DELETE - IT'S IMPLEMENTED ALREADY

DELETE - IT'S IMPLEMENTED ALREADY
Attachments
modded SPP.png
(545.94 KiB)
original wfm.png
(420.84 KiB)
original wf.png
(411.99 KiB)
original spp.png
(363.88 KiB)
modded wfm.png
(585.54 KiB)
modded wf.png
(589.19 KiB)
  • Votes +14
  • Project StrategyQuant X
  • Type Feature
  • Status Fixed
  • Priority Normal

History

k
#1

Karish

02.04.2021 14:10

Task created

k
#2

Karish

02.04.2021 14:10
Voted for this task.
k
#3

Karish

02.04.2021 14:36

Attachment original ranking.png added

Attachment modded ranking.png added

original ranking.png
(235.73 KiB)
modded ranking.png
(270.99 KiB)

We would also want to use the Ranking's Fitness function to use those 3: SPP, WF, WFM


Screenshots attached.

l
#4

lemming78

02.04.2021 19:25
Voted for this task.
b
#5

bentra

02.04.2021 19:47
Voted for this task.
b
#6

bentra

02.04.2021 19:51
SPP can be median based but what about the WF? portfolio of all the oos combined? best oos? worst oos? 
k
#7

Karish

02.04.2021 20:00

Attachment Example.png added

Example.png
(172.92 KiB)

Bentra, What i am asking for is to allow us to use the following (screenshot attached to this msg),

You dont have to use those, but it atleast must be there for us who do want to use those, it is already implemented into SQ but not specifically to those functions...., would be a game changer for me atleast to use those features with those tests.

k
#8

Karish

02.04.2021 20:06
Another thing i already said inside our DIscord channel is this:

When you find a strategy with some parameters and it already entered your Databank based on your filters criteria,

Why the hell would you like to use ALL available parameters randomly to re-test with SPP/WF/WFM?,
It will take you ages!!,


Why wont we just have an optional way to use MIN / MAX / STEP in % of our Original parameters to do those?,

it would be more efficient this way, in a more logical way, and a more fast way,


Besides that, i added a reply below the original head post here that we should also be able to Evolve our SPP/WF/WFM scores with the help of the searching engine to use the Fitness to Fit and aim to get higher scores for our requirements of SPP/WF/WFM.

This kinda seems to be forgotten or something IMO...


b
#9

beppil

02.04.2021 20:24
Voted for this task.
b
#10

bentra

03.04.2021 03:24
I meant for ranking fitness you can use the median of SPP but what to use from the wf?
JH
#11

Jabezz

03.04.2021 04:29
Voted for this task.
k
#12

Karish

03.04.2021 10:14
I meant for ranking fitness you can use the median of SPP but what to use from the wf?

Any score you'll like from rankings criteria ...
k
#13

Karish

03.04.2021 14:55

GSB got this already, you can use Random/Genetic finesses with Random/Nearest parameters on WF


https://www.youtube.com/watch?v=y2W2SUXu92k&list=UUxxvWcvtZUzb3xyvIpOYmlg&index=14



b
#14

bentra

03.04.2021 18:54
Yes here are some examples:
profit factor of the median of the SPP
sharpe ratio of the median of the SPP
stability of the %50 confidence level of the montecarlo
rdd of the 95% confidence level of the montecarlo
rdd  of the ____ of the WFM

I'm asking what part of the WFM? best OOS run? worst OOS run? All oos runs as a portfolio? Any other choices you can think of?
k
#15

Karish

03.04.2021 19:00
Probably we need to have multiple options,

but let say you had 4X4 WFM, each of the so its 16 different WFs overall,

we need some kind of a custom score metric, of all of them, so lets say 1 out of those 16 got a score of 0.62, ..... and so on,
so you take all the 16 Scores and divide it by the Number of WFM runs which is 16 and get the Average Total Score Value,

then we can just select to use it as a fitness criteria or something like that, i am sure the dev-team will figure something out.

k
#16

Karish

03.04.2021 19:08
Another very important thing i wanted to mention is by using SPP,  can we have an option to choose to do "Independent Value Optimizations"?
A perfect example was given in this video here:
https://youtu.be/q_02_zeFq4w

So for example we got 3 parameters...: 100, 100, 100.

we want to check the neighboring parameters of the 1st parameter first, leaving the other parameters UN-touched while we perform those optimizations,

etc, etc.., the video above explains it in perfection and perfect sense,


Because right now i assume we can only optimize all the parameters at-once all together which is kinda limiting us..,

an option to do "Independent Value Optimizations" would be great!, using SPP as it is, because we already got an awesome infrastructure built already inside SQ, so there is not much work needed.



Other videos that are doing the same'ish..:

https://youtu.be/-hRrfnVDo9Y


https://youtu.be/KtUZg0kn7lg


https://youtu.be/PaVIbaTRO7o


https://youtu.be/YLLqU_F7hms

MF
#17

Marti

04.04.2021 03:20
Voted for this task.
JK
#18

Insanity82007

04.04.2021 11:36
Voted for this task.
AA
#19

AngelDuz

06.04.2021 14:34
Voted for this task.
MO
#20

mareko

08.04.2021 13:43
Voted for this task.
JT
#21

TiNTa

03.05.2021 18:37
Voted for this task.
AP
#22

AP

12.08.2021 18:22
Voted for this task.
KB
#23

kbtech

12.08.2021 20:02
Voted for this task.
NS
#24

nicksim80

14.08.2021 03:52
Voted for this task.
JH
#25

jakehobbs

26.08.2021 11:27
Voted for this task.
k
#26

Karish

17.03.2024 21:27

Subject changed from Allow using Nearest parameters on optimizations like SPP/WF/WFM rather than Random parameters.. As Cross-Validation tests. to DELETE - IT'S IMPLEMENTED ALREADY

Description changed:

DELETE - IT'S IMPLEMENTED ALREADY

TT
#27

Tamas

11.04.2024 09:07

Status changed from New to Fixed


Votes: +14

Drop files to upload

or

choose files

Max size: 5MB

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

...
Wait please