|
PureBytes Links
Trading Reference Links
|
----- Message d'origine -----
De : Mark Brown <markbrown@xxxxxxxxxxxxx>
À : pierre.orphelin <pierre.orphelin@xxxxxxxxxx>
Cc : <omega-list@xxxxxxxxxx>
Envoyé : samedi 5 février 2000 17:31
Objet : Re[6]: Rocket Science
>
>
> TOO BAD - systems and indicators DO NOT give the same mathematical
> results in all Omega's product as proven by the article Stock and
> Commodities proved and provided screen shots proving such. Yea that
> gives me the kinda confidence I need to test systems. Now lets add a
> fuzzy thing on top of bad math calculations and what do we have? A
> crap shoot!
>
Even if it was true, ythis should not be a problem to neurofuzzy logic.
Look at the chart below:
It's the latest trade onCAC40 futures ( avery easy one).
The system has THREE classical indicators only.
Do you think that they have the same value along the trend ?
No, of course.
However, you do not have ant short signal and the trend is perfectly caught.
The need of double precision and the like for such systems is a joke,a s well as the supposed TS errors
( I do not remember of this article)
Even by adding a +-10% random component to the indicators used as inputs, I could
turn this into a decent winning neurofuzzy system...
> Living in reality...
>
Not the same, I bet.
A lot of things to learn ...
Sincerely,
-Pierre Orphelin
Neurofuzzy Logic tools for TradeStation
Free evaluation versions and competitive upgrades available
web: http://www.sirtrade.com
Attachment Converted: "f:\eudora\attach\snap41.gif"
|