When cycles fade from view 8


From mid-November last year until the end of May this year the US markets moved upwards in a very narrow channel, effectively etching a straight line onto the chart. It was a frustrating 27 weeks for a cyclic analyst, because of course we rely upon the evidence of the various cycles’ up and down movement to understand where we are, and what to expect next.

Hurst’s Cyclic Principles describe how multiple cycle combine to influence the price movement of financial markets, and one of the basic tenets is that cycles never disappear, but they can be overpowered by longer cycles (as I discussed at the time, such as in this post). It’s all very well knowing that the cycles are still there, but when we cannot see them it is of little comfort to understand that the markets are simply in the grip of a dominant cycle. After periods of dominant cycle influence, when we find ourselves staring at the resultant straight lines on the chart, we are left struggling to find a clear analysis, simply because it is difficult to place troughs during that time with any confidence.

That is when I find a process used by Hurst to “extract” the cycle information very useful. It is a process that Brian Millard also wrote about a great deal in books such as Channels & Cycles (a tribute to JM Hurst). The process is simple: you remove the effect of a particular cycle by first applying a displaced moving average to the data, and then you calculate the difference between price and that average. The resultant data effectively has the effects of one (or more) cycles removed from it.

Here is the S&P 500 with a displaced moving average of two years applied to it. (The average is displaced because it is plotted at the center point of the range of data used to calculate it).

The S&P 500 with a two-year displaced moving average

You can see that the displaced moving average reveals very clearly a long cycle (it has been projected up to the end of the data but I won’t tire you with those details). That long cycle is probably the 54-month cycle (or possibly a six-year cycle that I have written about before). That cycle is possibly the culprit (or one of the culprits) in the recent distention of price. And so I start by removing that cycle from the data. This is easily done by calculating the difference on each bar between the price and the displaced moving average. The result (loaded into Sentient Trader and analyzed) is this:

The S&P 500 with a long cycle removed

The big-picture analysis is (not very surprisingly, because we have removed only one cycle) the same analysis that we’ve been looking at here for some time. It is interesting to note the positions of the long 54-month cycle troughs, in August 2007 (as originally proposed by analyst “Airedale”) and October 2011. I’ve discussed reasons for that before (see this post), so let’s zoom in and take a look at the power of working with this data stripped of one potentially over-powerful dominant cycle:

A clearer cycle picture

Here you can see the cycle picture since July 2011. Stripping one cycle out of the data hasn’t removed all the imperfections of course (multiple cycles can be dominant at any one time), but the cycle picture does seem clearer. Here are the M-shapes of the 18-month cycle:

Clear M-shaped cycles

What is particularly apparent is the recent double-top formation around the 40-week cycle trough of late June 2013. Let’s zoom in further:

A closer look at the current 18-month cycle

Notice how stripping out the long cycle has exposed more detail in the recent price action, making the analysis easier to perform. Here is the actual price chart over the same time:

The actual price movement

The analysis is the same (or very similar), but it looks less convincing because of the way in which the cycle that we removed was distorting the price movement at this scale. Let’s zoom in even further, and look at the recent price action:

The recent 80-day cycle

A very clean 64-day cycle can be seen there (nominal 80-day).

I hope I have shown how useful it can be to remove a cycle from the data and to reveal the more subtle cycle action of other cycles. One can repeat this process, stripping away more cycles, and analyzing the results, but I am feeling confident enough with this analysis that it doesn’t warrant the effort.

One question remains: which cycle did we strip out? I used a two-year moving average which would have removed the next longer cycle, and so it was probably the 54-month cycle, but it would also have removed some of the effect of a six-year cycle (if there is one). In fact it doesn’t really matter which cycle was removed. The purpose of the exercise was to reveal the overpowered shorter cycles, and that was achieved.

I wish you all profitable trading.


About David Hickson

I have been trading for over 20 years, but only had any success after discovering Hurst's cyclic principles. Unable to find any software to speed up the analysis process I created Sentient Trader software, which now pretty much does all the analysis for me. I am a film maker and a TV director, but nowadays I mostly provide consultation services to professional traders and fund managers, helping them to integrate Hurst analysis into their trading. I'm South African and live with my family in Italy.


Leave a Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

8 thoughts on “When cycles fade from view

  • William

    David,

    You just described what I’ve been advocating for quite some time, the use of a bandpass filter to separate the dominant price waves. Technically what you did by subtracting the moving average from the price action is create a highpass filter. I use the same filter Hurst used and have separated every dominant price wave from the 4 year down to the thirty minute wave. Like yourself I created highpass CSV files for each dominant wave and loaded them into ST with spectacular results. It is still critical to have the correct parameters for the filters and the cutoff period and custom nominal model in ST. Otherwise you will be looking at garbage.

    Welcome Back!

    William

  • Jose

    Hello David and Willian

    I have gone in the same direction, subtracting cycles from the data again and again up to 10 day cycle, … In my case the problem is that the way we have developed the system needs to position next important peaka and lows “manually” (time and price level) by each subtracting, smoth changes in this positioning produce strong differences going down.

    Definetily I feel it is deeply interesting to find out where ST and this process can converge… I envision very interesting results

    Best for all

    • David Hickson Post author

      Hi Jose and William. It is a very useful process and I’d like to hear your thoughts about how ST could incorporate it more effectively. Many years ago when developing ST I spent several months building this into ST, and producing a “composite” of all the cycles projecting forward. I found it disappointing in terms of producing projections into the future – although very accurate of course in the past! For this reason I stopped building it into ST, because I found I got much better projections using the FLD Pattern Projections, and the sequence of FLD interactions that the FLD Trading Strategy is based upon. But I still go back to the process occasionally as described here.

  • derek frazier

    i agree with your post analysis appraisal David in that one can simply use fld interactions. the 80 day cycle was prominent with ST’s default analysis in SP500 over the last two months. price happened to interact with the 80 day FLD well