Ward Systems Group Adaptive Net Indicators are based on the work of Dr. Donald Specht. Specht’s work in turn has origins in an old (circa 1960) statistical method called the Parzen Window or the Parzen Kernel. Specht implemented his work as a neural network and called it the Probabilistic Neural Network (PNN). This is the form on which our Classifier Nets are based. Later Specht built the General Regression Neural Network (GRNN), which is the foundation of the Prediction Nets.

If you are familiar with Ward Systems Group general purpose neural network products, NeuroShell 2, NeuroShell Predictor, and NeuroShell Classifier, then you will recognize PNN and GRNN. Ward Systems Group Adaptive Nets are a different implementation of PNN and GRNN than those found in the general purpose products, an implementation that lends itself better to financial indicators.

In Specht’s work and in Ward Systems Group, Inc. general purpose products, the contributions are variously called contribution factors, smoothing factors, and sigmas.

A complete and rigorous description of how GRNN and PNN work (and hence how Adaptive Nets work) is beyond the scope of this software and beyond the scope of our technical support personnel. Anyone who wants such a description is referred to C.H. Chen’s book Fuzzy Logic and Neural Network Handbook copyright by McGraw-Hill Inc, 1996. Chapter 3 was written by Dr. Specht. The book is part of the McGraw-Hill Series on Computer Engineering, and is available in most university libraries and technical book stores.