|
PureBytes Links
Trading Reference Links
|
Nicholas and Rudolf ... plus others who have written
"... consistent statistical methodology ..."
Yes, how else are you suppose
to know or explain what is going on? If you use moving averages, how do
you explain what are they doing, i.e., is your system additive or
multiplicative? How does the compounding of "stuff" in your systems really
work?
Usually most "open systems" end up as black boxes because nobody really
knows what is going on, but we feel better because a few "well chosen
examples" to prove that it works are included.
There's a difference between "... consistent statistical methodology ..."
and choosing trading systems/models based on the statistical characteristics
of the time series. I personally feel more comfortable working with data
that I understand.
For example, 30 years of bean data times 7 contract months per year gives me
210+ different time series. These are easily sorted into bullish, bearish
and neutral series for example. Each grouping has it's own statistical
characteristics. The winner Vs loser spread trades for those series have
different profiles, different support and resistance levels and the trading
models are different.
The bottom line is that we have very hard working software programs. They
are suppose to be "labour saving" devices, that does not mean that they are
"no thinking" devices.
Hi Rudolf
Yes ... I model winners and losers separately also long and shorts.
"... PS: Can you give some practical hints on how to improve system
generalization? ..." no, because I would sooner fragment systems until I
reach a level where I can understand the detail of what is going on and the
inefficiency that I am trying to capture.
Best regards
Walter
PS: sorry I haven't have a chance to ask you about your XL workbooks that
you've been discussing with Angel.
----- Original Message -----
From: "Nicholas Kormanik" <nkormanik@xxxxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: Wednesday, January 26, 2000 1:54 PM
Subject: RE: Seasonals and technical analysis
| >
| >Walter writes:
| >Most of the discussion on the List is about ad-hoc methods and rules
| patched
| >together. There is no common methodology or statistical basis. Even the
| >"leaders" in the field TASC, Futures Mag, Metastock etc. work from a
"quasi
| >empirical" basis without an established method.
| >
|
| Another very interesting post, Walter.
|
| Your post begs the question: As serious traders, would it behoove us to
| strive for consistent statistical methodology as the basis for stock
| selection, and timing in entry and exit?
|
| Nicholas
|
|