For some time I have wondered if fractals might be used to select the important samples in a price time series, thus giving a fractal time-base. I ran across this paper that argues that such a set exists and seems to have a way to find that set.
This paper by Turiel points to a way to select this set using fractal decomposition. I have a problem trying to translate it into code so that it might be tested. In particular I need a way to find the Hurst exponents for each sample, and then sort the samples by exponent. I think that Turiel is claiming that the set with the smallest exponent makes up the most important set.
Anyone have any ideas?
This paper by Turiel points to a way to select this set using fractal decomposition. I have a problem trying to translate it into code so that it might be tested. In particular I need a way to find the Hurst exponents for each sample, and then sort the samples by exponent. I think that Turiel is claiming that the set with the smallest exponent makes up the most important set.
Anyone have any ideas?
Attached File(s)
Multi Fractal Geometry.zip
1.0 MB
|
1,393 downloads