possibility we have guarded against with our "profit preservation" or "trailing loss level" setup.

On Friday, the 6th of December, the expected upturn of the ten-week average occurs (as shown circled in Figure VIII-5). The extrapolation of the upturn is shown dotted, with an intersection with price motion at 18 1/8. (Remember to use the approximately equal time span between intersections of this average with the center line of the envelope which contains the component of which it is half-span as a guide to extrapolation. Alternatively, a 20-week average could be constructed and extrapolated to serve the same purpose.)

The move so far has been from 14 1/2 to 18 1/8. This is expected to be one-half of the total move, which now is targeted at 18 1/8 + 4 = 22 1/8. A tolerance of+10% of the total move is applied, providing 22 1/8 ±.8. Rounding off and shifting down by 1/8 (for convenience and to be conservative), sets 21, 22, and 23 as the lower, target, and upper limits respectively of the anticipated move. This prediction area is shown cross-hatched in Figure VIII-5.

On December 19th, the action of the 17th is shown to be the low that was expected in region "A," and valid uptrend Une VTL-3 is drawn. Four days later prices sweep into the prediction zone, and we have a choice to make. Either we can take profits now, or we can shift to the use of increasingly steep vaUd uptrend Unes and ride out the remainder of the move, if any. If the latter course is chosen, we set up the cross-hatched area at "B" as the expected low time zone for the previously noted 9.9-day component, and this low is firmly established by the price action of December 31st. VTL-4 is established on this day. But prices have now swept well past our prediction zone to a high of 24 3/4. We are now nine weeks along on the 18-week cycle, and time is running out. The same price action that set up VTL-4 allows us to draw VTL-5 (from low-to-low of the last two days motion). We determine that if prices continue to rise and remain above VTL-5 we will remain in the stock, but a puncture of this trend Ime on the downside will be interpreted as a sell signal. The following day this signal was flashed at a price of 23 7/8, 1 7/8 above our predicted target level. The use of increasingly steep valid trend lines based on very short duration components has extracted a tidy increment of profit for us.

The use of trend Unes formed by daily lows is not to be considered a standard practice. The elements of the situation that dictated this action here were:

• Price levels were already above our prediction .

• Price levels were rising above even our extrapolated envelope bounds.

• Cyclic time was running out on our trading cycle.

Such a "blowoff is normaUy followed by an equally rapid price contraction, and the steepest possible trend lines should be used to provide the take profit signal.

We bought at 15 1/2 on November 6, 1968, and sold m the vicinity of 23 7/8 on January 2, 1969. We made 49.7% profit on invested funds after costs in a total of 57 calendar days. The equivalent annual yield rate is 318% per year. Impressive as this seems, it must be kept in mind that stUl higher yields are possible if a selection is made of shorter cycUc components on which to trade. For example, this same issue was

Trading by Logic fmtead of by Guess 135

traded in the experiment described in the next section in which the objective of very short tradmg intervals was set. The result was a purchase at 16 1/2 and a sale at 17 1/2, in an elapsed time of one day. The ptMS percental gain was only 6%, and the net (after costs) was further reduced to 3.2%. However, on an annual basis this amounts to 1168% per year. The price paid for ttds kind of performance was the necessity of having many other fiilly analyzed stocks in reserve so that the funds could be put back to work elsevdtere hnmediately. By way of contrast, the previous example required time and effort tb be spent on one stock only, over a period of neariy two months.

One further comment is in order before leavii this example. The amount of analysis described was that which was /ust sufficient to resolve the decMon problems established by price action. If time is available, all the techniques described in previmis chapters should be applied and the results updated regularly. In some cases some of the methods wiB produce no new information, but the occasional situation where extra analytical effort avoids an incorrect decision makes the additional time spent well worth while.

A TRADING EXPERIMENT

The above issue was one of 35 that were analyzed and used in the first real-time trading experiment designed to apply the principles of the preceding chapters. This experiment was conducted from 28 October to 10 December 1968. The circumstances were as follows:

• Objectives were set at 10% return on invested capital in an average transaction interval of 30 days. Such a return, if continued without interruption, would compound $10,000 into $1,000,000 in approximately 44 months.

• It was decided that buy and profit-preservation signals would be used per the methods described, but that profits would be taken at the arbitrary level of 11.1%. In this manner it was planned to hold the transaction interval down and to establish as large a sample of trades as possible in the shortest possible interval of time.

• Operations were to take place from an office equipped to display the issues analyzed and tracked. A team of five was used to scan weekly charts, select issues for analysis, conduct analysis, and track prices for action signals.

• One member of the team used brokerage displays to track trades in the selected issues for action signals.

The results of this operation are aimmaiized as follows:

• Three short term "market" turns were successfully predicted as shown in Figure VIII-6.

• Two industry Group turns were predicted with the results shown in Figures VIII-7 and VHl-8.

• A total of 42 transactions was completed in 35 issues. Thirty-eight of these were successful, and four were failures, resulting in a success rate of 90.5%.

• The yield objectives were surpassed. Attained results were 8.9% net profit each 9.7 days. The equivalent annual yield rate achieved was 2474% per year as contrasted with the objective rate of 313.8%. The achieved rate of return, if continued without interruption, would compound $10,000 into $1,000,000 in approximately 15 months!

IOOOt

. DURATION OF £XPERII«NT

FIGURE -6

TRADING EXPERIMENT MARKET TURN PREDICTIONS OJIA-DAILY

DENOTES DATE,NATURE OF 1 PREDICTIONS

4 18 i 8 15 22 29 6 13 20 27

OCT.

NOV.

DEC,

JAN.

17 24 31 7

FEB.

Predicting Short-Term Moves In The Dow