|
PureBytes Links
Trading Reference Links
|
1. There are tests like the chi square test and the students t-test that
enable the user to determine if they have run enough tests to achieve a
required/desired confidence level.
2. My guess is that 20 to 30 trades should be enough to test the value of a
trading system. Few trading systems last "forever".
3. Some great advances in engineering have been made by non professionals;
some great advances have been made by engineers.
Lionel Issen
lissen@xxxxxxxxxxxxxx
----- Original Message -----
From: "Herman van den Bergen" <psytek@xxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: Monday, June 11, 2001 12:43 PM
Subject: Re: QQQ system test
> At 02:55 PM 6/11/01 +0200, you wrote:
> >Your number of tests are statistically insignificant and therefore
> useless in determining the worth of your system. Show me a report with at
> least 100 trades that have occurred in both up and down markets.>
>
> To discover and develop a new system one must have an open mind. Never
> shoot down any system on the basis of general performance, not only is it
> discouraging to the designer it also may prevent you from picking up new
> ideas that didn't have the opportunity to fully develop in the
> underperforming system.
>
> Throughout my life I have found more unique ideas among the inexperienced
> than among the professional (engineering in my case). The professional
> tends to be highly brainwashed in traditional approaches, so much so that
> they are often blind to the obvious.
>
> I wish you success with the development of your system. Perseverance will
> lead to success!
>
> Herman.
>
>
|