cotila1
Junior Member
Posts: 78
Registered: 8-5-2017
Member Is Offline
Mood: No Mood
|
|
@NQ - GSB settings share
I was exploring many settings to see the one that returns a lower degradation from IS days (see Avg. A) to OoS days (see Avg. B) on @NQ (day-trading).
Here the setting that returned me a degradation of about 10%.
Other settings I tested return higher degradation.
Personally I trend to explore first best setting that builds say max 10k systems, once I found it, then I start real building of more than 40k systems
with that setting.
Comments more than welcome.
TF: 29-30-31 Min bars.
#indic=5.
Data stream: NQ and $NDX.
SF=CloseLessPrevCloseDBPV.
Training=100% with Min-PF=1.3 and Min-#Trdes=120. Just on PF I've realized that high PF degrades much more than other metrics. I never go over 1.5.
Nth=1.
No S.L.-only EOD exit
Period: 2001-12/31/2016 - I prefer to leave out at least 2 years of OoS.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Youve done some great work. I hope to update the GSB methodology soon, but ive been stuck on some issues & made a critical mistake in the video that I
saw later. The videos going to be in the private forum first for feedback and comments from experienced GSB users first.
|
|
|
QuantDean
Junior Member
Posts: 6
Registered: 9-5-2017
Member Is Offline
Mood: No Mood
|
|
Thanks for sharing your experience with NQ - do you find that using 5 indicators may lead to greater overfitting? Has anyone here run some research on
the relationship between number of indicators and degradation?
|
|
|
rws
Member
Posts: 114
Registered: 12-6-2017
Member Is Offline
Mood: No Mood
|
|
I have been struggling with custom fit criteria in GSB and several other programs and found the custom fit formula has a big influence on OOS. There
is not a general rule because it also depends on the systems the program generates. To prevent from drawdown I always have a bit of that in the
formula.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Quote: Originally posted by QuantDean  | | Thanks for sharing your experience with NQ - do you find that using 5 indicators may lead to greater overfitting? Has anyone here run some research on
the relationship between number of indicators and degradation? |
be good to check using my cl benchmark.
Unless there is evident via a benchmark, im not in favour of 5 indicators.
Let the methodology with benchmark guide you in all things
|
|
|
![[*]](images/davis/default_icon.gif){kind=icon}
