Table 13.1 is representative of only a small portion of the data generated to make our initial determinations concerning a proper breakout time. Although it is possible using simple chart observation to generate this data, I have chosen to write a rather simple program on the Omega TradeStation platform to do so.

The delay column represents the time at which the data on this line was generated, in this case 60 minutes after the open of the trading session, or 10:30 a.m. eastern time. The next column tells us that 228 days were tested to generate the data. Of these 228 days, 162 had breakouts of only one side of the intraday range recorded 60 minutes into the session. This represents 71.05 percent of the days tested. Of these 162 days on which a one-sided breakout occurred, 73 breakouts occurred on the high side of the intraday range while 89 of these days had breakouts below the intraday range. Additionally, there were 32 days, or 14.04 percent, on which there were no breakouts of this early range. Finally, on 34 days, or 14.91 percent of the 228 days representing the sample, the market broke out of both sides of the intraday range calculated at the 60-minute point. By examining a table that contains this data for each tested time frame we are able to determine the best time for our breakout strategy.

So, how do we know that the 60-minute time frame is the primary breakout point for &Logic? At this point, we dont. Thats just the easy number I have used throughout the book to demonstrate the use of the Directional Day Filter. As you will see later, this prime time will vary between issues selected. To get a good handle on the best time to set our buy or sell stops I have run the program just mentioned against QLogic from January 1, 2000, through December 1, 2000.1 have generated numbers for each five-minute period starting at 15 minutes and running to 300 minutes after the open. This is the equivalent of testing each breakout strategy as it would have developed for each five-minute period beginning at 9:45 a.m. and ending at 3:30 P-M. eastern time. Table 13.2 reflects the results of testing all 58 breakout times.

Table 13.1 Sample Data from QLGC at 60-Minute Delay Point

Table 13.2 Complete Data Sequence for QLGC

(Continued)

Table 13.2 (Continued)

With the amount of data generated by such a process it is often helpful to graph the results to enable more efficient interpretation of the results. Since at this stage we are concentrating on the selection of the most profitable time to execute our breakout strategy, I have plotted the percentage of one-sided breakouts that occurred across all time frames tested.

From Figure 13.1 it is obvious that any time frame between 55 and 110 minutes after the open will work well for our system. The numbers tell us that on about 70 percent of the days tested, QLogic establishes its high or low for the entire day during roughly the first one and a half hours of trading. Additionally, this issue breaks out of its early range only on one side of the early intraday range on these days.

We have now established, based on the information in Figure 13.1, that we should enter the market on our breakout scenario sometime during the defined time frame. It is now necessary to time our entry in such a fashion as to limit our possibility of loss to an accept-

60.00 y-............................................................. ........ -H

70.00 lllllllillllll II

60.00 -i lll]lM II

s 50.00 IIIIIIIIIIIeIIIIIIiiIIIIIIIIillllllIllllllKIllllBllIIIIIII n. 40.00 -pi !!!!!!!]!!!!]!!!!!!!!!!!!!]!!!!!!!!!!!! Illllllli

3 lllllllllllllllllllllllllllllllllllllllll

JS 30.00 4flllllllllllllllllllllllllllllllllllllllllllllllll

«t llllllllllllllllllllllililllllllllllllllllll

20- IBIkbIII IIbIIbIIIiIiII .00 -U iIIbI I IIIIbbI 11 1111 1 11111 1111 1111 1 lllll lllllill III

ojoo IUJiMiMiM UiUiUiUiUi iUMiMiMiUiMrM**i*iUUiiUiiU(iMiMiUiU

Tlma Pally I

Figure 13.1 Graphing the breakout percentages from Table 13.2 reveals a rather wide range of breakouttimesthatnill be useful in setting the parameters!orour system.

able level. Since the possibility of a loss is the greatest in our scenario when the market is successful in breaking out of both sides of the early range, we should therefore choose a breakout time when a double breakout is less likely. The next chart graphs the double breakout percentages from the data table against the tested time frames.

Examining Figure 13.2 during these same time frames, the percentage of double breakouts is rather insignificant in comparison with the values running between 7.5 percent and 16.5 percent. From

Figure 13.2 Percentofdays exhibiting double breakouts is plotted as a function of the time of the breakout.

�124758205737305527748

Figure 13.1, our timing parameters should produce breakout trades on only one side of the intraday range on around 70 percent of the days. Also, from Figure 13.2, the chances of making a breakout on both sides of the early range and placing us in a stop loss position should only happen on less than 17 percent of the days traded.

To compare the percentages of days showing a single breakout to those showing a double breakout, I have plotted both occurrences in Figure 13.3.

Logically, the point at which the spread between our two plots on the chart is the widest will allow us, when using the matching delay time, to enter with the highest average percentage change of experiencing only a one-sided breakout and the lowest average chance of getting caught in a double breakout day and possibly being placed into a losing situation. Placing our breakout time at 90 minutes after the open of the market maximizes the potential for a successful onesided breakout while minimizing chances for a double breakout.

STEP 2: DETERMINE EXIT STRATEGY

This brings us to step two of the system development process. No system is complete without an exit strategy, both to take profits and to protect the system against significant losses. With this in mind, lets

80 w

I i i i i i i i i i i i i i i t...............i i i ~ i i i i i i i i i i i i-t i i i i

Figure 13.3 Timing our trade at 90 minutes after the open of the market in QLGC maximizes the possibility of a trade with a single-sided breakout while minimizing the chances of a double-sided breakout.

now look at additional numbers generated by the same program that gave us the figures quantifying the various breakout scenarios.

Since we are trading a breakout strategy, and assuming initially that we will be using a set target for our exit, it would be helpful to have an idea of how far the breakouts usually go at each breakout time. With this information we will have a better handle on which breakout time is the most profitable. We can also find an appropriate target level for trades, either long or short, that can be taken at each breakout time. In essence, the following study will impact both the entry and exit routines as we wish to enter at the most profitable times and exit with the highest target possible. Table 13.3 lists that information for QLogic.

To create the information in the table, our statistical program measures, on each day where there is only a single breakout, the maximum amount of profit that could have been realized had one been fortunate enough to close out the position at the perfect instant. In the case of a breakout on the upper side of the early intraday range, the program measures the dollars per share that could have been realized if one had purchased the breakout without any slippage and then sold the position at the exact high of the day. For a short position the program in a similar method measures the amount of profit that would have been possible selling the breakout and closing the position at the very low of the day.

Obviously, when we are trading in real time, only very occasionally will we be fortunate enough to sell the very high of the day or buy the very low of the day to exit a trade. These calculations are obviously done in retrospect on historical data and must be viewed in that light. These potential profits represent the absolute best that the system could have done on each breakout trade. While we will never expect our system to perform this admirably, these calculations do indeed give us an appreciation of the potential profitability of each given breakout scenario. It naturally follows that our profit potential will be greater if we consistently trade the breakout times that demonstrate the greatest gains on a historical basis.

When selecting a profit target always keep in mind that these numbers represent the best possible result for the system at each timing level. We cannot expect our system to do this well. Be sure to set any target levels accordingly.

Looking once again at Table 13.3, note how the potential prof-