Create a portfolio with WF results only

Hi,

Is it possible to select a certain WF setting for a strategy, then add that particular strategy to a portfolio?

For example, I want to merge 20 runs 20% OOS from Strat 1, and 30 runs 30 OOS% from strat 2.

It seems currently the ‘main backtest’ (not WF result) is merged.

Load the WFM strategies into Quant Analyzer and split the individual WF results out in there, then create portfolios with your chosen WF results

Insanity82007 wrote:

Load the WFM strategies into Quant Analyzer and split the individual WF results out in there, then create portfolios with your chosen WF results

Thank you, didn’t realize I was using an old version of QA.

Are you able to select only the OOS portion of the WF result inside QA?

No you can’t. This is a shortfall in QA.

What you can do though is make sure your OOS part of your WF strategy is never below a certain threshold and delete the rest of your WF results e.g. you can pick the following if you want at least 50% to be OOS in your analysis in QA.

You can then use recompute all stats to be 50% IS and 50% OOS and even if this doesn’t exactly match your WFM IS and OOS ratio’s this will still give you a reasonable enough basis to analyse which WFM cell or cells you want to pick for a particular strategy.

Lets say you have a WFM matrix with 2 to 4 runs with step 1 and 20% to 50% OOS with step 10%. If you export these strategies and want to have at least 50% of each of your results to be OOS then you would choose the single strategies in QA with the following combinations and ignore the others

2.50%

3.40% 3.50%

4.30%, 4.40%, 4.50%

So, for example, you would ignore 2.20% because this only gives about 33% OOS. For the same reason as you would ignore 2.30%, 2.40%, 3.20%, 3.30% and 4.20% as these are all under your 50% OOS threshold.

In this example your Matrix would be 12 cells and you would only chose 6 out of this.
1 Cell from a WFM matrix = 1 Simple Strategy in QA

If you decided to only use 33% OOS, then you will be able to use all 12 cells as part of your analysis and frankly I think 33% OOS is enough to clearly show you if the strategy is as stable IS as it is OOS.

Working with WFM results in either SQ or QA is a pain. It’s a shortfall of both applications because they don’t make this easy. I’ve developed custom scripts and snippets that I will share in posts following this one. Please keep in mind that these are customised to my naming conventions and way of working so you will need to evaluate the code and adjust the scripts to suit your needs.

All of these scripts are for QA.

The scripts go into C:\QuantAnalyzer4\scripts

The snippets go into C:\QuantAnalyzer4\extend\Snippets\com

 

 

Scripts

C:\QuantAnalyzer4\extend\Snippets\com\strategyquant\extend\DatabankActions

C:\QuantAnalyzer4\extend\Snippets\com\strategyquant\extend\DatabankColumns

C:\QuantAnalyzer4\extend\Snippets\com\strategyquant\extend\FitnessFunctions

Some of these scripts already exist as part of QA, so double check my scripts before replacing the existing ones.

Basically I either created new or modified existing scripts specifically to work with large volumes of WFM strategies.

Insanity82007 wrote:

Some of these scripts already exist as part of QA, so double check my scripts before replacing the existing ones. Basically I either created new or modified existing scripts specifically to work with large volumes of WFM strategies.

Wow that’s amazing, thank you for your generosity!