kiran
Junior Member
Posts: 19
Registered: 10-11-2017
Member Is Offline
Mood: No Mood
|
|
Controlling drawdown and Fitness functions ?
I'm tryingt the ES.30 data-set that came with GSB (new to GSB). Although i'm getting decent PF (3+) and Person coefficients, my drawdowns are >$4K
(hard to swallow, especially if capital allocated is $15-20K/contract). I'm using most default settings including the fitness function NP*AT.
1) Are there fitness functions that help reduce drawdowns and control downside? I like NP/(Sum of Areas of Drawdown) or (Avg of top 5 Drawdowns) -
any experiences with these?
2) Also, how do i insert a Divide operator in the fitness function? It seems to only multiply the factors in the fitness function.
thx
Kiran
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
you can try using a stop, but I would add it in after the system is needed.
You dont need divide as such. if you use dd, it will automatically divide.
I think the free gsbsys1es had < 4k dd.
Youve also done very few tests, and your machine is extremely slow.
A decent i7 should get you 2000 ++ systems per min.
YOu have way too many systems. Add the "tight" app settings metric.
You also should consider long short , not just long
|
|
|
kiran
Junior Member
Posts: 19
Registered: 10-11-2017
Member Is Offline
Mood: No Mood
|
|
I tried 2 fitness functions (NP*Avg Trade * Avg of 5 Drawdowns) and (NP*Avg Trade * VBasePlusMinusMedium), and they both had the issue that they
generate very few unique systems - using the same ES30m-UCCentralTime data that was shipped with GSB.
The first fitness only generated 8 unique systems out of 84K systems generated - see screenshot and settings file, and 2nd generated 1 unique for
every 12K systems. They also didn't help with reducing DD, as DD is still $3K-4K for these 8 systems.
- Is there some issue/bug that's introduced by other fitness functions, causing such few unique systems?
Other questions
- what is VBasePlusMinusMedium? Why is it better than Avg of 5 DDs?
Attachment: Login to view the details
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
I think at the heart of it you are in effect doing avetrade squared. That will screw everything. (not sure on this)
If youve got too few trades, remove the average trade factor but add 1 or 2 ticks under fitness slippage.
Dont confuse fitness what gsb looks for, vs what you look for.
Here is an analogy. If your primary goal is to be happy, your are likely to self centered, have poor relationships and be un-happy.
If your life has a purpose beyond your self, your more likely to be happy as a by-product.
Im summary dont place too much emphasis on DD
Dont use slippage on REPORTS with tight app settings.
They are separate but related issues.
Your easily not going to get under 3k dd for any system on es, though you could reduce it by having say a $1500 stop.
I add stops at the very last stage in TS, optimize on peak PF and dont get hung up on the exact best parameter, look for the best range of parameters.
Stops are not critical unless you go too tight
After doing a WF, likely dd can decrease as well.
|
|
|
![[*]](images/davis/default_icon.gif){kind=icon}
