|  JUL 1995

Signaling Change with Projection Bands by Mel Widner, Ph.D

V.13:07 (275-281): Signaling Change with Projection Bands by Mel Widner, Ph.D Projection bands, a new method using trading bands, projects market data forward along the data trend with the maxima and minima of the projections defining the band. The method provides another means of signaling potential changes for market direction relative to the trend. Buy low and sell high"" is the trader's tenet, and as a result, indicators that determine when prices may be low or high are very popular. Trading bands, which are envelopes drawn about price histories, are one such set of indicators designed to measure relative high and low prices. The envelopes are used to determine when prices are high or low relative to the price history. Trading bands evolved from simple envelope structures roughly sketched by hand to percentage boundaries about moving averages to statistical methods - for example, Bollinger bands - that are adaptive to changing market conditions. The latter method captures the average movement of the market by the moving average, and the randomness and cyclic behavior by the variation around this average. When the price is several standard deviations from the moving average, then it is at an extreme condition, a situation that often signals a reversal or sometimes a breakout. When combined with confirming information, buy and sell signals can be generated. Channel methods are an example of other methods with certain band characteristics. Channels are identified by connecting peaks and bottoms with straight lines that can be projected as a boundary around a selected price history. The rationale for this approach is that markets frequently move in a general direction with variations about this general movement between short-term support and resistance levels. Signals associated with touching channel boundaries or breaking out of the channel pattern are widely used and well documented. A recent STOCKS & COMMODITIES article by Bob McCullough provides a quantitative means to define channel boundaries using least-squares† fit analysis, noting that the channel boundaries often follow the same slope as the fit. Channel methods require that the analyst must make some selections, such as start and end times. In addition, long periods are often required to define the channel, and it is sometimes too late to take action once the pattern is clear.

by Mel Widner, Ph.D.