- This topic is empty.
-
I am running optimization checks on the MOC.
The last time I ran these tests was Jan this year when I was building the system.
What I’ve noticed is the metrics have shifted. The CAGRs have dropped 1-3%, including for the long term 2005 – present backtests. The biggest difference is the drawdowns have increased quite a bit, mainly because of June. (eg one system max DD was 14% between 2005 – present, which is now 20%).
The parameters I started with are still testing well during the current optimization tests, but there has been a shift … here is an example …
I’ve been comparing the Jan 19 – Jan 22 tests with the July 19 – July 22 tests I ran.
What I’m thinking of doing is maintaining the current params and retesting again in 6 months time, when there is more data, because the June data has skewed results.
My question is …
How are others handling the June data when finding their optimal params?
Am I data mining by keeping my original params and waiting 6 months ?
I think the wise move is to leave it as is, just thought I would ask the q ….
There might be an element of data mining by waiting 6 months, but considering what June did to the data …….In all params it is the DD that skews the results. Some DD graphs are all over the shop. But when I look at the original param graphs on the current optimization tests, they are still
in a flat spot.CAGR example
Jan 19 – Jan 22 left, July 19 – July 22 on right)
DD example
(July 19 – July 22 left, Jan 19 – Jan 22 right)
- You must be logged in to reply to this topic.