When entering a position solely using the transaction weight (as a different approach to a rotation strategy), the results will show large variations with some oscillators, but not with others. E.g. with SZO and RVI there will be large differences, while RSI will always be identical.
I tried changing various settings, adding "fake" conditions like "buy if open > 0", restarting WL and even using a different PC. None of that changes anything.
I assume, this could also influence "normal" systems.
I tried changing various settings, adding "fake" conditions like "buy if open > 0", restarting WL and even using a different PC. None of that changes anything.
I assume, this could also influence "normal" systems.
Rename
I'm getting zero variations with the Building Blocks strategy powered by either SZO or RVI for the Transaction.Weight. Could you change the Scorecard from Finantic's to Basic and retry?
Using the same settings as above:
And:
And:
Signal overview of 2 consecutive runs. The weight values for each symbol are exactly the same. However, quantity largely varies.
If "Retain NFS Positions" is not checked, then the results are always different, and if checked, they will be the same
What Build number are you running?
I have Build 35. But the same as the author describes - the results are always different unless "Retain NFS Position" is noted.
QUOTE:
If "Retain NFS Positions" is not checked, then the results are always different, and if checked, they will be the same
I made my test with Retain NSF unchecked and the results were identical.
I used build 34 and now 35.
I just tried it with "retain NSF positions" as proposed by fred9999 and in this case the results will indeed be always the same. But not without.
I just tried it with "retain NSF positions" as proposed by fred9999 and in this case the results will indeed be always the same. But not without.
The point is in the indicator itself. Despite the fact that its period is 14, it starts from the 42nd bar. If you change the "StartIndex" from 14 to 45 in the strategy, then everything is OK.
Or so, if too lazy to convert to C# :)
Fred9999, that is indeed spot-on! I would never have figured that out, I guess.
The work around for building blocks is setting a start date a bit later. Then the results will always be the same. It's something like this:
The work around for building blocks is setting a start date a bit later. Then the results will always be the same. It's something like this:
The issue is that the BB looks for indicator parameters with the word “period” to try and infer the start index. We can open an issue internally to beef up this process and make it work with indicators such as SZO.
We have it taken care of for Build 36.
Hi Glitch, thank you for help.
Would be great, if build 36 comes as soon as possible, cause both workarounds are not working (tested by some simple strategies, for example by using StochK for WEIGHT).
LG Marko :)
Would be great, if build 36 comes as soon as possible, cause both workarounds are not working (tested by some simple strategies, for example by using StochK for WEIGHT).
LG Marko :)
Your Response
Post
Edit Post
Login is required