Given the assumptions, there is only one reason why the distribution of
trades can look as it does in Figure 2.2: System 2 has become more and more prof-
itable over time, which could be a very dangerous situation, because the system
might be optimized too hard to the most recent market conditions.
Personally, having seen Figures 2.2 to 2.3, and knowing nothing else about
the systems, I would go with the last alternative (Figure 2.3), because the similar-
ities of the trades over the entire time period indicate that it has worked equally as
well all the time, which the other two obviously have not.
A system that has worked equally as well all the time will also have its
parameters set in such a way that each trade influences the final parameter set-
tings to an equal degree as all the other trades, no matter the underlying mar-
ket conditions, such as the direction of the trend. A system in which
performance has fluctuated over time (increased or decreased) will have its
parameters set in such a way that the trades from the most profitable period
have influenced the final parameter settings more than the trades from the less
profitable period.
In the case of the three systems above, this means that Systems 1 and 2 are
more likely to start to underperform (which might already be the case with System
1), and even cease to work completely, when market conditions change.CHAPTER 2 Calculating Profit 17
FIGURE 2.1
System 1 distribution of trades and profits.