|
PureBytes Links
Trading Reference Links
|
> Subject: Re: Natural hour bars
> Date: Fri, 10 Mar 2000 19:48:49 -0600
> From: Lawrence Price <lprice1023@xxxxxxxxxxxxx>
> To: Glenn Suprenard <glennts@xxxxxxxxxxxxxxxxxxx>,
> Omegalist <omega-list@xxxxxxxxxx>
>
>I have experimented with different length bars in many systems. I find that
>often there is a length of bar that seems to work with a particular stock
>and a particular system. I doubt that this has anything to do with other
>traders and their signals but rather to do with the peculiar juxtaposition
>of chart and system that is created by a bar length. It is an intriguing
>subject. I have wished that I were capable of programing a way to optimize
>bar length for a symbol.
>
>Lawrence Price
You almost guessed it - if you change the bar interval AND you are
using length-based indicators (and who isn't?) then EVERYTHING CHANGES!
The bars that produced the initial results are no longer there.
A moving average of 50 bars on a 5-minute chart does not cover/compare
the same bars as on a 10-minute chart. In this case it covers only half
as much time. What can happen in 250 minutes? Is it not obvious that
this will affect your results?
I have re-written several indicators to be time-based rather than
length-based. This was so that I do not have to calculate and change
the indicator length when I change the bar interval. (Of course they
actually ARE length-based, but the indicator does the calculation).
Also, I like thinking in terms of time (say, a 2-hour moving average)
rather than an abstract length (12 bars ? 15? 22? 36? whuzat?)
I can change the bar interval, and a crossing stays in the same place
(+/- the bar interval).
donc
|