|
PureBytes Links
Trading Reference Links
|
Jim,
On Fri, 15 Dec 2000 13:13:47 EST, you wrote:
>Determining flatness and testing how effective trading the flat part of the
>optimization curve can be done through walk forward analysis.
I never came across "walk forward analysis" as a well established
procedure, even if I'm rather familiar with optimization in general
(not specialized to trading) over the last 20 years or so.
>Trading an optimization spike can be dangerous. Spikes can be caused by
>a few trades that turn out extremely well because of price shocks to the
>market.
Of course you are right, but this phenomenon imo can be much better
described by the "minimum drawdown" concept or even better by the
"probability distribution of (wins and) losses".
>Genetic algorithms are effective however they can optimize to a spike unless
>some degree of flatness is programmed into the fitness function.
Genetic algorithms as an _optimization method_ cannot made be
responsible for a "bad" _optimization function_ .
mfg rudolf stricker
| Disclaimer: The views of this user are strictly his own.
|