative strength (RS). This technique is particularly useful during periods of nonconfirmation between the two averages, when RS can offer a useful clue as to how the discrepancy will be resolved. One such example occurred in the summer of 1998 when the DJIA made a marginal new high. Chart 21.13 shows that the Transports remained above their 40-week MA but the average had already violated a secondary uptrend, thereby indicating potential weakness. As it turned out, when the Industrials reached their new high, the Transports rallied back to the extended trendline that they had previously violated. However, the real tip-off that the Transports were unlikely to confirm the Industrials came from the fact that the RS line had crossed below its MA and secondary up trendline in April 1998. The 26-week ROC of relative strength also violated an up trendline. Thus, at the time when the Industrials were making a new high in July, the Transport RS line was declining and well below its MA. Finally, the ROC was unable to rally above zero, which represented an additional sign of vulnerability.

  CHART 21.13 DJ Transports, 1996–1998 and Three Indicators

  The Dow Jones Utility Average

  The Dow Jones Utility Average comprises 15 utility stocks drawn from electric utilities, gas pipelines, telephone companies, etc. This average has historically proved to be one of the most reliable barometers of the Industrials. This is because utility stocks are extremely sensitive to changes in interest rates and interest rates generally lead the overall stock market. Interest rate changes are important to utility stocks for two reasons. First, utility companies require substantial amounts of capital because they are usually highly financed with debt relative to equity. As interest rates rise, the cost of renewing existing debt and raising additional money puts pressure on profits. When interest rates fall, these conditions are reversed and profits rise. Second, utility companies generally pay out their earnings in the form of dividends so that these equities are normally bought just as much for their yield as for their potential capital gain. When interest rates rise, bonds, which are also bought for their yield, fall in price and thus become relatively more attractive than utilities. As a result, inve
stors are tempted to sell utility stocks and buy bonds. When interest rates fall, the money returns once again to utility stocks, which then rise in price.

  Major Technical Principle Since changes in the trend of interest rates usually occur ahead of reversals in the stock market, the Utility Average more often than not leads the DJIA at both market tops and market bottoms.

  Generally speaking, when the Utility Average flattens out after an advance or moves down while the Industrials continue to advance, it is usually a sign of an imminent change in trend for the Industrials. Thus, the Utilities led the Industrials at the 1937, 1946, 1953, 1966, 1968, 1973, and 1987 bull market peaks. Conversely, at the 1942, 1949, 1953, 1962, 1966, 1974, and 1982 bottoms, the Utilities made their bear market lows ahead of the Industrials. At most major juncture points, the Utilities coincided with the Industrials, and occasionally, as at the 1970 bottom and the 1976 top, the Utilities lagged. Chart 21.14 shows that for the most part since the 1970s they have led, but lagged at the 2000 peaks and very slightly at the 2007 top.