Signal Classification with FFT

[gvanto]

Hey I was wondering if anyone has played around with using a (fast) fourier transform (FTT) to try and gauge the movement / volatility of the SPI future (for example) on an on-going basis.

I'm using the R statistical package to play around a bit (http://ubuntuforums.org/showthread.php?t=932892), got some stuff going to try and identify slower swings (trends -> microtrends) as opposed to high-frequency smaller movements (volatile non-trending) with some mixed success.

I'm thinking of this as a way of identifying ranges, possibly for a breakout strategy.

IF someone has thoughts on this, would be cool to perhaps share some ideas ...

Happy trading (in a volatile market!)

Gerry
fft newbie

 

[rub92me]

Looks like you are trying to reinvent/refine the wheel a bit here. Nothing wrong with that as a mathematical exercise/challenge of course
FFT is a standard indicator in chart/trading packages such as Metastock and Amibroker. The effectiveness of any indicator in terms of trade set-up can easily be tested in those. I've never tried it for FFT myself though.

 

[gvanto]

Cheers rub92me,

Hope tradeStation has it, in the process of setting it up ...

I like using Java & R however, to test my strategies on.

Will see what FFT can conjure up in terms of helpful trending / non-trending advice

Cheers
G