Re: sliding window problem

[Miroslaw Kwasniak]

On Tue, Dec 14, 2004 at 11:33:39PM -0800, charles sugihara wrote:
> i used
> retval(i) = sum(returns)/(wins+losses);
> and it is not returning any errors, now i am checking if it is 
> computing what i want.

Which is equivalent  to retval(i) = mean(returns) in case  non-zero returns
(for action != 0).

> thank you so much for the help, if you can't tell i am a inexperienced 
> programmer essentially working with existing code and trying to modify 

My loopless approach so called vectorization  (currently it assumes non-zero
returns, but it can easily modified) can save your time even >100 times (for
large inst_range).

> it.  my next mod will be to compute a sharpe calculation at various 
> window sizes, then i will be working on correlation coefficients.  i 
> expect i will be posting more problems soon.

I wrote in previous msg (comments in my code) about unfinished window
problem. My neigbour did blockwise poisson proces simulation and something
linke not unfinished window caused large bias.

Mirek



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------