Catch Time In An Indicator
Variable-Interval Moving Averages
by R.G. Boomers
Time is the most difficult variable to capture in an indicator. Here's a way to have a responsive, adjustable-length moving average without a lot of high-level math.
Simple moving averages are, perhaps,
the best-known and utilized tool of technical analysis. Yet they do have
limitations. One such limitation becomes painfully apparent when you try
to choose an interval for the moving average. After all, different intervals
have different strengths; a long interval for the moving average is great
for smoothing random fluctuations, while a short interval improves early
detection of turning points.
Unfortunately, these two design considerations are incompatible. Wouldn't it be great if you could use a long moving average when that works best, a short moving average when that works best, and all the moving averages in between when they work best? Could a moving average that changes with market conditions be possible?
Yes, it could. Perry J. Kaufman's adaptive moving average (KAMA) and Tushar Chande's variable index dynamic average (VIDYA), as explained in the April 2001 Technical Analysis of STOCKS & COMMODITIES, use exponential moving averages that change with market conditions. If it can be done with exponential moving averages, why not with the easier-to-understand simple moving average? Let me tell you about VIMA, the variable-interval moving average.
WHAT DOES VIMA DO?
The VIMA indicator allows the user to vary the interval of a moving
average. With VIMA, the moving average interval is varied as the price
of a stock gets too high or too low. As a stock moves away from its extremes,
the moving average interval is varied in the opposite direction. The moving
average is varied proportionally to the degree that the stock is either
too high or too low. That's all there is to it.
SIDEBAR FIGURE 1: Price differences. VIMA uses the differences between prices to assemble an oscillator, which in turn is used for moving average lengths. The varying pairs of moving averages are backtested to find the most profitable and most stable.
It sounds simple, but for it to work, VIMA must have an accurate method for determining when a stock is too high or too low. You might think this is impossible and in actuality, it is. However, Vima does have a method that gives it a try.
To do this, VIMA utilizes a special kind of overbought/oversold indicator. Ordinary overbought/oversold indicators use a single time interval, such as 10 days. Next, with a normal overbought/oversold indicator, all the price differences are figured for each 10-day interval over a few hundred days for the datapoints under consideration. The result indicates the maximum change over the interval to the upside and the maximum change to the downside. When the stock starts to vary by approximately the known maximum to the upside, the stock is considered to be getting near overbought. When the stock starts to vary near the maximum to the downside, the stock is considered to be getting near oversold. You can even use sophisticated models of statistical analysis to accomplish this.
It all sounds great in theory. Unfortunately, it does not work well in practice. As it turns out, for any given interval, any stock can suddenly fluctuate more than average without enough other longer intervals getting high or low enough to actually cause the stock to be overbought or oversold. For example, the 20-day interval might be at the low end of an average fluctuation to the downside when the 10-day interval suddenly jumps up. Just because the 10-day interval became overbought does not mean the 20-day interval is overbought. This is important, as the indicator seems to take many intervals being either too high or too low to make a change in price direction probable. One interval by itself becoming overbought or oversold does not work in practice. That is why I designed VIMA to use a large number of intervals.
...Continued in the July 2001 issue of Technical Analysis of STOCKS & COMMODITIES
R.G. Boomers is a stock market behaviorist residing in a remote, rural, largely unpopulated, and unexplored portion of darkest Arkansas. He can be reached at rgboomers@yahoo.com.
Excerpted from an article originally published in the July 2001 issue of Technical Analysis of STOCKS & COMMODITIES magazine. All rights reserved. © Copyright 2001, Technical Analysis, Inc.
Return to July 2001 Contents