Technical Analysis: Moving Averages, Bollinger Bands

Moving Averages

The most basic quantitative indicator is the moving average (MA). Just like it sounds, an MA shows how the average price of a security (or in this case, a currency pair) changes over time. As a technical tool, it is useful for a few key reasons. First of all, it smooths price data. Because it is an average calculation, significant fluctuations in the underlying exchange rate generate smaller fluctuations in the moving average. By eliminating noise, an MA may provide a clearer picture of a trend than a plain price chart. Secondly, MAs can guide position entry and exit. When compared to the underlying currency pair (or to other MAs) it may confirm the start of a bullish or bearish trend. Of course, it’s important to understand that an MA is intrinsically a following indicator. That means that it will only generate trading signals after potential trends have already begun to take shape.

There are a handful of different types of MAs. While conceptually the same, they are calculated using slightly different methods in order to satisfy different objectives. The simple moving average (SMA) is an arithmetic average of prices. Since all prices in the series are given equal weight, an SMA line is usually the smoothest type of MA. On the other hand, since old prices are treated the same as new prices, SMAs take longer to register sudden changes in underlying prices. Indeed, you can see from Figure 6-11 that it takes longer for the SMA (represented by the dashed line) to reflect the start of both the uptrend and the downtrend in the EUR/USD.

Figure 6-11. Comparison of SMA, EMA, and SMMA

In contrast, a weighted moving average (WMA) assigns the greatest importance to the most recent price and incrementally decreases the weight to every point thereafter, such that the oldest prices receive the least weight. An exponential moving average (EMA) takes a similar approach, but weightings decrease exponentially rather than in even increments. EMAs register shifts in trends almost immediately. This hypersensitivity can be a strength—because it enables traders to profit from trends at their inception—but also a weakness, in the form of false signals. Finally, smoothed moving averages (SMMAs) aim to further eliminate noise (aberrant price spikes) so that only the raw trend remains. Of the three main types of MAs, SMMAs are the smoothest but are also the slowest at registering trends.

Most charting software is programmed to calculate MAs based on closing prices, though some can be rejiggered to incorporate high/low price data as well. The only parameter that these programs will ask traders to supply is the duration/number of prices. Including more points will result in a flatter MA. This is immediately apparent in Figure 6-12, which shows 5-day, 20-day, and 60-day MAs for the same underlying currency pair.

Figure 6-12. Altering the number of prices affects the moving average’s appearance

MAs can be calculated for any interval of time. Changing the chart’s time unit (from five minutes to one day, for example) will naturally produce a different MA. As I explained in the previous section, you should focus on a length of time that is consistent with your trading time horizon. Even if it produces stronger signals, an MA based on five-minute data will not really help you if you are planning to hold positions for a month. As for the ideal number of price points that should be included in the moving average, there is no right answer. Some forex gurus swear by the set of 4, 9, and 18. Others prefer 7, 21, and 90. What’s most important is that when looking at multiple MAs, the different periods should be spaced out so that they can produce clear signals.

In fact, the best way to utilize the MA as a trading tool is to look at multiple time periods simultaneously. Figure 6-12 shows how the EUR/USD was range bound for several months before it dropped precipitously. If I had developed a rule to sell whenever the short-term MA (5 days) dips below the long-term MA (60 days), I would have received an excellent signal just as the EUR/USD had begun to drop. On the other hand, this rule also produced two false signals and would have basically prevented me from capturing any part of the massive 1000 PIP upside correction that followed! While it might be possible to tweak the number of days in each MA to improve robustness, this example shows that there is no such thing as a surefire technical trading rule.

Moving Average Envelopes and Bollinger Bands

There are a handful of other technical indicators that are derived from the MA. The MA envelope, for example, is grounded in the idea that MAs can be used to identify points of support and resistance. The theory is that asset prices will never stray too far from a trend, designated in this case by the MA itself. When a currency pair rises too far above or falls too far below its MA, it could be an indication that a reversal is imminent. In addition, when a pair completely breaches the walls of the envelope, it could signal a breakout.

To plot an MA envelope, the first step is to plot the MA itself. In Figure 6-13, I used a 10-day SMA. Next, select the percentage above and percentage below the MA that will form your envelope. The exact percentage will depend on your investing horizon and the volatility in the currency pair that you are observing and will most likely be arrived at through trial and error. (There are no golden numbers for MA envelopes that apply universally to all currencies.)

Of course, you need to select percentages that are meaningful. If the pair bumps up against the envelope too frequently, you will receive false signals. If the envelope is too wide, however, the currency pair will never breach it, and you won’t receive any signals at all. I solved this problem by plotting two envelopes in Figure 6-13. The two dashed lines are respectively 1.5% above and below the 10-day moving average, which is represented by the center most solid line, while the solid outer lines delimit a 2% envelope.

Figure 6-13. AUD / USD with 10-day MA envelopes

Based on this configuration, the MA envelopes produce an abundance of sell signals. For the first two months, when the pair is range bound, the AUD/USD moves through the 1.5% resistance but bounces off the 2% resistance. For whatever reason, the lower envelope doesn’t provide such strong support. Then the pair completely smashes through this support on two separate occasions, providing two good opportunities to sell. On the way back up, it smashes through the resistance the first time around but bumps up against it the second time.

Bollinger bands take the idea of the MA envelope one step further. Since the upper and lower bounds of an MA envelope are fixed in percentage terms, the width never changes. Bollinger bands, in contrast, narrow and expand in synch with actual market conditions. That’s because they are configured as a function of volatility. When a pair is range bound, the Bollinger bands form a tight envelope around it. When a sudden upside or downside move takes place, the Bollinger bands widen proportionately.

This unique characteristic of Bollinger bands is reflected in Figure 6-14, which depicts the same AUD/USD pair as in Figure 6-13. When the pair is range bound, the Bollinger bands provide the same support and resistance as the MA envelopes in Figure 6-13. In Figure 6-14, however, the breakouts are accompanied by a widening of the band (due to an increase in volatility), underscoring the momentum that has coalesced around the new trend. Bollinger bands, then, are especially useful for forecasting breakouts. In general, the steeper the expansion of the band, the stronger the trend.

Bollinger bands can be adjusted just like MA envelopes. Instead of keying in a percentage, however, you need to select the number of standard deviations (also known as volatility) away from the mean that will form the upper and lower bounds of the band. As with MA envelopes, establishing a golden number may take some trial and error.

Despite their overall effectiveness, MA envelopes and Bollinger bands are not without weaknesses. Namely, they are basically useless when trends change suddenly. You may have noticed that the breakouts depicted in Figure 6-13 and Figure 6-14 actually started to take place before they were picked up by the Bollinger bands. On the one hand, the trend reversal that I have pointed out in Figure 6-14 below caused the MA to turn upward almost immediately.

Figure 6-14. Bollinger bands of 1.5, 2, and 2.5 standard deviations

At the same time, it wasn’t until the pair had exhausted most of its upward momentum that it finally crashed through the upward bound of the dashed Bollinger band. By this point, most of the profit potential had disappeared.

Parabolic Stop and Reverse

The Parabolic Stop and Reverse (Parabolic SAR) is one of the easiest technical tools to understand and interpret, but it is also among the least reliable. It is based on the notion that once trends form, they need to build momentum rapidly. Otherwise, investors will lose interest, and the trend will peter out just as quickly as it started. The indicator uses a complex formula to predict the beginnings and ends of trends, both of which are represented by a series of dots that appear directly on the price chart. Buy at the beginning of an uptrend, where the dots are below the actual price and rising, and sell when the Parabolic SAR switches to downtrend, where the dots are above the price point and falling. To build in a margin of error, perhaps you might consider waiting until the trend has reached three periods (three dots) before acting.

If only it were that simple. In Figure 6-15, the Parabolic SAR has identified seven discrete trends, each of which is separated by vertical lines. As can be seen by the actual and predicted trends (represented by the grey and black lines that I painted on for illustrative purposes), the Parabolic SAR was ultimately more often wrong than it was right!

Figure 6-15. Example of the Parabolic SAR

Oscillators

Oscillators represent a distinct category of technical indicators—one that is more complex and potentially more profound. Oscillators work by normalizing asset price data to a scale (of 0 to 100, for example) so that overbought and oversold conditions can easily be identified. Most charting software will enable you to view multiple oscillators simultaneously by plotting individual oscillators below the main price chart.

There are a handful of ways in which readings from oscillators can be interpreted and utilized. First, when a value reaches an extreme level, it is supposed to indicate that investor sentiment has also reached an extreme level and that a correction is imminent. Second, when an oscillator crosses over from positive territory into negative territory (and vice versa), it suggests that a trend is about to reverse. Finally, divergences between oscillators and the underlying currency pairs may signal that a trend is about to come to an end. There may also be additional interpretations, but these three are most common.

As for which approach is best, it depends not only on the specific oscillator, the asset in question (in this case forex), and prevailing market conditions, but also on the person that is performing the analysis. Some experts harp on crossovers, while others insist that divergences provide the best signals. Still others may promote the use of oscillators in combination with other indicators. As with other technical analysis indicators, there is no singular or correct way to incorporate oscillators into one’s trading strategy.

With all of this in mind, let’s look at a few of the most popular oscillators.

Moving Average Convergence Divergence

The Moving Average Convergence Divergence (MACD), a popular leading oscillator, represents the difference between two exponential moving averages (EMAs) of different durations. The theory is that when the short-term MA suddenly crosses a long term MA, their intersection could signal the start of a trend. In order to enhance the MACD’s signaling power, it is plotted against an MA of itself in the form of a histogram.

Before your head starts to spin, let’s look at a concrete example. You can see from the USD/CAD chart in Figure 6-16 that when the 12-day EMA crosses below the 26-day EMA, the MACD line plotted below similarly moves into negative territory. By itself, this could be interpreted as a signal to sell. When the MACD crosses below the 9-day MA of itself, the bar chart also undergoes a crossover from positive to negative territory, and this produces yet another sell signal. All of these signals are indicated by vertical dotted lines.

There are several additional observations that can be made about the MACD. First of all, the default settings are 12, 26, and 9 days (for the short-term EMA, long-term EMA, and MACD MA, respectively). You can easily change these parameters using charting software, which will obviously produce slightly different signals. Secondly, the MACD is prone to false signals since it may hover at an extreme level (or move back and forth between positive and negative territory) for many successive periods. Sure enough, the USD/CAD continues rising shortly after the second sell signal in Figure 6-16. Third, the imminent reversal signaled by the MACD may not take place for several periods after sentiment reaches an extreme level, exposing traders to risk in the interim. Any trader that put in a sell order for the USD/CAD following the final sell signal in Figure 6-16 may very well have experienced losses before the CAD/USD began to trend downward.

Figure 6-16. MACD in practice

Stochastic Oscillator

A stochastic oscillator uses changes in momentum as a basis for predicting changes in price. Specifically, it seeks to establish where the current price of an asset stands relative to its range over a recent period of time. This calculation (%K) is then compared to a moving average of itself (%D). The chartist must select both the number of periods (typically 14 or 20) for the %K calculation and the number of periods (typically 3 or 5) for the %D calculation. This is known as a fast stochastic and is shown in the middle panel in Figure 6-17. Those that are not satisfied with the fast stochastic and/or have longer trading horizons can utilize a slow stochastic, which simply uses the fast stochastic as its starting point (also known as its %K) and performs yet another moving average. This slow stochastic is typically smoother and should generate fewer false signals. It is displayed in the bottom panel in Figure 6-17.

Figure 6-17. Using fast and slow stochastics to identify buy and sell opportunities

Stochastic oscillators fluctuate between 0 and 100, and most technicians use the thresholds of 20% and 80%, respectively, as basis for identifying oversold and overbought conditions. Since a currency pair may remain at an extreme level for quite some time as a result of sustained buying or selling pressure, it makes sense to wait until the stochastic has reversed—when the %D crosses the %K—before acting. It’s also worth looking at where the stochastic currently reads relative to the halfway point. Below 50% implies that the pair is trading in the bottom half of its recent range and suggests bearishness. The opposite is true for readings above 50%.

Consistent with Figure 6-17, fast stochastics typically produce more signals (and more noise) than slow stochastics. That being said, each time the fast stochastic rose above 80 and then contracted sharply, the EUR/JPY followed suit. The dozen or so declines in the fast stochastic to below 20 meanwhile seem to coincide nicely with declines in the EUR/JPY While the slow stochastic produces even fewer signals—potentially causing its followers to miss good trading opportunities—most of these signals are quite accurate. In short, you should understand that increasing the number of periods should produce clearer but fewer signals.

Relative Strength Index and Commodity Channel Index

The Relative Strength Index (RSI) is similar to the stochastic oscillator. The RSI formula normalizes changes in momentum to an index from 0 to 100, where readings above 70 and below 30 represent overbought and oversold conditions, respectively. Whereas a stochastic indicator compares the most recent closing price with a trading range for a given number of periods, the RSI merely examines only upward and downward movements in price for a given period.

If the number of upward price movements exceeds the number of downward price movements, the RSI will increase. In theory, when the RSI crosses one of its twin thresholds (70 and 30, typically), it signifies that momentum has reached an extreme level and a reversal may be imminent. In addition, the appearance of resistance (support) in an RSI when the underlying currency is sharply rising (falling) implies difficulty sustaining upward (downward) momentum and could similarly herald a correction.

The commodity channel index (CCI) also measures momentum, but it does so by comparing the current price level to a simple moving average of itself over a given number of periods. Fluctuations between -100 and +100 are considered normal while movements outside of that band hint toward a reversal. For the best signal, it’s advisable to wait until the CCI crosses back through one of the boundary lines before acting.

The only variable traders need to input into their charting software to compute either an RSI or CCI is the number of periods. Naturally, a lower number will generate more sensitive readings. In Figure 6-18, I used 14 periods and 20 periods, respectively, for the RSI and CCI, which are the default settings in most charting software.

Figure 6-18. Downward reversals in the USD/CHF take place when the RSI crosses 70 and the CCI falls back below 100

As you see from the chart above, the accuracy and simplicity of the RSI are impressive, which explains why it is one of the most popular technical indicators. The two biggest reversals in the USD/CHF coincide with an RSI peak of slightly above 70 (indicated by the vertical black lines). Figure 6-18 also illustrates how using two indicators together can produce especially robust signals. Since the CCI rises above 100 and falls below -100 on several occasions (as indicated by the vertical gray lines), it helps to have another indicator with which to compare it. When used together, the RSI and CCI yield two very strong sell signals, both of which are followed by retracements in the USD/CHF. In fact, these two indicators are beeping loudly at the present (rightmost end of the chart), suggesting that another correction might come soon!

Awesome Oscillator

There are actually hundreds of different technical indicators and an infinite number of iterations and ways to combine them. In fact, when researching this book, I came across a handful that I had never even heard of before, and anyone with an imagination and a basic programming ability could create a new one. How about the Kritzer Index?

In fact, the awesome oscillator may very well have been invented by a technical analyst with too much time on his hands. It compares the 34-period MA with the 5-period MA, and the result is a histogram that moves back and forth across a 0-line. When the oscillator crosses firmly through this line, it generates a buy signal. The inventor of the awesome oscillator has also suggested that two peaks (or troughs) might also provide a strong signal.

Unfortunately, based on the way the awesome oscillator is constructed, it inherently provides concurrent (rather than advance) signals. In other words, when the 5-day MA crosses the 34-day MA, it may already be too late to buy. This is clearly evident in Figure 6-19; the strongest buying signal doesn’t come until after the massive 10% correction has already taken place.

Figure 6-19. Awesome oscillator produces signals that lag actual price movements

Summary

As you may have sensed, this overview represents only the tip of the technical analysis iceberg. The indicators that I selected for inclusion in this book are those that I believe are most compatible with trend trading and fundamental analysis. Consider that entire books have been written not only about technical analysis but also about specific aspects of technical analysis. For those of you that plan to approach trading from technical perspective, I would certainly recommend delving deeper into the subject on your own.

I have tried to present technical analysis in a way that is straightforward and intuitive. While it’s not necessary to memorize the formulas for calculating the various indicators in your technical arsenal, it nonetheless is important to understand how they are calculated. If you can avoid taking the indicators at face value, you will be rewarded with a fuller understanding of what you are seeing in their readouts.

In concluding this chapter, I would like to offer a couple of caveats regarding technical analysis. First, charts and technical indicators often produce unclear or conflicting signals. For that reason, it’s worth using a couple of indicators together in order to optimize their effectiveness. Recall that in Figure 6-18, the RSI and CCI produced incredibly accurate signals when used together. At the same time, don’t get carried away and try to develop technical trading rules that are based on too many indicators. Figure 6-20 takes this idea to a comical extreme. The chart is so cluttered that it’s hardly even possible to see the underlying movements in the EUR/USD, let alone to make a reasonable interpretation and open a position!

Figure 6-20. Extreme example of a chart with too many indicators

Second, consider that the flexibility of technical analysis is a potential pitfall as much as it is a benefit. While it might seem convenient that technical analysis can theoretically be applied to all asset prices at all times, this could lead to arbitrariness and laziness. In other words, it’s important to tweak the parameters of individual indicators and to experiment with different combinations of indicators until you find one that seems to fit the particular currency pair at a particular time, as well as your particular strategy.

Finally, technical analysis is far from fool proof. To be sure, it’s very easy to find examples of currency behavior that accord perfectly with the signals produced by technical indicators. However, there are just as many counterexamples. That’s because technical indicators are not really designed to predict the future; all they can do is reorient the way that we see the past. They can convert seemingly random currency movements into smooth lines and indexes that are easy to interpret so that you might have a better idea of what is apparently happening in the present. As for what will happen next, well, that is a different story altogether.