EndPointFFT

EndPointFFT(DataArray, Periods, Threshold)

DataArray - The data array the the end point fast fourier transform will be performed on.

Periods - This is must be an even number more then 8 specifying the maximum bars the end point fft will be calculated on.

Threshold - The threshold percent noise filter a value between 0 and 1. Amplitudes that are threshold percent lower then the largest are considered noise and removed.

The regular FFT of a price series recalculates on each new bar of data making it useless for trading and just a sophisticated curve fit. The EPFFT takes the FFT at each bar, filters the noise out of the price series using a unique noise filter in the frequency domain, and creates a one day ahead projected price value. This process is repeated for each bar of data. The projected one day ahead value is processed to reduce any jumps between bars and plotted. This process gives a more robust noise filtered signal than a single frequency (dominant cycle) procedure. For trading purposes we are interesting mostly in the trend of the plot and more importantly the turning points.