Reason number four is the order of the systems in the first place. In this case,
we started by adding the stops to all systems and markets, traded in any market
condition. We did so because we wanted the systems to work on average equally
well all the time. However, we could have defined the market condition under
which we want the system to operate first, and then optimized the stops for that
condition only. As it is now, the settings for the stop and exit lengths and times also
are influenced by conditions under which we wouldn’t have traded anyway, given
the filters we just added.
From a practical point of view, this means that most stops, exits, and time
frames probably are a little too tight to be optimal, given the market conditions
under which they are supposed to operate. During the bad periods, the systems try
to make a buck as quickly as possible or stop out a loser as soon as possible.
During good periods, on the other hand (those supposedly identified by the fil-
ters), the systems really should make the profits run a little longer and keep the
stops a little further away from the entry price, to give the systems a better chance
to make use of the favorable conditions. They apparently don’t manage to do this
as things are right now.
Aside from these reasons, adding a filter will almost always also reduce the
risk–reward ratio, simply because the filter also reduces the number of trades that
go into the calculation for this system measure. There isn’t much we can do about
that. To achieve smoother returns through diversification, we have to make room
for several systems and markets to be traded simultaneously. To do so we need to
trade each system–market combination less frequently by somehow filtering out
some of the trades. The fewer trades produced by a specific system–market com-
bination, the less we can say about its outcome in the future.
Let’s look through the results for the select system and filter combinations
that we will take with us to the money management analysis in Part 4. Remember
that all systems are long only.RS System No. 1 as the Filter
Table 21.3 shows that the stop-loss version of the meander system, together with
RS system No. 1 as the filter, didn’t do much worse than the filterless version,
depicted in Table 20.3. In fact, given the anticipated decrease in performance in
general and in the risk–reward ratio in particular, the only true negative is that the
percent of time spent in the market didn’t decrease as much as desired. The mean-
der system definitely has some serious potential.
The same holds true for the trailing-stop versions of the volume-weighted
average system (Table 21.4) and the Harris 3L-R pattern variation (Table 21.5).
The only true negative here is their relatively low risk–reward ratios, compared to
the original versions in Tables 20.6 and 20.8, respectively. One good thing with
these systems, as with the stop-loss version of the expert exits system (Table 21.6),260 PART 3 Stops, Filters, and Exits