gfxParticipant
Average
hi all,
I do want to optimise the Backtesting of a system using two variables. They are both coefficient from 0 to 1 with a step of 0.1
So I do set my variables C1 = 0 to 1 step 0.1 and C2 = 0 to 1 step 0.1 … and trigger the back testing.
But in the algorythm, it only makes sens to have C2 > C1. So I get results like
Gain C1 C2
10’000$ 0.5 0.1
10’000$ 0.5 0.2
10’000$ 0.5 0.3
….
And the backtesting takes quite a long time for combinaisons which are not relevant.
I don’t want to set C1 = 0 to 0.5 step 0.1 and C2 = 0.5 to 1 step 0.1.
It does not make sense as I want to back text also C1 = 0.7 & C2 = 0.8
I thought of placing a condition in the code ‘If C2 > C1 then process’ to filtrer those case. But still, it does not prevent the backtesting to run. I will just return no trade results but taking the same time.
So here is the question .. that you might have understood already.
How do I set the conditions that the BackTesting instance will only be triggered when C2 > C1 ?
Thks,
Gfx.
Sadly, you can’t (that I know of).
You can make several runs where you vary the one until under the start of the other.
Thus first a run with C1 from 0 to 0.2 step 0.1 and C2 from 0.3 to 1.0 step 0.1.
Then a run with C1 from 0 to 0.4 step 0.1 and C2 from 0.5 to 1.0 step 0.1.
Finally a run with C1 from 0.8 to 10 step 0.1 and C2 also from 0.8 to 1.0 step 0.1. This one will have the usual doubles you (visually) need to filter out.
Now if PRT could provide us with some sort of filter for this (like your proposed If C2 > C1 but at the “parameter-selection” level …
🙂
gfxParticipant
Average
Sadly it confirms what I thought …
I’ll slice it 🙂 hoping that PRT reads you and come up soon with an ‘advanced’ backtesting module for efficient results 🙂