admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
The making of GSB sysm2.1ES
This was the first es system made on a really crude proof of concept 29,30,31 min bars. Everything post 10/31/2007 was out of sample, and it was never
optimized after this date. It was built 3/2018, live tradeded 3/26/2018
I am not saying this should be done, but im saying it could be done.
Build systems nth all , dates 1/1/2000 to 10/31/2007
Then look for out of sample results that are good.
3 oscillators were used, operand * only. closeD filter. data1 es only/
1 tick and $2.4 used in this report each side.
Then look for some good results post 2007
System was built with zero slippage and commision
Attachment: Login to view the details
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
the locked code is on the ts app store. 30 day trial.
https://tradestation.tradingappstore.com/products/GSBSYSTEM2...
reports below are zero slippage & commission
pre 10/31/2007
post 2007
|
|
|
cotila1
Junior Member
Posts: 78
Registered: 8-5-2017
Member Is Offline
Mood: No Mood
|
|
I am using live the GSB sysm2.1ES over the last few months and I confirm the very positive track record!
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
as a experiment, I tried to make GSB find gsbsys2.1 by only allowing the 3 oscillators it used.
I didnt spend a lot of time on ut but the system I liked best had 39500 net proft pre 10/2007 compared to $50425 of gsbsys2.1
Looking later I see a system that had a little better metrics
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Here was the post 2007 results.
Still a bit down from GSBsysm2.1es of np $133525 pf 2.11 vs 101254 pf 1.82
but still very except-able.
When I wf the system pre 2007 data, and applied parameters to the post 2007 data, results were fractionally improved.
Note I only considered systems wiht verification of 8/8. GSB sysm2.1 did not have that luxury when it was made.
It would also be valid to make systems with only 30 min bars, but still do verification. Average results of 29,30,31 will normally be less than 30 min
bars, and I am comparing these results to gsbsysm2.1es on 30 min bars
|
|
|
![[*]](images/davis/default_icon.gif){kind=icon}
