A HaJf-Span Average "Hold Short" Signal

OTHER USES FOR HALF- AND FULL-SPAN AVERAGES

GrapMcal cyclic analysis should, of course, accompany the use of all techniques. In this case the two can interact very favorably. For example, looking back at Figure VI-6 we see tiiat there is absolutely no doubt whatsoever about how the 20-week channel should be drawn. The 20-week moving average very nicely slices through the center of sohdly visible peaks and valleys which provide points on the desired envelope.

You will find some issues in which the situation is not this clear. In such cases, "eyeball" the stock chart to get a rough feel for the dominant component. Construct a half- and full-span average based upon your best estimate of the duration of this cycle. You will find that plotting these avera makes identification of channel bounds very much easier. After your channel is constructed, you will be able to refine your estimate of trading cycle duration. If this is significantly different from the original, you will then want to reconstruct the half- and full-span averages based upon your revised durations.

Quite often a properly constructed set of averages will signal an envelope turn before either real or non-real time envelopes could poibly provide this information. Alloys Unlimited provides a typical example of this.

In Figure VI-5, it is impossible to know whether or not the 20-week channel is pivoting upward (about the 32 1/8 low) without the aid of the moving averages. In fact, the cycle high at 48 3/4 is lower than the preceding one at 49 3/4, leading to a suspicion tiiat the channel direction is still down.

Inspection of the 20-week moving average tells us exactiy what is happening. The trend here is definitely up-telling us that the sum of all components of duration greater than 20 weeks is moving up. Since this is the center Une of the 20-week channel, this channel has to have bottomed with the 32 1/8 low and must now be curving upward. The approximate amount of curvature is obtained from one-half channel width measurements from the graphically smoothed 20-week moving average. The envelope bounds so defined give envelope analysis estimates of 20-week cycle moves that are in good agreement with those obtained by observation of ten-week moving average turns. Note that this advance information was completely unobtainable from envelope analysis alone until a subsequent cyclic top was formed and confirmed at 52 718, a full three weeks later!

From this example it is seen that the graphical and computational techniques can be used to aid and complement one another.

The same loc that leads to the use of half- and fuU-span moving averages to predict amounts of moves can be used to generate back-up time information as well.

As shown in previous paragraphs, a half-span average reversal agnals the half-way point in a cycUc move in terms of magnitude. Now, when the half-span average reaches mid-channel, the price of the stock has preceded it down to the theoretical low for the move. But in predicting magnitude variation, we found it necessary to extrapolate the half- and full-span averages to intersection, which by definition is a best estimate of the point where the half-span avera is at mid-channel. The theoretical low (in the predicted price zone) should take place one-half the span of the half-span average later.

Lets use a ten- and 20-week (half- and full-span respectively) case as an example.

The ten-week average has just topped out. We extrapolate both half- and full-span averages to intersection with the stock price. Note the calendar date on the chart at which this occurs. One-half the half-span average span is one-half of ten, or five. We expect the predicted low to occur five lime uiuts from the estimated intersection point date.

Return now to Figure VI-1 and VI-2. In Figure VI-1, note the time of estimated mtersection. You now predict zone entrance and high five weeb later. From Figure VI-2 it is seen that zone entiy and bottom-out took place nine weeks later. This an example of about the poorest estimate correlation to be expected. In this case reliance would have had to be placed on other techniques to assure results.

From Figures VI-3 and Vl-4 we note intersection at 42 and expect bottom-out five weeks later. Actud zone entry took place three weeks later, with a low achieved six weeks from intersection (an example of excellent correlation).

In Figures VI-5 and VI-6 intersection occurred at 43 3/4. Top-out was anticipated five weeks later. Zone entry occurred three weeks later with top-out at five weeks.

The reason for a larger percentage spread m timing is the existence of shorter duration components which tend to swell or depress prices as they peak or low out. Neverthele the method is of some use when allied with all of the other techniques available.

Generally speaking, graphical envelope analysis and valid trend Imes can be depended upon for the most accurate and detailed information regarding move termination timing.

NOW TURN YOUR MOVING AVERAGES INSIDE OUT

Youre familiar now with the principal characteristics of a moving average, and how the price-motion model generates criteria for getting the most from them. In this section, you will find they have still more uses.

Lets ask ourselves the question: Is there useful information in what a moving average throws away?

We recall that a moving average "smooths" data. It does this by reducing the magnitude of short duration fluctuations, while permitting the longer ones to remain. In short, it throws away the short fluctuations.

What a Moving Average Throws Away Can Be

Recovered By Subtracting The Average From The Price-Motion Data

An example will clarify how this works. Suppose you form an -week moving average of the weekly mean prices of a stock. This means that in the moving average the magnitude of any cyclic component of duration exactly equal to 11 weeks is reduced to zero. Thus, the average contains only cyclic components present in the price motion which have durations longer than 11 weeks. Subtracting the average from the price motion removes aD these longer duration fluctuations, leaving only those shorter than 11 weeks that did not appear m the average (or were "thrown away").

The only critical item to remember (as always) is to match the movmg average and price data points properly before subtracting-taking into account the half-span time lag previously described,

USE THE INVERSE HALF-SPAN AVERAGE TO IMPROVE YOUR TIMING

There are many ways in which you can make use of the inverse moving average. One of the most important of these is in connection with the half-span average concept. Lets see how it can work for you.

Direct your attention to Figure Vl-8. Here is a weekly plot of Alloys Unlimited