You are on page 1of 17

1 | Page

Secrets of financial market forecasting

Introduction to the modified Hurst nominal model

There has been much debate about the usefulness of cyclical analysis in the financial markets during recent
years. Most remain skeptical until a healthy correction begins to unfold in the markets and I speak of the
stock market in particular. People temporarily begin to gain interest in cyclical analysis in order to find some
way to explain the ‘unexpected’ decline that is unfolding. All is once again forgotten until the next significant
correction occurs and the cycle continues. This analytical approach has found better hospitality in the
commodity market considering that commodity cycles are more evident and do not experience much high
translation and hence they are obvious. The irony is that market speculators and academics accept that the
corrections are most likely due to cyclical forces but refuse to accept that the advances are due to the same
forces as they continue to believe their obsolete models of fundamentals driving prices rather than mood
driving prices. In this short manual I will share with you my discoveries in the field of cyclical analysis in
relation to the financial markets. This paper is not meant to be for sale rather it is meant for the few
individuals that I believe are worthy of aquiring this knowledge in hope that they use it responsibly.

J.M. Hurst’s work on cyclical analysis in terms of the financial markets is certainly one of the most
comprehensive explanations in print. I have expanded upon J.M. Hurst’s ideas myself by adding several larger
cycles that are longer than the infamous 54 year cycle discovered by Nikolai Konratieff in the 1920s. The new
cycles discovered are indeed present in historical prices of stocks and commodities as the principle of
commonality would suggest. I also expanded on J.M. Hurst’s work by added a phenomena I call ‘a similar
cyclical circumstance’ which looks at cycles within cycles or ‘wheels within wheels’. This new discovery, after
some tweaking, allows us to obtain unsmoothed projection lines based on past price history that correlate
with the future by over 90 percent. The similar cyclical circumstance phenomena also allows us to forecast
with 90% accuracy when fat tails are likely to occur in the market under consideration provided we have
enough price history. The similar cyclical circumstance phenomena also answers the long debated and
problematic principle of variation which is the sudden change of cycle length relative to the current average
period. In fact most of our forecast will be based on an approximate of the length of the cycle we are
attempting to forecast rather than the recent average length of the cycle. This will be looked at in greater
depth once we proceed through this manual.

Let us now begin by first stating what the average period of the cycles being studied are. Hurst has presented
cycles as short as 5 days to as long as 18 years. I believe he made no mention of the 54 year cycle that Nikolai
Konratieff discovered which I certainly cannot take credit for. The great thing about cyclical analysis is that
the largest cycle in the model is related to its one larger cycle by a factor of two or three hence if one has
the data he can easily determine what the larger cycles are based on FLD analysis , VTL analysis and spectral
analysis. I believe I can take credit for discovering three larger cycles than the 54 year wave. Those are the 162
year, 324 year and 972 year. If one knows what cycle we are in, taking into context the larger cycles as large
as one could go, he will know what lies ahead for the future of nations, wars, economics and markets. An
example would be: if we are currently in the second 54 year cycle of the first 162 year cycle of the second 324
year cycle of the 972 year cycle. One only needs to find that same cycle in the past to know with extremely
high accuracy what lies ahead in terms of prices and events. This is why aquiring the skill of isolating cycles is
extremely important to practice this lost, forbidden and mystical art.
2 | Page

The table above is a visual illustration of the modified and simplified nominal cyclic model. The largest cycle
that I am aware of is the 972 year cycle. I would like to state that these cycles have nothing to do with
‘astrological’ cycle periods. The cause of the cycles is not of interest to us at this current point in time. What
we need to be concerned with is the fact that these periods are averages of cycles that are very much present
in the price histories of commodities and stocks. What is also worth emphasizing is that these cycles’ presence
in the price chart is due to the mood of the public, world, and speculators which are simply reflected in the
barometers of social mood which are the stock market or commodity market indexes.
3 | Page

Methods of cycle isolation

The analytical process begins with the isolation of component cycles of the price history. It is important to
note that this process is the most important process in this analytical approach and MUST be done correctly
in order to gain accurate results. We will go through this process step by step by isolating the cycles in the
Gold market as an example. It is important to isolate the cycles as far back as the data goes rather than simply
on 180 price bars as presented by J.M. Hurst. If variation in cycle length causes confusion, the price history can
be dissected into portions in order to aid the cycle isolation process mentioned below. The reason for phasing
the entire price history is to be able to use the phenomena of similar cyclical circumstances. If one wants to
obtain a projection line of the current 20 week cycle one will have to take its position in context of as many
of the larger cycles as the data allows and find this 20 week cycle’s counterpart in the past. This 20 week
cycle (could be several rather than only one) found in the price history will be used as our input to obtain a
projection line to forecast direction, a model to forecast volatility and price targets based on amplitude ratios.

Steps of cycle isolation

1. Obtaining accurate price data of weekly/daily duration as far back as possible


2. Utilize spectral analysis and wavelet technology in Timing Solution software in order to find the most
prominent cycle that is related to the modified nominal cyclic model
3. Obtain an approximate for the other cycles in the modified nominal cyclic model based on
harmonicity
4. Put together the initial cyclic model of the instrument under consideration
5. Begin the isolation process of the larger cycles by starting from a prominent low making sure that the
harmonics of the large cycle are present and evident using the harmonic wave tool present in Timing
Solution software
6. Step down to the smaller cycles and use their approximate length obtained from spectral analysis to
aid in the isolation.
7. Make sure the harmonics of the cycles isolated are present utilizing the harmonic wave tool in Yiming
Solution
8. Continue the process till the chart is completed ‘phased’
4 | Page

Phasing example: Gold market

Step 1

Obtaining an accurate price history of the Gold market is not challenging. I have obtained my price history
from http://stooq.com. This website has a suprisingly large database of historical price data for many
commodities, currencies and most global stock indices. I would very much recommend this source since they
have data that goes back centuries in some cases. We will look at first look at the weekly chart of Gold
followed by the daily plot.

Step 2, 3 & 4

The picture above is the spectrum module’s results of a weekly chart of Gold starting from 1971. You can
replicate this result in Timing Solution’s spectrum module with the settings above. The most prominent peak
in the periodogram has a period of approximates 7.5 years. This periodogram peak corresponds to the
nominal 9 year cycle. Here we are given a hint that the cycles are most likely running shorter than average in
5 | Page

the prices of Gold. After validating this cycle using the wavelet module we will begin to determine the
approximate lengths of the other cycles based on the principle of harmonicity.

The wavelet diagram above confirms our interpretation of the periodogram. since this cycle has been active
from the beginning of this time series. Let us now put together what Hurst called the initial cyclic model

Nominal cyclic model periods Actual cyclic periods


18 years 14.98 years
9 years 7.49 years
54 months 44.93 months
18 months 14.98 months
40 weeks 32.56 weeks
20 weeks 16.28 weeks
10 weeks 8.14 weeks
5 weeks 4.07 weeks
6 | Page

Step 5, 6, 7 & 8

In the following charts the 18 year, 9 year, 54 month and 18 month cycles will be isolated. I have checked the
smaller harmonics of the 18 month cycle to make sure the right troughs are selected. I chose not to present
such detailed analysis here in order to save time and space. I write this manual assuming that you have
acquired the skill of phasing analysis. For more detailed explanation on phasing kindly refer Hurst’s cycles
course.

The chart above is a daily plot of the spot price of Gold. The horizontal lines are of the length of the nominal
9 year cycle which is currently running short at approximately 7.49 years as per the peak in the periodogram.
Notice how accurate the cycle length extracted is. All the horizontal lines dipicted significant troughs in price.
Before arbitrarily labelling the toughs based on this rough estimate we would have to make sure the harmoics
are present within the isolated cycles. We know that the nominal 9 year cycle should ideally be divided in two
54 month cycle that should each have three 18 month cycles. Let us now take a look at each cycle individually
in order to determine the validity of the potential troughs that were isolated with the aid of the initial cyclic
model. Other horizontal lines of the lengths of the other cycles in the initial cyclic model should also be used
in order to isolate the subdivisions of the 9 year nominal wave. I enjoy using the harmonic wave tool in timing
solution rather than horizontal lines which are the preferred course for people who are new to phasing
analysis
7 | Page

The chart above shows the phasing analysis of the first nominal 9 year wave off the low in 1977. We can see
that the 9 year cycle subdivided into two 54 month cycles each of which subdivided into three 18 month
cycles. We can now move on to the next cycle.

The second 9 year nominal wave’s subdivisions are quite clear as well. Notice how nicely the second harmonic
wave depicts the trough of the first 54 month cycle within this nominal 9 year wave. The first 18 month cycle
of the second 54 month cycle is longer than average but is a good fit in context of the larger cycles. It is
worth mentioning that the 9 year cycle trough formed a straddle since it was not the extreme low within the
cycle, Identifying a straddle is not difficult if one looks at the subdivisions prior to making any conclusions on
the location of the trough of a particular cycle. We can now move on to the next cycle.
8 | Page

The chart above shows the first 54 month cycle of the following nominal 9 year wave. Notice how nicely the
third harmonic of the harmonic wave tool depicts the troughs of the 18 month waves within this 54 month
cycle. Let us now take a look at the bigger picture to see how this 54 month cycle fits in context of the 9 year
cycle.

As evident on the chart above, the second harmonic of the harmonic wave tool of the 9 year wave nicely
depicts the trough of the 54 month wave. It is worth mentioning that the 9 year cycle trough formed a
straddle once again and was not the extreme low within the cycle.
9 | Page

The nominal 9 year wave presented above has subdivided ideally into two 54 month cycles. The trough of the
first cycle is exactly at the midpoint of the cycle. Each of the 54 month cycles has subdivided into three
nominal 18 month waves.

The final 9 year cycle has the trough of the 54 month cycle exactly at the midpoint of the cycle. The
harmonics of the 54 month cycle are ideal where each of the 18 month cycles are almost equal in length.
Spectral analysis cycle isolation would have led to profitable trades in this cycle since there was little variation
within this 9 year nominal wave.
10 | P a g e

Obtaining the indicators

Now that that the cycles have been isolated it is time to obtain our indicators which include

1) A projection line based on past price history


2) A detrended zigzag neural network model to help us depict turning points
3) An up/down percent neural network model to help us in knowing which parts of the cycle are most
profitable
4) A volatility neural network model to tell us which parts of the cycle are likely to be most volatile

One critical factor needs to be adressed before calculating our models and that is the length on which all the
models above will be based. Due to the principle of variation the past price history on which the four models
are based need to be adjusted to the approximate length of the cycle being forecasted. There are several way
one can approximate the upcoming cycle

1) Based on the current average of the cycle being forecasted the shorter the sample number the better
(last 3 cycles, 5 cycles, 8 cycles etc)
2) Based only on the last cycle (The is sometimes chosen due to the fact that periods of cycles change slowly
with time)
3) Based on the ratio of the same cycle to the one that preceeded it or the mean ratio (last X cycles) under
a similar cyclical circumstance although we need a significant amout of historical data to chose this
method.
 20 week ratio – find its its counterparts ratio with its preceding cycle or mean ratio in the previous 54
month cycle
 40 week ratio - find its its counterparts ratio with its preceding cycle or mean ratio in the previous 9
year cycle
 18 month ratio - find its its counterparts ratio with its preceding cycle or mean ratio in the previous 18
year cycle
 54 month ratio - find its its counterparts ratio with its preceding cycle or mean ratio in the previous
54 year cycle
 9 year ratio - find its its counterparts ratio with its preceding cycle or mean ratio in the previous 162
year cycle ratio
 Etc.
4) Find the ratio of the minor cycles of the current cycle with the cycle’s most recent counterpart’s minor
cycles then multiply the cycle’s counterparts length by the ratio in order to estimate the length of the
cycle being forecast (allow a lead/lag window incase the phasing is not 100% accurate) This is done to
obtain the highest possible correlation with what has already passed of the current cycle or just before its
beginning. This is the preffered way to obtain an approximate of the current cycle’s length. (One can
code the computer to automatically alter the cycle’s most recent counterpart’s length and project it from
the appropriate low with a certain lead/lag window in order to obtain the highest possible correlation
with what has passed of the cycle or just prior. The rest of the cycle will remain in high correlation with
the projection line (superposition of the cycle’s counterparts of modified length) as presented below
11 | P a g e

Synced minor cycles

Correlation of the rest of the cycle


12 | P a g e

After choosing the appropriate method in determining the approxamite length of the upcoming cycle. It is
now time to start cliping parts of the price history. We first need to determine which cycle we are in to know
which parts of the price history we are interested in. We know from the phasing analysis above that we are
currently in the first 18 month cycle of the first 54 month cycle of the first 9 year cycle of the 18 year cycle.
Now that we know where we are in the heirchy of cycles we would go back in history in order to clip the first
18 month cycle of the 18 year cycle in the past (similar cyclical circumstance) this can be done by setting
Tmin/Tmax from the beginning to the end of each individual cycle then right click and save the chart as text.
We will use the cycles extracted as an input for our indicators. Keep in mind that the cycles isolated will be of
slightly different lengths due to the principle of variation. We should now do the following steps:

1) Extend or shrink the cycles evenly to match our approximation of the length of the cycle we are
forecasting (if isolated correctly once overlayed will be very similar after unifying the length)
2) Form an algebric sum of the cycles (which should now be of equal length) which will be used to aid us in
determining the direction of the market under consideration
3) Append the individual cycles into one continious series
4) Load up the algebric sum as an additional chart in Timing Solution or load it up in the intermarket analysis
portion of ‘pattern’ for statistical analysis or back testing. Make sure you select the coincide index
5) Load the appended series into Timing Solution and extract the cycle with how many overtones you wish
even if there is no peak in the periodogram then copy it into the clipboard
6) Run the neural network module and select the detrended zigzag, volatility & up/down percent as outputs
Use what is in the clipboard as an input and train on all pervious bars.

Let us go through the process step by step


13 | P a g e

Step 1

First we need to decide which method we will use in order to approximate the length of the upcoming
cycle. In this case I will base it on the length of the last cycle. Now let us extend/shrink the price history to
our desired length using the extend program

Do the steps above for how many ever cycles you have extracted (in our case it is two). Don’t be mislead
by the extend button which can either extend a series or shrink it depending on the time difference
between the one set by you and the original series.
14 | P a g e

Step 2 & 3

After forming a superposition for a projection line and after appending the series for neural network
models (the append button will do the combining in alphabetical order by the way) we can now begin
using Timing solution once again.

Step 4
15 | P a g e

I have loaded up the algebric some from the ‘Add’ button as another chart. In order to do that kindly find
the chart button next to the print button on the second tool bar. Notice the correlation with the annual
cycle. This gives us reason to believe that our isolation was accurate and our estimate of the upcoming
cycle is likely accurate as well. Another way to find out if our isolation is accurate is by overlaying the
altered cycles over each other to see if they are highly correlated. Considering that the cycles were under
a similar cyclical circumstance and now have the same length the correlation is likely to be quite high
between them.

The chart above shows both cycles overlayed. Unfortunately I am unable to obtain a correlation
coeficcent for the two but it is obvious that their correlation is quite high. This is the kind of correlation
that can be expected between the superposition presented earlier and future price action provided that
the approximate length of the cycle isn’t too far off. I would like to stress that if one has the data then the
ratio method of obtaining the approximate length is the best method in my opinion although further
research will be conducted.
16 | P a g e

Step 5

We are only using two periods in our calculations hence spectral analysis was unable to pick up the cycle. We
know that it is present so we can select the period manually and move the selected cycle with the amount of
overtones required into the clipboard. Open the neural network module and choose the cycles as the input
and which ever indicator you chose as an output. The accuracy of your results will be astonishing for the
following reasons

1) We chose the cycles that are most similar in context of the larger cycles hence the trend component is
very similar
2) We unified their length and matched it to the approximate length of the upcoming cycle to make sure
the calculations are as close to accurate as possible
3) The projection line obtained syncs up the smaller cycles of the cycles under a similar cyclical circumstance
in order for them to match the smaller cycles within the cycle we are attempting to forecast hence the
correlation coeficcient should be quite high
17 | P a g e

Turning point indicator

Up/down percent & turning point indicator

Sychronozed zones in blue boxes

Ahmed Farghaly

You might also like