Uncommon signals from Common Indicators I
Indicators have been around for 50 + years or so in the markets. They started out being quite simple affairs; mainly moving averages as calculators, let alone affordable computers were still in the eye of the beholder stage. The 1970's saw an increase in both participation in trading the various markets and as a consequence an increase in both books and indicators. Always a chicken and the egg routine. Charting programmes bagan to appear in the early 1980's. This was about the time when I became involved in the futures markets as well. The internet was still not around, especially in its current iteration. Bulletin boards were the go in terms of contact over long distances.
Now in 2012 the scenario has changed dramatically, as it should and continues to morph into something else. There are more indicators around now than one can deal with, computers and internet and data are virtually throw-away items. How things have changed.
The one thing that has remained is the thirst for real knowledge, as opposed to mere information. And there is a mile (or kilometer) difference between the two. Part of my goal with life and hence with the Profit from Patterns venture is to pass on what I have found to be helpful and valuable in my 30 year journey through the markets.
The following is from an article I penned 10 years ago or so. The content and subject matter goes back about 25 years or so. Enjoy the read.
UNCOMMON RSI SIGNALS © 2003 Ivan Krastins aka The Investment Educator
I am very confident that many of may have heard of the Relative Strength Index (RSI) of J Welles Wilder Jr fame, or know the indicator or even actually use it for their own analysis or trading. My article will deal with actually understanding the RSI. Rest assured that there is a mile (whoops … a kilometre) of difference between ‘knowing’ and ‘understanding’ something. Whilst the name J Welles Wilder Jr is quite well known, I had the pleasure of being exposed to another Wilder, namely Frank Wilder.
Frank was the programmer for his more famous brother and through Frank I gained some tremendous insights into actually understanding his brother’s indicators. Frank was responsible for a charting package that I used extensively in my workshops in the 1980’s – One Day at a Time. It is really pleasing to see that it still exists and even provides the ability to display a version of my Hybrid Fractal Charts.
A term, or concept, that is commonly applied with the RSI is divergence and is seen to be the main signal provided by this indicator. It is also a concept that is often applied rather loosely resulting in a less than optimal outcome.
The formula for the RSI uses only the closing price, so it is preferable, at least in the early stages of gaining a thorough understanding of the application of the RSI, to display a line chart of the close (as opposed to a bar or candlestick chart) of the market being traded or analysed together with the RSI. This will help you not to fall into the trap of seeing a signal that does not actually exist. One property of the RSI is that it always follows the direction of the underlying market.
In other words, if the market closes higher than the previous period, the RSI will also be higher; if the market closes lower than the previous period the RSI will also have a lower value than on the previous period. From this you can see that the RSI can never actually diverge (move in the opposite direction) from the price. It is a visual illusion to believe that the RSI does move in the opposite direction to the price! You will see from this artcile that what do diverge are the peaks and troughs of the price and the RSI!
As, broadly speaking, the RSI is a momentum indicator, it measures the relative strength (or momentum) of two movements in price. The fact that these two movements in price should be related to each other is often an overlooked aspect of applying the RSI. It is a bit like comparing apples with apples. I shall now spell out a very strict (and hence testable) application of a bearish divergence signal for the RSI. The chart below features the daily DJIA with a 7 period RSI.
1. The market must have a completed movement up (a close, a higher close and then a lower close).
2. The RSI must register a reading above 70.
3. The market must then move down (one or more lower closes). *
4. The RSI must not register a reading below its equilibrium level (50 percent level). *
5. The market must then have one more movement up and exceed its previous highest close.
6. The RSI must move above the 70 level but should not go above its previous high reading (Point 2)
At this stage the RSI is not confirming the new high in the market. This is still not a completed divergence sell signal, rather merely a potential divergence sell signal. There is one more requirement to go.
7. The market must register a lower close than a previous period, thereby causing the RSI also to move down.
As long as the RSI has not registered a higher reading than it did at Point 2, a textbook bearish divergence, or in Ivan'speak, non-confirmation, is complete. The real significance of my non-confirmation interpretation will become obvious later in this article.
* The fact that the decline of the market (Point 3) did not result in the RSI going below its 50 level confirms that the second movement up in price is related to the first movement up. Had the RSI broken below its equilibrium level, the two movements in price would not be seen to be related. This is an important aspect in my experience in getting the best results when using the RSI to provide an early warning of a potential top in a market.
The daily chart of the All Ordinaries Index on the left with a 7 day RSI beneath I believe illustrates this point quite clearly. As the Index rises from A to B in one movement, the RSI also rises to register a reading above 70. When the Index then declines to C, the RSI proceeds to go down beneath the 50 level.
The trend-following application of the RSI would now deem that the trend was actually down. (I should note here that all momentum indicators can be used as trend-following indicators and vice versa. More on that in an article on uncommon signals from moving averages.)
Whilst the price does go on to mark a higher close at D, this movement up consists of two movements up (as marked by the solid blue line). The RSI registers a reading above 70 and below its reading at B. Not surprising, as the movement of the Index from C to D is not related to its movement from A to B (because the RSI went below 50) the sell ‘divergence’ between the Index and the price did not see the price fall.
Rather it merely went sideways for a few days and then broke back to the topside.
According to the designer of the RSI, J Welles Wilder Jr, the signal to act is provided by the RSI itself. The RSI needs to register a reading lower than at 4. This is referred to as the ‘failure swing’ point.
I have observed over time that using a timing tool, either of the five provided by bar charts or some candlestick chart signals tends to allow you to take a position earlier than waiting for the RSI to break its failure swing point. For this, the market must close lower and in many cases close much lower, thereby necessitating the use of a large stop loss. And hence a larger risk.. (In this article I will be referring to both bar chart and candlestick chart based timing tools, notwithstanding the fact that I have not used bar charts as examples.)
Additionally, I have found that the better divergence signals come when there are less than ten periods between the two peaks in price and correspondingly also between the two peaks in the RSI. In fact, the fewer the periods between those two peaks, the stronger the signal seems to be. Again, with the software available in the market, it would not be difficult for you to backtest this concept on the market, or markets, that you are involved in, or even wish to trade.
Naturally, in the case of correctly defining a RSI buy non-confirmation or bullish divergence, you would be dealing with lows in price and lows in the RSI. In brief, both lows in the RSI ought to be below the 30 level; the second low in the RSI ought to be higher than the first low and the peak between those two lows in the RSI should be below the equilibrium level of 50. At the same time, the closing price of the market needs to make a low, rally from that low close and then close below its first low close. Again, the actual signal to adopt a long stance in that market, according to its author, would occur when the RSI penetrates it failure swing high point.
The next chart features a textbook bullish divergence (A) with the two lows in the RSI being merely three periods apart. Both lows of the RSI are below the 30 level and the failure swing high point is well and truly below the 50 level. In fact, the RSI did not even manage to break above the 30 level. Nor is it a prerequisite that it should.
Daily closing line chart of Dow Jones Industrial Average with a 7 period RSI below
I believe that it is important to reinforce the textbook examples of the use of an indicator, especially when starting out on the journey of learning something new. Using these tightly defined parameters with the RSI you will be able to backtest the concept quite easily. Additionally, you will see how infrequently this indicator actually provides a signal.
Naturally, you will still need to come up with initial, trailing and maybe even time-based stop loss ideas, exit rules, position sizing concepts and general money management routines before you have a robust, rule-based trading system. And you thought that trading was an easy way to earn a living! Well, once you do have a set of rules to follow, just like when driving a car, trading can be simple and easy. I almost forgot … and boring!
The chart above also provides one example of an uncommon RSI sell signal at B. This occurs when the RSI provides a reading below 70, then just one, or a maximum of two consecutive readings, above 70 and the next one is below 70 again. It is as if the market just manages to poke its head into an overbought zone (above 70) and then retreats. The actual short entry may be based on the period that sees the RSI register its first reading below 70, or indeed, you could again use a chart signal (either bar chart or candlestick chart based) to time your entry into a market.
For fairly obvious reasons I have termed this a RSI extreme signal. For a buy, the RSI needs to register above 30, then below 30 for one or two periods only, and then back above 30. By combining this idea with a chart signal or even a setup, the probabilities of success increase.
In fact, you may discover that simply placing a buy on stop order above the high of the first period that has the RSI register below 30 provides you with some interesting and useable results. On the flipside for a sell, a sell on stop order below the first period that sees the RSI go above 70 also seems to coincide with turning points. And is that not exactly what the RSI is designed to do? After all it is a reversal indicator! It was designed by Welles Wilder Jr to pick tops and bottoms in markets way back in the late 1970’s.
Weekly candlestick chart of the All Ordinaries Index with a 7 period RSI below
A) Not only does the RSI highlight the fact that the market was oversold for the first time (by being below 30), the period is also one of the high probability chart signals that I wrote about last year. Thank you to those readers who contacted me about their experiences with this idea.
B) This time the market was not accompanied by a chart signal and did not break the low of the period that had the RSI first register above 70. The following period both saw the RSI below 70 again, as well as being an inside period. Inside periods indicate that a market is indecisive, with neither the buyers nor the sellers being able to gain the upper
hand. Had a short position been taken on the close of this inside period, or even on the break of the low on the following period, a loss would have ensued.
C) This time the required reading by the RSI was accompanied by one of the high probability setups that seems to perform and perform. I refer to it as invisible support and resistance in my book, Listen to the Market. It is interesting to note that the RSI reading was 70.26 this period whilst three periods prior, the RSI registered 70.00. The requirement is for the RSI to be above 70.00 (i.e. greater than 70.00), so there was no signal three periods ago. Rules can, and I believe, do need to be so precise and black and white.
D) The final example on this chart of an uncommon application of the RSI again features the RSI going from above 30 to below 30 for just one period, and being accompanied by the inverse of the setup that was seen at C. The market then embarked on a rally complete with a hammer signal that could have been used to add to the initial long position.
From the previous example covering ten months of weekly data you saw that there were only four opportunities highlighted by this uncommon application of the RSI. Many traders would not have the discipline and patience to wait so long for so few signals to act. Is it the frequency of the opportunity or the probability of success that really counts?
Applying my Universality Test (in the last issue of this Journal) one could look for the same RSI opportunity on a variety of markets and a variety of timeframes, including Hybrid Fractal Charts (HFC). This would naturally have the effect of providing more potential trades.
The next uncommon RSI signal is based on the term that I used earlier in this article instead of divergence. That was non-confirmation. It is also a simple concept that you can backtest to see whether it suits your requirements in trading and to see how it performs in your market or markets of choice. This uncommon application of the RSI I have dubbed angle non-confirmation.
This may be a bit of a mouthful at first, but the name really does describe the signal rather well. For a sell the RSI must first go above 70, then retreat to no lower than 50 and then once again climb up above 70. This time however, the RSI needs to exceed its previous high reading. At the same time, the market needs to register a high close, followed by one or more lower closes and then the market needs to also exceed its previous high close. Now you draw a line connecting the two peaks in price and do the same with the two peaks in the RSI.
The final step is to compare the angles of these lines. If the angle between the RSI peaks is not as steep as the angle of the line between the closes, then all the requirements have been met. This occurs at A on the chart below. As with the common application of RSI bearish divergence, when the RSI breaks below its failure swing low at B, a short stance in the market is signaled by this approach. Please note that the price does not have to break below its previous swing low.
However, that is not to say that you could not add that requirement to your own application of this RSI concept.
Daily line chart of the Dow Jones Industrial Average with a 7 period RSI below it.
With an angle non-confirmation buy signal, you require the RSI to move below 30 and then move up for at least one period without going above 50. It is not a requirement for the RSI to go either above 30 or not, but merely to not go above the 50 level. The final requirement is for the RSI to go down lower than it did the first time. As all this is happening, the market needs to register a low close followed by a higher close (for one or more periods) and then to close lower than it did the first time.
The next step is to compare the angle of the line drawn connecting the two low closes with the angle of the two corresponding lows in the RSI. If the RSI line is not as steep as the line on the market closing prices, then it is all systems go for a possible low in price to be in place. You can see that marked on the chart at C. When, or if, the RSI breaks above its failure swing high point marked D, this technique is signalling a low to be in place.
Other triggers can be used such as the various chart signals that I have mentioned before. As with most aspects of life and trading, when there a number of signals coinciding in providing a signal, the higher the probability of success. You may have noticed on the previous chart that the angle non-confirmation buy signal coincided with a Fibonacci retracement level of the rally from the mid-August
lows to the early September highs.
The chart on the left illustrates a textbook bullish angle non-confirmation, as well as the ability of a chart signal to highlight a potential turn before the RSI does. The second last close on the chart is clearly lower than the close at A. You have to look at the numeric values of the RSI to establish that at B the RSI was 22.74 and at the previous swing low (corresponding with A on the price chart) the RSI was 22.86. Hence, even though on the chart it appears that the two lows of the RSI are equal, closer examination reveals that the second one is lower.
By comparing the angle of the two lines, you can see than the line on the RSI is not as steep as the one on the price chart. The last perod closing higher than the previous period causes the RSI to move up and break above its failure swing high point at C. Job done by an uncommon application of the RSI. Again, from my perspective, detail with trading, as with life, does make a difference to the outcome.
It is interesting to note that the second last period was a high probability candlestick chart signal (a spinning bottom) having a ‘small’ real body with the upper and lower shadows being ‘large’ in comparison to the real body. I am confident that your own testing of this timing tool will reveal exactly what the commonly used and vague terms, ‘small’ and ‘large’ really mean. One really has to look at the actual price data to see this clearly. This was then followed by the last period being a high probability bar chart buy signal (pivot point).
These signals are akin to the road signs that we see regularly when we are driving. Generally we tend to obey them and ocassionally we choose not to! Markets also have their own road signs, so to speak.
The next chart is a larger chart of the same market with the same bullish angle non-confirmation signal marked on it. That signal occurred at the end of bear market in the Japanese stockmarket that lasted 14 years. The Nikkei topped at 38,915.90 in 1989 and made its final low at 7,697.88 in 2003. That was a decline of 31,308.02 points, or 80.45%.
History has shown that one can experience falls of this magnitude when a bubble bursts. The Dow Jones fell by 89% after the 1920’s bubble burst, whilst the Australian stockmarket fell by a smaller amount during the Great Depression.
In order to spot changes of this size, you really do need to examine large timeframes. The chart featured on this page is a monthly chart. There is even a strong case to be made for using even larger timeframes such as 3 monthly, 6 monthly and even yearly charts. The larger the timeframe that you are using, the larger the potential movement that you are able to trade.
So far we have looked at two uncommon signals from the RSI that required the RSI to register an overbought reading or an oversold reading (the RSI extreme signal). In other words, the RSI had to be above the 70 line to confirm that the market was overbought, or to be below 30 showing that the market was oversold.
The next, and final uncommon RSI signal occurs when the RSI is between those two levels. As the RSI spends most of its time between 30 and 70, with rare sorties either above 70 or below 30, it would be logical to believe that there would be more signals to be found by this last uncommon application of the RSI.
With this uncommon application of the RSI, you need to focus on what the RSI does first and then look to the price chart to see if the market is confirming the RSI movement. Let’s take some time now to walk through 5 odd years of price action in the Nikkei. The area market by the ovals shows you where the previous chart fits into the big picture.
Monthly candlestick chart of the Nikkei with a 7 period RSI below it.
Notice how when the RSI breaks above its previous swing high (A) that occurred in the prerequisite area, between 30 and 70, the Nikkei at B did not break its corresponding high (also A). This set up the desireable internal non-confirmation signal with this application of the RSI. The moving average on the chart confirms that the market is in downtrend, thereby reinforcing the sell signal.
Two periods (months) later the market presented a high probability sell setup and the following month also provided a bar chart sell signal (a pivot point). The market then continued to decline for 10 more months before the angle non-confirmation buy signaled a possible low.
When the RSI breaks below C, a swing low, the market at D is nowhere near breaking its corresponding low also marked C. On this occasion, the candlestick chart provided a timing tool that could have been used to actually trigger a long position. Additionally, the moving average was pointing upwards indicating that an uptrend was is force and that buys were warranted for those who are dedicated trend-followers.
The last example on this chart occurs when the RSI breaks its previous swing low at E and the market does not break its corresponding swing low point. The RSI value at E was 47.32 and the period that breaks that sees the RSI register a reading of 45.80. Whilst it does take a keen eye to see the break on the chart, in real time you would actually check the values of the RSI with your charting software. As both readings of the RSI were between 30 and 70, this qualifies as an internal non-confirmation buy signal. Again the moving average was pointing up, adding more weight to the potential buying opportunity.
The very next month provided a bar chart buy signal (a reversal) that could have been used for the timing of a long position.
Whilst I have been using a 7 period RSI in all these examples, in reality any period RSI can be used in the same fashion. The smaller the number of periods used for any indicator, the more sensitive and reactive that indicator becomes to price movements. The concepts remain exactly the same and can be applied in the same fashion. The larger the number of periods used, generally the fewer the opportunities that are presented and, arguably, the stronger the resultant movement.
From the next three charts you can see that a textbook bearish divergence (A) is provided by all the RSI’s – 5, 7 and 14.
When Welles Wilder Jr originally released this indicator to the trading world, he actually used 14 as his suggested number of periods for the RSI.
I believe later he started using a 7 period one as it is more sensitive to price movements.
The same is true of all indicators. The smaller the number in the sample, the more reactive the indicator is to movements in price.
Both the 5 and 7 period RSI also provide one of the uncommon signals (the extreme signal), with the 5 period giving a buy at B and then a sell at C.
Not surprisingly, the 7 period RSI only provides one, a sell, at B.
Not surprisingly as it is slower to react to the price movements. The 14 period RSI is the slowest of all.
The 14 period RSI is the only one to provide another of the uncommon signals.
At B, the RSI has broken lower than its previous swing low, whilst the price is certainly above its corresponding low.
In other words, the price is not confirming the new low in the RSI, and hence is seen as a potential buying opportunity.
That magic word of divergence strikes again!
By looking at these 3 examples of using different periods with the RSI, you can see that there is a case for using more than just one.
The more signals, the more potential trades, the better.
There are two fairly obvious statements with the previous sentence.
The first one is pretty obvious. As long as on balance they produce a positive result.
The second one has to do with the practical aspect of the markets.
This becomes an important consideration when you are looking to harness the power of compound to the maximum.
As long as you can incrementally increase your bet size.
I should stress that it is impossible to ever achieve the maximum in the real world!
The FX and stock markets do provide virtually unlimited scaleability. Again limited by practical real world limitations.
The futures markets with their fixed contract size make this a tad more challenging to do.
The Experiment that we are conducting is designed to see how far we can push the limits in relation to compounding in the real world.
This final example features two of the uncommon RSI signals that I have discussed in this article. At A, the All Ordinaries Index registers an extreme oversold reading for the first time on the chart. The period that created that extreme low reading in the RSI is neither a chart signal to adopt a long position in the market, nor a setup for a buy.
Given the observation that the market opened at the highs and closed virtually at the lows for that week, and closed below previous support levels, it would have been reasonable to deduce that the sellers had been the dominant force.
Also, given the size of the range, it would have been reasonable to deduce that the market players were rather bearish and probably short in the market.
So what should have the market done on the following period?Weekly candlestick chart of the All Ordinaries Index with a 7 period RSI
It should have continued its fall, naturally! It did not, and the RSI was below 30 for a second period. This meant that the uncommon extreme application of the RSI was still in buy mode.
Additionally, a high probability buying setup (one of the twelve L.I.V.E. Setups) was also present on the All Ordinaries Index.
And rally the market did, without breaking a weekly low until the RSI registered an overbought reading the first time.
The angle non-confirmation sell signal at B also coincided with the second time the RSI registered above the critical 70 level for an extreme overbought sell signal. Probabilities of success do increase when more than one technique signal a possible trade simultaneously.
The following week featured a sell timing tool in the form of a pivot point and formal entry being signalled by the RSI going below it failure swing low of 4 weeks earlier (C).
None of the concepts I have presented todate dealing with uncommon and common applications of indicators are difficult or hard to master. I believe that to be able to decide whether an indicator is suitable to be included in your arsenal of trading tools you do need to have more than just a passing knowledge of it.
It is only with a true understanding of them that you are in a position to make a valid judgement.
The RSI itself has many other aspects that have been beyond the scope of this article to fully explain. One such idea is to apply the RSI to the opening price, thereby being able to create a daytrading tool. More on that another time.
I trust you are beginning to grasp the real benefits of looking at the same palm tree as everyone else ...
and seeing it from a different angle!
Enjoy your journey.