Pattern recognition techniques are often subjective (this includes candlestick techniques). Oscillators are mathematically derived techniques which offer a more objective means of analyzing the market. They are widely used and are the basis of many computerized trading systems.
Oscillators include such technical tools as the relative strength index, stochastics, and momentum. Oscillators can
serve traders in at least three ways:
- Oscillators can be used as divergence indicators (that is, when the price and the oscillator move in different directions). They can warn that the current price trend may be stalling. There are two kinds of divergence. A negative, or bearish, divergence occurs when prices are at a new high, but the oscillator is not. This implies the market is internally weak. A positive, or bullish, divergence is when prices are at a new low but the oscillator does not hit a new low. The implications are that the selling pressure is losing steam.
- As overbought/oversold indicators, oscillators can notify the trader if the market has become overextended and, thus, vulnerable to a correction. Using an oscillator as an overbought/oversold indicator requires caution. Because of how they are constructed, oscillators are mainly applied in lateral price environments. Using an oscillator as an overbought/oversold indicator when a new major trend is about to commence can cause problems. If, for example, there is a break above the top of a congestion band, it could indicate the start of a new bull leg and the oscillator could stay overbought while prices ascend.
- Oscillators can confirm the force behind a trend’s move by measuring the market’s momentum. Momentum measures the velocity of a price move by comparing price changes. In theory, the velocity should increase as the trend is underway. A flattening of momentum could be an early warning that a price move may be decelerating.
Use oscillator signals to place a position in the direction of the dominant trend. Thus, a bullish oscillator indication should be used to buy, if the major trend is up, and to cover shorts, if the major trend is down. The same idea applies to a sell signal vis-a-vis an oscillator. Do not short on a bearish oscillator signal unless the prevailing trend is heading
south. If it is not, a bearish oscillator signal should be used to liquidate longs.
THE RELATIVE STRENGTH INDEX
The Relative Strength Index (RSI)’ is one of the most popular technical tools used by futures traders. Many charting services plot the RSI and many traders closely monitor it. The RSI compares the relative strength of price advances to price declines over a specified period. Nine and 14 days are some of the most popular periods used.
How to Compute the RSI
How to Use RSI
The two main uses of RSI are as an overbought/oversold indicator and as a tool to monitor divergences.
As an overbought/oversold indicator, the RSI implies that the market is overbought if it approaches the upper end of this band (that is, above 70 or 80). At that point, the market may be vulnerable to a pullback or could move into a period of consolidation. Conversely, at the lower end of the RSI range (usually below 30 or 20), it is said to reflect an oversold condition. In such an environment, there is a potential of a short covering move.
As a divergence tool, RSI calculations can be helpful when prices make a new high for the move and the RSI fails to make a concurrent high. This is called a negative divergence and is potentially bearish. A positive divergence occurs when prices make a new low, but the RSI does not. Divergences are more meaningful when RSI oscillator readings are in overbought or oversold regions.
Exhibit 14.1 displays both a bullish positive and a bearish negative RSI divergence which helped these candlestick readings. At the time of the January 24 price low, the RSI was 28%. On January 31, a new price low for the move occurred. The RSI then was 39%. This was noticeably higher than the 28% RSI value of January 24. New price lows and a higher RSI level created a bullish positive divergence. Besides the positive divergence, the white line of January 31 engulfed the prior black candlestick. This built a bullish engulfing pattern.
A doji star arose on March 14. The next session created a candlestick similar to a hanging man. (The lower shadow was not long enough for it to be a classic hanging man, though.) At the time of these potentially bearish candlestick indicators, the RSI was also sending out a warning alert. Specifically, the new price peaks of March 15 and 16 were mirrored by lower RSI readings. This is bearish negative divergence. The market made another price surge on March 21, and although this was a new price high, RSI levels continued to decline. The result was a pullback to the March support area of $1.11.
In Exhibit 14.2 the decline that began with the bearish engulfing pattern stopped at the piercing pattern. The constructive outlook implied by this piercing pattern was reinforced by the positive divergence of the RSI. Some technicians also use trendlines with RSI. In this case, the RSI uptrend support line held in spite of new price lows on March 29.
Exhibit 14.2 illustrates another reason to use candlesticks as a compliment to the RSI. Candlesticks may give a bullish or bearish signal before the additional confirmation sometimes needed by the RSI. Specifically, some technicians will view the RSI as giving a bullish signal if two steps occur. The first is the aforementioned positive divergence. The next is that the RSI has to move above its prior high. In this example, that would mean a move above the April 20 RSI level (A). Based on this procedure the bullish signal would have been given at point B. However, by joining the bullish candlestick indication (the engulfing line) with the RSI’s positive divergence, the bullish signal would have been apparent a few days earlier.
Doji are a warning signal during uptrends. But, like all technical clues, they can sometimes mislead you. One way to filter out the misleading clues is to add other technical tools. Exhibit 14.3 illustrates the use of the RSI as a tool of confirmation. A bearish shooting star and a set of doji lines appeared in the middle of May (time frame A). These sig-naled the end of the prior uptrend, at least temporarily, as the market moved into a lateral range for the next few weeks. After this respite, a rally pushed prices to new highs at time frame B. These highs were 100 points above where they were at time frame A. Yet, at time frame B, the RSI was where it was at A. This reflected a flagging of the markets internal strength. The harami at B sounded more warning sirens.
After a 100-point setback from the highs at time frame B, another rally ensued. This rally touched the 3000 level at time frame C. These new highs were sharply above prices at time frame B but the RSI was noticeably less. This bearish divergence at time frame C accompanied with the shooting star, the doji, and the hanging man indicated the internally weak structure of the market, even though prices touched new highs.
The stochastic oscillator is another popular tool used by futures technicians. As an oscillator, it provides overbought and oversold readings, signals divergences, and affords a mechanism to compare a shorter-term trend to a longer-term trend. The stochastic indicator compares the latest closing price with the total range of price action for a specified
period. Stochastic values are between 0 and 100. A high-stochastic reading would mean the close is near the upper end of the entire range for the period. A low reading means that the close is near the low end of the period’s range. The idea behind stochastics is that, as the market moves higher, closes tend to be near the highs of the range or, as the market moves lower, prices tend to cluster near the lows of the range.
The “100” in the equation converts the value into a percentage. Thus, if the close today is the same as the high for the period under observation, the fast %K would be 100%. A period can be in days, weeks, or even intra-day (such as hourly). Nineteen, fourteen, and twenty-one periods are some of the more common periods.
Because the fast %K line can be so volatile, this line is usually smoothed by taking a moving average of the last three %K values. This three-period moving average of %K is called the slow %K. .Most technicians use the slow %K line instead of the choppy fast %K line. This slow %K is then smoothed again using a three-day moving average of the slow %K to get what is called the %D line. This %D is essentially a moving average of a moving average. One way to think of the difference between the %K and %D lines is too view them as you would two moving averages with the %K line comparable to a short-term moving average and the %D line comparable to a longer-term moving average.
How to Use Stochastics
As mentioned previously, stochastics can be used a few ways. The most popular method is to view it as a tool for showing divergence. Most technicians who monitor stochastics use this aspect of divergence in conjunction with overbought/oversold readings.
Some technicians require another rule. That rule is to have the slow %K line cross under the %D line for a sell signal, or for the slow %K to move above the %D for a busy signal. This is comparable to the bullish (bearish) signal of a faster moving average crossing over (under) the slower moving average. For instance, to get a buy signal, the market must be oversold (25% or less for %D), there is a positive divergence and, the %K line is crossing above the %D line.
Looking at Exhibit 14.4, the doji session of January 3 should give you pause. A doji following a long white candlestick “ain’t pretty.” The doji session made new price highs as they pushed above the December highs. But stochastics did not echo these price highs with concomitant highs, so this was bearish negative divergence. It was an important affirmation of the bearish signal sent on the doji day.
Besides the divergence, some technicians look for crossovers of the %K and %D lines. See Exhibit 14.5. In mid-1989, copper based out via a hammer. Another series of hammer lines materialized in early 1990. Was this a sign of another base? The answer was more than likely yes because of what the stochastic evidence told us. Hammer B made new lows as it broke under the lows from hammer A. Yet, at hammer B there was a higher stochastic reading than at hammer A. This was a positive divergence. The implications were an abating of selling pressure.
There was also a positive crossover as the more volatile, solid %K line crossed above the dotted %D line (see arrow). This crossover is considered more significant if it is from oversold readings (that is, under 25%). That is what occurred here. So, in early 1990, there were a series of hammer like lines and a positive divergence with a positive crossover during an oversold market. A confluence of technical indicators that were strong clues that the prior downtrend had ended.
As illustrated by Exhibit 14.6, April 12 and 16 formed a dark-cloud cover. The black candlestick session on April 16 pushed prices above the former highs in March. Thus prices were at a new high, but stochastics were not. The dark-cloud cover and the negative divergence were two signs to be circumspect about further rallies. The next downleg was corroborated by a negative crossover when the faster %K line crossed under the slower %D line (as shown by the arrow).
I do not often use candlesticks with British Pound futures because, as can be seen in Exhibit 14.7, many sessions are small real bodies or doji. This is in addition to the frequent gaps induced by overnight trading (this is also true of other currency futures). Nonetheless, at times, there are candlestick signals that bear watching especially when confirmed
with other indicator2.
During the week of March 19, a variation of a morning star arose. Normally, it is best if the third white candlestick of the morning star pattern pushes well into the first session’s black body. This white candlestick did not. Before deciding how much importance to place on a variation of a pattern, scan the other technical evidence. At the time of the morning star (or actually its variation), prices touched new price lows; stochastics did not. This was a bullish positive divergence which was soon confirmed when the %K line crossed over the %D line. Consequently, the fact that the morning star pattern was less than ideal should have only given you temporary pause about calling for a bottom. Stochastic indicators provided plenty of added proof to this outlook.
Momentum, also called price velocity, is a measurement of the difference between the closing price today and the closing price a specified number of days ago. If we use a ten-day momentum we compare today’s close to that of ten days ago. If today’s close is higher, the momentum is a positive number on the momentum scale. If today’s close is lower than that of ten days ago, the momentum is a minus figure. Using the momentum index, price differences (the difference between today’s close and that of whatever period you pick) should rise at an increasing rate as a trend progresses. This displays an uptrend with increasingly greater momentum. In other words, the velocity of the price changes is increasing. If prices are rising and momentum begins to flatten, a decelerating price trend is in effect. This could be an early warning that a prior price trend could end. If the momentum crosses under the 0 line, it could be
construed as a bearish sign, above the 0 line, as bullish.
Momentum is also handy as an overbought/oversold indicator. For instance, when the momentum index is at a relatively large positive value the market may be overbought and vulnerable to a price pullback. Momentum usually hits its peak before prices. Based on this, a very overbought momentum oscillator could be presaging a price peak.
In Exhibit 14.8, the long-legged doji in January was a warning for the longs to be careful. Further reason for caution was that prices produced new highs on this doji session, yet the momentum was noticeably lower than at the prior high in late November (A). More proof that a downtrend could start was apparent when the momentum fell under 0 in early February.
Another use of momentum is to provide a yardstick for overbought or oversold levels (see Exhibit 14.9). In this heating oil chart, observe how an oscillator reading of around +200 (that is, the current close is $.02 above the close ten days ago) represents an overbought environment. An oversold state exists for this market when the momentum oscillator is -400 points or $.04 under the close of ten days prior. At the 200-point overbought level, continuation of the prior rally is unlikely and the market should either trade sideways or backoff. The odds of a top reversal with an overbought momentum reading are increased if there is bearish candlestick confirmation. In this regard, in February an overbought
momentum level is coupled with an evening star and then a harami cross. Another overbought oscillator in early April joined another evening star pattern. Hammers A and B accompanied the oversold momentum levels in March and April. At these points, further selloffs were unlikely and either sideways action or rallies could unfold to relieve the oversold nature of the market.