Scribd
Upload
43 views4 pages

LETTER

FDGHBJKJL

Uploaded by

Vahid
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
43 views4 pages

LETTER

FDGHBJKJL

Uploaded by

Vahid
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Stocks & Commodities V.

3:1 (43-44): Letters to Technical Analysis

Letters to Technical Analysis

This column is our means of communicating with our subscribers and serves as an outlet for questions
and answers. Please feel free to write about anything that you would like to know more (or less) about or
pass along to the rest of us. We all run across trading techniques, services and products that have proven
to be useful to some degree under different market conditions. Without a constant source of new ideas
and subject matter coming from our readers this magazine would not exist. Each issue is designed to hold
at least one idea or clarify one subject that should justify your subscription cost.
Dear Jack,
I just finished a long conversation with Stuart Pahn of Dean Witter in Chicago in which we talked about
forecasting the price time series and he referred me to your article on MEM forecasting in the December
1984 issue (Technical Analysis of Stocks & Commodities). After reading the article, I noticed how
similar your work and approach is to mine. There are some basic differences though and I wonder if you
would like to know about them? Please ask Stuart if you (would) like to find his opinion on what I
present here.
The maximum order I use for forecasting is 2. That is "two" not nine or any higher order. This is because
a difference equation of second order is an oscillator: it is the discrete version of a second-order
differential equation which models a sinusoidal signal. A Hartley oscillator is second order because it
contains a capacitor and a coil. The market is the same because of time delays in trading and opinion
variables. Instead of making only one forecast as you do, I make forecasts on each separate oscillation
and then add it all together. That is, forecasts within spectral bands.
The equation is thus:

y(i)= b1(i)*y(i-1)= b2(i)*y(i-2)= process noise(i)


Try this on syntyhetic data where you have only one cycle plus noise! You will find that a very much
improved forecast can be made by forecasting each component separately and then adding the outputs.
This is because the true structure of the market is a sum of cycles caused by very-much independent
trading groups.
As a final note, please let me know if you have a copy of the spectral analysis program that has an
advantage of Fourier. I have done FFT for about 15 years and am curious that you may be low-pass
filtering the data before taking the spectrum. Did you intend to publish the MESP program in the

Article Text Copyright (c) Technical Analysis Inc. 1


Stocks & Commodities V. 3:1 (43-44): Letters to Technical Analysis

December issue? In any case, I would like to study your spectral program and send you my comments if
you would like to send it. I don't have the July 84 issue; should I order it?
Sincerely yours,

Jeff Gordon

Dear Mr. Gordon,


This letter is in response to your letter of December 19 to Technical Analysis of Stocks & Commodities,
In which you raise a number of issues regarding data forecasting based on past data cycle analysis. I
shall begin by noting that the Maximum Entropy forecasting method (December 1984 issue of Technical
Analysis of Stocks & Commodities magazine) is based on the Maximum Entropy method of Spectrum
Analysis which is described in the January and March 1984 issues of Technical Analysis of Stocks &
Commodities magazine. I am enclosing a copy of the March article "Optimizing the Maximum Entropy
Method" to illustrate certain features of this method. The January overview article on MEM and the July
issue containing the MEM fit and Spectrum code are available as back issue of Technical Analysis of
Stocks & Commodities magazine. (Ed.: Volume 2, 1984, 280 pages, $129.)
You are apparently using a forecasting method which consists of
A) Identifying all of the important cycles in the data based on past history,
B) Forecasting each cycle separately using a second order oscillator model, and
C) Adding the cycle forecasts to obtain a composite data forecast.
This seems to be a reasonably sound method for forecasting past cyclical data. However, the MEM
procedure has two advantages over this type of method, based on past experience:
* First, The author's MEM code allows one to predict both cyclical market forecast and trend line error
channels, which give a visual guide to the accuracy of the forecast versus time. (This is an important
legacy of the Box-Jenkins forecasting school.)
* Secondly, this MEM code is implemented by optimizing an objective criterion function and requires
little subjective knowledge or pattern recognition in order to obtain forecasts, unlike the Box-Jenkins
methodology which requires considerable learning time for successful results.
Finally, I would like to add that the MEM technique is inherently a high resolution method of spectrum
analysis, i.e. one can resolve spectral peaks that are close together with less data than is needed for
Fourier Analysis. However, with noisy time series data such as occurs in analyzing daily commodity
contracts, it appears to be necessary to preprocess the data in order to obtain reliable, high resolution
MEM spectra. (This is the essence of the enclosed March 1984 Technical Analysis of Stocks &
Commodities article.)
I hope this discussion has proven constructive, and I would encourage you to publish the results of your
forecasting methods in Technical Analysis of Stocks & Commodities magazine for others to appreciate.
Yours Respectfully,

Anthony Warren, Ph.D.

Article Text Copyright (c) Technical Analysis Inc. 2


Stocks & Commodities V. 3:1 (43-44): Letters to Technical Analysis

Technical Analysis Staff Writer

Dear Sir,
We note with interest Mr. John Sweeney's article in your July/August issue regarding the COMMODEX
system. In his closing remarks he mentions that his detailed research is limited by his lack of back issues
of the COMMODEX bulletins.
We have enclosed a copy of our brochure describing our ACCUDEX trading system which we supply on
a disc ready-to-run on Apple computers. It will duplicate all the COMMODEX trading recommendations
for as long back as the user has data discs. In addition, it has the unique feature that the user can adjust all
the moving averages in the COMMODEX program to whatever number of days he wishes and use these
in his daily trading recommendation print-out. They can be set to be different for every commodity
contract and can be tested automatically by using the powerful historical analysis section of the program.
It occurred to us that our program might help Mr. Sweeney in his excellent and thorough research and
also other readers who are familiar with the COMMODEX system which is probably one of the oldest yet
most reliable and consistent system available on the market. ACCUDEX is reasonably priced at only
$600 prepaid overnight to anywhere in the USA.
Yours sincerely,

Raymond E. Green

Technicom, Inc.

Dear Mr. Hutson:


Received your December sample issue but hesitate subscribing due to the following points...
The magazine seems to be strongly biased towards financial futures and stocks far out of proportion to
the actual open interest figures. Coverage of non-financials, i.e.: agricultures, seems to be just about
ignored.
I have traded every market and I find the financials far more difficult to trade the the Ag's. Technical
action, trends, patterns, cycles and seasonals all work much better in the Ag's. In addition, the Ag's are
less volatile (overall) and therefore losses can be better limited plus stops are not hit as often.
Also, RE: Page 5 Liquidity article. Many of the rankings are totally wrong! In particular, liquidity in
copper is very poor as far as fill prices go...ask any broker or trader about copper and it will be ranked
near the bottom of the list. It has high open interest - but volume is due to a few large traders who whip
the price on stop orders and fast market orders. Also, cotton is even worse than copper - both should be
near the bottom. Bad fills are the "order of the day" in cotton and have been so for many years. Also,
T-bills and sugar are ranked far too high. Also, live cattle and wheat (Chicago) are far too low in the
rankings.
I trade cattle and wheat a lot and find the fills could not be any better. Slippage is rarely more than one

Article Text Copyright (c) Technical Analysis Inc. 3


Stocks & Commodities V. 3:1 (43-44): Letters to Technical Analysis

tick and never more than 2 ticks in cattle (5 points) or 1 tick (1/4 cent) in wheat - this works out to $8 and
$12.50 respectively. In fact, most of the time on both stop and market orders I will pay 1 tick ($4) in
cattle and get the exact last sale price in which works out to zero slippage. The fills in soybeans almost
always result in 1/2 cent slippage and in T-bills one or two ticks ($25 to $50). Yet by some strange math
beans and bills are shown to have about seven times better fills than cattle or wheat. That is simply not
true! I know from lots of experience!

D. Green.

Editor: I agree. The chart is based on daily and weekly data, for slipage to be correctly reflected you
must examine intraday prices. I hope you will agree that a speculator would do best by not trading
outside this "Liquidity" list.

Editor

Gentlemen:
Some time back I purchased Vol. 1 programs for Apple ][ and tried to use the programs today. I am not a
member of CompuTrac.
How can I use the studies? My data is from CSI in CSI format.
Sincerely,

R. E. Walker

Editor: The programs were made available on disk to save the user the trouble of typing and no input or
output (graphics) was provided. We will soon be distributing a programming system for Apple and IBM
(later) computers that will run these studies and encourage new study development by our computer
orientated subscribers.

Article Text Copyright (c) Technical Analysis Inc. 4

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy