[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Metastockusers] Re: Stationarity and Real World application of statistics



PureBytes Links

Trading Reference Links

ramble...don't worry about it, we may just figure it all out
here...what's a few extra K in my inbox anyway, bits are cheap.....

Yeah, I am personally not a trader at all.  I am just a developer.  I
am trying to wrap my head around it all and start from the best point.
To me, this starts with indicators.  In fact, I think the questions of
math surrounding position sizing and everything past indicators has
been "solved" (or at least settled into a few distinct camps)...So my
study turn to indicators.  The classic ones just make no sense to me.
 The things have to be "smarter" than that.  The answer has to be in
some sort of "adaptive" sense.  But adaptive how?  

The degrees to which an indicator as simple as an average or smoother
can be "adaptive" or dynamic, is quite enormous.  It starts with the
period over which the average is calculated (i.e.-bars) and the
"shape" of the filter. (Box for SMA, Triangle for WMA, and up to
advanced designs like Laguerre, etc.).  

Then to be truely adaptive, the period and the shape should change
depending on some input (shorter periods in noisy environments, and
shorter periods when shorter cyles exist are the two most common). 
The tools used to measure these cycles and volatility can then be
adaptive themselves, and so on...

We have some examples we will be posting very soon where we use a
signal-to-noise ratio (which is kind of a measure of volatility) as a
variable input to some of Ehlers functions.  We then use these as
inputs to other Ehlers' functions, as well as some ASI functions, and
the result is some indicators that show some adavantages over Ehlers'
stuff directly from his books.

The end result is Ehlers' indicators with one more input made adaptive.  

The end result as a whole is a system that always looks at an input
and asks, "what effects this?" and tries to make a logical and
mathematically sound conclusion.  However, we know that regardless of
all of this theory, you throw it out the window if it doesn't look
good...In general, I think this is the way better indicators are made.
  In the end, the more sound they are, the more you can build upon
them because you can't extend logic on top of bad-logic...that doesn't
work. 

I think people that use metastock are somewhat limited NOT by their
understanding of the math, but by a limiting language that encourages
the use of static tools...which is why we built these products;
because we could not accomplish what we desired with Metastock code
alone (and when we could it was slow and hard to maintain).  Instead
of do it all hard coded in dlls, we decided to expose some lower level
tools that would leave some flexibility to do new things...

p.s. - You guys appear to be right about Dr. (as I like to call him)
Ehlers'.  We will see about changing our references...How does John
Ehlers Esquire sound?  Also, the man is getting up there in age and he
is still working...several jobs....some people just like to work...




 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/Metastockusers/

<*> To unsubscribe from this group, send an email to:
    Metastockusers-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/