|
PureBytes Links
Trading Reference Links
|
Hello Walter,
Chuck LeBeau/David Lucas, Nelson Freeburg and the late Bruce Babcock all say
the more the better.
The more trades the test generates, the more statistical reliable will be
the results.
This is important because it relates to the probabilty that your historical
results will repeat in the future.
That is the whole purpose of your test isn't it...
The absolute mimimum they agree on is 30 tests, to have any significant
statistical reliabilty.
Best wishes,
Frans
At 12:39 PM 11-02-99 -0500, you wrote:
>One question has arisen about the number of individual tests in a system
>test of a stock or other tradeable.
>
>Some system tests seem to be set up or "ball parked" for 100 to 200
>individual tests, others are for 500+ individual tests and others at the
>5,000+ level mainly acording to the number of optimized variables and
>optimization levels set.
>
>Aside from obvious considerations such as system time and number of
>tradeables to be tested, does anyone have a rule of thumb, preference, words
>of wisdom, pipeline to the Great Trader ... etc.?
>
>Best regards
>
>Walter
>
>
>
|