SOMETHING RELIABLE
First off, in contrast to, say, trendlines, it is important to recognize that averages are a mathematical construct. Therefore, they will be computed with robotic catatonia, no matter what the current hysteria. While trendlines, arcs and circles are usually personal visual constructs, it is their mathematical abstraction that gives indicators their attraction and validity. Though an indicator can be thrown by data quirks that a human might -- might! -- ignore, the good side to this unquestioning use of data is that the indicator ignores nothing, an error that humans, prone to ignoring bad news anyway, routinely commit.
Second, the typical moving average tells you what the prices were -- past tense -- in the middle of the group of prices you're using. (As an exercise, you might think about what it would mean to take an average of prices not continuous in time. Then consider sampling past prices rather than recording and using them all.) The value computed by adding up all the prices and dividing by the number of prices is meant to be representative of the middle of the group of prices, given that the prices are ordered by time. Since the prices are always from a continuous series of days (or weeks, months, hours), this means the value you compute today is representative of the value x days back -- in fact, half the days back.
Typically, folks compare the value from several days back to today's price. Averages are a simple way of telling you if today's price is higher or lower than it was umpteen days ago. Recalling the fundamental approach of buying higher highs and higher lows (or shorting lower lows and lower highs), averages are truly basic indicators that will do you no wrong.
A simple average weights each price equally, but things needn't stop there. Estimating the middle value of a series of prices seems to lend itself to creativity. Traders and analysts have changed the computations to weight current, past or median values more or less heavily, the idea being that whatever is at the end of the series is weighted more heavily and so is more important in trading. Though there have been a fair numbe of thse schemes, I'm not aware of any having been shown to be more effective or successful than simple averages.
Another, more popular, approach has been to include the entire data series in computing the current value. A so-called exponential moving average (EMA) achieves this by blending the past average with today's value. (The term exponential, according to MESA developer John Ehlers, is the shape of the computation's impulse response -- its behavior after a single, sharp value is given to it.) The EMA does this by taking a percentage (referred to as alpha) of the average computed yesterday and adding it to a percent of today's price. The two percentages must sum to 100 to achieve equanimity for most mathematicians. (Consider the alternative and what that might mean!)
For now, think in a straightforward manner. Which way are prices going? While skeptics have concluded that averages are too simple to win in the marketplace, it's rare to find indicators that produce significantly better results in trending markets.
John Sweeney is STOCKS & COMMODITIES'Technical Editor.
