THE HAnDbook of TEcHnicAl AnAlysisthey are somewhat perplexed as to why the moving average is underperforming.
We also see that the equity returns associated with point 2 may represent moving
averages with varying equity curve behaviors. The best choice of moving aver-
ages would therefore be those associated with equity curves A and C. Both curves
display increasing returns over the in‐sample test period and we may assume that
this behavior should persist at least for some duration before failing when tested
with out‐of‐sample data. We also observe that although curve C displayed nega-
tive returns (below the initial capital level), the moving average associated with
equity curve C is still a viable choice as it has a greater potential to be profit-
able during the out‐of‐sample test period than moving averages associated with
curves B, D, and E.
This simple example shows that optimizing a system merely to produce the
greatest returns is not the correct approach. It would be better to select the system
associated with an up‐sloping equity curve during the test period, regardless of the
size of returns it produces over the test period.
Besides just testing for positive equity curve performance, the trader may also
test the system for the largest percentage drawdown. It would be pointless to
select a system with a positive equity curve performance that is accompanied by
extreme and volatile drawdowns. See Figure 29.4.
In Figure 29.4, we now see that the moving average associated with equity
curve A is displaying an extreme drawdown event, indicated at point X. This sug-
gests the trader should in fact select a moving average associated with positive
equity curve performance that is not overly volatile. It is interesting to note that
the moving average associated with equity curve B displays the greatest return
when compared to the equity curves associated with the other moving averages.
Figure 29.3 Equity Curve of Various Moving Average Returns during the In‐Sample
Test Period.