|
DLL's for
Maximum Entropy Method (MEM) DLL ( Cycle
analysis)
USAGE:
The MEM code can be used to input the main period into an
oscillator (as the look back period), to calculate the phase of the price data
series, as a look back period for a channel breakout system...
The MEM prediction by itself can be astonishing as well as
very deceptive, according the the spectrum stability over time.
Usually, you will need the main period and eventually the
Power value to use this code into a trading system.
As for all of our DLL's accepting data series as inputs
to the code, you can load anything you want into the input array( raw prices, or
indicators). These values can even be changed during time, as the analysis is performed
again on the data series loaded into the input array. This gives more
flexibility (as
you can change the data series by a single line of Easy Language), and the price to pay is
some lack of speed.
EXAMPLES:
Printed using TradeStation™ 4 by TradeStation Technologies
Click on the above thumbnail image to get the full
size picture.
MEM DLL applied
to the raw close.
The very bottom chart represents the Power spectrum
for the first and 2nd frequencies
The upper part shows the values of the period of these frequencies.
one may remark that when the power spectrum shows high reading,
cycles may exist in the original data serie (white rectangles).
A linear trend usually show high values of the period and a low reading
of the Power spectrum (between the white windows to the left).
A trend with a cyclical component shows high period readings,
but also a power spectrum of some importance (rightmost white window).
Of course, these descriptions are somewhat arbitrary,
and they are there to explain the concept.
The main period graph (blue line) is somewhat jerky,
and you will certainly need to smooth it by using a short term moving average.
rinted using TradeStation™ 4 by TradeStation Technologies
Click on the above thumbnail image to get the full
size picture.
MEM DLL applied here
to the RSI oscillator.
.
The RSI is displayed in subgraph2.
The bottom subgraph displays the Power Spectrum (4 main frequencies) as colored
histograms
and the related 4 periods of importance color matching (lines on the
top of the subgraph3).
Even with this one, the cyclical behaviour appears on the same areas of the
original bar chart.
The main period (blue line) has been smoothed with a 3 bars moving average.
Both graphs use the same 10 min SP500 future data.
Ordering: see the ORDER section on this web
PRICE is 240€
SEE OUR SPECIAL PRICE OFFER FOR ALL OUR
TRADESTATION DLL's.
EC users must ad 19.6%VAT or provide their international VAT
number when ordering.
Please provide your name , address, TradeStation or Supercharts Realtime
Block Number.
TradeStation Code ©1996-1998 Philippe Lembo- HLG Gestion &
Pierre Orphelin-Sirtrade International .
Sirtrade International 110, avenue du President Wilson -F93100 Montreuil. FRANCE.
|
 |
MEM analysis
|
 |
MEM is a short name for Maximum Entropy Method, based on
the JP Burg algorithm.
Also know as MESA (commercial name of the adaptation made and sold by John Ehlers)
It's use in technical analysis has been popularized by Anthony Warren and John
Ehlers, contributors to Technical Analysis of Stocks and Commodities Magazine.
A lot have been said about this code and its ability to extract cyclical components
from a price data serie, and the weakness of the FFT vs MEM code has been widely spread
by these authors.
Nevertheless, extracting tradable cycle information from price data is not an easy
piece of cake, as cycles vanishes very quickly, change and are often combined into complex
wave shapes.
This MEM algorithm extracts the five main cycle
components, the Power spectrum of each five fundamental frequency (sorted by magnitude),
and has the ability to be used as a predictive tool by projecting in the future using the
current MEM coefficients.
|
 |
|
I
|
