Moving averages don't work all of the time. They do their best work when the market is in a trending phase. They're not very helpful during trendless periods when prices trade sideways. Fortunately, there's another class of indicator that performs much better than the moving average during those frustrating trading ranges. They're called oscillators and we'll explain them in the next chapter.

One of the problems encountered with the moving average is choosing between a fast or a slow average. While one may work better in a trading range market, the other may be preferable in a trending market. The answer to the problem of choosing between the two may lie with an innovative approach called the "adaptive moving average."
Perry Kaufman presents this technique in his book Smarter Trading. The speed of Kaufman's "adaptive moving average" auto­matically adjusts to the level of noise (or volatility) in a market. The AMA moves more slowly when markets are trending side­ways, but then moves more swiftly when the market is trending. That avoids the problem of using a faster moving average (and getting whipsawed more frequently) during a trading range, and using a slower average that trails too far behind a market when it is trending.
Kaufman does that by constructing an Efficiency Ratio that compares price direction with the level of volatility. When the Efficiency Ratio is high, there is more direction than volatility (favoring a faster average). When the ratio is low, there's more volatility than direction (favoring a slower average). By incorpo­rating the Efficiency Ratio, the AMA automatically adjusts to the speed most suitable for the current market.

We've presented a lot of variations on the moving average approach. Let's try to simplify things a bit. Most technicians use a combination of two moving averages. Those two averages are usually simple averages. Although exponential averages have become popular, there's no real evidence to prove that they work any better than the simple average. The most commonly used daily moving average combinations in futures markets are 4 and 9, 9 and 18, 5 and 20, and 10 and 40. Stock traders rely heavily on a 50 day (or 10 week) moving average. For longer range stock market analysis, popular weekly moving averages are 30 and 40 weeks (or 200 days). Bollinger Bands make use of 20 day and 20 week moving averages. The 20 week average can be converted to daily charts by utilizing a 100 day average, which is another use ful moving average. Channel breakout systems work extremely well in trending markets and can be used on daily, weekly, and monthly charts.

The first edition of this book included the results of extensive research produced by Merrill Lynch, which published a series of studies on computerized trading techniques applied to the futures markets from 1978-82. Extensive testing of various moving aver­age and channel breakout parameters was performed to find the best possible combinations in each futures market. The Merrill Lynch researchers produced a different set of optimized indicator values for each market.
Most charting packages allow you to optimize systems and indicators. Instead of using the same moving average in all mar­kets, for example, you could ask the computer to find the moving average, or moving average combinations, that have worked the best in the past for that market. That could also be done for daily
and weekly breakout systems and virtually all technical indicators included in this book. Optimization allows technical parameters to adapt to changing market conditions.
Some argue that optimization helps their trading results and others that it doesn't. The heart of the debate centers on how the data is optimized. Researchers stress that the correct procedure is to use only part of the price data to choose the best parameters, and another portion to actually test the results. Testing the opti­mized parameters on "out of sample" price data helps ensure that the final results will be closer to what one might experience from actual trading.
The decision to optimize or not is a personal one. Most evi­dence, however, suggests that optimization is not the Holy Grail some think it to be. I generally advise traders following only a handful of markets to experiment with optimization. Why should Treasury Bonds or the German mark have the exact same moving averages as corn or cotton? Stock market traders are a different story. Having to follow thousands of stocks argues against opti­mizing. If you specialize in a handful of markets, try optimizing. If you're a generalist who follows a large number of markets, use the same technical parameters for all of them.

There are other alternatives to the moving average as a trend-fol­lowing device. One of the best known and most successful of these techniques is called the weekly price channel or, simply, the weekly rule. This technique has many of the benefits of the mov­ing average, but is less time consuming and simpler to use.
With the improvements in computer technology over the past decade, a considerable amount of research has been done on the development of technical trading systems. These systems are mechanical in nature, meaning that human emotion and judg­ment are eliminated. These systems have become increasingly sophisticated. At first, simple moving averages were utilized. Then, double and triple crossovers of the averages were added. The averages were then linearly weighted and exponentially smoothed. These systems are primarily trend-following, which means their purpose is to identify and then trade in the direction of an existing trend.
With the increased fascination with fancier and more com­plex systems and indicators, however, there has been a tendency to overlook some of the simpler techniques that continue to work quite well and have stood the test of time. We're going to discuss one of the simplest of these techniques—the weekly rule.
In 1970, a booklet entitled the Trader's Notebook was pub­lished by Dunn & Hargitt's Financial Services in Lafayette, Indiana. The best known commodity trading systems of the day were computer-tested and compared. The final conclusion of all that research was that the most successful of all the systems test­ed was the 4 week rule, developed by Richard Donchian. Mr.
Donchian has been recognized as a pioneer in the field of com­modity trend trading using mechanical systems. (In 1983, Managed Account Reports chose Donchian as the first recipient of the Most Valuable Performer Award for outstanding contributions to the field of futures money management, and presents The Donchian Award to other worthy recipients.)
More recent work done by Louis Lukac, former research director at Dunn & Hargitt and currently president of Wizard Trading (a Massachusetts CTA) supports the earlier conclusions that breakout (or channel) systems similar to the weekly rule con­tinue to show superior results. (Lukac et al.)*
Of the 12 systems tested from 1975-84, only 4 generated significant profits. Of those 4, 2 were channel breakout systems and one was a dual moving average crossover system. A later arti­cle by Lukac and Brorsen in The Financial Review (November 1990) published the results of a more extensive study done on data from 1976-86 that compared 23 technical trading systems. Once again, channel breakouts and moving average systems came out on top. Lukac finally concluded that a channel breakout system was his personal choice as the best starting point for all technical trading system testing and development.
The 4 Week Rule
The 4 week rule is used primarily for futures trading.
The system based on the 4 week rule is simplicity itself:
1. Cover short positions and buy long whenever the price exceeds the highs of the four preceding full calendar weeks.
2. Liquidate long positions and sell short whenever the price falls below the lows of the four preceding full calendar weeks.
The system, as it is presented here, is continuous in nature, which means that the trader always has a position, either long or short. As a general rule, continuous systems have a basic weakness. They stay in the market and get "whipsawed" during trendless market periods. It's already been stressed that trend-following systems do not work well when markets are in these sideways, or trendless phases.
* See Bibliography
The 4 week rule can be modified to make it noncontinu­ous. This can be accomplished by using a shorter time span—such as a one or two week rule—for liquidation purposes. In other words, a four week "breakout" would be necessary to initiate a new position, but a one or two week signal in the opposite direc­tion would warrant liquidation of the position. The trader would then remain out of the market until a new four week breakout is registered.
The logic behind the system is based on sound technical principles. Its signals are mechanical and clearcut. Because it is trend following, it virtually guarantees participation on the right side of every important trend. It is also structured to fol­low the often quoted maxim of successful trading—"let profits run, while cutting losses short." Another feature, which should not be overlooked, is that this method tends to trade less fre­quently, so that commissions are lower. Another plus is that the system can be implemented with or without the aid of a computer.
The main criticism of the weekly rule is the same one lev­eled against all trend-following approaches, namely, that it does not catch tops or bottoms. But what trend-following system does? The important point to keep in mind is that the four week rule performs at least as well as most other trend-following sys­tems and better than many, but has the added benefit of incred­ible simplicity.
Adjustments to the 4 Week Rule
Although we're treating the four week rule in its original form, there are many adjustments and refinements that can be employed. For one thing, the rule does not have to be used as a trading system. Weekly signals can be employed simply as anoth­er technical indicator to identify breakouts and trend reversals. Weekly breakouts can be used as a confirming filter for other tech­niques, such as moving average crossovers. One or 2 week rules function as excellent filters. A moving average crossover signal could be confirmed by a two week breakout in the same direction in order for a market position to be taken.
Shorten or Lengthen Time Periods for Sensitivity
The time period employed can be expanded or compressed in the interests of risk management and sensitivity. For example, the time period could be shortened if it is desirable to make the sys­tem more sensitive. In a relatively high priced market, where prices are trending sharply higher, a shorter time span could be chosen to make the system more sensitive. Suppose, for example, that a long position is taken on a 4 week upside breakout with a protective stop placed just below the low of the past 2 weeks. If the market has rallied sharply and the trader wishes to trail the position with a closer protective stop, a one week stopout point could be used.
In a trading range situation, where a trend trader would just as soon stay on the sidelines until an important trend signal is given, the time period could be expanded to eight weeks. This would prevent taking positions on shorter term and premature trend signals.
The 4 Week Rule Tied to Cycles
Earlier in the chapter reference was made to the importance of the monthly cycle in commodity markets. The 4 week, or 20 day, trading cycle is a dominant cycle that influences all markets. This may help explain why the 4 week time period has proven so suc­cessful. Notice that mention was made of 1, 2, and 8 week rules. The principle of harmonics in cyclic analysis holds that each cycle is related to its neighboring cycles (next longer and next shorter cycles) by 2.
In the previous discussion of moving averages, it was pointed out how the monthly cycle and harmonics explained the popularity of the 5, 10, 20, and 40 day moving averages. The same time periods hold true in the realm of weekly rules. Those daily numbers translated into weekly time periods are 1, 2, 4, and 8 weeks. Therefore, adjustments to the 4 week rule seem to work best when the beginning number (4) is divided or multiplied by 2. To shorten the time span, go from 4 to 2 weeks. If an even shorter time span is desired, go from 2 to 1. To lengthen, go from
4 to 8. Because this method combines price and time, there's no reason why the cyclic principle of harmonics should not play an important role. The tactic of dividing a weekly parameter by 2 to shorten it, or doubling it to lengthen it, does have cycle logic behind it.The 4 week rule is a simple breakout system. The original system can be modified by using a shorter time period—a 1 or 2 week rule—for liquidation purposes. If the user desires a more sen­sitive system, a 2 week period can be employed for entry signals. Because this rule is meant to be simple, it is best addressed on that level. The 4 week rule is simple, but it works.(Charting packages allow you to plot price channels above and below current prices to spot channel breakouts. Price channels can be used on daily,

The reader should not overlook using this technique in longer range trend analysis. Longer range moving averages, such as 10 or 13 weeks, in conjunction with the 30 or 40 week average, have long been used in stock market analysis, but haven't been given as much attention in the futures markets. The 10 and 40 week moving averages can be used to help track the primary trend on weekly charts for futures and stocks.Some Pros and Cons of the Moving Average
One of the great advantages of using moving averages, and one of the reasons they are so popular as trend-following systems, is that they embody some of the oldest maxims of successful trading. They trade in the direction of the trend. They let profits run and cut losses short. The moving average system forces the user to obey those rules by providing specific buy and sell signals based on those principles.
Because they are trend-following in nature, however, mov­ing averages work best when markets are in a trending period. They perform very poorly when markets get choppy and trade sideways for a period of time. And that might be a third to a half of the time.
The fact that they do not work that well for significant periods of time, however, is one very compelling reason why it is dangerous to rely too heavily on the moving average tech­nique. In certain trending markets, the moving average can't be beat. Just switch the program to automatic. At other times, a nontrending method like the overbought—oversold oscillator is more appropriate. (In Chapter 15, we'll show you an indicator called ADX that tells you when a market is trending and when it is not, and whether the market climate favors a trending moving average technique or a nontrending oscillator approach.)

Moving Averages As Oscillators
One way to construct an oscillator is to compare the difference between two moving averages. The use of two moving averages in the double crossover method, therefore, takes on greater sig­nificance and becomes an even more useful technique. We'll see how this is done in Chapter 10. One method compares two expo­nentially smoothed averages. That method is called Moving Average Convergence/Divergence (MACD). It is used partially as an oscillator. Therefore, we'll postpone our explanation of that technique until we deal with the entire subject of oscillators in Chapter 10.
Some Pros and Cons of the Moving Average
One of the great advantages of using moving averages, and one of the reasons they are so popular as trend-following systems, is that they embody some of the oldest maxims of successful trading. They trade in the direction of the trend. They let profits run and cut losses short. The moving average system forces the user to obey those rules by providing specific buy and sell signals based on those principles.
Because they are trend-following in nature, however, mov­ing averages work best when markets are in a trending period. They perform very poorly when markets get choppy and trade sideways for a period of time. And that might be a third to a half of the time.
The fact that they do not work that well for significant periods of time, however, is one very compelling reason why it is dangerous to rely too heavily on the moving average tech­nique. In certain trending markets, the moving average can't be beat. Just switch the program to automatic. At other times, a nontrending method like the overbought–oversold oscillator is more appropriate. (In Chapter 15, we'll show you an indicator called ADX that tells you when a market is trending and when it is not, and whether the market climate favors a trending moving average technique or a nontrending oscillator approach.)
Moving Averages As Oscillators
One way to construct an oscillator is to compare the difference between two moving averages. The use of two moving averages in the double crossover method, therefore, takes on greater sig­nificance and becomes an even more useful technique. We'll see how this is done in Chapter 10. One method compares two expo­nentially smoothed averages. That method is called Moving Average Convergence/Divergence (MACD). It is used partially as an oscillator. Therefore, we'll postpone our explanation of that technique until we deal with the entire subject of oscillators in Chapter 10.
Some Pros and Cons of the Moving Average
One of the great advantages of using moving averages, and one of the reasons they are so popular as trend-following systems, is that they embody some of the oldest maxims of successful trading. They trade in the direction of the trend. They let profits run and cut losses short. The moving average system forces the user to obey those rules by providing specific buy and sell signals based on those principles.
Because they are trend-following in nature, however, mov­ing averages work best when markets are in a trending period. They perform very poorly when markets get choppy and trade sideways for a period of time. And that might be a third to a half of the time.
The fact that they do not work that well for significant periods of time, however, is one very compelling reason why it is dangerous to rely too heavily on the moving average tech­nique. In certain trending markets, the moving average can't be beat. Just switch the program to automatic. At other times, a nontrending method like the overbought–oversold oscillator is more appropriate. (In Chapter 15, we'll show you an indicator called ADX that tells you when a market is trending and when it is not, and whether the market climate favors a trending moving average technique or a nontrending oscillator approach.)
Moving Averages As Oscillators
One way to construct an oscillator is to compare the difference between two moving averages. The use of two moving averages in the double crossover method, therefore, takes on greater sig­nificance and becomes an even more useful technique. We'll see how this is done in Chapter 10. One method compares two expo­nentially smoothed averages. That method is called Moving Average Convergence/Divergence (MACD). It is used partially as an oscillator. Therefore, we'll postpone our explanation of that technique until we deal with the entire subject of oscillators in Chapter 10.
The Moving Average Applied to Other Technical Data
The moving average can be applied to virtually any technical data or indicator. It can be used on open interest and volume figures, including on balance volume. The moving average can be used on various indicators and ratios. It can be applied to oscillators as well.

We'll cover the Fibonacci number series in the chapter on Elliott Wave Theory. However, I'd like to mention here that this myste­rious series of numbers—such as 13, 21, 34, 55, and so on—seem to lend themselves quite well to moving average analysis. This is true not only of daily charts, but for weekly charts as well. The 21 day moving average is a Fibonacci number. On the weekly charts, the 13 week average has proven valuable in both stocks and com­modities. We'll postpone a more in depth discussion of these numbers until Chapter 13.

Many market analysts believe that time cycles play an important role in market movement. Because these time cycles are repetitive and can be measured, it is possible to determine the approximate times when market tops or bottoms will occur. Many different time cycles exist simultaneously, from a short term 5 day cycle to Kondratieff's long 54 year cycle. We'll delve more into this fasci­nating branch of technical analysis in Chapter 14.
The subject of cycles is introduced here only to make the point that there seems to be a relationship between the underly­ing cycles that affect a certain market and the correct moving averages to use. In other words, the moving averages can be adjusted to fit the dominant cycles in each market.
There appears to be a definite relationship between mov­ing averages and cycles. For example, the monthly cycle is one of the best known cycles operating throughout the commodity mar­kets. A month has 20-21 trading days. Cycles tend to be related to their next longer and shorter cycles harmonically, or by a factor of two. That means that the next longer cycle is double the length of a cycle and the next shorter cycle is half its length.
The monthly cycle, therefore, may explain the popularity of the 5, 10, 20, and 40 day moving averages. The 20 day cycle measures the monthly cycle. The 40 day average is double the 20 day. The 10 day average is half of 20 and the 5 day average is half again of 10.
Many of the more commonly used moving averages (including the 4, 9, and 18 day averages, which are derivatives of 5, 10, and 20) can be explained by cyclic influences and the har­monic relationships of neighboring cycles. Incidentally, the 4 week cycle may also help explain the success of the 4 week rule, covered later in the chapter, and its shorter counterpart—the 2 week rule.

Bollinger Bands differ from envelopes in one major way. Whereas the envelopes stay a constant percentage width apart, Bollinger Bands expand and contract based on the last 20 days' volatility. During a period of rising price volatility, the distance between the two bands will widen. Conversely, during a period of low market volatility, the distance between the two bands will contract. There is a tendency for the bands to alternate between expansion and contraction. When the bands are unusually far apart, that is often a sign that the current trend may be ending. When the distance between the two bands has narrowed too far, that is often a sign that a market may be about to initiate a new trend. Bollinger Bands can also be applied to weekly and monthly price charts by using 20 weeks and 20 months instead of 20 days. Bollinger Bands work best when combined with overbought/oversold oscillators that are explained in the next chapter. (See Appendix A for addi­tional band techniques.)
Centering the AverageThe more statistically correct way to plot a moving average is to center it. That means to place it in the middle of the time peri­od it covers. A 10 day average, for example, would be placed five days back. A 20 day average would be plotted 10 days back in time. Centering the average, however, has the major flaw of producing much later trend change signals. Therefore, moving averages are usually placed at the end of the time period cov­ered instead of the middle. The centering technique is used almost exclusively by cyclic analysts to isolate underlying mar­ket cycles

The simplest way to use Bollinger Bands is to use the upper and lower bands as price targets. In other words, if prices bounce off the lower band and cross above the 20 day average, the upper band becomes the upper price target. A crossing below the 20 day average would identify the lower band as the downside target. In a strong uptrend, prices will usually fluctuate between the upper band and the 20 day average. In that case, a crossing below the 20 day average warns of a trend reversal to the downside.

This technique was developed by John Bollinger. Two trading bands are placed around a moving average similar to the enve­lope technique. Except that Bollinger Bands are placed two stan­dard deviations above and below the moving average, which is usually 20 days. Standard deviation is a statistical concept that describes how prices are dispersed around an average value. Using two standard deviations ensures that 95% of the price data will fall between the two trading bands. As a rule, prices are con­sidered to be overextended on the upside (overbought) when they touch the upper band. They are considered overextended on the downside (oversold) when they touch the lower band.

The usefulness of a single moving average can be enhanced by surrounding it with envelopes. Percentage envelopes can be used to help determine when a market has gotten overextended in either direction. In other words, they tell us when prices have strayed too far from their moving average line. In order to do this, the envelopes are placed at fixed percentages above and below the average. Shorter term traders, for example, often use 3% envelopes around a simple 21 day moving average. When prices reach one of the envelopes (3% from the average), the short term trend is considered to be overextended. For long range analysis, some possible combinations include 5% envelopes around a 10 week average or a 10% envelope around a 40 week average. (See Figures 9.8a-b.)

The moving average is essentially a trend following device. Its pur­pose is to identify or signal that a new trend has begun or that an old trend has ended or reversed. Its purpose is to track the progress of the trend. It might be viewed as a curving trendline. It does not, however, predict market action in the same sense that standard chart analysis attempts to do. The moving average is a follower, not a leader. It never anticipates; it only reacts. The mov­ing average follows a market and tells us that a trend has begun, but only after the fact.
The moving average is a smoothing device. By averaging the price data, a smoother line is produced, making it much easi­er to view the underlying trend. By its very nature, however, the moving average line also lags the market action. A shorter mov­ing average, such as a 20 day average, would hug the price action more closely than a 200 day average. The time lag is reduced with the shorter averages, but can never be completely eliminated. Shorter term averages are more sensitive to the price action, whereas longer range averages are less sensitive. In certain types of markets, it is more advantageous to use a shorter average and, at other times, a longer and less sensitive average proves more useful. (See Figure 9.1b.)
Which Prices to Average
We have been using the closing price in all of our examples so far. However, while the closing price is considered to be the most
important price of the trading day and the price most commonly used in moving average construction, the reader should be aware that some technicians prefer to use other prices. Some prefer to use a midpoint value, which is arrived at by dividing the day's range by two.
Others include the closing price in their calculation by adding the high, low, and closing prices together and dividing the sum by three. Still others prefer to construct price bands by averaging the high and low prices separately. The result is two separate moving average lines that act as a sort of volatility buffer or neutral zone. Despite these variations, the closing price is still the price most commonly used for moving average analysis and is the price that we'll be focusing most of our attention on in this chapter.
The Simple Moving Average
The simple moving average, or the arithmetic mean, is the type used by most technical analysts. But there are some who question its usefulness on two points. The first criticism is that only the peri­od covered by the average (the last 10 days, for example) is taken into account. The second criticism is that the simple moving aver­age gives equal weight to each day's price. In a 10 day average, the last day receives the same weight as the first day in the calcula­tion. Each day's price is assigned a 10% weighting. In a 5 day aver­age, each day would have an equal 20% weighting. Some analysts believe that a heavier weighting should be given to the more recent price action.
The Linearly Weighted Moving Average
In an attempt to correct the weighting problem, some analysts employ a linearly weighted moving average. In this calculation, the closing price of the 10th day (in the case of a 10 day average) would be multiplied by 10, the ninth day by nine, the eighth day by eight, and so on. The greater weight is therefore given to the more recent closings. The total is then divided by the sum of the multipliers (55 in the case of the 10 day average: 10 + 9 + 8 + .. . + 1). However, the linearly weighted average still does not address the problem of including only the price action covered by the length of the average itself.
The Exponentially Smoothed
Moving Average
This type of average addresses both of the problems associated with the simple moving average. First, the exponentially smoothed average assigns a greater weight to the more recent data. Therefore, it is a weighted moving average. But while it assigns lesser importance to past price data, it does include in its calculation all of the data in the life of the instrument. In addi­tion, the user is able to adjust the weighting to give greater or lesser weight to the most recent day's price. This is done by assigning a percentage value to the last day's price, which isadded to a percentage of the previous day's value. The sum of both percentage values adds up to 100. For example, the last day's price could be assigned a value of 10% (.10), which is added to the previous day's value of 90% (.90). That gives the last day 10% of the total weighting. That would be the equivalent of a 20 day average. By giving the last day's price a smaller value of 5% (.05), lesser weight is given to the last day's data and the average is less sensitive. That would be the equivalent of a 40 day mov­ing average.
The computer makes this all very easy for you. You just have to choose the number of days you want in the moving aver­age-10, 20, 40, etc. Then select the type of average you want—simple, weighted, or exponentially smoothed. You can also select as many averages as you want—one, two, or three.
The Use of One Moving Average
The simple moving average is the one most commonly used by technicians, and is the one that we'll be concentrating on. Some traders use just one moving average to generate trend signals. The moving average is plotted on the bar chart in its appropriate trad­ing day along with that day's price action. When the closing price moves above the moving average, a buy signal is generated. A sell signal is given when prices move below the moving average. For added confirmation, some technicians also like to see the moving average line itself turn in the direction of the price crossing. (See Figure 9.3.)
If a very short term average is employed (a 5 or 10 day), the average tracks prices very closely and several crossings occur. This action can be either good or bad. The use of a very sensitive aver­age produces more trades (with higher commission costs) and results in many false signals (whipsaws). If the average is too sen­sitive, some of the short term random price movement (or "noise") activates bad trend signals.While the shorter average generates more false signals, it has the advantage of giving trend signals earlier in the move. It stands to reason that the more sensitive the average, the earlier the signals will be. So there is a tradeoff at work here. The trick is to find the average that is sensitive enough to generate early sig­nals, but insensitive enough to avoid most of the random "noise." Let's carry the above comparison a step further. While the longer average performs better while the trend remains in motion, it "gives back" a lot more when the trend reverses. The very insensitivity of the longer average (the fact that it trailed the trend from a greater distance), which kept it from getting tangled up in short term corrections during the trend, works against the trader when the trend actually reverses. Therefore, we'll add another corollary here: The longer averages work better as long as the trend remains in force, but a shorter average is better when the trend is in the process of reversing.
It becomes clearer, therefore, that the use of one moving average alone has several disadvantages. It is usually more advan­tageous to employ two moving averages.
How to Use Two Averages to Generate Signals
This technique is called the double crossover method. This means that a buy signal is produced when the shorter average crosses above the longer. For example, two popular combinations are the 5 and 20 day averages and the 10 and 50 day averages. In the for­mer, a buy signal occurs when the 5 day average crosses above the 20, and a sell signal when the 5 day moves below the 20. In the latter example, the 10 day crossing above the 50 signals an uptrend, and a downtrend takes place with the 10 slipping under the 50. This technique of using two averages together lags the market a bit more than the use of a single average but produces fewer whipsaws. The Use of Three Averages, or the Triple Crossover Method
That brings us to the triple crossover method. The most widely used triple crossover system is the popular 4-9-18-day moving average combination. The 4-9-18 method is used mainly in futures trading. This concept was first mentioned by R.C. Allen in his 1972 book, How to Build a Fortune in Commodities and again later in a 1974 work by the same author, How to Use the 4-Day, 9-Day and 18-Day Moving Averages to Earn Larger Profits from Commodities. The 4-9-18-day system is a variation on the 5, 10, and 20 day moving average numbers, which are widely used in commodity circles. Many commercial chart services publish the 4-9-18-day moving averages. (Many charting software packages use the 4-9-18-day combination as their default values when plotting three averages.).How to Use the 4-9-18-Day Moving Average System
It's already been explained that the shorter the moving average, the closer it follows the price trend. It stands to reason then that the shortest of the three averages—the 4 day—will follow the trend most closely, followed by the 9 day and then the 18. In an uptrend, therefore, the proper alignment would be for the 4 day average to be above the 9 day, which is above the 18 day average. In a downtrend, the order is reversed and the alignment is exact­ly the opposite. That is, the 4 day would be the lowest, followed by the 9 day and then the 18 day average. (See Figures 9.7a-b.)
A buying alert takes place in a downtrend when the 4 day crosses above both the 9 and the 18. A confirmed buy signal occurs when the 9 day then crosses above the 18. This places the
4 day over the 9 day which is over the 18 day. Some intermin­gling may occur during corrections or consolidations, but the general uptrend remains intact. Some traders may take profits during the intermingling process and some may use it as a buy­ing opportunity. There is obviously a lot of room for flexibility here in applying the rules, depending on how aggressively one wants to trade.
When the uptrend reverses to the downside, the first thing that should take place is that the shortest (and most sensitive) average—the 4 day—dips below the 9 day and the 18 day. This is only a selling alert. Some traders, however, might use that initial crossing as reason enough to begin liquidating long positions. Then, if the next longer average—the 9 day—drops below the 18 day, a confirmed sell short signal is given.

The following pages contain examples of long term weekly and monthly charts (Figures 8.2-8.12). The drawings on the charts are limited to long term support and resistance levels, trendlines, per­centage retracements, weekly reversals, and an occasional price pattern. Be aware, however, that anything that can be done on a daily chart can also be done on a weekly or monthly. We'll show you later in the book how the application of various technical indicators to these long term charts is accomplished, and how sig­nals on weekly charts become valuable filters for shorter term tim­ing decisions. Remember also that semilog chart scaling becomes more valuable when studying long range price trends.((ssss))

Long term charts are not meant for trading purposes. A distinc­tion has to be made between market analysis for forecasting pur­poses and the timing of market commitments. Long term charts are useful in the analytical process to help determine the major trend and price objectives. They are not suitable, however, for the timing of entry and exit points and should not be used for that purpose. For that more sensitive task, daily and intraday charts should be utilized.

A question often raised concerning long term charts is whether or not historic price levels seen on the charts should be adjusted for inflation. After all, the argument goes, do these long range peaks and troughs have any validity if not adjusted to reflect the changes in the value of the U.S. dollar? This is a point of some controversy among analysts.
I do not believe that any adjustment is necessary on these long range charts for a number of reasons. The main reason is my belief that the markets themselves have already made the necessary adjustments. A currency declining in value causes commodities quoted in that currency to increase in value. The declining value of the dollar, therefore, would contribute to rising commodity prices. A rising dollar would cause the price of most commodities to fall.
The tremendous price gains in commodity markets during the 1970s and declining prices in the 1980s and 1990s are classic examples of inflation at work. To have suggested during the 1970s that commodity price levels that had doubled and tripled in price should then be adjusted to reflect rising inflation would make no sense at all. The rising commodity markets already were a mani­festation of that inflation. Declining commodity markets since the 1980s reflect a long period of disinflation. Should we take the price of gold, which is now worth less than half of its value in 1980, and adjust it to reflect the lower inflation rate? The market has already taken care of that.The final point in this debate goes to the heart of the tech­nical theory, which states that price action discounts everything, even inflation. All financial markets adjust to periods of inflation and deflation and to changes in currency values. The real answer to whether long range charts should be adjusted for inflation lies in the charts themselves. Many markets fail at historic resistance levels set several years earlier and then bounce off support levels not seen in several years. It's also clear that falling inflation since the early 1980s has helped support bull markets in bonds and stocks. It would seem that those markets have already made their own inflation adjustment

It's especially important to appreciate the order in which price charts should be studied in performing a thorough trend analysis. The proper order to follow in chart analysis is to begin with the long range and gradually work to the near term. The reason for this should become apparent as one works with the different time dimensions. If the analyst begins with only the near term picture, he or she is forced to constantly revise conclusions as more price data is considered. A thorough analysis of a daily chart may have to be completely redone after looking at the long range charts. By starting
with the big picture, going back as far as 20 years, all data to be con­sidered are already included in the chart and a proper perspective is achieved. Once the analyst knows where the market is from a longer range perspective, he or she gradually "zeros in" on the shorter term.
The first chart to be considered is the 20 year monthly chart. The analyst looks for the more obvious chart patterns, major trend-lines, or the proximity of major support or resistance levels. He or she then consults the most recent five years on the weekly chart, repeat­ing the same process. Having done that, the analyst narrows his or her focus to the last six to nine months of market action on the daily bar chart, thus going from the "macro" to the "micro" approach. If the trader wants to proceed further, intraday charts can then be con­sulted for an even more microscopic study of recent action.

Price patterns appear on the long range charts, which are inter­preted in the same way as on the daily charts. Double tops and bot­toms are very prominent on these charts, as are head and shoul­der reversals. Triangles, which are usually continuation patterns, are frequently seen.
Another pattern that occurs quite frequently on these charts is the weekly and monthly reversal. For example, on the monthly chart, a new monthly high followed by a close below the previous month's close often represents a significant turning point, especially if it occurs near a major support or resistance area. Weekly reversals are quite frequent on the weekly charts. These patterns are the equivalent of the key reversal day on the daily charts, except that on the long range charts these reversals carry a great deal more significance.

The most striking features of long range charts is that not only are trends very clearly defined, but that long range trends often last for years. Imagine making a forecast based on one of these long range trends, and not having to change that forecast for several years!
The persistence of long range trends raises another inter­esting question that should be mentioned—the question of ran­domness. While technical analysts do not subscribe to the theory that market action is random and unpredictable, it seems safe to
observe that whatever randomness does exist in price action is probably a phenomenon of the very short term. The persistence of existing trends over long periods of time, in many cases for years, is a compelling argument against the claims of Random Walk Theorists that prices are serially independent and that past price action has no effect on future price action.

An innovative solution to the problem of price continuity was developed by Robert Pelletier, president of Commodity Systems, Inc., a commodity and stock data service (CSI. 200 W. Palmetto Park Road, Boca Raton, FL 33422), called the Perpetual Contract.TM ("Perpetual ContractTM " is a registered trademark of that firm.)
The purpose of the Perpetual ContractTM is to provide years of futures price history in one continuous time series. That is accomplished by constructing a time series based on a constant forward time period. For example, the series would determine a value three months or six months into the future. The time peri­od varies and can be chosen by the user. The Perpetual ContractTM is constructed by taking a weighted average of two futures con­tracts that surround the time period desired.
The value for the Perpetual ContractTM is not an actual price, but a weighted average of two other prices. The main advan­tage of the Perpetual ContractTM is that it eliminates the need for using only the nearest expiring contract and smoothes out the price series by eliminating the distortions that can take place dur­ing the transition between delivery months. For chart analysis pur­poses, the nearest-month continuation charts published by chart services are more than adequate. A continuous price series, how­ever, is more useful for back-testing trading systems and indicators. A more complete explanation of ways to construct continuous futures contracts is provided by Greg Morris in Appendix D.

Of all the charts utilized by the market technician for forecasting and trading the financial markets, the daily bar chart is by far the most popular. The daily bar chart usually covers a period of only six to nine months. However, because most traders confine their inter­est to relatively short term market action, daily bar charts have gained wide acceptance as the primary working tool of the chartist.
The average trader's dependence on these daily charts, however, and the preoccupation with short term market behavior, cause many to overlook a very useful and rewarding area of price charting—the use of weekly and monthly charts for longer range trend analysis and forecasting.
The daily bar chart covers a relatively short period of time in the life of any market. A thorough trend analysis of a market, however, should include some consideration of how the daily market price is moving in relation to its long range trend struc­ture. To accomplish that task, longer range charts must be employed. Whereas on the daily bar chart each bar represents one day's price
action, on the weekly and monthly charts each price bar repre­sents one week's and one month's price action, respectively. The purpose of weekly and monthly charts is to compress price action in such a way that the time horizon can be greatly expanded and much longer time periods can be studied.Construction of Continuation Charts for Futures.
THE IMPORTANCE OF LONGER RANGE PERSPECTIVE
Long range price charts provide a perspective on the market trend that is impossible to achieve with the use of daily charts alone. During our introduction to the technical philosophy in Chapter 1, it was pointed out that one of the greatest advantages of chart analysis is the application of its principles to virtually any time dimension, including long range forecasting. We also addressed the fallacy, espoused by some, that technical analysis should be limited to short term "timing" with longer range forecasting left to the fun­damental analyst.
The accompanying charts will demonstrate that the prin­ciples of technical analysis—including trend analysis, support and resistance levels, trendlines, percentage retracements, and price patterns—lend themselves quite well to the analysis of long range price movements. Anyone who is not consulting these longer range charts is missing an enormous amount of valuable price information.
CONSTRUCTION OF
CONTINUATION CHARTS FOR FUTURES
The average futures contract has a trading life of about a year and a half before expiration. This limited life feature poses some obvi­ous problems for the technician interested in constructing a long range chart going back several years. Stock market technicians don't have this problem. Charts are readily available for individ­ual common stocks and the market averages from the inception of trading. How then does the futures technician construct longer range charts for contracts that are constantly expiring?
The answer is the continuation chart. Notice the emphasis on the word "continuation." The technique most commonly employed is simply to link a number of contracts together to provide continuity. When one contract expires, another one is used. In order to accomplish this, the simplest method, and the one used by most chart services, is to always use the price of the nearest expiring contract. When that nearest expiring contract stops trading, the next in line becomes the nearest contract and is the one plotted.
Other Ways to Construct Continuation Charts
The technique of linking prices of the nearest expiring contracts is relatively simple and does solve the problem of providing price continuity. However, there are some problems with that method. Sometimes the expiring contract may be trading at a significant premium or discount to the next contract, and the changeover to the new contract may cause a sudden price drop or jump on the chart. Another potential distortion is the extreme volatility expe­rienced by some spot contracts just before expiration.
Futures technicians have devised many ways to deal with these occasional distortions. Some will stop plotting the nearest contract a month or two before it expires to avoid the volatility in the spot month. Others will avoid using the nearest contract alto­gether and will instead chart the second or third contract. Another method is to chart the contract with the highest open interest on the theory that that delivery month is the truest rep­resentation of market value.
Continuation charts can also be constructed by linking specific calendar months. For example, a November soybean continuation chart would combine only the historic data pro­vided by each successive year's November soybean contract. (This technique of linking specific delivery months was favored by W.D. Gann.) Some chartists go even further by averaging the prices of several contracts, or constructing indices that attempt to smooth the changeover by making adjustments in the price premium or discount.

That concludes our coverage of volume and open interest, at least for now. Volume analysis is used in all financial markets—futures,
options, and stocks. Open interest applies only to futures and options. But, since futures and options are traded on so many stock market vehicles, some understanding of how open interest works can be useful in all three financial arenas. In most of our discussions so far, we've concentrated on daily bar charts. The next step is to broaden our time horizon and to learn how to apply the tools we've learned to weekly and monthly charts in order to perform long range trend analysis. We'll accomplish that in the next chapter.

Options traders use open interest and volume put/call figures to determine extremes in bullish or bearish sentiment. These senti­ment readings work best when combined with technical measures such as support, resistance, and the trend of the underlying mar­ket. Since timing is so crucial in options, most option traders are technically oriented.

Volume figures for the options markets are used essentially the same way as in futures and stocks—that is, they tell us the degree of buying or selling pressure in a given market. Volume figures in options are broken down into call volume (bullish) and put volume (bearish). By monitoring the volume in calls versus puts, we are able to determine the degree of bullishness or bearishness in a mar­ket. One of the primary uses of volume data in options trading is the construction of put/call volume ratios. When options traders are bullish, call volume exceeds put volume and the put/call ratio falls. A bearish attitude is reflected in heavier put volume and a higher put/call ratio. The put/call ratio is usually viewed as a con­trary indicator. A very high ratio signals an oversold market. A very low ratio is a negative warning of an overbought market.

Our coverage of open interest has concentrated on the futures markets. Open interest plays an important role in options trading as well. Open interest figures are published each day for put and call options on futures markets, stock averages, industry indexes, and individual stocks. While open interest in options may not be interpreted in exactly the same way as in futures, it tells us essen­tially the same thing—where the interest is and the liquidity. Some option traders compare call open interest (bulls) to put open interest (bears) in order to measure market sentiment. Others use option volume.market.

It is possible to chart the trends of the three market groups, and to use those trends to spot extremes in their positions. One way to do that is to study the net trader positions published in Futures Charts (Published by Commodity Trend Service, PO Box 32309, Palm Beach Gardens, FL 33420). That charting service plots three lines that show the net trader positions for all three groups on a weekly price chart for each market going back four years. By providing four years of data, historical comparisons are easily done. Nick Van Nice, the publisher of that chart service, looks for situations where the commercials are at one extreme, and the two categories of traders at the other, to find buying and selling opportunities (as shown in Figures 7.13 and 7.14). Even if you don't use the COT Report as a primary input in your trading decisions, it's not a bad idea to keep an eye on what those three groups are doing.

The guiding principle in analyzing the Commitments Report is the belief that the large commercial hedgers are usually right, while the traders are usually wrong. That being the case, the idea is to place yourself in the same positions as the hedgers and in the opposite positions of the two categories of traders. For example, a
bullish signal at a market bottom would occur when the com­mercials are heavily net long while the large and small traders are heavily net short. In a rising market, a warning signal of a possi­ble top would take place when the large and small traders become heavily net long at the same time that the commercials are becoming heavily net short.

Our treatment of open interest would not be complete without mentioning the Commitments of Traders (COT) Report, and how it is used by futures technicians as a forecasting tool. The report is released by the Commodity Futures Trading Commission (CFTC) twice a month—a mid-month report and one at month's end. The report breaks down the open interest numbers into three cate­gories—large hedgers, large speculators, and small traders. The large hedgers, also called commercials, use the futures markets primarily for hedging purposes. Large speculators include the large commodity funds, who rely primarily on mechanical trend-following systems. The final category of small traders includes the general public, who trade in much smaller amounts

One final situation not covered so far that deserves mention is the type of dramatic market action that often takes place at tops and bottoms—blowoffs and selling climaxes. Blowoffs occur at major market tops and selling climaxes at bottoms. In futures, blowoffs are often accompanied by a drop in open interest during the final rally. In the case of a blowoff at market tops, prices suddenly begin to rally sharply after a long advance, accompanied by a large jump in trading activity and then peak abruptly. (See Figure 7.12.) In a selling climax bottom, prices suddenly drop sharply on heavy trading activity and rebound as quickly. (Refer back to Figure 4.22c.)

Let's summarize some of the more important elements of price, volume, and open interest.
1. Volume is used in all markets; open interest mainly in futures.
2. Only the total volume and open interest are used for futures.
3. Increasing volume (and open interest) indicate that the current price trend will probably continue.
4. Declining volume (and open interest) suggest that the price trend may be changing.
5. Volume precedes price. Changes in buying or selling pres­sure are often detected in volume before price.
6. On balance volume (OBV), or some variation thereof, can be used to more accurately measure the direction of vol­ume pressure.
7. Within an uptrend, a sudden leveling off or decline in open interest often warns of a change in trend. (This applies only to futures.)
8. Very high open interest at market tops is dangerous and can intensify downside pressure. (This applies only to futures.)
9. A buildup in open interest during consolidation periods intensifies the ensuing breakout. (This applies only to futures.)Increases in volume (and open interest) help confirm the resolution of price patterns or any other significant chart developments that signal the beginning of a new trend

The rules for interpreting open interest changes are similar to those for volume, but require additional explanation.1. With prices advancing in an uptrend and total open inter­est increasing, new money is flowing into the market reflec2.If, however, prices are rising and open interest declines, the rally is being caused primarily by short covering (holders of los­ing short positions being forced to cover those positions). Money is leaving rather than entering the market. This action is considered bearish because the uptrend will probably run out of steam once the necessary short covering has been completed. (See Figure 7.10.)
3.With prices in a downtrend and open interest rising, the technician knows that new money is flowing into the market, reflecting aggressive new short selling. This action increases the odds that the downtrend will continue and is consid­ered bearish. (See Figure 7.11.)
4. If, however, total open interest is declining along with declining prices, the price decline is being caused by discour‑ting aggressive new buying, and is considered bullish.
aged or losing longs being forced to liquidate their positions.
This action is believed to indicate a strengthening technical situation because the downtrend will probably end once open interest has declined sufficiently to show that most losing longs have completed their selling.
Let's summarize these four points:
1. Rising open interest in an uptrend is bullish.
2. Declining open interest in an uptrend is bearish.
3. Rising open interest in a downtrend is bearish.Declining open interest in a downtrend is bullish.
Other Situations Where Open Interest Is Important
In addition to the preceding tendencies, there are other market situations where a study of open interest can prove useful.
1. Toward the end of major market moves, where open interest has been increasing throughout the price trend, a leveling off or decline in open interest is often an early warning of a change in trend.
2. A high open interest figure at market tops can be considered bearish if the price drop is very sudden. This means that all of the new longs established near the end of the uptrend now have los­ing positions. Their forced liquidation will keep prices under pres­sure until the open interest has declined sufficiently. As an exam‑
ple, let's assume that an uptrend has been in effect for some time. Over the past month, open interest has increased noticeably. Remember that every new open interest contract has one new long and one new short. Suddenly, prices begin to drop sharply and fall below the lowest price set over the past month. Every sin­gle new long established during that month now has a loss.
The forced liquidation of those longs keeps prices under pressure until they have all been liquidated. Worse still, their forced selling often begins to feed on itself and, as prices are pushed even lower, causes additional margin selling by other longs and intensi­fies the new price decline. As a corollary to the preceding point, an unusually high open interest in a bull market is a danger signal.
3. If open interest builds up noticeably during a sideways con­solidation or a horizontal trading range, the ensuing price move inten­sifies once the breakout occurs. This only stands to reason. The mar­ket is in a period of indecision. No one is sure which direction the trend breakout will take. The increase in open interest, however, tells us that a lot of traders are taking positions in anticipation of the breakout. Once that breakout does occur, a lot of traders are going to be caught on the wrong side of the market.
Let's assume we've had a three month trading range and that the open interest has jumped by 10,000 contracts. This means that 10,000 new long positions and 10,000 new short posi­tions have been taken. Prices then break out on the upside and new three month highs are established. Because prices are trading at the highest point in three months, every single short position (all 10,000 of them) initiated during the previous three months now shows a loss. The scramble to cover those losing shorts nat­urally causes additional upside pressure on prices, producing even more panic. Prices remain strong until all or most of those 10,000 short positions have been offset by buying into the market strength. If the breakout had been to the downside, then it would have been the longs doing the scrambling.
The early stage of any new trend immediately following a breakout is usually fueled by forced liquidation by those caught on the wrong side of the market. The more traders caught on the wrong side (manifested in the high open interest), the more severe the response to a sudden adverse market move. On a more
positive note, the new trend is further aided by those on the right side of the market whose judgment has been vindicated, and who are now using accumulated paper profits to finance additional positions. It can be seen why the greater the increase in open interest during a trading range (or any price formation for that matter), the greater the potential for the subsequent price move.
4. Increasing open interest at the completion of a price pattern is viewed as added confirmation of a reliable trend signal. The break­ing of the neckline, for example, of a head and shoulders bottom is more convincing if the breakout occurs on increasing open inter­est along with the heavier volume. The analyst has to be careful here. Because the impetus following the initial trend signal is often caused by those on the wrong side of the market, sometimes the open interest dips slightly at the beginning of a new trend. This ini­tial dip in the open interest can mislead the unwary chart reader, and argues against focusing too much attention on the open interest changes over the very short term.

The level of volume measures the intensity or urgency behind the price move. Heavier volume reflects a higher degree of intensity or pressure. By monitoring the level of volume along with price action, the technician is better able to gauge the buying or selling pressure behind market moves. This information can then be used to confirm price movement or warn that a price move is not to be trusted. (See Figures 7.3 and 7.4.)
To state the rule more concisely, volume should increase or expand in the direction of the existing price trend. In an uptrend, vol­ume should be heavier as the price moves higher, and should decrease or contract on price dips. As long as this pattern contin­ues, volume is said to be confirming the price trend.
The chartist is also watching for signs of divergence (there's that word again). Divergence occurs if the penetration of a previ­ous high by the price trend takes place on declining volume. This action alerts the chartist to diminishing buying pressure. If the volume also shows a tendency to pick up on price dips, the ana­lyst begins to worry that the uptrend is in trouble.
Volume as Confirmation in Price Patterns
During our treatment of price patterns in Chapters 5 and 6, volume was mentioned several times as an important confirming indicator. One of the first signs of a head and shoulders top occurred when prices moved into new highs during the formation of the head on light volume with heavier activity on the subsequent decline to the neckline. The double and triple tops saw lighter volume on each suc­cessive peak followed by heavier downside activity. Continuation
patterns, like the triangle, should be accompanied by a gradual drop off in volume. As a rule, the resolution of all price patterns (the breakout point) should be accompanied by heavier trading activity if the signal given by that breakout is real. (See Figure 7.5.)
In a downtrend, the volume should be heavier during down moves and lighter on bounces. As long as that pattern con­tinues, the selling pressure is greater than buying pressure and the downtrend should continue. It's only when that pattern begins to change that the chartist starts looking for signs of a bottom.
Volume Precedes Price
By monitoring the price and volume together, we're actually using two different tools to measure the same thing—pressure. By.Volume Precedes Price
By monitoring the price and volume together, we're actually using two different tools to measure the same thing—pressure. By..the mere fact that prices are trending higher, we can see that there is more buying than selling pressure. It stands to reason then that the greater volume should take place in the same direction as the prevailing trend. Technicians believe that volume precedes price, meaning that the loss of upside pressure in an uptrend or down­side pressure in a downtrend actually shows up in the volume fig­ures before it is manifested in a reversal of the price trend.
On Balance Volume
Technicians have experimented with many volume indicators to help quantify buying or selling pressure. Trying to "eyeball" the vertical volume bars along the bottom of the chart is not always precise enough to detect significant shifts in the volume flow. The simplest and best known of these volume indicators is on balance volume or OBV. Developed and popularized by Joseph Granville in his 1963 book, Granville's New Key to Stock Market Profits, OBV actually produces a curving line on the price chart. This line can be used either to confirm the quality of the current price trend or warn of an impending reversal by diverging from the price action.
Figure 7.6 shows the price chart with the OBV line along the bottom of the chart instead of the volume bars. Notice how much easier it is to follow the volume trend with the OBV line.
The construction of the OBV line is simplicity itself. The total volume for each day is assigned a plus or minus value depending on whether prices close higher or lower for that day. A higher close causes the volume for that day to be given a plus value, while a lower close counts for negative volume. A running cumulative total is then maintained by adding or subtracting each day's volume based on the direction of the market close.
It is the direction of the OBV line (its trend) that is impor­tant and not the actual numbers themselves. The actual OBV val­ues will differ depending on how far back you are charting. Let the computer handle the calculations. Concentrate on the direc­tion of the OBV line.
The on balance volume line should follow in the same direc­tion as the price trend. If prices show a series of higher peaks and market.
troughs (an uptrend), the OBV line should do the same. If prices are trending lower, so should the OBV line. It's when the volume line fails to move in the same direction as prices that a divergence exists and warns of a possible trend reversal.
Alternatives to OBV
The on balance volume line does its job reasonably well, but it has some shortcomings. For one thing, it assigns an entire day's vol­ume a plus or minus value. Suppose a market closes up on the day by some minimal amount such as one or two tics. Is it reasonable to assign all of that day's activity a positive value? Or consider a situation where the market spends most of the day on the upside, but then closes slightly lower. Should all of that day's volume be
given a negative value? To resolve these questions, technicians have experimented with many variations of OBV in an attempt to discover the true upside and downside volume.
One variation is to give greater weight to those days where the trend is the strongest. On an up day, for example, the volume is multiplied by the amount of the price gain. This technique still assigns positive and negative values, but gives greater weight to those days with greater price movement and reduces the impact of those days where the actual price change is minimal.
There are more sophisticated formulas that blend volume (and open interest) with price action. James Sibbet's Demand Index, for example, combines price and volume into a leading market indicator. The Herrick Payoff Index uses open interest to measure money flow. (See Appendix A for an explanation of both indicators.)
It should be noted that volume reporting in the stock mar­ket is much more useful than in the futures markets. Stock trad­ing volume is reported immediately, while it is reported a day late for futures. Levels of upside and downside volume are also avail­able for stocks, but not in futures. The availability of volume data for stocks on each price change during the day has facilitated an even more advanced indicator called Money Flow, developed by Laszlo Birinyi, Jr. This real-time version of OBV tracks the level of volume on each price change in order to determine if money is flowing into or out of a stock. This sophisticated calculation, how­ever, requires a lot of computer power and isn't readily available to most traders.
These more sophisticated variations of OBV have basical­ly the same intent—to determine whether the heavier volume is taking place on the upside (bullish) or the downside (bearish). Even with its simplicity, the OBV line still does a pretty good job of tracking the volume flow in a market—either in futures or stocks. And OBV is readily available on most charting software. Most charting packages even allow you to plot the OBV line right over the price data for even easier comparison. Other Volume Limitations in Futures
We've already mentioned the problem of the one day lag in reporting futures volume. There is also the relatively awkward practice of using total volume numbers to analyze individual con­tracts instead of each contract's actual volume. There are good reasons for using total volume. But how does one deal with situa­tions when some contracts close higher and others lower in the same futures market on the same day? Limit days produce other problems. Days when markets are locked limit up usually produce very light volume. This is a sign of strength as the numbers of buyers so overwhelm the sellers that prices reach the maximum trading limit and cease trading. According to the traditional rules of interpretation, light volume on a rally is bearish. The light vol­ume on limit days is a violation of that principle and can distort OBV numbers.Even with these limitations, however, volume analysis can still be used in the futures markets, and the technical trader would be well advised to keep a watchful eye on volume indications

Let's begin by placing volume and open interest in their proper perspective. Price is by far the most important. Volume and open interest are secondary in importance and are used primarily as con­firming indicators. Of those two, volume is the more important.
VolumeVolume is the number of entities traded during the time period under study. Because we'll be dealing primarily with daily bar charts, our main concern is with daily volume. That daily volume is plotted by a vertical bar at the bottom of the chart under the day's price action

Volume can be plotted for weekly bar charts as well. In that case, total volume for the week would simply be plotted under the bar representing that week's price action. Volume is usually not used, however, on monthly bar charts.
Open Interest in Futures
The total number of outstanding or unliquidated contracts at the end of the day is open interest. In Figure 7.2, open interest is the solid line plotted on the chart under its corresponding price data for the day, but above the volume bars. Remember that official volume and open interest figures are reported a day late in the futures markets and are, therefore, plotted with a one day lag. (Only estimated volume figures are available for the last trading day.) That means that each day the chartist plots the high, low, and closing price bar for the last day of trading, but plots the offi­cial volume and open interest figures for the previous day.
Open interest represents the total number of outstanding longs or shorts in the market, not the sum of both. Open interest is the number of contracts. A contract must have both a buyer and a seller. Therefore, two market participants—a buyer and a seller—combine to create only one contract. The open interest figure reported each day is followed by either a positive or negative number showing the increase or decrease in the number of con­tracts for that day. It is those changes in the open interest levels, either up or down, that give the chartist clues as to the changing character of market participation and give open interest its fore­casting value.
How Changes in Open Interest Occur. In order to grasp the signifi­cance of how changes in the open interest numbers are interpret­ed, the reader must first understand how each trade produces a change in those numbers.
Every time a trade is completed on the floor of the exchange, the open interest is affected in one of three ways—it increases, decreases, or stays unchanged. Let's see how those changes occur.
Change in
Buyer Seller Open Interest
1. Buys new long Sells new short Increases
2. Buys new long Sells old long No change
3. Buys old short Sells new short No change
4. Buys old short Sells old long Decreases
In the first case, both the buyer and seller are initiating a new position and a new contract is established. In case 2, the buyer is initiating a new long position, but the seller is merely liq­uidating an old long. One is entering and the other exiting a trade. The result is a standoff and no change takes place in the number of contracts. In case 3, the same thing happens except this time it is the seller who is initiating a new short and the buyer who is only covering an old short. Because one of the traders is entering and the other exiting a trade, again no change is pro­duced. In case 4, both traders are liquidating an old position and the open interest decreases accordingly.
To sum up, if both participants in a trade are initiating a new position, the open interest will increase. If both are liquidat­ing an old position, the open interest will decline. If, however, one is initiating a new trade while the other is liquidating an old trade, open interest will remain unchanged. By looking at the net change in the total open interest at the end of the day, the chartist is able to determine whether money is flowing into or out of the market. This information enables the analyst to draw some con­clusions about the strength or weakness of the current price trend.
General Rules for Interpreting Volume and Open InterestThe futures technician incorporates volume and open interest information into market analysis. The rules for the interpretation of volume and open interest are generally combined because they are so similar. There are, however, some distinctions between the two that should be addressed. We'll begin here with a statement of the general rules for both. Having done that, we'll then treat each one separately before combining them again at the end.If volume and open interest are both increasing, then the current price trend will probably continue in its present direction (either up or down). If, however, volume and open interest are declining, the action can be viewed as a warning that the current price trend may be nearing an end. Having said that, let's now take a look at volume and open interest separately..

This concludes our treatment of price patterns. We stated earlier that the three pieces of raw data used by the technical analyst were price, volume, and open interest. Most of what we've said so far has focused on price. Let's take a closer look now at volume and open interest and how they are incorporated into the analytical process.

The principle of confirmation is one of the common themes run­ning throughout the entire subject of market analysis, and is used in conjunction with its counterpart—divergence. We'll introduce both concepts here and explain their meaning, but we'll return to them again and again throughout the book because their impact is so important. We're discussing confirmation here in the context of chart patterns, but it applies to virtually every aspect of technical analysis. Confirmation refers to the comparison of all technical sig­nals and indicators to ensure that most of those indicators are pointing in the same direction and are confirming one another.
Divergence is the opposite of confirmation and refers to a situation where different technical indicators fail to confirm one another. While it is being used here in a negative sense, diver­gence is a valuable concept in market analysis, and one of the best early warning signals of impending trend reversals. We'll discuss the principle of divergence at greater length in Chapter 10, "Oscillators and Contrary Opinion."

In the previous chapter, we treated the head and shoulders pat­tern at some length and described it as the best known and most trustworthy of all reversal patterns. The head and shoulders pat­tern can sometimes appear as a continuation instead of a reversal pattern.
In the continuation head and shoulders variety, prices trace out a pattern that looks very similar to a sideways rectangu­lar pattern except that the middle trough in an uptrend (see Figure 6.11 a) tends to be lower than either of the two shoulders. In a downtrend (see Figure 6.11b), the middle peak in the consol­idation exceeds the other two peaks. The result in both cases is a head and shoulders pattern turned upside down. Because it is turned upside down, there is no chance of confusing it with the reversal pattern.

The measured move, or the swing measurement as it is sometimes called, describes the phenomenon where a major market advance or decline is divided into two equal and parallel moves, as shown in Figure 6.10a. For this approach to work, the market moves should be fairly orderly and well defined. The measured move is really just a variation of some of the techniques we've already touched on. We've seen that some of the consolidation patterns, such as flags and pennants, usually occur at about the halfway point of a market move. We've also mentioned the tendency of markets to retrace about a third to a half of a prior trend before resuming that trend.
In the measured move, when the chartist sees a well-defined situation, such as in Figure 6.10a, with a rally from point A to point B followed by a countertrend swing from point B to point C (which retraces a third to a half of wave AB), it is assumed that the next leg in the uptrend (CD) will come close to duplicating the first leg (AB). The height of wave (AB), therefore, is simply measured upward from the bottom of the correction at point C.

The rectangle formation often goes by other names, but is usually easy to spot on a price chart. It represents a pause in the trend dur­ing which prices move sideways between two parallel horizontal lines. (See Figures 6.9a-c.)The rectangle is sometimes referred to as a trading range or a congestion area. In Dow Theory parlance, it is referred to as a line. Whatever it is called, it usually represents just a consolidation period in the existing trend, and is usually resolved in the direc.tion of the market trend that preceded its occurrence. In terms of forecasting value, it can be viewed as being similar to the sym­metrical triangle but with flat instead of converging trendlines.
A decisive close outside either the upper or lower bound­ary signals completion of the rectangle and points the direction of the trend. The market analyst must always be on the alert, how­ever, that the rectangular consolidation does not turn into a rever­sal pattern. In the uptrend shown in Figure 6.9a, for example, notice that the three peaks might initially be viewed as a possible triple top reversal pattern.
The Importance of the Volume Pattern
One important clue to watch for is the volume pattern. Because the price swings in both directions are fairly broad, the analyst should keep a close eye on which moves have the heavier volume. If the rallies are on heavier and the setbacks on lighter volume, then the formation is probably a continuation in the uptrend. If the heavier volume is on the downside, then it can be considered a warning of a possible trend reversal in the works.
Swings Within the Range Can Be Traded
Some chartists trade the swings within such a pattern by buying dips near the bottom and selling rallies near the top of the range. This technique enables the short term trader to take advantage of the well defined price boundaries, and profit from an otherwise trendless market. Because the positions are being taken at the extremes of the range, the risks are relatively small and well defined. If the trading range remains intact, this countertrend trading approach works quite well. When a breakout does occur, the trader not only exits the last losing trade immediately, but can reverse the previous position by initiating a new trade in the direction of the new trend. Oscillators are especially useful in side­ways trading markets, but less useful once the breakout has occurred for reasons discussed in Chapter 10.
Other traders assume the rectangle is a continuation pat­tern and take long positions near the lower end of the price band
in an uptrend, or initiate short positions near the top of the range in downtrends. Others avoid such trendless markets altogether and await a clearcut breakout before committing their funds. Most trend-following systems perform very poorly during these periods of sideways and trendless market action.
Other Similarities and Differences
In terms of duration, the rectangle usually falls into the one to three month category, similar to triangles and wedges. The vol­ume pattern differs from other continuation patterns in the sense that the broad price swings prevent the usual dropoff in activity seen in other such patterns.
The most common measuring technique applied to the rectangle is based on the height of the price range. Measure the height of the trading range, from top to bottom, and then project that vertical distance from the breakout point. This method is similar to the other vertical measuring techniques already men­tioned, and is based on the volatility of the market. When we cover the count in point and figure charting, we'll say more on the question of horizontal price measurements.Everything mentioned so far concerning volume on break­outs and the probability of return moves applies here as well. Because the upper and lower boundaries are horizontal and so well defined in the rectangle, support and resistance levels are more clearly evident. This means that, on upside breakouts, the top of the former price band should now provide solid support on any selloffs. After a downside breakout in downtrends, the bot­tom of the trading range (the previous support area) should now provide a solid ceiling over the market on any rally attempts

The wedge formation is similar to a symmetrical triangle both in terms of its shape and the amount of time it takes to form. Like the symmetrical triangle, it is identified by two converging trend-lines that come together at an apex. In terms of the amount of time it takes to form, the wedge usually lasts more than one month but not more than three months, putting it into the inter­mediate category.
What distinguishes the wedge is its noticeable slant. The wedge pattern has a noticeable slant either to the upside or the downside. As a rule, like the flag pattern, the wedge slants against the prevailing trend. Therefore, a falling wedge is considered bullish and a rising wedge is bearish. Notice in Figure 6.8a that the bullish wedge slants downward between two converging trendlines. In the downtrend in Figure 6.8b, the converging trendlines have an unmistakable upward slant.Wedges as Tops and Bottom Reversal Patterns
Wedges show up most often within the existing trend and usually constitute continuation patterns. The wedge can appear at tops or bottoms and signal a trend reversal. But that type of situation is much less common. Near the end of an uptrend, the chartist may observe a clearcut rising wedge. Because a continuation wedge in an uptrend should slope downward against the prevailing trend, the rising wedge is a clue to the chartist that this is a bearish and not a bullish pattern. At bottoms, a falling wedge would be a tip-off of a possible end of a bear trend.
Whether the wedge appears in the middle or the end of a market move, the market analyst should always be guided by the general maxim that a rising wedge is bearish and a falling wedge is bullish.

and unusually emotional. Because this pattern also represents an unusual amount of public participation, it most often occurs at major market tops. The expanding pattern, therefore, is usually a bearish formation. It generally appears near the end of a major bull market.
FLAGS AND PENNANTS
The flag and pennant formations are quite common. They are usu­ally treated together because they are very similar in appearance, tend to show up at about the same place in an existing trend, and have the same volume and measuring criteria.
The flag and pennant represent brief pauses in a dynamic market move. In fact, one of the requirements for both the flag
and the pennant is that they be preceded by a sharp and almost straight line move. They represent situations where a steep advance or decline has gotten ahead of itself, and where the mar­ket pauses briefly to "catch its breath" before running off again in the same direction.
Flags and pennants are among the most reliable of contin­uation patterns and only rarely produce a trend reversal. Figures 6.6a-b show what these two patterns look like. To begin with, notice the steep price advance preceding the formations on heavy volume. Notice also the dramatic drop off in activity as the con­solidation patterns form and then the sudden burst of activity on the upside breakout.
Construction of Flags and PennantsThe construction of the two patterns differs slightly. The flag resembles a parallelogram or rectangle marked by two parallel trendlines that tend to slope against the prevailing trend. In a downtrend, the flag would have a slight upward slope.The pennant is identified by two converging trendlines and is more horizontal. It very closely resembles a small symmet­rical triangle. An important requirement is that volume should dry up noticeably while each of the patterns is forming.
Both patterns are relatively short term and should be com­pleted within one to three weeks. Pennants and flags in down­trends tend to take even less time to develop, and often last no longer than one or two weeks. Both patterns are completed on the penetration of the upper trendline in an uptrend. The breaking of the lower trendline would signal resumption of downtrends. The breaking of those trendlines should take place on heavier volume. As usual, upside volume is more critically important than down­side volume. (See Figures 6.7a-b.)
Measuring Implications
The measuring implications are similar for both patterns. Flags and pennants are said to "fly at half-mast" from a flagpole. The flagpole is the prior sharp advance or decline. The term "half-mast" suggests that these minor continuation patterns tend to appear at about the halfway point of the move. In general, the move after the trend has resumed will duplicate the flagpole or the move just prior to the formation of the pattern.
To be more precise, measure the distance of the preceding move from the original breakout point. That is to say, the point at which the original trend signal was given, either by the penetra­tion of a support or resistance level or an important trendline. That vertical distance of the preceding move is then measured from the breakout point of the flag or pennant—that is, the point at which the upper line is broken in an uptrend or the lower line in a downtrend.
SummaryLet's summarize the more important points of both patterns.1. They are both preceded by an almost straight line move (called a flagpole) on heavy volume.
2. Prices then pause for about one to three weeks on very light volume.
3. The trend resumes on a burst of trading activity.
4. Both patterns occur at about the midpoint of the market move.
5. The pennant resembles a small horizontal symmetrical tri­angle.
6. The flag resembles a small parallelogram that slopes against the prevailing trend.
7. Both patterns take less time to develop in downtrends.Both patterns are very common in the financial markets

The descending triangle is just a mirror image of the ascending, and is generally considered a bearish pattern. Notice in Figures 6.4a and b the descending upper line and the flat lower line. This pat­tern indicates that sellers are more aggressive than buyers, and is usually resolved on the downside. The downside signal is regis­tered by a decisive close under the lower trendline, usually on increased volume. A return move sometimes occurs which should encounter resistance at the lower trendline.
The measuring technique is exactly the same as the ascending triangle in the sense that the analyst must measure the height of the pattern at the base to the left and then project that distance down from the breakdown point.
The Descending Triangle as a Top
While the descending triangle is a continuation pattern and usu­ally is found within downtrends, it is not unusual on occasion for the descending triangle to be found at market tops. This type of pattern is not that difficult to recognize when it does appear in the top setting. In that case, a close below the flat lower line would sig­nal a major trend reversal to the downside.The Volume Pattern
The volume pattern in both the ascending and descending trian­gles is very similar in that the volume diminishes as the pattern works itself out and then increases on the breakout. As in the case of the symmetrical triangle, during the formation the chartist can detect subtle shifts in the volume pattern coinciding with the swings in the price action. This means that in the ascending pat­tern, the volume tends to be slightly heavier on bounces and lighter on dips. In the descending formation, volume should be heavier on the downside and lighter during the bounces.
The Time Factor in Triangles
One final factor to be considered on the subject of triangles is that of the time dimension. The triangle is considered an intermediate pattern, meaning that it usually takes longer than a month to form, but generally less than three months. A triangle that lasts less than a month is probably a different pattern, such as a pen­nant, which will be covered shortly. As mentioned earlier, trian­gles sometimes appear on long term price charts, but their basic meaning is always the same.

The ascending and descending triangles are variations of the sym­metrical, but have different forecasting implications. Figures 6.3a and b show examples of an ascending triangle. Notice that the upper trendline is flat, while the lower line is rising. This pattern indicates that buyers are more aggressive than sellers. It is consid­ered a bullish pattern and is usually resolved with a breakout to the upside.
Both the ascending and descending triangles differ from the symmetrical in a very important sense. No matter where in the trend structure the ascending or descending triangles appear, they have very definite forecasting implications. The ascending triangle is bullish and the descending triangle is bearish. The sym­metrical triangle, by contrast, is inherently a neutral pattern. This does not mean, however, that the symmetrical triangle does not have forecasting value. On the contrary, because the symmetrical .triangle is a continuation pattern, the analyst must simply look to see the direction of the previous trend and then make the assump­tion that the previous trend will continue.
Let's get back to the ascending triangle. As already stated, more often than not, the ascending triangle is bullish. The bullish breakout is signaled by a decisive closing above the flat upper trendline. As in the case of all valid upside breakouts, volume should see a noticeable increase on the breakout. A return move back to the support line (the flat upper line) is not unusual and should take place on light volume.
Measuring Technique
The measuring technique for the ascending triangle is relatively simple. Simply measure the height of the pattern at its widest
point and project that vertical distance from the breakout point. This is just another example of using the volatility of a price pat­tern to determine a minimum price objective.The Ascending Triangle as a BottomWhile the ascending triangle most often appears in an uptrend and is considered a continuation pattern, it sometimes appears as a bottoming pattern. It is not unusual toward the end of a down­trend to see an ascending triangle develop. However, even in this situation, the interpretation of the pattern is bullish. The break­ing of the upper line signals completion of the base and is con­sidered a bullish signal. Both the ascending and descending trian­gles are sometimes also referred to as right angle triangles

The symmetrical triangle (or the coil) is usually a continuation pattern. It represents a pause in the existing trend after which the original trend is resumed. In the example in Figure 6.1a, the prior trend was up, so that the percentages favor resolution of the triangular consolidation on the upside. If the trend had been down, then the symmetrical triangle would have bearish implications.
The minimum requirement for a triangle is four reversal points. Remember that it always takes two points to draw a trendline. Therefore, in order to draw two converging trendlines, each line must be touched at least twice. In Figure 6.1a, the tri­angle actually begins at point 1, which is where the consolida­tion in the uptrend begins. Prices pull back to point 2 and then rally to point 3. Point 3, however, is lower than point 1. The upper trendline can only be drawn once prices have declined from point 3.
Notice that point 4 is higher than point 2. Only when prices have rallied from point 4 can the lower upslanting line be drawn. It is at this point that the analyst begins to suspect the he or she is dealing with the symmetrical triangle. Now there are four reversal points (1, 2, 3, and 4) and two converging trendlines.
While the minimum requirement is four reversal points, many triangles have six reversal points as shown in Figure 6.1a.This means that there are actually three peaks and three troughs that combine to form five waves within the triangle before the uptrend resumes. (When we get to the Elliott Wave Theory, we'll have more to say about the five wave tendency within triangles.)
Time Limit for Triangle Resolution
There is a time limit for the resolution of the pattern, and that is the point where the two lines meet—at the apex. As a general rule, prices should break out in the direction of the prior trend some­where between two-thirds to three-quarters of the horizontal width of the triangle. That is, the distance from the vertical base on the left of the pattern to the apex at the far right. Because the two lines must meet at some point, that time distance can be mea­sured once the two converging lines are drawn. An upside break­out is signaled by a penetration of the upper trendline. If prices remain within the triangle beyond the three-quarters point, the triangle begins to lose its potency, and usually means that prices will continue to drift out to the apex and beyond
The triangle, therefore, provides an interesting combina­tion of price and time. The converging trendlines give the price boundaries of the pattern, and indicate at what point the pattern has been completed and the trend resumed by the penetration of the upper trendline (in the case of an uptrend). But these trend-lines also provide a time target by measuring the width of the pat­tern. If the width, for example, were 20 weeks long, then the break­out should take place sometime between the 13th and the 15th week. (See Figure 6.1d.)
The actual trend signal is given by a closing penetration of one of the trendlines. Sometimes a return move will occur back to the penetrated trendline after the breakout. In an uptrend, that line has become a support line. In a downtrend, the lower line becomes a resistance line once it's broken. The apex also acts as an important support or resistance level after the breakout occurs. Various penetration criteria can be applied to the breakout, simi­lar to those covered in the previous two chapters. A minimum penetration criterion would be a closing price outside the trend-line and not just an intraday penetration.
Importance of Volume
Volume should diminish as the price swings narrow within the triangle. This tendency for volume to contract is true of all con­solidation patterns. But the volume should pick up noticeably at the penetration of the trendline that completes the pattern. The return move should be on light volume with heavier activity again as the trend resumes.
Two other points should be mentioned about volume. Asis the case with reversal patterns, volume is more important onthe upside than on the downside. An increase in volume is essen‑tial to the resumption of an uptrend in all consolidation patterns.The second point about volume is that, even thoughtrading activity diminishes during formation of the pattern, aclose inspection of the volume usually gives a clue as to whether
the heavier volume is occurring during the upmoves or down-moves. In an uptrend, for example, there should be a slight ten­dency for volume to be heavier during the bounces and lighter on the price dips.
Measuring Technique
Triangles have measuring techniques. In the case of the sym­metrical triangle, there are a couple of techniques generally used. The simplest technique is to measure the height of the ver­tical line at the widest part of the triangle (the base) and mea­sure that distance from the breakout point. Figure 6.2 shows the distance projected from the breakout point, which is the tech­nique I prefer.
The second method is to draw a trendline from the top of the base (at point A) parallel to the lower trendline. This upper channel line then becomes the upside target in an uptrend. It is possible to arrive at a rough time target for prices to meet the upper channel line. Prices will sometimes hit the channel line at the same time the two converging lines meet at the apex.