 |
 |
 |
 |
|
 |
 |
21
It fits a trigonometric curve to each occurrence of the cycle in the data. The amplitude of the measured cycle
is compared to the amplitude expected by chance. This comparison gives a direct means, through normal
probability measurements, to arrive at a mathematical measure of genuineness.
The measure of genuineness or regular occurrence of a cycle will be highest when there is stability in both
the amplitude and timing of a cycle. For this reason, the Bartels test is the most valuable of the statistical
tests for examining data for the regular occurrence of a cycle.
The Bartels test is negatively impacted when there are less data and thereby fewer cyclic occurrences.
Results are often poor (less reliable) when there are fewer than 700 data points, and diminishing results can
be seen when there are fewer than 1800 to 2000 data points. Further, cycles longer than one-third of the
data often show similar, poor Bartels scores. Better overall results are generally gained with more data, at
least 2000 data points (the Techsignal™ default value), and a greater number of cyclic occurrences up to a
maximum of between 5,000 to 8,000 data points. At that point with so much data, phase and amplitude
measures may begin to be less applicable or in synch with the current time frame and most recent data.
However, in these situations Techsignal™ allows you to re-phase or re-synch such cycles. Typically this
situation occurs with the longer cycles which can get out of synch with current data more easily. So, by
eyeball, Techsignal™ allows you to move or adjust slightly the position of the cycle to be more in line with
the current price data, while keeping the period and amplitude the same. This adjustment is made by
moving the most recent significant high or low in the individual long cycle a few periods forward or backward
(using the triangle icons of the left toolbar), to become more in line with the most recent high or low in the
price data. Such movement may put the cycle out of synch with earlier historical data but often more in line
with the more recent price data.
The concept is that events too many years ago or periods ago that shaped a cycle’s behavior then may be
less important in the current time frame. Once such example is the corn market in commodities which in the
1800s was a 130-day cycle crop until the advent of the reaper, fertilizer, irrigation measures, and hybrid
seed when corn became a 90-day cycle crop. So, on a weekly bar plot, the character of a cycle nearly
10,000 periods ago (190 years) may be less valuable, out of synch or not even applicable in today’s time
frame.
Amplitude and Phase (characteristics of a cycle)
The default amplitude is the expected number of y-axis points or units that the market or data should move
from the zero line (see scale on right side of graph). The amplitude of a cycle is relative to the amplitude of
other cycles but not necessarily to market points itself. Techsignal™ uses all of the data in a data series for
determining the amplitude of the cycle.
The Phase is determined by the date of the first occurrence of a high in the data. A phase of 8 for example
would mean the first high in the data occurred on the 8th data point in the data series. Techsignal™ uses
only the last third of the data of a data series in determining the phase of a cycle, and has found this method
is the best approach for synching the data behavior of a cycle to the most recent occurrences of a cycle.
Adjustments to Amplitude and Phase
Both Phase and Amplitude can be adjusted. The reason for allowing the researcher or user to adjust the
phase or amplitude is that sometimes shorter cycles have little impact on larger cycles. Thus, if you want to
see the presence of the shorter cycles in a synthesis of cycles, the amplitude of the shorter cycles can be
increased. Enter a number in the Amplitude adjustment cell of the Grid.
Phases can also be adjusted by shifting the entire cycle a bit forward or backward in time to better deal with
current conditions. As this minor shifting is done by eyeball, you have to be careful you don’t become guilty
of curve fitting where you move the cycle so much as to make the cycle meaningless. See Bartels test
above for more discussion.
The phase can be adjusted in two ways. One way is by typing a number in the Phase Adjustment column of
the grid, in the appropriate cell (that takes some practice to learn what numbers to try); a second method is
by using the four triangle icons on the left-hand Toolbar to move the cycle. The double triangle pointing
right moves the cycles plot 10 data points to the right, and the left double triangle icon moves it 10 to left.
The single triangles move the cycle plot one data point in the appropriate direction. A phase can not be
adjusted an amount that is greater than the period of the cycle. The program will not allow any adjustment
that is not acceptable. Care should be used so that you don't curve fit.
 |