46 PART 1 How to Evaluate a System
second signal unanalyzed. (To learn more about this system and its trailing stop
version, see Active Tradermagazine, February 2002 to June 2002.)
Once the system and all its signals is fully tested, the random number gener-
ator can be turned off, so that the system can be tested according to normal prin-
ciples. Figure 3.8 shows a few sample trades in Microsoft, using the original
model with the random number generator (RNG) set on. Compare these trades
with the signals in Figure 3.7 and you will find that the RNG prevents you from
taking all signals generated by the system, which means that the RNG version of
the system will not always trade the first signal in a series of signals.
As discussed more thoroughly in Part 3, one way to alter the system’s character-
istics is to vary the stops and exits. Figure 3.9 also shows a few trades generated with
the RNG set on, but also with a different set of exits than those used for Figure 3.8. In
the case of Figure 3.9, the stops and exits are much tighter than those for Figure 3.8,
which makes the system trade more frequently. Compare the trades with those in Figure
3.8 and the signals in Figure 3.7, and you will see that they are different. However, note
also that another reason for the differences between Figures 3.8 and 3.9 is that the RNG
is on in both cases; it therefore generates the trades in a random manner.46
FIGURE 3.9
Microsoft trading RNG, but different exits.