|
PureBytes Links
Trading Reference Links
|
Rters,
By now everybody interested in this technique would have loaded
a copy of the indicator.
Let me explain what the program does. The program builds an internal
swing table dynamically based on the input parameter QTime and QPrice.
Then depending on the constraints provided will combine the pivots
using a set of ratios in this case 1.618 and 2.618. The default will
only combine pivots of the same type, top to top and bottom to bottom.
For those who wants to understand the logic of what it does before using
it here is a short pseudocode of an unconstraint projection :-
for each pivot in the pivot list(outer pivot)
consider all pivots later than the above and of the same type(inner
pivot)
i.Multiply the period between the outer pivot and inner pivot
by a set of ratios
ii.Add (i) to the start of the outer pivot and tally the result in
a table.
display the result of the table
Now this is very important. The quality of the projections is influenced
by the pivots set used. Pivots which are unrelated to each other when
combined with each other will generate substandard result.
Inserting the indicator on an arbitrarily selected market with existing
settings will generally result in bad predictions. The variety of inputs
provided serves to allow the user to influence the pivots sets use.
On the daily Aussie SPI, projections seems to work better with calendar
day mode. This would tend to suggest that the influence of the dynamic
cycles projections seems to exert itself on this market on a continuous
basis, .eg it doesn't stop for weekends and holidays. Although on
different market the outcome could be different.
A QTime of 15 which translate to approximately 3 weeks of filter to
each pivot, would expose the major pivots. And these major pivots are
related to each other in the 'natural sense'.
It is true that all pivots are related to past pivots. But it is not
true to say that a pivot is related to ALL the past pivots. Therefore
without using the correct set of pivots with the exclusion of unrelated
pivots any projections made should be doubtful.
Looking at the gif posted earlier you would be able to see that the
histogram levels are quite sparsed. At first glance this would indicate
that the level of noise have been substantially reduced. Reducing noise
without bringing out the good ones is no good either. If you
look carefully going back several years, many of these levels actually
coincides with good CIT in this market. Because the histogram levels
have been
filtered any levels above a certain threshold can be considered as high
probability time for change in trend. Most people would make the mistake
in this instance of looking only at the tallest bar within the screen.
I have illustrated several astrological techniques in several other
posts
that can be used in predicting market directions such as Zain-Bradley,
Intraday Bradley and George Bayer's Five Fold Horoscope. Although most
of my work have been done on the local market these techniques are
suppose
to be for the US markets!! I would also want to stress that
non-astrological
techniques if used properly would also provide good result.
The dates projected also agrees with the astrological dates
the most recent being Sep 11th, what a coincidence? Bottom line is
both type of methods work except the astro techniques gives you
the pivot type as well(above chance).
Remember these are techniques of analysing market directions it is not
a trading plan. How successful these technique is to anyone is very
dependant on the trader's personality.
BTW, I forget to mention that this indicator can also be used for SC but
it needs to be version 4. Version 5 of TS will not work because it's
32 bit and all 16 bit DLL is not compatible.
Clement
schong wrote:
>
> RTers,
>
> Anyone who is interested may download a copy of the Dynamic
> Time Projection indicator from :-
>
> http://home.mira.net/~schong/dyntime/dyntime.zip
>
> Attached is a gif of how it looks like on the Aussie market.
> The indicator employs a very common and well known timing
> algorithm.
>
> I have another indicator which I call Dynamic Time Modeller
> which allows any time cycles to be projected from any pivots whether
> it's adjacent or alternate pivots, past or forward. I will post this
> indicator to the forum in due course. This indicator allows
> any timing model to be built.
>
> This indicator takes all combinations of top to top and
> bottom to bottom pivots as determined by the QTime and QPrice
> parameter and projects them using a ratio of 1.618 and 2.618
> forward. The ratios are fully customisable and weightable but
> I won't get into it now.
>
> All pivots that has at least QTime bars to the left and right
> of it and have at least QPrice points or percentage(if less than
> 1) would qualify as a valid pivot.
>
> To use calendar days mode specify CalrType=2, any other value is
> interpreted as trading bars.
>
> Let all other values to default. To look at the forward
> predictions you are required to padd dummy bars into your data.
> Make sure the dummy bars have negative volumes and negative open
> interest. The indicator need these values to properly recognise
> the real end of bar as it is has been written to work with an
> automatic Data Extender to overcome the restriction of TS not being
> able to plot forward. Failure to put in the negative values will
> result in the incorrect build of the swing pivot table.
>
> A brief descriptions of the other inputs :-
>
> Project :- How many bars to project from the last pivot
> BarOrPvt:- Control the units used by MinBar, Maxbars and MaxBack
> as either number of bars or number of pivots
> MaxBacks:- Maxmimum number of bars/pivots to look back. A value
> of -1 means use all available data.
> MinBars :- Minimum number of bars/pivots between two pivots
> MaxBars :- Maximum number of bars/pivots between two pivots. A
> value of -1 means no restrictions within the context
> of MaxBacks.
>
> Download instructions :-
>
> 1. Unzip DynTime.zip into your local drive
> 2. Copy fwdbars.dll and dyntime.dll to your TS "\omega\prog" directory
> 3 Transfer the Dynamic Time Count from the DynTime.Ela archive file
> into TS
> 4. Insert the indicator and don't forget to padd your data out.
>
> Clement
>
> ---------------------------------------------------------------
> [Image]
|