RE: Easier way to out-of-sample test

["neo" <neo1@xxxxxxxxx>]

Herman

Very Interesting. The other observation is that this system does not appear
to be stable over time at any opt value.

Dennis


-----Original Message-----
From: owner-metastock@xxxxxxxxxxxxx
[mailto:owner-metastock@xxxxxxxxxxxxx]On Behalf Of Herman van den Bergen
Sent: Wednesday, November 07, 2001 7:35 PM
To: Metastockusers@xxxxxxxxxxxxxxx; metastock@xxxxxxxxxxxxx
Subject: Easier way to out-of-sample test


Hi,

Here's a way to evaluate your opt variables over time that I think is
easier and more throrough than out-of-sample testing. Using a macro like I
posted earlier (28 Oct 2001) it takes only a few minutes to make the charts.

Add one opt variable to your list and intialize it 1990-2001 step 1. Then
add the statement "AND year()=optn" to your trading rule. In addition to
the year opt, change only one additional opt variable at a time )keep tests
< 1000) and produce separate 3D charts for each opt variable tested.

You'll get charts something like the ones attached, Excel gives you lots of
options to play with them. It is easy to see whether opt variables are
stable over time. If each year optimizes to the same value you are in luck.

Happy development,
Herman.