trades can be discarded in real-time trading will be required in order to develop a profitable strategy.

On the other hand, the slower, less sensitive settings will usually be more effective in isolating the prime turning points in the market but their formulas create the actual signal a few bars later thad would be ideal.

Even though these delayed signals may be more accurately placed, the fact that they are a bit behind the prime market turning points causes our system to sacrifice several points of possible system profit. The trade-off then becomes one of trying to select the best of the numerous entries placed by the faster moving averages or being satisfied by the slower average signals, which admittedly will decrease the profit potential of the trade.

Combining these two diverse market measurements can increase the effectiveness of the entire process by utilizing both the quick action from the faster settings and the more accurate determinations from the slower settings.

Think of this again as a filtering process. We first look at the slow settings for our indicator, expecting, in the case of an anticipated buy signal, that the plot of the indicator will fall to a relatively low level prior to issuing a buy order. When the slow averages are in this low position, then, and only then, will we consider the activity of the faster average setting. By using the slower average to filter out the signals from the faster average we can gain the advantage of the accuracy of the slow average while still being able to use the quicker entry provided by the faster average. Another way to think about this concept is to consider the slower average as a qualifier of sorts. The situation does not qualify for a trade until the market gets itself into a favorable condition as defined by the slower indicator setting.

To take maximum advantage of this filtering process, it is usually preferable to utilize multiple overbought and oversold thresholds rather than single values for these levels. Using a wider overbought and oversold zone for the slower averages allows the qualifying plot a bit more latitude, therefore giving the faster moving average that actually triggers the trade increased opportunities to do so. Conversely, these situations seem to generate more accurate trades when a comparatively narrow range for these two critical values is used with the faster average.

Although this concept is obviously applicable to any oscillator in-

dicator, it seems to be a bit more effective with stochastic. For this reason we will utilize this indicator for the initial examination of this important filtering concept.

First, to illustrate the qualifier concept using an indicator with a relatively low sensitivity setting, well look at a chart of i2 Technologies Inc. (ITWO) during a relatively nontrending day (Figure 6.1). Applied to this chart is a 45-bar Fast D plot, one of the major components of most stochastic studies. The overbought and oversold levels are placed at 70 and 30, respectively.

In the lower graph in Figure 6.1 you will notice the plot of the Fast D indicator with a 45 sensitivity setting. Note that the plots are programmed to display a thicker line when the values for the indicator are below the oversold level or above the overbought level. The use of this particular configuration of the Fast D indicator would allow trades only in the areas where the line appears darker. Again, buys are taken only with the indicator in the oversold area and sells are allowed only when the indicator tells us the market is overbought. As

? Tr-ideSlotion Char! - (1TW0J i2 TeiJinulogies Inc LAST-1

rrWOLAST-1 min 10/27/2000 C=I67.250 -&.75G -3.88% 0=170.563 H=173.125 L=166.625 V=2692100

FastD(70.1) 41.84 30.00 70.00

170.000 168.000 166.000 164.000 162.000 160000

10127 8:56 S:21 9:46 l6:11 l0:36 li:01 11:26 Hi 51 12:16 12:41 1:06 1:31 1:56 2-21 2:46

Figure 6.1 The slow componentofthe dualstoclmsticindicator, a 45-period Fast D, is shown here with the overbought threshold at 70 and the oversold level at 30. Trades from the faster setting of this dualsignalindicator can be taken only when the plotofthe slower setting is in the area designated by the thicker black line.

Chart created with TradeStation® 2000i by Omega Research, Inc.

you can see, this setting of the indicator allows a fairly liberal amount of the chart to accept trades.

In Figure 6.2 we can see what happens when we change the overbought and oversold levels to 80 and 20. Note that the areas in which trades are now allowed has been rather significantly restricted when compared to the 70 and 30 settings used earlier. Now lets look at Figure 6.3, where the values are moved to 90 and 10. Incidentally, these are the settings we will be using for the faster moving plot of our indicator that will actually trigger the trades.

In this plot the areas where trades could be initiated have been restricted to points that are of little use for this demonstration. Looking at this chart one would be tempted simply to use this technique to actually place trades. However, recall that oscillator indicators work best on sideways days. Also recognize that this is an almost perfect sideways day, specifically chosen for the purpose of demonstrating the dual setting concept for the use of these indicators. Although the trades that could have been placed on this chart look good, be assured that this is not a consistent observation when

(TWO LAST-, min; ,Of27/20O0 C=167.250 -6.750 -3.80% C=170.S63 H=U3.126 L=166.625 «=2883100!

¹7 8:56 931 1 11 10:36 1101 l i-26 Tl:5l l2:16 *12 4l ,

1:31 1-56 2:21 2:46

figure 6.2 the amountoftime during which a trade may be generated is reduced significantly when the overboughtthreshold is raised to 80 and the oversold level is lowered to 20.

figure 6.3 reducing the threshold areatothe area above 90 for a sell and below 10 for a buy restricts trading possibilities to an almost useless level when the slow setting is used as shown here. these levels will be used effectively by the faster settiags for the dual settiag indicator. chart created with tradestation® 2000i by omega research,inc.

applied over a longer time frame where uptrending and downtrend-ing days are also considered.

Now lets shift gears slightly and consider the faster component of our combination indicator. Figure 6.4 shows the seven-bar Fast D indicator on the same chart of i2 Technologies Inc. (ITWO).

The Fast D indicator as displayed with the seven-bar sensitivity is using the same 70 and 30 overbought and oversold thresholds. These levels are the same ones as were shown to be the most effective for the same indicator using the 45-bar sensitivity setting. When comparing this chart to previous charts used to demonstrate the activity of the slower indicator settings, you will notice that many of the same entry areas are identified. Also note that there are many more entry points selected, several of which appear to be far ahead of the ideal entry point. Also, when observing this chart, understand that the actual entry point generated from this routine will fall on the first bar whose data has caused the corresponding indicator to plot with a thicker line. Although the highlighted plots may indeed encompass several prime trading areas, careful observation

The buy and sell arrows appearing on the chart identify the precise entry points for this demonstration trading system. They are generated by the following rules.

For a buy:

1. The Fast D plot with a 45-bar setting must be below 30.

2. The Fast D plot with a 7-bar setting must:

Be below 10.

Turn up on a closing basis while still below 10. For a sell:

1. The Fast D plot with a 45-bar setting must be above 70.

2. The Fast D plot with a 7-bar setting must:

Be above 90.

Turn down on a closing basis while still above 90.

Figure 6.7 details the anatomy of a sell and a buy signal as generated by these trading rules.

Lets first examine the sell signal that appears on the chart. The first event that must occur is that the stochastic plot created with the 45-bar length must rise above the 70 level and remain there until a signal is completed. Again this plot is shown with a heavier line while it is above the 70 level and therefore in a position to qualify the trade. Secondly, the lighter tracing representing the stochastic using the seven-bar length must pass above the 90 level. Finally, this line must turn down on a closing basis for the short sale to be generated.

The buy signal is created in the same manner. First, the 45-bar length line must be below the 30 level. Next, the seven-bar length stochastic plot must pass below 10 and then turn up on a closing basis. The buy signal is placed on the chart after all three of these conditions are satisfied.

Note on Figure 6.7 that a considerable section of this chart has been qualified as a possible buying area by the 45-bar stochastic setting. In fact, the trader would have been alerted for a possible purchase beginning at 9:46 a.m. when the line passed below the 30 level. This buy window remains open until the line pulls up and out of the oversold range at 10:06 a.m. Simply looking for the 45 period plot line

1 t.i.itoSI.Ut«n l.h.,nt (ltWQ\a h:ctmoli>ri":s Inc LrVJT-1 i

ITWOmST-l min 10J27/2000 C=167-250 -6.750 -3.80%

...... ,3 ............ ...........(-172.000

Fasip(70.1)..41S4 30.00 70.00........ .........

Downturn - Sell Signal

Sell »

Buy i

166.000 164.000 160.000

Figure 6.7 Trading signals are generated by the dual oscillator routine when the slower setting is above or below its specific threshold, while the faster setting also exceeds its threshold levels andturnsthe opposite direction on a closing basis.

Chart created with TradeStation® 2000i by Omega Research, Inc.

to turn higher after dropping below the 30 threshold line would have generated buy signals well ahead of the prime entry point on several occasions. Waiting for the same line to cross back above the 30 threshold line would have resulted in an entry well past the best entry level. Our actual buying point resulted from the confluence of the prescribed actions of both stochastic plots. All of the buy and sell signals plotted on the chart that shows the entire day of trading (Figure 6.6) were generated in this manner. Using the two separate, diverse settings of our stochastic indicator allowed us in this case to make a long entry quite close to the low of the market drop.

Are all entries using this technique this accurate? Of course not. This particular chart was specifically chosen to demonstrate the three steps necessary to create these trading signals. As you recall from previous discussions, these groups of indicators work best on sideways days, of which this is one. Lets now look at this combina tion on a few trending days. Figure 6.8 is the same chart that we have used previously to illustrate system and indicator behavior on an uptrending day.