<- Back to Tasks

[126] WFM result strange and different confirm

Dear SQ team

I try to test WFM many times for check this problem.
I know there are 3 types Walk-forward type can be select for test.
When I choice Simulated IS, Simulated OOS and Simulated IS, Exact OOS it always happen different results in my databank with the same strategy.(it will use different parameter item for test every time)
But when I select the Exact IS, Exact OOS, the parameter item will lock by this type. So, I can get the same results with the same strategy with the same setting.
Is it normal?
If yes, I think it will be a problem when I use Simulated IS, Simulated OOS and Simulated IS, Exact OOS because I will get different results in my every test.
It will cause me difficult to reference the WFM report passed or failed.
Thank you.

Attachments
WF Matrix - 0_GJ_H1_B_Strategy 02416945.sqx
WF Matrix - 0_GJ_H1_B_Strategy 02416945.sqx
(1.17 MiB)
1646b0b65a34abf7c3754a732acff8fa.jpg
1646b0b65a34abf7c3754a732acff8fa.jpg
(304.19 KiB)
WF Matrix - 0_GJ_H1_B_Strategy 02416941.sqx
WF Matrix - 0_GJ_H1_B_Strategy 02416941.sqx
(980.25 KiB)
WF Matrix - 0_GJ_H1_B_Strategy 02416946.sqx
WF Matrix - 0_GJ_H1_B_Strategy 02416946.sqx
(1.17 MiB)
WF Matrix - 0_GJ_H1_B_Strategy 02416944.sqx
WF Matrix - 0_GJ_H1_B_Strategy 02416944.sqx
(1.03 MiB)
WF Matrix - 0_GJ_H1_B_Strategy 02416942.sqx
WF Matrix - 0_GJ_H1_B_Strategy 02416942.sqx
(980.77 KiB)
WF Matrix - 0_GJ_H1_B_Strategy 02416943.sqx
WF Matrix - 0_GJ_H1_B_Strategy 02416943.sqx
(1.03 MiB)
WF Matrix - 0_GJ_H1_B_Strategy 0241694.sqx
WF Matrix - 0_GJ_H1_B_Strategy 0241694.sqx
(953.65 KiB)
  • Votes 0
  • Project StrategyQuant X
  • Type Bug
  • Status Fixed
  • Priority Normal

History

SC
#1

fxchess

31.03.2020 13:57

Task created

TB
#2

Tomas Brynda

31.03.2020 14:42
Original task https://roadmap.strategyquant.com/tasks/sq4_5794
MF
#3

Mark Fric

02.04.2020 12:42

Status changed from New to Waiting for information

sorry, but I cannot reproduce this behavior. I get different results for each strategy doesn't matter if I use simulated+simulated or exact+exact method.


It depends also on how many steps you choose - if the max. steps is bigger than all combinations then it shouldn't use genetics and it should produce the same results because it will try all the possibilities.


Can you try to explain the problem better and perhaps also attach your Optimizer project config?



SC
#4

fxchess

02.04.2020 16:29

Attachment WFMOptimizeGBPJPY.cfx added

Dear Mark

Thank you for your reply, here is my config with optimizer

I only can get the same result in Exact IS, Exact OOS. 

Otherwise will get different result often.

I am just want to confirm is it normal with this setting?

Thank you.

MF
#5

Mark Fric

23.04.2020 07:51

Status changed from Waiting for information to Fixed

I don't see any error. The same results are more related to brute force / genetic approach - which are used depending on if your totam number of possible optimization combinbations is smaller or higher than max steps.
Votes: 0

Drop files to upload

or

choose files

Max size: 5MB

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

...
Wait please