<- Back to Tasks

B130dev1 - Jforex Engine (0 accepted per hour) vs MT4 Engine (120 accepted per hour): Same build config

I use this build config. in multiple projects across multiple instruments for some time now with the MT4 engine. It generates on average 120 accepted strategies per hour for this pair.



Changing the engine to jforex produces zero strategies per hour.


I can imagine there might be some difference between the two engines in build speed etc (MT4 generates around 600K/120 accepted per hour). Jforex 300k/0. But I would imagine the ratio of accepted strargeies to be about the same over the long term, all other things remaining the same.


The building blocks used in both cases are only those compatible wothjforex.




Attachments
MT4 vs JFOREX Engine.cfx
MT4 vs JFOREX Engine.cfx
(31.43 KiB)
  • Votes +3
  • Project StrategyQuant X
  • Type Bug
  • Status Duplicate
  • Priority Normal
  • Assignee Mark Fric
  • Milestone Build 130
  • Category Backend

History

N
#1

nathan

19.11.2020 15:31

Task created

N
#2

nathan

19.11.2020 15:44
Digging deeper....turning off the cross check (high bt precision) produces strategies....albeit at an unrealistic rate (70k accepted per hour) hinting at other issues.


I think it might be due to using the 1 min tick sim data iso selected tf data precision. Selected TF works if used in both build and the cross check, but using 1 min tick sim in either = 0 strategies.


However using selected TF across the build task results in very odd equity curves (straight lines with thousands of trades) and an acceptance rate of 40K per hour which is nice, but totally unrealistic :)

N
#3

nathan

19.11.2020 17:53
Playing with it more, I think the jforex engine is only working with certain types of data? Tick; Precision Tick Sim 1 min and Tick; Precision Real are the only two the jforex engine appears to work with. Other precisions don't work, and data type M1 doesn't appear to work. is this correct/intentional?
TB
#4

Tomas Brynda

19.11.2020 18:56

Assignee changed from Tamas to Tamas

N
#5

nathan

19.11.2020 20:18
Done some more testing; I think Precision: Selected Timeframe Only doesn't work with the jForex engine. For both Tick and M1 data.
N
#6

nathan

20.11.2020 09:57

Attachment equity.jpg added

equity.jpg
equity.jpg
(200.68 KiB)
These are the sort of equity charts that are produced (at a rate of 70k per hour) if precision: Selected timeframe only is used in the build with M1 or Tick data.
TT
#7

Tamas

20.11.2020 10:03

Assignee changed from Tamas to Mark Fric

Milestone changed from None to Build 130

Category changed from None to Backend

N
#8

nathan

24.11.2020 11:08
Voted for this task.
N
#9

nathan

26.11.2020 14:44
Tested in v130dev2 - behavior is the same. I 'think' it might be ignoring ranking filters and/or automatic filters when selected timeframe is used in the build with the jForex engine.
DR
#10

mentaledge

27.11.2020 13:51
Voted for this task.
MF
#11

Mark Fric

03.12.2020 10:36

Status changed from New to Waiting for information

same as with other JForex task - we cannot reproduce it, it works for us. Could you pack your whole SQ folder and share it for us somehow?
N
#12

nathan

03.12.2020 11:01
Done!


Also added some additional info to ticket 7324



MF
#13

Mark Fric

03.12.2020 11:44

Status changed from Waiting for information to Duplicate

Attachment image-0.png added

Attachment image-0.png added

Attachment image-0.png added

image-0.png
image-0.png
(70.30 KiB)
image-0.png
image-0.png
(70.30 KiB)
image-0.png
image-0.png
(70.30 KiB)
duplicate of this, we'l be solving together: https://roadmap.strategyquant.com/tasks/sq4_7324
Votes: +3

Drop files to upload

or

choose files

Max size: 5MB

Not allowed: exe, msi, application, reg, php, js, htaccess, htpasswd, gitignore

...
Wait please