|
|
On Exit Strategies, 09/05/05
Hypothesis:
The distance from 'normal' that a system enters the market determines
the optimal exit strategy for that system.
For example, a system that enters after a large standard deviation move,
will find its biggest reward in a longer duration exit such as holding
for the highest high of the last 'n' days.
The same system that enters
after a smaller standard deviation move will find its biggest reward
after a short duration exit such as a single day or a first profitable close.
Some counting:
I tested this on the Auto-Composite system which has two components with
longer range exit strategies. Each component enters the market in succession
so I tested using a faster exit on the first component and a longer range
exit on the second component.
Results:
After modifying the exit strategies based on the hypothesis, the z score for the
Composite improved from 5.4 to 6.2 and the
control charts
smoothed out a bit, especially the rolling 20 trade P&L. The internal rate of return
for the system improved from 43% to 46% annually.
An interesting hypothesis.
|