There has been a lot of discussion recently about whether the cycles are “running short” relative to Hurst’s nominal model. The answer to that question depends on one’s definition of nominal model.
A little discussed fact is that Hurst created two nominal models, a nominal time domain model (54 months,18 months, etc.) and a nominal spectral domain model which he stated was of “considerably more character” than the time model. The difference between the two is best illustrated with an example.
The first two charts below are monthly charts of the DOW from 1900 to present. The price wave at the bottom of the chart is the 4 year wave extracted using Hurst’s bandpass filter design derived from his spectral model.
The chart below plots the period of each oscillation (solid black line) over the past 100+ years. As you can see the period of the wave is constantly changing. The yellow dotted line is a smooth five oscillation running moving average of the wave period. The dashed red line is 54 months and the solid red line is the historical average of about 44 months. During the time period of Hurst’s analysis, the average period of the price wave was approximately 53 months, hence the 54 month nominal period. However the chart shows that from 1900 to 1945 and from 1975 to today the average period is shorter.
In order to calculate the average period of the 4 year wave to create a top-down nominal model one must decide how many oscillations to average and what time span is to be used, hence the title “Dynamic Nominal Model.” Fortunately the modulations are very, very slow which probably only require adjusting every 15 years or so.
Therefore if one defines his or her nominal model based on the average period of the 4 year wave from 50 years ago, then the cycles are running short. If one defines his or her nominal model based on the average of the most recent five or six oscillations of the 4 year wave, then the cycles are running normal.
Profitable trading,
William
11 thoughts on “Dynamic Nominal Model”
Very, very interesting post Willian!
If we use the Initial Ciclyc Model (ICM) instead of the Nominal Model (NM) on ST over monthly DJIA with the entry point in 1974, we see that he current average length of the 54 month cycle is 46,4 months although I’m not agree with the positioning of the latest 54 month’s trough.
It is a pity not to be able to post pictures on comments!
Thanks for sharing Willian
Best
José
Hi Jose,
I believe my comment conclusively demonstrates that the current nominal model is too long with respect to the long waves if you accept the premise that Hurst’s bandpass filter output accurately reflects the wave. The average period of the 4 year wave from the 1974 low to the November 2012 low is approximately 42 months. I don’t know if ST’s phasing algorithm is flexible enough to accommodate such a large deviation from its default values.
The 46 month average period figure you reported appears to be the average of 10 oscillations. I would suggest creating a custom nominal model with a 4 year wave period closer to 42 months. I would start with 52 months and then reduce it in increments of 2 months until it matches the filter output of 11 oscillations.
Good luck,
William
Hi William,
This is a great post. It raises more than a few issues regarding Hurst’s nominal model and what we abide by as Hurst analysts. This bandpass filter view is challenging my approach to Hurst cycles.
One question I have is: How many oscillations do you count for the 4 year cycle between 1974 and 2009? I have this as a full 36 year Hurst cycle (eg. Dewey’s 35.56). Yet there appear to be more than 8 full oscillations.
My interpretation of Hurst’s work as taught in the course is that we as analysts (as does Sentient Trader) work with cycles as they appear trough to trough, and that the sum of many different cycles will deliver respective major nested lows etc. So while the oscillations of the 4 year per your bandpass filter may indicate a low, this does not always coincide with the extreme low see in price. Examples are the period in late 70s and mid 80s, which are timeframes I talked about in my Jan. post.
One final comment. When I speak about cycles expanding or contracting, I refer to what is apparent with a cycle measured trough to trough. We know that cycles expand and contract as they are dynamic. For example, in the 1974 – 1982 period, there was a complete 9 year cycle that ran just under 8 years in length. This was the first 9 year cycle in a new 36 year cycle and was thus bullish (and it contracted to some degree – again per one’s view of this cycle trough to trough). I expect the same for the current 9 year cycle that started 2009 which should bottom more or less in 2016.
Thanks for sharing your work.
cheers,
john
Hi John,
I’ll try to address the issues you raised without being too verbose.
The use of bandpass filters doesn’t really change anything. They merely fine tune the period of the price wave, among other things. Generally speaking, it is mathematically impossible for the output of the filters to phase to all of the absolute lows. This results in the waves exhibiting a slightly more complex, albeit more accurate, harmonic relationship over long spans of time. Perfectly synchronized lows and a simple harmonic relationship by a factor of two among the waves is an artificial, but useful, simplification Hurst developed in the cycles course. We sometimes forget that his cyclic principles are “strong tendencies”, and not hard-coded facts. Remember in 1970 there were no desktop computers, Excel, or trading platforms! It was impossible for the average trader to plot a simple moving average let alone the output of a high resolution digital filter!
According to the output of the filter there were 10 oscillations of the 4 year wave from 1974 to 2009. In my opinion there is misconception about the waves in the market longer than the 4 year wave. On page 149 of Profit Magic Hurst clearly states that the 4 year wave is the longest wave in his price-motion model. Artificial graphical constructs can be plotted connecting every other 4 or 8 oscillations, but they do not necessarily constitute a price wave. Filter analysis shows that the waves longer than 4 years lack the regularity and cohesiveness of the shorter waves. There is one possible exception which I may post in the near future.
I agree that if the 4 year wave continues to oscillate at approximately its average period over the last 20 or 30 years, we should see the next low in 2016.
Hope this helps.
William
Hi William,
If Hurst thought the 4 year wave was the longest in his price motion model, why did he include the 9 yr wave on page 152? From the look of this chart, the 9 yr wave is clearly dominant. Is this the exception you spoke of?
Curt
Its like yoս read my thօughts! You seem to understand so much
about this, like you wrote the ebook iո it or somеthinɡ.
I believe that you just can do witɦ a ffew p.c. to
fߋrce the message house a bit, hoաeveг instead of that, that is wonderful blog.
A great read. І will definitely Ƅe back.
Thanks!
I will have a lot more to add in the future. Spread the word.
William
Quisiera saber mas sobre este tipo de análisis ciclico , como funciona , siempre me he basado en el análisis fundamental, chartismo , velas japonesas pero siempre llega a romperse esta ecuación facilmente, por eso estoy interesado en conocer este tipo de análisis , podrías decirme q debo hacer , muchas gracias
Estimado John
Lamentablemente para nosotros, hispanohablantes, la gran mayoría de los recursos que puedes encontrar sobre la teoría cíclica de JM Hurst está en inglés, incluido este blog multicontribuido por expertos analistas cíclicos internacionales. Hice un pequeño resumen en castellano sobre la esencia del análisis cíclico de JM Hurst q
perdón, … que puedes encontrar en http://www.analisisciclico.com
Un cordial saludo