WHAT WORKS, WHAT DOESN'T, AND WHY
Overbought (OB) and Oversold (OS) are among the least understood market conditions that traders grapple with. Most lose money attempting to utilize what they know about the subject. This is not surprising, because we're getting into the use of coincident and Leading Indicators and very few traders are properly prepared for the challenge these concepts present. Because of the level of misunderstanding, instead of narrowly defining what I use and how I use it, my approach will be to discuss the broad topic of Oscillators in general: what works, what doesn't, and why.
Typical thinking about Oscillators can be summed up by the following comment. "Oscillators work in a consolidating market, but once a Trend starts, they don't work at all." While this thinking may be typical, it severely limits and distorts a wealth of important trading strategies. The idea behind the statement goes something like this. You can sell Overbought and buy Oversold, as long as the market is consolidating and... you would expect to make money. This "sound good, feel good" strategy implies that you can identify when a market is consolidating with enough certainty to place orders. Does anyone want to try the ADX Average Directional Movement Index1 as a means of making this judgment? While this approach may be acceptable for some ofyou, it's not for me. I
1 J. Welles Wilder Jr., New Concepts in Technical Trading Systems (Trend Research, 1978), hereafter cited parenthetically in the text as Wilder, New Concepts.
haven't found it to be sufficiently accurate in this context, particularly for intraday charts. What about the second part of this statement?
"Once a Trend starts, they (Oscillators) don't work at all." The idea here is that the initiation of orders against the prevailing Trend will likely result in losses, by your stops getting hit.
The real problem with the initial statement is inherent in the way works is defined. I intend to show you how the right Oscillator can be made to work for you, in a wide variety of market situations.
Before we discuss uses and potential benefits however, let's first discuss which Oscillators are most commonly used, and which are best to use in the context of an Overbought/Oversold indicator.
The Stochastic is one of the most consistently misused indicators in the trader's arsenal. Traders typically consider any move over 75 to be Overbought and any move under 25 to be Oversold. That's not what George Lane (its originator) teaches and that's exactly the opposite of what Jake Bernstein's research of the Stochastic Pop Indicator2 tells us. In fact, according to Jake's research, fully 50% of a strong market move can take place after the 75/25% barriers have been crossed!
In the daily treasury bond Chart 7-1, there are two places, marked by vertical lines, where you would have been killed buying in excess of 25% Oversold. Note, this is the case even though I have used the more typical (stronger) 14 period Stochastic, rather than what I described to you in CHAPTER 5. Further complicating the issue for new traders is the fact that in strongly trending markets, the Stochastic may never see these extreme (75%/25%) levels on typical retracements of an ongoing Trend. If you were waiting for these levels, you'd likely never get the opportunity to sell in a strong down trend or buy in an strong up trend.
2 Jake Bernstein, Short Term Trading in Commodity Futures, (Probus Publishing Company, 1987), hereafter cited parenthetically in the text as Bernstein, Short Term Trading.
Other traders use the MACD (Moving Average Convergence Divergence) to indicate extremes in market Movements, or worse yet, as a divergence tool. As you know from CHAPTER 5, this is a cleverly designed and highly capable Trend-indicating Oscillator, not an Overbought/Oversold tool. There is, however, an innovative technique (Bernstein, Short Term Trading) of using the distance between the slow and fast lines of the MACD as an Overbought and Oversold indicator. In my opinion, however, there is a much better way to accomplish this.
THE RELATIVE STRENGTH INDEX (RSI):
The RSI is not it. Although far better than the Stochastic or the MACD for OB/OS analysis, this important indicator was created by Welles Wilder (Wilder, New Concepts) to nave universal appeal and easy application across markets. He certainly achieved those objectives, but for the more sophisticated trader something was lost. Since the RSI is normalized at +/- 100, like the Stochastic, it squashes strong market moves. If the Oscillator is at 95 and a big up move continues, it only has 4.9999 points left to go.
When the RSI achieved 96.50 on the daily coffee chart below, where could it go? As the move exploded in price, the RSI went down to 93.78, while the Detrended Oscillator increased from 9.41 to 16.19. A seven period input was used for both indicators in this example. On the right side of the chart, we see the Detrend at a whopping value (in relative terms, over four times the original amount of 9.41). The RSI was actually less than it was originally at 89.00. Discernible values of Overbought and Oversold relative to price are of critical importance. Why not use an indicator that plainly illustrates them?
Also, consider that Overbought in coffee has a different character than Overbought in corn, which has a different character than Overbought in the S&P! This character doesn't have the elbow room to show itself using the normalized RSI.
THE COMMODITY CHANNEL INDEX (CCI):
Finally, there's the Commodity Channel Index (CCI). I have the least amount of criticism for the CCI, probably because it does a fairly good job of approximating the Oscillator I use. Although Donald Lambert's3 development of this indicator was tied to Trend and cycle work, most traders who use this indicator use it as an Overbought/Oversold tool. The CCI is not normalized to +/- 100 and therefore requires more understanding to employ. This is likely why it is not used (or misused) extensively. Although the CCI has value, I believe the Detrended Oscillator achieves significantly better results.
3 Donald Lambert, "Commodity Channel Index: Tool for Trading Cyclic Trends", Technical Analysis of Stocks & Commodities magazine, July/August 1983, page 120-122.
THE DETRENDED OSCILLATOR:
The Detrend is an indicator that's been around a long time. I don't know who the originator was or when it was developed. The Detrend attempts to measure variations of price about a zero line which represents the Trend, hence Detrended. We define the Trend as a given Moving Average, then we mathematically make that average constant, or the zero line.
The formula for the Detrend is simple:
Detrended Oscillator = Close minus Moving Average.
A reasonable variation is the high or low, minus a given Moving Average as depicted on Chart 7-5.
Some of my colleagues believe incomprehensible math equates with genius, and therefore to profits. I always believe in keeping things as simple as possible. Since I did my research on the indicator in the early 80s on an 8088 processor, keeping the math simple was a practical, as well as a philosophical consideration .
I approached research on the Detrend in the same way as I approached research on the DMAs. I observed the quality of the indicator's usefulness, in trading situations, over a broad spectrum of data. I used none of the typical optimization techniques popularized some years later.
In observing literally thousands of data sets, with a wide variety of combinations of the detrend (simple, weighted, exponential, and mathematical MAs, of the median, high, low, close, etc.), my final conclusion for the best data sets were:
1. The close (today) minus a three day simple Moving Average of the close.
2. The close (today) minus the seven day simple Moving Average of the close.
Of the two data sets, the 7 day MA of the close is clearly the most useful in the context I apply it. I still use both data sets, however, particularly under Strategy 1, described below.
Aside from the profits generated from this laborious enterprise, the most gratifying aspect is that I see no reason to change the parameters today, over 15 years later!
USING THE DETRENDED OSCILLATOR:
Now, let's talk about how to use this powerful and versatile Oscillator in a variety of easy-to-apply strategies.
When your position reaches 70, 80, 90, or 100% of average Overbought/Oversold, take your profit.
The key considerations for employment of Strategy 1 are:
The Time Frame we use to calculate Overbought/Oversold and... the definition of what is meant by average Overbought/Oversold.
Here's where experience comes in. I always calculate OB/OS levels on a daily basis, i.e. daily data, even though 80% of my trades are off a five minute chart. Let me put this a little differently, so there is no misunderstanding. I never use intraday charts to calculate OB/OS levels for determining Logical Profit Objectives, even though intraday charts are where my position may be taken. To determine OB/OS levels, I look back over the Oscillator peaks and valleys. I consider about six months of the most recent daily data.
Average Overbought/Oversold is a value judgment, not a strict mathematical calculation. If I have three Overbought peaks as in Chart 7-3 with values of 96.85, 101.00, and 100.70, I'd take approximately 98.00 as an average Overbought.
I'll typically have my order resting in the market at a price equivalent to approximately 90% of the Oscillator's average Overbought level. At that point, I say "adios" to the trade. You might choose a lower or higher percentage. A resting day order is always best to take advantage of unexpected news or large traders pushing the market around for their own purposes. If it doesn't get hit, you cancel it or just let it expire.
Okay, so you have the Oscillator value in mind where you wish to take your profit. You can't call the floor and tell them to take you out at a seven day Detrend of 88 (90% of 98); you need a price. To get that price ahead of time, you need the Oscillator Predictor™ which I'll describe in some detail at the end of this chapter and in Appendix G. If you don't have the Oscillator Predictor™ to precalculate price levels which correspond to the Detrended Oscillator levels you wish to act on, you have another option. Some analysis software (Aspen Graphics™ and TradeStation®, among them) will allow you to set an alert at a given level on an indicator. You hear a beep, you exit the trade. While this is acceptable, the problem with setting an alert on the indicator is that you might miss the trade by the time you hear the alert, act on it, and contact the floor. These price levels are by definition unstable. These prices typically do not last for long, unless the market is in a runaway mode. It is conceivable that if you give a price order when the alert goes off, that you will only get out in runaway markets, i.e. those situations that would enhance your profit by staying in! So, if you use an indicator alert to get your exit point, simply exit at the market, and hope for the best. Those Mercedes and Jags in the basement parking garage of the Chicago Merc aren't there by accident. Market orders are one of the reasons the locals are able to buy them, so beware.
If profits were taken each time extremes were reached in Overbought, as shown on Chart 7-3, you would not suffer the draw downs of the subsequent pullbacks. These pullbacks are where most traders would have been stopped out by improperly tightening their stops. These pullbacks are for reentering against Fibonacci retracement levels. They are not for exiting!
When you employ the strategy of utilizing Logical Profit Objectives properly, your percentage of winning trades should increase dramatically, but you may not be there if the market really takes off and keeps going. You could, of course, hold multiple contracts and take Logical Profit Objectives on only a portion of the contracts you hold. It might interest you to know that I have run parallel accounts in which I have taken LPOs on all positions, partial positions, and none at all. Over time, the hands down winner is taking LPOs on all positions.
There's a corollary for Strategy 1. Let's say you're trading an intraday Time Frame and you are using Fibonacci Profit Objectives. The strategy would be to take close-in Objectives (COP's), when operating at or near these extremes in price, as defined by the Detrended Oscillator. A variation of this technique is used even by floor traders I've taught. Their actions in the pit change significantly as these Overbought and Oversold levels are approached. Your actions can change too. Think about it. If a market reaches 70% to 90% of average Overbought on a given day, it is likely that the market will meet resistance and at a minimum, consolidate for the next several days. Under these circumstances, avoid "buy stopping" old highs. Look for, and position yourself, on dips intraday, against Fibonacci support areas. Then, immediately exit against old highs, when the price nears average Overbought again, or when you get close to a Fibonacci Objective, whichever comes first. Remember, the price levels that produce OB/OS will change each day. This is a dynamic condition and dynamically calculated market levels are generally hands down winners over those that are statistically calculated, such as fixed money stops.
Now, I understand that some of you may hesitate to take Logical Profit Objectives because you lack adequate entry techniques that will get you back into the market, once you exit. Part of that problem is addressed below. Most of it will be answered when we study advanced Fibonacci Analysis, DiNapoli Levels™, CHAPTERS 9, 10, and 11.
The Detrended Oscillator can be used as a filter for any entry technique.
I had a high percentage of poor entries before I knew about, and understood, high quality entry techniques. An entry is not only poor if it ends up in losses, it's also poor if it puts you under significant pressure before the market goes your way. I dramatically reduced these unfortunate situations by the use of the Detrended Oscillator to determine what value of average Overbought/Oversold was apparent at my prescribed entry levels. If the price level at entry exceeded approximately 65% of Overbought/Oversold, I simply wouldn't take the trade. If the signal stayed in effect the following day, I would again look at the Detrend, to see if the trade was now "safe." If you're unsure of how to set up the Detrend on your equipment, you can try this. Go to your Oscillator Set Up menu, take the one day Moving Average of the close (which is the close), minus the seven day simple Moving Average of the close. That should work... then, just look at the value of the Oscillator at the time your entry signal is given and see how Overbought or Oversold the market is, before you act.4
4 Ifyou have the Oscillator Predictor™, these levels can be calculated ahead of time and are automatically printed out in the support resistance table, a feature of the CIS TRADING PACKAGE, TIMESAVER function.
Consider Chart 7-4 and imagine a simple system where you would be long when price is above the MA on close and short when the price is below the MA on close. I have pictured a non-displaced 12 day simple MA because that's typical of what many traders use. Two buy signals have been selected for this example.
The thinking here is simple and obvious. Ifyou take a buy signal in a highly OB situation (unsafe), it is likely to produce more pain than one taken at a reasonable (safe) level of OB/OS. In this case, both signals would have produced profit for you, if you weren't bored or frightened out of the trade, i.e. if you stayed with the system criteria. If, however, the system criteria carried an intraday stop, or if the stop was tightened, the unsafe entry may have produced a significant loss.
I recognize that volatility breakout players might be upset by this strategy, but I think there is nothing wrong with letting the market calm down a bit before entry. Will you miss some good trades? Of course! Will you miss being stopped out many times? Of course! Will the net effect be positive? I think it will. But, test this strategy on your own and see what you think. Many traders are in this game for excitement, not for profit. Some traders can comfortably tolerate 30% or 40% winning trades. I can't. You need to figure out where you fit in.
An OB/OS level can be used for stop placement.
If you observe the price corresponding to the value of maximum Overbought or maximum Oversold, you can simply put your stop behind that level by some comfortable margin, a few 32nds in the bonds or perhaps 50 points in the S&P. But, you have to be careful how the order reads. If you are using the close minus Moving Average, as suggested above, you should have a stop close only. If you want a physical, intraday stop, use the high or low minus the Moving Average, depending upon whether you are short or long the market. These Oscillator values will be respectively higher and lower than the Oscillator calculated using the close, and will produce correspondingly different stops. (See Chart 75). I'll guarantee you one thing, your stop won't be located where other stops are, unless by accident. Also, your stop is dynamic, every day it's moved. It should go without saying, that you're within money management parameters, and that whatever signal placed you in the trade remains in effect.
ill/15 112/1112/13 It9?4 11/17 12/1 12/M ft/1 Q/H H/lH/ll f5/2 |5/16 | Pailtf I
The best time to use Strategy 3 is when you have a lot of confidence in your entry techniques. This implies that you don't want to be bothered by close stops and that you choose to give these methods the time and space they require to work. I'll give you two possible examples.
Larry Williams has come up with a variety of high risk, potentially high profit, non-judgmental systems, usually based on some form of pattern recognition. One problem however, is that some of these systems carry a "stop close" only, or no stop at all on the day of entry. This makes some systems users understandably uncomfortable. An alternative would be to hide your (intraday) stop, as suggested in Strategy 3.
When we get to Fibonacci techniques, you will see on occasion, the desirability of an initial, far away, or disaster stop. Strategy 3 can provide an answer to the question, where should the stop be placed? This stop is almost never hit. If the original entry signal is negated, I simply exit "at market," or at the first retracement in the Direction of my entry. Then I cancel the disaster stop.
Realize that if the market is Oversold, a maximum Overbought stop will be miles away. But if the market is nearing Overbought, the stop will be relatively much closer. You can adjust the level of stop placement by using a lower percentage of average Overbought and Oversold, but I would suggest not going below 70%.
CHAPTER 6 discussed the Directional Indicator called Stretch. It uses an Overbought/ Oversold maximum Oscillator level, in combination with a major Fibonacci resistance or support area, to locate a trade entry range. This is Strategy 4. Admittedly, countering an existing Trend can be risky. It is, nonetheless, worthwhile because the combined strength of these two powerful indicators is substantial.
Since we haven't covered Fibonacci analysis yet, I will jump ahead a bit, with the following explanation. I suggest that you reread this after completing CHAPTERS 8 through 11, ifyou don't understand it now. The example above, Chart 7-6, showing daily bonds, was chosen for a couple of reasons. The cursor is placed on the initial Stretch sell. From that point, price reacted down to Fibonacci support and then achieved a Logical Profit Objective (COP) on the upside which was in Agreement with the .618 retracement of the down move from A to B. This is the same Fibonacci Retracement area that helped to give us the initial Stretch sell. To reiterate, if at approximately the same price of maximum Overbought, you have a significant Fibonacci resistance area, act on this combined level as an entry range to sell
IMPORTANT POINTS TO NOTE:
The initial Stretch sell gave us two hard down days. It was the setup signal for the break.. Do you realize how much you can make on a high probability break, where all the intraday trends are in your favor and you can margin up on your selling? If you have high confidence, you can pile up on size and walk away with the loot. You don't need a five point down move to score big; what you do need is patience for the set up and confidence in what is unfolding. In this case, the intraday down trend after the resistance was met, would have supported an aggressive short position, all the way to a support level which was measurable ahead oftime by techniques covered in CHAPTERS 9 and 10 .
On the rally back to the marginal new high, we only managed to achieve a move back to major Fibonacci resistance, as noted above. The Detrended Oscillator was still high and you then had an opportunity for a second shot at the short side. When we determine ahead of time, the level of Overbought or Oversold, corresponding to a given price level, we can make an informed and confident decision about how we wish to handle a given trade.
Every rose has its thorn, and stop placement, utilizing Stretch, can be prickly, since you are countering an existing move. If, for example, your entry sell level was at approximately the close minus MA maximum level, a safe stop may need to be placed at the high minus MA maximum level or behind another, more distant Fibonacci area. I never place money stops in the market. If a safe stop exceeds my money stop, I just don't take the trade.
Since we're on the topic of Oscillators, now is a good time for a prohibition. Don't use an Oscillator to price divergence as an entry technique, unless you have an excellent means of filtering it. In our last example, there was a divergence signal that worked between the initial high and the rally back high, but this is not a high probability signal. Even for the best of Oscillators this can, and often does, spell disaster, particularly for the new participant. Look back at Chart 7-2, as well as other charts contained herein. Price and Oscillator divergences abound, and the market continues to stop out the divergence players. Looking back in time, you're sure to find divergences that work, but when you're going forward in time, as in real trading, the accuracy of this technique just isn't there.
Special applications of the Detrended Oscillator to determine major Trend changes.
There are a variety of such applications. I'll offer one example. The idea behind this strategy is that long term Detrended Oscillator breakouts can be more significant than long term price breakouts. Look at Chart 7-7, monthly gold.
For the first time since the 1980 break in the gold market, we have an Oscillator reading, monthly basis, which significantly surpassed previous peaks, corresponding to run ups in price. I am not concerned with the divergence the Oscillator has made versus price, but rather with the fact that the Oscillator value has surpassed its previous rally high after a long move down. Note also that when we are measuring momentum in this context, we use the high minus Moving Average or the low minus Moving Average, rather than the close. This is because momentum in this context is attempting to measure the maximum push behind the market, rather than its strength at a given point in time.
This indicates that the bear market in gold is over and that we are likely to be in for a period of prolonged consolidation or an up move. To confirm this supposition we would like to see the 47.71 level significantly surpassed in the next gold rally. It would also be nice to see the pullback in gold contained in an ascending pattern. So far this has not occurred.
I've alluded to volatility breakout techniques in CHAPTER 1 and under Strategy 2, in this chapter. The idea behind the success of such methods is that volatility peaks precede price peaks and there is validity in this reasoning. The gold example is really a variation of this. While I will not endeavor to cover fully the details of how I handle extremes in volatility, I will make the following comments so you can understand my general approach. First, you don't know you have an extreme (let's say twice average Overbought or Oversold) until after the fact, so I prefer to take my profits as I near Overbought or Oversold, as stated earlier. When, in hindsight, I see a true breakout in volatility, I employ the techniques in CHAPTERS 9, 10, 11, and 13, to enter in the Direction ofthe breakout. Second, I attempt to filter any such breakout to eliminate blowoffs which will produce amazing volatility breakouts but by definition, the termination of any extreme in price.
Finally, as I have suggested, if you are an intraday player, use daily Overbought/Oversold data for calculation of Detrended Oscillator or Oscillator Predictor™ levels. If you trade daily-based, pay attention to daily and weekly Overbought/Oversold levels. If you trade weekly-based, i.e. using a weekly chart, be aware of both weekly and monthly Oscillator levels. Long term mutual fund switching can be greatly enhanced by using these techniques .
Above all, please keep in mind a market axiom I have been stressing for years. It is inherent in my market approach and should be gleaned from the afore-mentioned rules.
LOSS OF OPPORTUNITY IS PREFERABLE TO LOSS OF CAPITAL!
THE OSCILLATOR PREDICTOR™:
In the early 80s, I decided that I needed a means of capturing profit that was more efficient than any I had seen to date. At that time, I was unaware of Fibonacci Expansion Analysis and although the Displaced Moving Averages I had developed gave me reasonable entries, I gave back more "paper profit" than I wanted to in the exit strategies I was using at that time. Any paper profit was to me, my profit. I had assumed risk to achieve that profit. I had done considerable work to enter the trade initially and I didn't want the market taking any of it back! The problem simply stated was, "How could I exit on extremes in price, rather than wait for a DMA crossover?" Since I was convinced the Detrended Oscillator was my best Overbought/Oversold tool, and using my background as an engineer, I reasoned that a set of parametric equations could be created which would produce, a day ahead of time, the price level that would correspond to an OB/OS condition in the market. I approached my programmer, George Damusis, with the problem. He crawled off to his office for two weeks and after applying his considerable talents to the challenge, the mathematics behind the Oscillator Predictor™ were created and the study was graphically programmed into the CIS TRADING PACKAGE.
Consider what this discovery meant to me. I could accurately forecast a full day ahead of time what level of price would produce a logical (historically-based) profit, a businessman's profit. When you take Logical Profit Objectives, your percentage of winning trades can't help but increase. The main problem in taking logical profits is the lack of knowledge traders have in attempting to re-enter the market at a lower risk point. This issue is addressed in CHAPTERS 8 through 13.
Appendix G shows an example of how the Oscillator Predictor™ works in practice.
Before we get into the Fibonacci work, let's do a quick summary of the overall plan, (CHAPTER 3), and see what we've learned so far:
THE NECESSARY ASPECTS OF A SUCCESSFUL TRADING APPROACH
1. Money and self management
See the Bibliography and the Reference material sections.
References on market mechanics have been made as we have progressed to this point. When appropriate, more references will be made. For more information, see the reference materials.
3. Trend and Directional analysis
Lagging and Coincident Indicators theory has been covered. We've learned how best to identify Trend.
We've learned about certain very powerful Directional signals that overrule Trend.
4. Overbought/ Oversold evaluation
The theory has been covered.
We've learned how to effectively filter and quantify trades. We've learned how to take certain Logical Profit Objectives.
THE BOTTOM LINE IS:
WE NOW KNOW WHETHER TO BE LONG, SHORT, OR OUT OF A GIVEN MARKET, AND WE HAVE A MEANS OF DETERMINING WHETHER OR NOT A TRADE WELL HAVE A REASONABLE EXPECTATION OF A GAIN.
5. Market entry techniques (Leading Indicators)
Next we'll see how to position ourselves as safely as possible, within a market that we have chosen to enter according to the above criteria. We will also investigate powerful stop placement techniques.
6. Market exit techniques (Leading Indicators)
Then, we will cover additional methods of determining Logical Profit Objectives.
Was this article helpful?