Data Filtering Methods For Technical Analysis
by ANTHONY WARREN, Ph.D./Technical Analysis staff writer
Data filtering methods such as moving averages are quite prevalent in many trading systems. In this
article we will discuss a methodology for selecting appropriate data filters based on Fourier spectrum
analysis, and the basics on how to use smoothing filters for trading decisions. In the course of this article
we will attempt to familiarize the reader with data filtering from a frequency domain point of view (i.e.
cyclic). The reader need not be concerned with the mathematics of Fourier analysis, but should begin to
appreciate this methodology as a powerful tool for understanding and applying many trading systems.
LINEAR FILTERING: FREQUENCY DOMAIN THEORY
Suppose we desire to apply data filtering to time series x(i) i=1...N, such as the stock market daily
closes. In general, a linear filter is defined by a fixed set of weights (w(j) j=1...M), where M is the length
of the data filter. The output of the filter is a second time series