back start next
[start]                                          [ 42 ]          
Table 13.3 Average
| || |
| || |
itability of the system decreases steadily as we go further into the day. Selecting a proper time for trading the breakout of the Directional Day Filter for our security now becomes a matter of trade-offs between the percentage of days during which a successful breakout occurs versus potential profitability versus possibility of a stop-out in the case of a double breakout. It is in the sorting of these possibilities that once again the style and risk-carrying ability of the individual trader comes into play; each user defines this particular strategy according to his or her individual preferences.
STEP THREE: SET STOP PROTECTION
As always, we must always carry stop protection of some nature on all trades. My preference, as detailed extensively earlier, is to implement a trailing stop placement governed by the appearance of market-generated support and resistance as detailed in several instances in this book.
STEP FOUR: SELECT STOCK
Near the end of this book you will find an appendix containing data for a number of popularly traded stock issues. These numbers are generated by the same program that created the tables we have used to build the system. This data is presented both to demonstrate the manner in which different securities respond to the breakout scenario and to give you information from which you may construct your own trading scenario.
I have constructed a series of charts that describe the response of a portfolio of 80 stock issues to the aforementioned breakout scenario. This list is compiled from issues on which I have been asked to do analyses by various clients, and therefore the stocks are quite randomly chosen. The 80 stocks included in the test portfolio are listed in Table 13.4.
Figure 13.4 graphs the response of the entire database to the optimal breakout time. The percentages plotted on the y-axis reflect the percent of days tested that had breakouts only on one side of the early range. Generally, the most profitable times run from 65 to 95 minutes after the open.
We can gain considerable valuable trading information from the
First, you can see that for the entire 80-stock database that the high or low of the day was established on 75 percent of the days tested in the first one and a half hours of trading.
Furthermore, we also can establish that on each of these days there was at least one new high or one new low made on the opposite side of the established high or low.
We can also state that the Directional Day Filter was accurate in its prediction of the direction of the one-sided breakout with an accuracy approaching 75 percent.
Table 13.4 Test Portfolio
80.00 -70 -OO -60.00 -50.00
« 40.OO -
% . -
Figure 13.4 Graphing test resultsverifies that the issues included in the sample placed tbeir high or Ion for tbe day within 90 minutes of tbe market open on 75 percentofthe days tested.
Now lets look at the profit potentials for the entire database for both the long and the short trades generated by the breakout strategy (Figure 13.5).
This graph details the average number of high and low breakouts that occurred across the entire database for the entire testing period for each breakout time frame tested.
Closely observing this graph, it is immediately obvious that there were more trades generated by our breakout system when trading the short side of the market. Close observation of the Data Appendix will also reveal that the profit potential during this time frame was greater for the trades generated on the short side. While this revelation is certainly due in part to the negative situation of the Nasdaq market during the testing period, it has been my observation that the short side is usually more productive regardless of the market posture in general. We need only to turn once again to market psychology for an explanation of why this is true.
Most people, when trading securities in particular, do so mostly from the buy side of the market. We all like to be long. As a consequence, especially in the case of nervous, inexperienced investors, a sharp down move can create a general market panic as traders bail out of their positions in fear of substantial losses. For this reason the
| || |
\v 80 0 -
| || |
| || |
| || |
| || |
Figure 13.5 Wbile tbere is slightly more profit potentialfortbe sbort side during tbe test period, one must coasider tbe overall negative bias of tbe market during thetestperiod before drawing any long-term conclusions.
market movements that accompany a downside breakout can frequently be violent and nasty, often driving the market lower than normal fundamentals would dictate, at least for the moment. Even though the market soon realizes this fact and usually recovers from this overbought condition, there still remains on the chart of this activity a sudden spike to the downside. Many of these days will fully recover, often actually finishing with net gains for the day.
The point is that, even though we see higher numbers as potential profits for the short side of our breakout strategy, this doesnt necessarily mean that this particular security was in the tank all day. Many of these potential profits are the result of days that show a significant spike down as the intraday range is broken on the down side and there is a price recovery later. With all of the above in mind, one should always at least consider either taking profit on your short trades when one of these spikes shows up, or at least move the trailing stop a bit closer to the market.
Figure 13.6 plots the number of days during which there was a single, a double, or no breakout for each time frame tested, again for the entire database.
At the 15-minute time frame, one would expect very few days to have established either side of the eventual daily range at this very
Figure 13.6 The prime time for our breakout trade occurs when the gap between the single breakout plot and the double breakout line is at its widest. This occurs as the no breakout line crosses the double breakout line.
early time in the session. Hence, you will notice the "no breakout" line on the graph, which reports the number of days on which both the high and low were established at each respective time frame, is very nearly zero as the graph begins. As you would expect, the line rises rather regularly as the day progresses and the daily range widens. As the range widens so does the likelihood of the boundaries of the current range being the high or low of the day when all is said and done for the session, thus the increasing number of days at each time frame where no breakouts occur.
The reverse is true for days experiencing double breakouts. Since the intraday range at the early breakout times is relatively narrow when compared to the eventual final daily range, one would expect a greater number of these days to show breakouts on both sides of the range for the early periods. As the day progresses and the intraday range widens, fewer and fewer time frames are expected to show a break of both sides of the range.
The line labeled "single breakouts" is the same representation as seen on the bar graph earlier in this chapter (Figure 13.4) that shows the number of days for each time period that have established either their high or low at that point.
Also, as one would expect, the number of one-sided breakouts peaks as the lines plotting the double breakouts and zero breakouts cross each other. If nothing else, this chart, taken from actual price charts of 80 stock issues over an 11-month period, is of interest as it validates our earlier theories concerning early range breakouts and the likelihood of one side of the daily range being established rather early in the day.
Before you get too comfortable with these representations from the entire database, dont forget that these graphs are created from actual chart-generated statistics from the entire SO-issue database. These numbers are averages of all the issues tested. The point is that they are averages. As mentioned on multiple occasions, each issue responds differently to any technical analysis scenario, including this system. Figure 13.7 represents the varying responses of selected issues to a single system parameter.
This graph represents the first 31 issues in the database, chosen in alphabetical order, as they respond to the system as measured by the average profit potential of the long breakout at the 60-minute time frame. As you can see, even in this relatively small sample, the
[start]                                          [ 42 ]