About Stability

What is stability in overview? How is this figure comes up?

Hello,

it is a proprietary formula which uses linear regression. It basically evaluates how straight the equity is. If it was a straight rising line it would be Stability = 1

Greetings Tomas,

I was comparing str result between SQ & QA, sorting them by stability. While other statistics looks OK, stability is quite different, 

I’ve attached the screenshot and one of the str here.

The attached str shows a 0.9 stability in SQ, but in QA shows only 0.69

Austin

File: QA.JPG

File: SQ.JPG

Greetings Tomas,

 

I was comparing str result between SQ & QA, sorting them by stability. While other statistics looks OK, stability is quite different, 

I’ve attached the screenshot and one of the str here.

 

The attached str shows a 0.9 stability in SQ, but in QA shows only 0.69

 

Austin

Interesting to know if the calculation is different between both or if the trade counts snippet interfere somewhere. Thanks for raising the issue.

Hello,

thanks for reporting this. We will look at it what can be done for fix

Hello,

the Stability in SQ on your screenshot is probably the Stability (In-Sample) while in QA it takes (In-Sample+ Out Of Sample) into consideration hence the difference

Thanks Tomas.

The str was created using in/out sample, in sample between 2006 – 1st Jul 2012, out of sample was set between 2nd Jul 2012 – 2015.

I retest the str again with 100% in sample data between 2012-2016, save it as another file and open up again on SQ and QA.

So there is no out of sample period calculation for QA.

Which is accurate? SQ or QA.

Austin

Hi,

I guess there is no answer to my query? Which result should i trust, QA or SQ?

And i did a test on SQ.

The stability calculation itself in SQ between  IS & OOS is in-consistence. See attached.

IMO none is to be trusted. I have to stop using stability.

File: Capture.JPG

If you run the same strategy on the same date range in SQ and QA with no out of sample data – is there a difference?

This is what sparks me to bring up this issue earlier. Retest STRs are all 100% in sample with same date range, and there is differences.

Hello,

yes, I also see small differences in stability value for the same periods of data. It can be caused by the stabiity formula itself for some reason. It uses regression and other indicators. Mark has been notified about this