|
PureBytes Links
Trading Reference Links
|
----- Original Message -----
From: "rudolf stricker" <lists@xxxxxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: Thursday, February 07, 2002 12:49 PM
Subject: Re: adaptive systems - how to define appropriate time windows ?
>
> John,
>
> On Wed, 6 Feb 2002 14:40:52 -0500, you wrote:
>
> >I'm not sure what you're looking for. What are you trying to adapt to?
>
> Sorry for the confusion... I shouldn't have mentioned my system to be
> adaptive, because my understanding of adaptiveness may be a little bit
> different from yours you gave in your second posting. In a general
> sense, _any_ system _adapts_ some kind of information to other
> information, eg by some indicators, but this is not an "adaptive
> system" for me.
>
> My question was:
> >What criterions (to read from the equity's historical data) could help
> >to define "the best" time period?
> At this time, I use a fixed-size time window, but I'm looking for
> appropriate measures useful to control dynamically my time window.
>
> > If you are trying to adapt your system to the current cycle length look
at some
> >of Dr. John Elhers indicators that measure cycle length.
>
> Cycle length (of option prices) is seemingly not a very helpful
> measure for my data because of the rather short (and heavily changing)
> cycles that i see in my option prices.
I would look at the cycle length of the underlying, not the option.
So concentration on one or a
> few cycles would not provide enough statistical material to do a solid
> system modeling job. - Or would you recommend to use a constant
> (rather high) number of cycles to represent the time window?
>
> >If you're trading
> >options why not make the system adaptive to current or historical
> >volatility.
>
> Do you recommend to use "market volatility" to control my time window?
> ... i cannot see, that this would make so much sense... high
> volatility => small time window? - Volatility imo has much too much
> "volatility", and my time window should be more stabile to provide
> consistent system results over time.
Some people do use volatility that way. However I look to use it to
determine whether to be a buyer or seller of options. If volatility is way
above some normal range I like to sell volatility and if it's below some
normal range I will buy volatility.
>
> >Personally I don't believe a "best" time period exists. That is
> >too restrictive for me.
>
> That's why I'm looking for parameters to control my time window
> dynamically.
>
> >A strongly trending stock will probably have a long
> >cycle length while that same stock may have a very short cycle length if
it
> >is in a trading range.
>
> So you recommend the cycle length of the underlying (ie DAX in my
> case) if I understand correctly? - Seems to me like a good idea to
> start... What (eg MS) indicator should I use to measure the current
> ("macro"-) cycle length to control my time window?
I'm not completely comfortable with cycle length ideas because you're always
trying to use the past to predict the future. On the other hand that may be
the best approach anyway. I would like to look at setting up some sort of
statistical test to determine the likelyhood that the most recent cycle will
be profitable. Sort of like generating a confidence factor number that I
believe in. When the confidence factor is high maybe I buy more contracts
and less contracts when low. Someone on this list posted a Hilbert transform
that was supposed to measure cycle length. I will look and see if I still
have it. I toyed around with it for awhile and may try to work on it again
in the future. I looked in my Metastock folder and didn't see the function.
I would think it could be found on the Stocks and Commodities magazine
website or on John Ehlers website. If I find it I'll post it.
Regards
John Manasco
>
> mfg rudolf stricker
> | Disclaimer: The views of this user are strictly his own.
>
>
|