READ-ONLY

The forum is now a read-only archive.

For bug reports & platform questions → [email protected]

Our community lives on Discord and YouTube — come join us!

WFM as Robustness Testing

3 replies

David Turner

Subscriber, bbp_participant, customer, community, sq-ultimate, 9 replies.

Visit profile

3 years ago #281247

Hi all,

From the Lab course I’ve been using the WFM as a robustness test. In short, building then testing strategies through the various robustness tests, through different MC tests and then using WFM as a final robustness test.

During this WFM test – which obviously takes a long time per strategy – I’m finding that most of them are failing, which I’m happy to accept, just wondering if anyone has any suggestions on workflow that means that I’m operating in the most efficient way?

Doing a single strategy at a time (so that the correct settings can be applied for 20,000 tests) takes quite a while, any tips on how to put forward the best strategies? Or is it just simply a case of running through them and accept that it takes a while to reject / accept?

Thanks in advance

Dave

0

Massimo Scapini

Subscriber, bbp_participant, customer, community, sq-ultimate, 59 replies.

Visit profile

3 years ago #281278

I am using WFM quite extensivly and I have run a small study to evaluate which is the best configuration to use. You can see details in my previous post here

https://strategyquant.com/forum/topic/which-wfm-configuration-to-use/

I basically do the following:

  1. I run the process in batch using the suggested configs, and the Selected TF precision. I discard those that fail
  2. I re-run the process for those remaining, using 1min precision to select the best optimization to be used for the future. I discard those that fail

Of course the method you use can highly depend on your PC. I have a powerful one (24 threads, 64GB RAM) and it runs almost 24×7. With a less powerful PC you must probably skip Step 2, even though I found meaningful differences between 1Min and TF Precison as far as Best Parametrization is concerned.

 

 

 

2

David Turner

Subscriber, bbp_participant, customer, community, sq-ultimate, 9 replies.

Visit profile

3 years ago #281284

Thank you so much Massimo, that’s really useful, thank you.

Another question (or two!) then… you say in step 2, “to select the best optimization to be used for the future”…

Once you’ve ran the WFM test, do you then find some parameters from the test and apply those to the strategy?

How do you select those parameters please?

Many thanks again,

Dave

0

Massimo Scapini

Subscriber, bbp_participant, customer, community, sq-ultimate, 59 replies.

Visit profile

3 years ago #281285

I hope this helps

https://youtu.be/0vjD17aTsjw

1

Viewing 3 replies - 1 through 3 (of 3 total)