Trading against a strong dominant cycle

Occasionally, one observes an instrument that is behaving poorly in all opposition to what it should be doing according to proper Hurstonian principles. Perhaps all the signs point to a trough forming, and yet, no matter how one performs an analysis the blasted thing rudely continues to rise up.

Of course we know the culprit is the dastardly SIGMA-L. The maledictive sum of all cycles that can’t be properly counted using our methods.

What is the sensible approach in this case? Usually I dismiss the instrument as being bothersome and move on to something that is more agreeable. However lately I’ve been wondering if I should seek out these bad boys of the markets, and get swept up in their irresistible trend. If they’re so powerful as to resist what the cycle should be doing just imagine what they’ll do when the cycle is finally in their favour.

Or is this madness because the risk of a reversal is higher than usual? Or maybe I’ve missed the point entirely and it is my cyclic analysis that needs improvement.

Yes I dont know what the FLD interactions are anymore but the trough timing seems to work in conjunction with FLD projections short term. Perfect example is BTCUSD 8000ish target right now. Working with top stocks with options.

Agree with you, The melting up of equities including all the top stocks is indicative of psuedo-trend IMO.

Im happy to see what Facebook did overnight. Massive volume of share buying on a huge dip, then new highs. Pretty hilarious action. Never before seen stuff. 36 year cycle?

Everyone should be aware of TBT. Its moving higher. TLT lower.


From my experience it is usually due to your model being incorrect. There will always be a certain amount of modulation, but an accurate model should be able to compensate for it. That is why I only trade instruments with the least amount of modulation.

In my opinion Sigma-L is not the culprit. It can be calculated with a high degree of accuracy at nowtime, assuming your model is accurate. Just don’t project it into the future where the error increases exponentially! The usual suspect is a high degree of amplitude modulation but if you are not using a “mathematical” approach, it is difficult to compensate for it.

To which instrument are you referring in your comment? My are moving with (relative) machine like precision.


It was meant as a general comment not in reference to any specific instrument, but what got me thinking about it was the behaviour of AUDUSD today. I trade currencies mostly. I can predict their movements down to the minute with good accuracy in most cases.

I am for the moment long EURUSD, GBPUSD, NZDUSD, and they are behaving. AUDUSD is not as much. I thought if it were to decline it should so at the end of yesterday, instead it did so today. Perhaps I am just demanding too much of my model. If I widen the wave I’m searching for I get the pattern I expect but then other periods show a skipped/extended cycle. Is your model not susceptible to those?

I agree that one shouldn’t project too far into the future, and that shorter term trading produces better results; I believe Hurst said the same.

Forex pairs are a slightly different cyclical animal. This is due to the base currency/quote currency relationship. This means that a peak of one is the trough of the other. On the surface that seems to run afoul of one of Hurst’s principles concerning synchronized troughs and unsynchronized peaks. From my experience it means that forex pairs are subject to a high level of modulation resulting in a complex harmonic relationship among the price waves.

While my approach allows me to compensate for such modulation, theoretically it makes “predicting” price movement more difficult, which is why I don’t do it. AUDUSD is not one of my instruments but I will take a quick look at it to satisfy my curiosity and post a chart here.

Pseudo trend is just a camouflage of “I do not understand what is going on” and of “Market does not behave as I have thought”

same thing for ellioticians “extension” meaning the Wave went much higher as forecasted

1 Like

This seems to present a bullish picture, does it not, Alain?

Scarlet pip, bullish ???

Bullish in itself has no real sense if you do not define the Timeframe

Here the Time Unit is 1 minute and the period 1 day

Bullish in itself has no real sense if you do not define the Timeframe

Yes, I agree.

I’m just asking based on how the envelopes are drawn what you’d expect from this graph. What was your meaning when you posted?

Price is at the bottom of the middle and outer channels and is rising up from the inner channel. We typically expect price to be rejected at the envelope boundaries, no?

Would you be willing to post a graph of one such machine-like precisely moving instrument so I can compare standards? Again, it may simply be that I am demanding too much.

However, at the risk of becoming persona non grata on this forum, I am lately becoming more cycle-skeptic.

I have been reading Mandelbrot’s ‘Mis-behaviour of Markets’ and it cites the example of Feller’s research into the patterns that emerge from random independent data. For example a coin flipped 10,000 times will produce distinctly cyclical patterns, even though we can’t expect that the prior coin tosses are affecting the future ones, as we do with price movements.

If you look at the graph on the 21st page of this document (you’ll have to rotate it or your screen or your head) you’ll see what appears to be a cycle beginning with a trough on the 2000th flip, a peak on the 4000th, another trough after the 6000th. But eventually the cycle breaks down. This is a pattern I see in financial markets, just enough of a sine wave to say something is there, but then, the cycle breaks down and we must recalibrate. I wonder if we are chasing phantasms.

Whenever a cycle fails to meet our expectations we can say it is caused by pseudo-trend, sigma-L, or instruments that are simply not cyclical, or it was a problem with our model, or a phasing error. There is always something, but have we considered that the cycles are simply ‘pressure eddys’ caused by the reflection principle, and are part of randomness? I’m not saying it is so, just saying it needs to be considered.

The predictions posted by the venerable members of this site, if you track them as I have, are very nearly 50% accurate which is to say their ‘point’ when it can be divined, turns out to be wrong as often as right. Which is better than you will find elsewhere, but not enough to completely abandon other methods of enquiry.

Note that I’m not saying that members aren’t having success in their trading, but should it be attributed to other tools of the gambling profession, such as good risk management, or is it really from forecasting skill?

Hi scarletpip.

For all the time I’ve been posting on the various Hurst forums I’ve been an advocate of the “mathematical” approach to cyclical analysis. I think that description is a misnomer since all forms of technical analysis are mathematical. I have found Hurst’s analytical techniques to be the most accurate model of the movement of financial instruments for trading and investment purposes. In my opinion, once you intuitively understand the underlying mathematics, it becomes self-evident. The most difficult part is just believing it.

I clearly have a different perspective on the interpretation of Hurst’s work than the majority of individuals in the Hurst community. The differences are not that great but the obvious key is to have correct model of market movement. By “correct” I mean profitable over long periods of time. I do not harbor any “expectations” with respect to the market, therefore I’m never surprised or disappointed. I just take what the market gives me. I trade action signal to action signal, period.

My “machine precision” comment was just in reference to my interpretation of Hurst’s work. For example, somewhere on this forum I posted a chart showing the 4 year wave over the last 50 years based on Hurst’s filter parameters. I just took the average percentage gain and period of the 4 year price wave and created a projection box from the February 2016 low, the most recent 4 year low. I did the same thing with the 20 week price wave. The price action had moved into both boxes, albeit slightly translated from the 4 year average peak which is to be expected. All of my “coal mine canaries” had stopped singing. The dead give away was the ridiculously steep up trend line of the 5 week price wave. I dialed down to my short action signal and was good to go. I could not possibly have “forecast” that break and subsequent decline with any accuracy, but I didn’t need to.

I hope that helps a little.


1 Like

I always find your posts helpful, thanks.

I do not remember why I Posted this graph !

But the system I use is just a reading grid for analysis and trading.

As the 3 Envelopes were in a Downtrend we could suspect it would continue.

IF Price Action was UP as the Underlying Trend was down we could assume that the counter-move would not statistically reach the Upper Limit of the Envelope 256 TU (brown) but around 70 / 75 % of the Amplitude of the Envelope.

A significant break of this Upper Liit would signal a change in trend (within the parameters used)

Good post ScarletPip, thank you.

Not sure if I am a venerable member or not but I would say my success rate with Hurstonian methodology at what I would call the ‘swing’ lows and peaks (80 day nominal and above) is around 75%. I would have ditched Hurst some time ago if it only resulted in half the trades being correct. I purely use twitter as a record of such and I have been posting ‘then and now’ comparisons recently to recall the predictions etc… Certainly I get some wrong but I rarely get the direction wrong - more often the degree and resulting amplitude. If you move down to intraday level it gets a lot more difficult. Sure, Hurst stated in PM that the lower the cycle the higher the yield but it gets extremely tricky intraday - of course Hurst had no access to this in his time. I would say there is somewhat more of a random walk intraday…but not totally random!

Having used many of the established ‘technical analysis’ tools available (EW, Fibs, S&R, pure PA etc), JM Hurst’s work really stood out. It was something that was superior because it neatly explained the others (think double bottoms / head and shoulders / moving averages) in a way that made perfect, elegant sense.

I often challenge people on twitter to explain why they post arbitrary moving averages like the ‘50 SMA’ or ‘200 DMA’ in their charts. Many just follow the crowd and stick it on as it looks good! When you understand Hurst’s methods, which present the MA as a simple low pass filter and a tool to extract cyclical elements, the arbitrary MA, a stalwart of technical analysis, becomes ridiculous!

Keep at it!

1 Like

Thanks David,
You were one of the members I was thinking of in my un-scientific assessment. Your Twitter posts are worth reading. You mentioned the 80 day nominal and above, and maybe that’s the timeframe I should be focusing upon.

I should add I haven’t made up my mind about Hurst, there’s no debating that it has more promise than other forecasting methods. I just wonder if I wouldn’t have more success by focusing less on the future and more on the present, using a stochastic method, however arbitrary, or some combination of those approaches.

What timeframe are you trading?

20 day and below, 5 day preferred.

Should be ok, I haven’t dabbled in that timeframe myself. I combine medium term swing trades (40/80 day cycle) with intraday trading on a very select group of instruments.

I’ve been thinking about this for a while. After David_F made me aware of how many phasing errors can be introduced with the default automatic phasing of ST, I’ve begun producing phasing analyses by hand for my favorite instruments. The exercise is recommended for anyone new to cycles.

One thing I’ve wondered is if placing the cycle peak or trough at the price high or low is not always optimal. Is it better to place them at the envelope peaks and troughs, and to place them in the ‘lacunae’ between the legs of of a peak or trough rather than trying to find an insignificant prominence/promontory near where the expected peak or trough should be.

Doing this could avoid some nonsense caused by using the average distance between points as the expected future peak/trough, because points will conform more closely to the nominal model and have less variance with the mean distance(s). I’ll post some charts later that show what I mean and see if the forecasts produced by such charts are superior to what ST can manage on its own.