Double Stochastic

Markets do trend from time to time. Those of use who use indicators or oscillators that are either bounded, or simply go to and stay in an area that is perceived to be overbought, often have difficulty jumping on board an extended trend. Those who are trend followers simply buy and hope the trend persists. Oscillator traders are having a hard time when the trend followers are having fun. Sometimes a clear signal does not occur to announce the start of a trend. Sometimes a trade is exited at a price target that turns out to be just the beginning of a trend, and the indicator or oscillator stays overbought, not allowing a new entry point. This can be very frustrating. I tried to find a solution. Refer to the following two charts.

ds1.gif
Here is the Energy ETF on a daily chart. It is displayed through the date that this is being written. Many other charts display a similar trending pattern. In this case you can see the adaptive CCI staying above the +100 line for most of two months.

ds2.gif
Here is the same chart with a Stochastic, smoothed with a moving average, as it is in most charting programs. Again, it stayed above the upper, overbought reference line even longer than the CCI. Getting to overbought was only the beginning of the move. Viewing the upper reference line as being overbought is only one interpretation of using a Stochastic indicator. George Lane never regarded the upper level as overbought when the indicator started to garbage, as he called it. He would look to a longer time frame to see where the longer term cycle was before he would conclude an overbought condition. But going to a longer time frame didn’t help with the short-term decision of when to jump onto the bandwagon.

One possibility is just to use a very short-term parameter on a bounded indicator like the RSI or Stochastic. If you put a 4 or 5 input parameter on an RSI you’ll get plenty of movement with a small amount of deceleration in momentum. I tried this idea but wasn’t satisfied.

I read somewhere about an indicator called the StochasticRSI. I was intrigued, but I disliked the RSI. I figured I would try the same concept substituting a Stochastic for the RSI. It seemed kind of redundant, but it showed promise. It was the Stochastic Stochastic. Upon further research I came to Walter Bressert’s web site and I saw an indicator that looked very similar. He had quite a bit of information on his new indicator. He didn’t have the formula published, but when I put in his parameters on my indicator they were almost identical. I don’t know if we have the same formula, but it is close enough to say they are essentially identical. I suspect the moving average I use is different, but other than that they function the same way. I have been using the 10 period input parameter as Mr. Bressert suggests. Refer to the follow chart:

ds3.gif
Here is the same XLE chart as in the previous two examples. You can see that when the CCI and the Stochastic stayed in the overbought zone, the Stochastic Stochastic, or we can just call it the Double Stochastic, descended from the overbought zone and came all the way down to the oversold zone. It can do this at times when there is very little down price movement. The almost constant cycling back and forth between the upper and lower reference lines to me displays a cyclic tendency. The Double Stochastic is not a cycle indicator by any means. It is not measuring a cycle or projecting a cycle. It is simply following a tendency toward a down cycle that is often imperceptible on the price bars. In an instance like in the charts above, a normal Stochastic often won’t react enough, or with enough amplitude to descend to a low enough level to feel comfortable entering the trade, even with a very short parameter. The Double Stochastic can turn down from an overbought area and prices can still be climbing, but when the cycle tendency bottoms out and then turns up in the direction of the trend, often a good entry point presents itself. The three magenta circles are good examples of the Double Stochastic going all the way to the opposite reference line while the regular Stochastic in the previous chart stayed over the upper line. The pivot just to the left of the first circle almost made it as well.

ds4.gif
Walter Bressert also would display the 5 period Double Stochastic on the same sub-graph as the 10 period. Here it is displayed on the same XLE chart. The black line is the 10 period and the white line is the 5 period. I’m not sure how Mr. Bressert uses the 5 period line, but I like to use it when the market is just starting a move and the 10 period Double Stochastic gets to the overbought line in a hurry but fails to come back down. In this case often the 5 period Double Stochastic will make a slingshot action by briefly hooking the lower line as the 10 period stays over the top line. Here is a perfect example. It almost made the same type of slingshot action a few bars later. Of course, the idea can be reversed for downtrends. There was almost a slingshot action on the sell side a few bars prior to the middle circle, but the trend was up. This only seems to work in the direction of the trend.

A couple of other uses of the 5 period Double Stochastic are illustrated in the next two magenta circles. The middle circle shows the 5 and 10 period lining up perfectly. It appears two cycles are in synch which gives me added confidence in entering on this pullback. You can see how shallow the pullback was on the prices, but both Double Stochastics lined up and caught the next impulse up very well. The circle to the right is similar, except there was a very slight micro-W formation on the 5 period (white line). I like to see this pattern as well, as it gives me added confidence. I also like to see the 5 period hook inside the 10 period and lead the way, as it did in this case. Compare that to the upturn that occurred in between the two magenta circles. There was a very slight micro-W and it did hook the lower line, but the 10 period started to lead, that is the 5 period started from outside the 10 period. The 5 period eventually went above the 10 period and led the way up. The two signals had about the same outcome. However, I find the example highlighted by the right circle to be the more reliable of the two.

As I said earlier, most of my useage of the Double Stochastic is simply to find entry points within existing trends. I do sometimes look for confirmation on divergences. I primarily look at the adaptive CCI that I describe elsewhere on this blog. However, I don’t find the CCI to be a very good divergence indicator. The Stochastic, as well as the Double Stochastic are much superior. The Stochastic formula relates the current bar price to a range of prices, whereas the CCI simply averages the high, the low, and the last, or close. If prices closed sharply higher, but the closing price was at the very bottom of a long bar with a long tail above it, the CCI would regard that as positive and probably point upward. A Stochastic would take more of a dim view of this condition and would probably roll over and turn down. There’s a higher chance of this happening with a divergence pattern, while the CCI is going upward as if all is well.

ds5.gif
The above chart shows a three drives to a high pattern in the XLE. The indicator under the prices is the same 10 period Double Stochastic as in the previous charts, except I smoothed it with a short term linear regression curve. I sometimes use this for divergence analysis. It smoothes out the bumps a bit, and of course can lag behind a little. It actually can have a bit more amplitude at times than the unsmoothed Double Stochastic.

I don’t encourage the use of this indicator to generate initial trading signals, but it can give excellent signals to enter on pullbacks within a trend, when often a trend is too strong to offer visual pullbacks on the price bars. There are probably many other uses that one can discover with some experimentation.

8 thoughts on “Double Stochastic

  1. Hi there!
    Interesting article! How do you set up a double stochastic like that? I have a normal charting package and can make changes to the Stochastics.
    I’d appreciate your help!
    Thanks!
    Hank

  2. I’ve been using the 10 period double stochastic (DS) to trade the QQQQ, by waiting for the 60 minute DS to turn up thru 10 (or down thru 90) then entering each time the direction of the 15 minute DS aligns with the direction of the 60 minute DS (i.e when the 60 and 15 minute DS’s are both rising or falling). But as your article addresses, I have been missing extended rallies as the DS’s stay overbought (or oversold). So per your example, I’ve overlaid both the 5 and 10 period DS’s (in both timeframes), and will see if that makes a difference in terms of:
    1. Good Long entries using 5 DS dips when the 10 DS is ‘overbought’
    2. Using the 5 DS for a slighly earlier long entry (i.e while the 10 DS is less than 10 and flat, but not yet turned up).

    It looks like you use 80/20 for your upper/lower crossovers (vs my 90/10). I’ll have to watch for which works better with both DS’s.

    Also combining two DS’s in two time frames may create confusing signals or may add additional confirmation …. time will tell.

  3. Hi,
    The writeup(s) are well conceptualized and content is very helpful in expressing the viewpoints. Great Work, Appreciate the same.
    On the issue of not giving any parameter settings etc., i agree that these are proprietory studies and one should set up what’s workable for them, however if you could provide any info or links to obtain such parameter settings in order to further understand the subject shall be just great. Once again, great work.

  4. Read your article with great interest.
    I have access to the standard Stochastics on eSignal with inputs %K, %K, %D for modifying it.
    What are the parameters for the %K, %K, %D for the Double Stochastic.
    Thanks….Daniel

  5. The double stochastic is a different formula than a regular stochastic. Changing the parameters of a regular stochastic won’t simulate the double stochastic. You need to actually imped the regular stochastic within another stochastic formula. You might be able to find the formula already written in eSignal if you google around or check the forums.

Comments are closed.